• subsurface;
  • biomass;
  • groundwater;
  • aquifer


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results and discussion
  6. Conclusions
  7. References

There is abundant evidence for widespread microbial activity in deep continental fractures and aquifers, with important implications for biogeochemical cycling on Earth and the habitability of other planetary bodies. Whitman et al. (P Natl Acad Sci USA, 95, 1998, 6578) estimated a continental subsurface biomass on the order of 1016–1017 g C. We reassess this value in the light of more recent data including over 100 microbial population density measurements from groundwater around the world. Making conservative assumptions about cell carbon content and the ratio of attached and free-living microorganisms, we find that the evidence continues to support a deep continental biomass estimate of 1016–1017 g C, or 2–19% of Earth's total biomass.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results and discussion
  6. Conclusions
  7. References

Microbial life in continental aquifers and deep fractures constitutes a large carbon reservoir and may play an important role in global biogeochemical cycles. However, little is known about its total size, diversity, activity or distribution across time and space. In a widely cited paper, Whitman et al. (1998) estimated a continental subsurface biomass on the order of 1016–1017 g C and a marine subsurface (i.e. subseafloor) biomass on the order of 1017 g C. Kallmeyer et al. (2012) found that Whitman et al. had overestimated marine subsurface biomass by 1–2 orders of magnitude. Hence, a reassessment of continental subsurface biomass is timely. Here, we attempt this reassessment using recent measurements of groundwater microbial population density, cell carbon content and the ratio of free-living to attached microorganisms in groundwater.

Whitman et al. (1998) calculated terrestrial subsurface prokaryotic cell numbers in the top 4 km of groundwater by three methods based on weakly constrained estimates for key global parameters. Using cell counts measured in a range of marine and terrestrial unconsolidated sediments and extrapolated below 600 m (on the assumption of a logarithmic decline with depth), they obtained a total of 2.5 × 1029 cells: a minimum value because unconsolidated sediments represent only a small part of the continental subsurface. Secondly, using rock porosity and cell volume as a percentage of pore space, they obtained a total of 2.2 × 1030 cells. Finally, using the global volume of groundwater, an average number of unattached cells per volume of water (‘cell density’), and a ratio of grain-surface-attached to unattached cells, they obtained 2.5 × 1030 cells. Despite the limited scope of the minimum value calculated from unconsolidated sediments, the cell numbers and associated biomass were reported to range from 2.5 to 25 × 1029 cells, and 22 to 215 Pg of carbon (1 Pg = 1015 g), assuming a rather large cell mass of 172 fg (50% C, i.e. 86 fg C). In a study of marine subsurface cell numbers, Kallmeyer et al. (2012) adopted the arithmetic mean of these upper and lower biomass estimates for the terrestrial subsurface, that is, 119 Pg C. Fry et al. (2009) raised the lower estimate for the number of cells to 6 × 1029 on the assumption that there is no decline in cell density with depth, contrary to the inference made by Whitman et al. (1998) from submarine trends.

From a small data set, Whitman et al. (1998) calculated an average density of free cells of 1.54 × 105 mL−1 in groundwater and assumed that only 0.058% of cells are free, while the rest are attached to grain surfaces, yielding a total of 2.5 × 1030 cells in a groundwater volume of 9.5 × 1021 mL. For an assumed cellular carbon content of 86 fg C, this represents 215 Pg carbon. Using more recent data, we reassess the three critical parameters of groundwater cell density, proportion of unattached cells and cell carbon content, bearing in mind that cells in the deep subsurface typically exhibit adaptation to starvation survival (Amy et al., 1993; Kieft et al., 1997).


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results and discussion
  6. Conclusions
  7. References

Groundwater cell density

We compiled c. 120 individual measurements of unattached microbial cell densities (cells mL−1) from 24 studies representing continental aquifers and fracture waters at depths ranging from 10 to 3600 m (Table 1). To determine a global average groundwater cell density as a function of depth, the arithmetic mean cell count was calculated for each 250-m-depth interval for each sampling site. The global arithmetic mean in each depth bin was then calculated from these local means. Binning in this way was intended to clarify global trends by mitigating the heavy sampling bias towards shallower depths and moderating the influence of large or multiple data sets representing single sampling sites (by allowing each sampling site to count only once towards each 250-m bin). In fact, only the shallowest six depth bins, spanning 1500 m, were sampled in four or more distinct sites (Table 2). Mean cell densities for these bins are shown in Fig. 1 (bars).

