Long-Term Sustainability of a High-Energy, Low-Diversity Crustal Biome

See allHide authors and affiliations

Science  20 Oct 2006:
Vol. 314, Issue 5798, pp. 479-482
DOI: 10.1126/science.1127376


Geochemical, microbiological, and molecular analyses of alkaline saline groundwater at 2.8 kilometers depth in Archaean metabasalt revealed a microbial biome dominated by a single phylotype affiliated with thermophilic sulfate reducers belonging to Firmicutes. These sulfate reducers were sustained by geologically produced sulfate and hydrogen at concentrations sufficient to maintain activities for millions of years with no apparent reliance on photosynthetically derived substrates.

Most subsurface microbial ecosystems examined to date (including subseafloor sediments, deep-sea hydrothermal vents, terrestrial sedimentary aquifers, and petroleum reservoirs) ultimately depend on sunlight. These studies have been mostly confined to depths of less than 1 km, and the ecosystems are either supported by photosynthetically produced electron donors and acceptors transported by groundwater or seawater with ages much less than a million years, or are in constant contact with oxygenated seawater migrating through the underlying fractured basaltic aquifer (14). Although two occurrences of autotrophic microbial communities have been reported to exist in <300-m-deep volcanic aquifers flushed with fresh meteoric water (5, 6), their long-term sustainability on H2 and isolation from photosynthetically produced substrates have not been demonstrated. Although the existence of subsurface microorganisms at depths greater than 1 km in pristine environments is well established (7), much is still unknown regarding the abundance, diversity, and sustainability of these microbial communities over geological time scales.

To determine the long-term sustainability of a deep terrestrial environment, we examined the microbial diversity and metabolic activity of a 3-to 4-km-deep fracture in the 2.7-billion-year-old Ventersdorp Supergroup metabasalt, in which fracture water ages of tens of millions of years (8), abundant abiogenic hydrocarbons (9), and radiolytically produced H2 (10) have been reported. To characterize the indigenous microbial composition and its principal respiratory pathway, we analyzed 16S ribosomal DNA (rDNA) sequences, aqueous and gas geochemistry, and stable and noble gas isotopic signatures of moderately saline groundwater emanating from a fracture zone 2.825 km below the land surface (kmbls) in the Mponeng gold mine, South Africa. This high-pressure water-bearing fracture was intersected during exploratory drilling ahead of a tunnel advancing into an unmined zone ∼100 m above the Ventersdorp Contact Reef (VCR) ore zone. The fracture water was initially sampled as soon as it was safe to do so (4 days after the fracture intersection), and three subsequent samples were obtained over a 54-day interval (Table 1) to monitor drilling contamination and possible changes of community structure and geochemistry as the fracture was de-watered and before being sealed by the mine's cementation team.

Table 1.

Geochemical and microbiological characteristics of fracture water and mining water. NA, not available; <d.l., below detection limit, which is 1 μM for O2 and 0.5 fg ml–1 for archaeal DNA; Ma, million years ago.

Sample no. (number of days since intersection)1 (4)2 (7)3 (15)4 (58)5
Sample name MP104E65X C-091602 MP104E65X C-091902 MP104E65X C-092702 MP104E65X C-110902 MP104E65XC-SW-091602
Origin Fracture water Fracture water Fracture water Fracture water Mining water
Water and gas flow rates (liters min-1) (borehole volumes) 40/2.4 (45) 10.9/0.8 (55) 8.2/1.2 (64) 2.3/1.7 (96) NA
pH 9.3 9.3 9.3 9.2 NA
Eh (mV) -330 -350 -340 -263 NA
Temperature (°C) >60 52 52 52 20
Formate (μM)View inline 7.6 NA 7.1 8.9 1.4
Acetate (μM)View inline 24.6 NA 22.5 35.7 5.9
Cl- (mM)View inline,View inline 54.1 NA 71.9 84.9 0.42
Br- (μM)View inline 125 NA 177 218 3.8
SO42- (μM)View inline 529 NA 900 1860 171
HS- (μM)View inlineView inline NA 1390 NA 1060 NA
H2 (μM)View inlineView inline 1940 2600 2090 3715 NA
CH4 (μM)View inlineView inline 8580 11800 9320 16600 NA
O2 (μM)View inlineView inline 6 <d.l. <d.l. <d.l. 285
δ2H-H2 (‰ VSMOW)View inline -684 -684 -688 -695 NA
δ13C-CH4 (‰ PDB)View inlineView inline -31.6 -31.7 -32.8 -33.2 NA
δ2H-CH4 (‰ VSMOW)View inline -364 -367 -366 -390 NA
4He model age (Ma) 20.9 ± 10.5 20.4 ± 10.2 15.8 ± 7.9 NA NA
40Ar model age (Ma) 16.3 ± 8.2 21.3 ± 10.6 16.9 ± 8.4 NA NA
134Xe model age (Ma) 18.7 ± 7.0 19.4 ± 3.8 21.0 ± 6.0 NA NA
136Xe model age (Ma) 21.6 ± 6.0 25.0 ± 3.8 23.8 ± 4.6 NA NA
Bacterial DNA (