Table 1. Details of groundwater cell density measurements used in this study
Host lithologyLocation(s)Cell count methodMetabolic groups recordedDepth (m)References
  1. BacLight and Syto-13 are epifluorescence-based techniques. Metabolic groups overlap. Some depth limits are interval midpoints.

  2. AO, acridine orange; DAPI, 4′,6′-diamino-2-phenylindol; FC, flow cytometry; A+, aerobic; A, anaerobic; SR, sulphate reducers; FeR, iron reducers; MnR, manganese reducers; NR, nitrate reducers; NX, nitrite oxidisers; HX, hydrogen oxidisers; DN, denitrifiers; SX, sulphur oxidisers; ?, inferred.

CarbonateParis, FranceDAPIA, autotrophs, heterotrophs, CH4-gens, SR800Basso et al. (2005)
Clastic sedimentary rocksKetzin, GermanyDAPIA, SR, ?CH4-gens647Morozova et al. (2010)
Shale, sandstoneNew Mexico, USAAOA, CH4-gens183–191Takai et al. (2003)
GraniteMizunami, JapanAOA+, A, heterotrophs, ?NR1169Fukuda et al. (2010)
Granite, sandstone, conglomerateTono, JapanAO, DAPI, SYBR Green IA+, A, heterotrophs, NR, autotrophic SR, SX, DN104–177Murakami et al. (2002)
MudstoneHokkaido, JapanAOHeterotrophs, ?CH4-gens37–480Kato et al. (2009)
Mudstone or sandstoneHokkaido, JapanAOA+, A, SR, CH4-gens, DN297–458Shimizu et al. (2006)
QuartziteGauteng, South AfricaFCA, SR, CH4-gens2830–3270Moser et al. (2005)
SandstoneWestern Queensland, AustraliaAOA, heterotrophs, HX, SR, CH4-gens937Kimura et al. (2005)
Sandstone, mudstoneOklo, GabonAOA+, A, heterotrophs, SR10–101Pedersen et al. (1996)
Dolomite, igneous rocksTransvaal, Free State and Gauteng, South AfricaDAPIA+, A, autotrophs, heterotrophs, SR, NX, SX, CH4-trophs1300–3600Borgonie et al. (2011)
Schist, ultramaficsOutokumpu, FinlandBacLightA+, A, autotrophs, heterotrophs, SR100–1500Itävaara et al. (2011)
BasaltSoutheastern Idaho, USAAONo data75–200O'Connell et al. (2003)
Gneiss, granite, granodioriteOlkiluoto, Kivetty, Hästholmen and Romuvaara, FinlandAO and DAPIA, autotrophic and heterotrophic acetogens and CH4-gens, SR, FeR,248–910Haveman et al. (1999)
GraniteLaxemar, SwedenAOHeterotrophs836–1039Pedersen & Ekendahl (1992a)
GraniteVästmanland, SwedenAOHeterotrophs, ?SR803–1105Pedersen & Ekendahl (1992b)
GraniteManitoba, CanadaAOA+, A, heterotrophs, SR, DN240Jain et al. (1997)
GraniteVästmanland, SwedenAOHeterotrophs816–1105Ekendahl & Pedersen (1994)
Granite/granodioriteÄspö, Avro, and Laxemar, SwedenAOHeterotrophs, CH4-gens129–860Pedersen & Ekendahl (1990)
Granite/granodioriteÄspö, SwedenAOA+, A, autotrophic and heterotrophic acetogens and CH4-gens, SR, FeR, MnR, NR171–978Hallbeck & Pedersen (2008)
MetabasaltNorth West, South AfricaSyto-13A, autotrophic SR, CH4-gens2825Lin et al. (2006)
BasaltWashington State, USAAOA+, A, heterotrophs, CH4-gens, SR, FeR, MnR, NR316–1270Stevens et al. (1993)
MetavolcanicsGauteng, South AfricaFCA, SR3100Kieft et al. (2005)
Dolomitic limestoneMontana, USAAOA, SR, CH4-gens1274–1492Olson et al. (1981)
Table 2. Depth distribution of cell density measurements and independent sampling sites
Depth interval (m)Number of cell density measurementsNumber of independent sampling sites
  1. a

    Sites within the Witwatersrand Basin, South Africa.


Figure 1. Cell density measurements from groundwater used in this study. Each point represents one measurement (or a mean of tightly clustered measurements from one study).

Download figure to PowerPoint

Groundwater volume and distribution

To convert the cell density–depth profile into total unattached cell numbers requires a known volume of groundwater and a groundwater volume–depth distribution so that the depth-binned cell density averages can be weighted by the proportion of groundwater present in each depth interval. Sokolov (1977) estimated a global groundwater volume of 2.43 × 1022 mL. Following Whitman et al. (1998), we assume that the entire volume of groundwater represents habitable space. Whitman et al. used a volume of 9.5 × 1021 mL groundwater in 4 km depth. We use a volume of 1022 mL for 2 km depth (Fry, 2005), which avoids spurious precision. We assume that the vertical distribution of this habitable space follows the simple porosity–depth relationship (compaction curve) given by Athy's law (Athy, 1930):

  • display math

where φ = porosity, φ0 = surface porosity, k = the compaction coefficient and z = depth. Because sandstone hosts a large proportion of the world's groundwater (Foster & Chilton, 2003), compaction coefficients were extracted from seven studies of marine and continental sandstones to determine the required weighting (Fig. 2).


Figure 2. Porosity profiles from seven studies of sandstones: (1) Mount Simon aquifer, USA (Person et al., 2010); (2) Submarine sandstones (Bahr et al., 2001); (3) New Jersey coastal plain, USA (Kominz & Pekar, 2001); (4) Terrestrial oil and gas reservoirs, USA (Maxwell, 1964); (5) Nigeria delta/coastal plain (Benjamin & Nwachukwu, 2011); (6) Mount Simon aquifer, USA (Medina et al., 2011); (7) Petroleum-bearing sandstones, USA (Chapman et al., 1984). The mean (black) closely matches Kominz & Pekar (2001).

Download figure to PowerPoint

If no weighting is applied, the mean cell density is 4.1 × 105 cells mL−1. Weighting by the minimum and maximum values of k yields total average densities 4.8 × 105 and 6.1 × 105 cells mL−1 respectively, assuming that the mean cell density in the 1750- to 1999-m-depth bin, for which no data were available, follows the logarithmic trend extrapolated from the 0- to 1749-m cell densities (R2 = 0.84). The average k value, 5.3 ×10−4 m−1, yields an average cell density of 5.2 × 105 unattached cells mL−1, an increase of c. 27% over the unweighted value. The final biomass estimate is increased proportionally.

Attached vs. unattached cells

The ratio of attached to unattached cells in the deep continental biosphere is poorly constrained. Using data from a single sandstone aquifer (Hazen et al., 1991), Whitman et al. (1998) determined that 0.058% of cells in groundwater are unattached, equivalent to an attached/unattached ratio of 1723. Considering only viable (not dead) cells, the same data yield a value of 0.22% cells unattached (attached/unattached = 454), based on 21 pairs of measurements in three wells. Several data sets have more recently become available from other aquifers. In a pristine groundwater site, Griebler et al. (2002) reported data equivalent to an attached/unattached ratio of about 1050 (mean of 7 ratios over 10 months). Alfreider et al. (1997) report data equivalent to a mean total/unattached ratio of 1010 from four groundwater wells, assuming 30% porosity. A simulated basalt aquifer system has yielded 99% biomass attached to the substrate (Lehman et al., 2001), that is, a ratio of about 100, but using a mean particle size coarser than sand. Similarly, a coarse (gravelly) natural aquifer yielded ratios of between 10 and 103 (Kölbel-Boelke et al., 1988). For a typical fine to medium sand, the particle surface area would be an order of magnitude greater and the resultant ratio also greater. However, still finer-grained clays and silts have been reported to support lower abundances of microorganisms than adjacent sands, perhaps because of restrictively small pore size (Sinclair & Ghiorse, 1989; Fredrickson et al., 1997).

Lower ratios have also been measured (ranging down to < 1; see review in Cozzarelli & Weiss, 2007), but for oil-bearing or contaminated aquifers where nutrient supplies are high. Where nutrient levels are limited, as expected in most of the deep biosphere, attachment favours survival (Marshall, 1988; Griebler et al., 2002). Indeed, the proportion of viable cells is higher in the attached than in the unattached population (Hazen et al., 1991). As the world's major aquifers occupy clastic sediments and sedimentary rocks ranging in particle size from silt to gravel (Foster & Chilton, 2003), we adopt a range from 102 to 103 to scale up from viable groundwater cell determinations to the whole sample; this results in an order-of-magnitude range in the final biomass estimate.

Carbon content of cells

The carbon content of cells varies widely, and indeed the biomass calculation of Whitman et al. (1998) uses values ranging from 10 to 100 fg C for different host environments, including 86 fg C for terrestrial subsurface cells. A recent synthesis of numerous studies of carbon contents in bacteria, not specific to the subsurface, indicates a majority of values in the range 20–100 fg C (Romanova & Sazhin, 2010). However, in a nutrient-limited deep biosphere, it is safer to adopt a mass typical for bacterial cells in starvation conditions, for which a mean and consistent value of 26 fg C has been determined from 10 cultivated strains (Troussellier et al., 1997). We therefore adopt this value, with the caveat that uncultured and environmental species may commonly have still smaller cell biomass.

Results and discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results and discussion
  6. Conclusions
  7. References


Weighting the extrapolated cell density–depth distribution (< 1749 m) by the distribution of groundwater yields an average cell density of 5.2 × 105 cells mL−1 in the top 2 km. This revises upwards the equivalent value of 1.54 × 105 determined by Whitman et al. (1998) and represents 5.2 × 1027 unattached cells according to a groundwater volume estimate of 1022 mL. Combining these estimates with an attached/unattached ratio ranging from 102 to 103 and a cellular carbon content of 26 fg C yields a total biomass of 14–135 Pg for the top 2 km of continental crust (Table 3). The relative insignificance of deeper biomass can be crudely illustrated by extrapolating the logarithmic groundwater volume–depth and cell density–depth curves to 5 km, which yields a total biomass c. 0.2% greater than the 2-km estimate (The assumption that cell density continues to decline logarithmically below 2 km is discussed in the next section). Excluding the two data points in the 1500–1749 depth bin also increases the total biomass by < 1%.

Table 3. Estimated key parameters determined in this study and by Whitman et al. (1998). The new average cell-density estimate is based on a global data set weighted by the estimated vertical distribution of groundwater
 Groundwater cell density (cells mL−1)Cell biomass (fg)Attached/unattached cells ratioGroundwater volume (mL)Total biomass (Pg C)
This paper5.2 × 10526102–103102214–135
Whitman et al. (1998)1.54 × 10510–10017239.5 × 102122–215

The new biomass estimate overlaps the range suggested by Whitman et al. (1998) for the top 4 km and represents 2–19% of Earth's total biomass (Kallmeyer et al., 2012). Using the marine subsurface biomass of 1.5–22 Pg estimated by Kallmeyer et al., we obtain the surprising result that continental subsurface biomass may be equal to or larger than the marine subsurface biomass, which amounts to 0.2–3.5% of Earth's total. Nevertheless, the continental estimate should be regarded as conservative given that:

  1. The data used in this calculation are skewed towards aquifers (which are most commonly of sand grain-size) and fractured crystalline rock (which are effectively coarse-grained). Finer-grained sediments are likely to have higher ratios of attached/unattached cells (Albrechtsen, 1994; Griebler et al., 2002), which may substantially raise the true biomass; on the other hand, clays and silts have been reported to support lower abundances of microorganisms than adjacent sands (Sinclair & Ghiorse, 1989; Fredrickson et al., 1997).
  2. Measurements from coal and hydrocarbon-producing systems, which provide exceptional carbon sources for heterotrophic communities, were not included. Such deposits probably do not host a large fraction of continental groundwater, but they may perhaps contribute significantly to the total biomass (given that microbial populations grow geometrically). Similarly, nutrient-contaminated groundwaters were not included.
  3. Most of the world's groundwater is stored in clastic sedimentary aquifers (Foster & Chilton, 2003), which contribute only 30% of the available cell density database. If the average cell density is extrapolated from these data alone, which include the highest reported values, the estimated biomass rises about twofold. However, there are too few data to say with confidence whether cell densities are consistently higher in any particular lithological context.

There are at least three other sources of major uncertainty attached to our estimate. Firstly, the wells from which groundwater samples are taken can themselves be a source of microbial contamination or a focus for biofilm formation. Basso et al. (2005) found that purging and mechanically cleaning an 800-m-deep well caused an order of magnitude decrease in the cell density of the groundwater it sampled. Sampling techniques and precautions taken to avoid contamination differ widely between studies. Secondly, cell counts were measured using fluorescent stains, which can sometimes infiltrate mineral particles, leading to falsely high cell counts (although some studies guard against this; Murakami et al., 2002). Thirdly, and perhaps most importantly, the geographical and geological distributions of cell count data do not reflect the variety of the Earth's continental crust or groundwater. The Fennoscandian shield and the Witwatersrand Basin of South Africa are over-represented, and most of the world's major aquifer systems are absent from the data set.

Cell density–depth distribution

By analogy with the marine subsurface biosphere, Whitman et al. (1998) expected groundwater cell densities to decrease logarithmically with depth below the continental surface, an inference recently challenged (Fry et al., 2009; Breuker et al., 2011). In our data set, the six 250-m averages in the top 1499 m, each representing at least four separate groundwater reservoirs, fit reasonably well to a logarithmic regression (R2 = 0.78), suggestive of a global trend over this depth interval. Only two reservoirs are sampled by the 1500- to 1749-m bin but, if included, these measurements continue the apparent logarithmic decline (R2 = 0.84). If the very low measurement of 2.0 × 102 cells mL−1 at 1700 m (Borgonie et al., 2011) is regarded as an outlier and excluded, R2 rises to 0.86 (increasing the final biomass estimate by < 1%).

Unfortunately, few groundwater cell density measurements are available from below this depth, and all hail from the same geological system, the Witwatersrand Basin of South Africa. This basin lies in a stable cratonic region and therefore maintains very low geothermal gradients (< 10 °C km−1; Omar et al., 2003) and hence potentially deeper habitable conditions than average continental crust. At depths > 3 km, Kieft et al. (2005) and Borgonie et al. (2011) recorded low cell densities (respectively 4 × 103 mL−1 at 3100 m and 3.4 × 103 mL−1 at 3600 m) consistent with the global logarithmic decline suggested by the shallower data set, although Borgonie et al. reported even lower cell densities at 1300 and 1700 m (3.0 × 103 and 2.0 × 102 mL−1 respectively; the latter not in the Witwatersrand). Inclusion of these data improves the goodness of fit of the depth bin averages to a logarithmic regression (R2 = 0.89). In stark contrast, Moser et al. (2005) obtained cell densities on the order of 106 mL−1 at c. 3200 and 3300 m, sharply reversing the apparent decline with depth.

It seems plausible that a global decline in cell densities through shallower depths might reflect the diminishing supply of photosynthetic organic matter from above, while deeper communities are supported independently by geochemical carbon sources and redox reactants. However, microbial communities in the Witwatersrand may be atypical for continental environments at these depths, even in a cratonic basin; local phenomena including abiogenic hydrocarbon generation and an anomalous radiolytic hydrogen flux may provide exceptional habitats for microorganisms (Lin et al., 2005; Borgonie et al., 2011). Hence, more data are needed, particularly from multi-kilometer deep settings geographically and geologically distinct from the Witwatersrand Basin, before a ‘typical’ continental cell density–depth profile can be constructed robustly. On balance, however, we cautiously suggest that the global data set does support a logarithmic decline with depth in groundwater cell densities, as postulated by Whitman et al. (1998). This is a distinct question from the distribution of biomass per se, which is shaped by the roughly exponential decline with depth in the availability of pore space and groundwater.

Future prospects

To further constrain deep continental biomass, a much wider range of depths, aquifer systems and lithological contexts must be sampled for measurements of cell population density and cell carbon content. The apparent decline with depth in both habitable space and groundwater cell density suggests that biomass below 2-km depth may contribute only marginally to the total. However, much more extensive drilling at these depths is necessary to clarify global trends, especially given the unexpectedly high cell counts c. 3 km deep in the Witwatersrand Basin. Many of the world's largest aquifer systems are missing from the present data set. The potential for the discovery of new organisms and new community structures is suggested by the identification of a fracture in the Witwatersrand dominated by a single previously unknown genus of bacterium (Chivian et al., 2008). The most poorly constrained variable in the present study is the ratio of attached and unattached cells; more work is needed to understand and predict how this ratio varies with lithology and community structure. There is also a pressing lack of quantitative data about the global volume of groundwater and its three-dimensional distribution.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results and discussion
  6. Conclusions
  7. References

A reappraisal of groundwater cell density, cell carbon content and the ratio of attached to unattached cells predicts a total subsurface continental biomass between 1016 and 1017 g carbon. Hence, the continental subsurface biomass may be similar to or even larger than the subseafloor biomass. The data appear to indicate a global logarithmic decline in groundwater cell density with depth. The biomass estimate is similar to that of Whitman et al. (1998) but should be regarded as conservative given the cautious choice of assumptions.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results and discussion
  6. Conclusions
  7. References
  • Albrechtsen H (1994) Distribution of bacteria, estimated by a viable count method, and heterotrophic activity in different size fractions of aquifer sediment. Geomicrobiol J 12: 253264.
  • Alfreider A, Krössbacher M & Psenner R (1997) Groundwater samples do not reflect bacterial densities and activity in subsurface systems. Water Res 31: 832840.
  • Amy PS, Durham C, Hall D & Haldeman DL (1993) Starvation-survival of deep subsurface isolates. Curr Microbiol 26: 345352.
  • Athy LF (1930) Density, porosity, and compaction of sedimentary rocks. AAPG Bull 14: 124.
  • Bahr DB, Hutton EW, Syvitski JP & Pratson LF (2001) Exponential approximations to compacted sediment porosity profiles. Comput Geosci 27: 691700.
  • Basso O, Lascourrèges J, Jarry M & Magot M (2005) The effect of cleaning and disinfecting the sampling well on the microbial communities of deep subsurface water samples. Environ Microbiol 7: 1321.
  • Benjamin U & Nwachukwu J (2011) Model compaction equation for hydrostatic sandstones of the niger delta. Ife J Sci 13: 161174.
  • Borgonie G, García-Moyano A, Litthauer D, Bert W, Bester A, van Heerden E, Möller C, Erasmus M & Onstott T (2011) Nematoda from the terrestrial deep subsurface of South Africa. Nature 474: 7982.
  • Breuker A, Köweker G, Blazejak A & Schippers A (2011) The deep biosphere in terrestrial sediments in the Chesapeake Bay area, Virginia, USA. Front Microbiol 2: Article 156.
  • Chapman DS, Keho T, Bauer MS & Picard MD (1984) Heat flow in the Uinta Basin determined from bottom hole temperature (BHT) data. Geophysics 49: 453466.
  • Chivian D, Brodie EL, Alm EJ et al. (2008) Environmental genomics reveals a single-species ecosystem deep within Earth. Science 10: 275278.
  • Cozzarelli I & Weiss J (2007) Biogeochemistry of aquifer systems. Manual of Environmental Microbiology (Hurst C, Crawford R, Garland J, Lipson D, Mills A & Stetzenbach L, eds), pp. 843859. ASM Press, Washington, DC.
  • Ekendahl S & Pedersen K (1994) Carbon transformations by attached bacterial populations in granitic groundwater from deep crystalline bed-rock of the Stripa research mine. Microbiology 140: 15651573.
  • Foster S & Chilton P (2003) Groundwater: the processes and global significance of aquifer degradation. Philos Trans R Soc Lond B Biol Sci 358: 19571972.
  • Fredrickson JK, McKinley JP, Bjornstad BN et al. (1997) Pore-size constraints on the activity and survival of subsurface bacteria in a late cretaceous shale-sandstone sequence, northwestern New Mexico. Geomicrobiol J 14: 183202.
  • Fry A (2005) Water Facts and Trends. World Business Council for Sustainable Development, Geneva.
  • Fry JC, Horsfield B, Sykes R, Cragg BA, Heywood C, Kim GT, Mangelsdorf K, Mildenhall DC, Rinna J & Vieth A (2009) Prokaryotic populations and activities in an interbedded coal deposit, including a previously deeply buried section (1.6–2.3 km) above∼ 150 Ma basement rock. Geomicrobiol J 26: 163178.
  • Fukuda A, Hagiwara H, Ishimura T, Kouduka M, Ioka S, Amano Y, Tsunogai U, Suzuki Y & Mizuno T (2010) Geomicrobiological properties of ultra-deep granitic groundwater from the Mizunami Underground Research Laboratory (MIU), central Japan. Microb Ecol 60: 214225.
  • Griebler C, Mindl B, Slezak D & Geiger-Kaiser M (2002) Distribution patterns of attached and suspended bacteria in pristine and contaminated shallow aquifers studied with an in situ sediment exposure microcosm. Aquat Microb Ecol 28: 117129.
  • Hallbeck L & Pedersen K (2008) Characterization of microbial processes in deep aquifers of the Fennoscandian Shield. Appl Geochem 23: 17961819.
  • Haveman SA, Pedersen K & Ruotsalainen P (1999) Distribution and metabolic diversity of microorganisms in deep igneous rock aquifers of Finland. Geomicrobiol J 16: 277294.
  • Hazen TC, Jiménez L, de Victoria GL & Fliermans CB (1991) Comparison of bacteria from deep subsurface sediment and adjacent groundwater. Microb Ecol 22: 293304.
  • Itävaara M, Nyyssönen M, Kapanen A, Nousiainen A, Ahonen L & Kukkonen I (2011) Characterization of bacterial diversity to a depth of 1500 m in the Outokumpu deep borehole, Fennoscandian Shield. FEMS Microbiol Ecol 77: 295309.
  • Jain D, Providenti M, Tanner C, Cord I & Stroes-Gascoyne S (1997) Characterization of microbial communities in deep groundwater from granitic rock. Can J Microbiol 43: 272283.
  • Kallmeyer J, Pockalny R, Adhikari RR, Smith DC & D'Hondt S (2012) Global distribution of microbial abundance and biomass in subseafloor sediment. P Natl Acad Sci USA 109: 1621316216.
  • Kato K, Nagaosa K, Kimura H, Katsuyama C, Hama K, Kunimaru T, Tsunogai U & Aoki K (2009) Unique distribution of deep groundwater bacteria constrained by geological setting. Environ Microbiol Rep 1: 569574.
  • Kieft TL, Wilch E, O'Connor K, Ringelberg DB & White DC (1997) Survival and phospholipid fatty acid profiles of surface and subsurface bacteria in natural sediment microcosms. Appl Environ Microbiol 63: 15311542.
  • Kieft TL, McCuddy SM, Onstott T, Davidson M, Lin L, Mislowack B, Pratt L, Boice E, Lollar BS & Lippmann-Pipke J (2005) Geochemically generated, energy-rich substrates and indigenous microorganisms in deep, ancient groundwater. Geomicrobiol J 22: 325335.
  • Kimura H, Sugihara M, Yamamoto H, Patel BK, Kato K & Hanada S (2005) Microbial community in a geothermal aquifer associated with the subsurface of the Great Artesian Basin, Australia. Extremophiles 9: 407414.
  • Kölbel-Boelke J, Anders E & Nehrkorn A (1988) Microbial communities in the saturated groundwater environment II: diversity of bacterial communities in a Pleistocene sand aquifer and their in vitro activities. Microb Ecol 16: 3148.
  • Kominz MA & Pekar SF (2001) Oligocene eustasy from two-dimensional sequence stratigraphic backstripping. Geol Soc Am Bull 113: 291304.
  • Lehman RM, Colwell FS & Bala GA (2001) Attached and unattached microbial communities in a simulated basalt aquifer under fracture-and porous-flow conditions. Appl Environ Microbiol 67: 27992809.
  • Lin L, Hall J, Lippmann-Pipke J, Ward JA, Sherwood Lollar B, DeFlaun M, Rothmel R, Moser D, Gihring TM & Mislowack B (2005) Radiolytic H2 in continental crust: nuclear power for deep subsurface microbial communities. Geochem Geophys Geosyst 6: Q07003.
  • Lin L, Wang P, Rumble D, Lippmann-Pipke J, Boice E, Pratt LM, Lollar BS, Brodie EL, Hazen TC & Andersen GL (2006) Long-term sustainability of a high-energy, low-diversity crustal biome. Science 314: 479482.
  • Marshall KC (1988) Adhesion and growth of bacteria at surfaces in oligotrophic habitats. Can J Microbiol 34: 503506.
  • Maxwell JC (1964) Influence of depth, temperature, and geologic age on porosity of quartzose sandstone. AAPG Bull 48: 697709.
  • Medina CR, Rupp JA & Barnes DA (2011) Effects of reduction in porosity and permeability with depth on storage capacity and injectivity in deep saline aquifers: a case study from the Mount Simon Sandstone aquifer. Int J Greenhouse Gas Control 5: 146156.
  • Morozova D, Wandrey M, Alawi M, Zimmer M, Vieth A, Zettlitzer M & Würdemann H (2010) Monitoring of the microbial community composition in saline aquifers during CO2 storage by fluorescence in situ hybridisation. Int J Greenhouse Gas Control 4: 981989.
  • Moser DP, Gihring TM, Brockman FJ, Fredrickson JK, Balkwill DL, Dollhopf ME, Lollar BS, Pratt LM, Boice E & Southam G (2005) Desulfotomaculum and Methanobacterium spp. dominate a 4-to 5-kilometer-deep fault. Appl Environ Microbiol 71: 87738783.
  • Murakami Y, Fujita Y, Naganuma T & Iwatsuki T (2002) Abundance and viability of the groundwater microbial communities from a borehole in the Tono uranium deposit area, central Japan. Microbes Environ 17: 6374.
  • O'Connell SP, Lehman RM, Snoeyenbos-West O, Winston VD, Cummings DE, Watwood ME & Colwell FS (2003) Detection of Euryarchaeota and Crenarchaeota in an oxic basalt aquifer. FEMS Microbiol Ecol 44: 165173.
  • Olson G, Dockins W, McFeters G & Iverson W (1981) Sulfate-reducing and methanogenic bacteria from deep aquifers in Montana. Geomicrobiol J 2: 327340.
  • Omar G, Onstott T & Hoek J (2003) The origin of deep subsurface microbial communities in the Witwatersrand Basin, South Africa as deduced from apatite fission track analyses. Geofluids 3: 6980.
  • Pedersen K & Ekendahl S (1990) Distribution and activity of bacteria in deep granitic groundwaters of southeastern Sweden. Microb Ecol 20: 3752.
  • Pedersen K & Ekendahl S (1992a) Assimilation of CO2 and introduced organic compounds by bacterial communities in groundwater from southeastern Sweden deep crystalline bedrock. Microb Ecol 23: 114.
  • Pedersen K & Ekendahl S (1992b) Incorporation of CO2 and introduced organic compounds by bacterial populations in groundwater from the deep crystalline bedrock of the Stripa mine. J Gen Microbiol 138: 369376.
  • Pedersen K, Arlinger J, Hallbeck L & Pettersson C (1996) Diversity and distribution of subterranean bacteria in groundwater at Oklo in Gabon, Africa, as determined by 16S rRNA gene sequencing. Mol Ecol 5: 427436.
  • Person M, Banerjee A, Rupp J, Medina C, Lichtner P, Gable C, Pawar R, Celia M, McIntosh J & Bense V (2010) Assessment of basin-scale hydrologic impacts of CO2 sequestration, Illinois basin. Int J Greenhouse Gas Control 4: 840854.
  • Romanova N & Sazhin A (2010) Relationships between the cell volume and the carbon content of bacteria. Oceanology 50: 522530.
  • Shimizu S, Akiyama M, Ishijima Y, Hama K, Kunimaru T & Naganuma T (2006) Molecular characterization of microbial communities in fault-bordered aquifers in the Miocene formation of northernmost Japan. Geobiology 4: 203213.
  • Sinclair JL & Ghiorse WC (1989) Distribution of aerobic bacteria, protozoa, algae and fungi in deep subsurface sediments. Geomicrobiol J 7: 1531.
  • Sokolov A (1977) World water resources: perspectives and problems. World Dev 5: 519523.
  • Stevens T, McKinley J & Fredrickson J (1993) Bacteria associated with deep, alkaline, anaerobic groundwaters in southeast Washington. Microb Ecol 25: 3550.
  • Takai K, Mormile MR, McKinley JP, Brockman FJ, Holben WE, Kovacik WP & Fredrickson JK (2003) Shifts in archaeal communities associated with lithological and geochemical variations in subsurface Cretaceous rock. Environ Microbiol 5: 309320.
  • Troussellier M, Bouvy M, Courties C & Dupuy C (1997) Variation of carbon content among bacterial species under starvation condition. Aquat Microb Ecol 13: 113119.
  • Whitman WB, Coleman DC & Wiebe WJ (1998) Prokaryotes: the unseen majority. P Natl Acad Sci USA 95: 65786583.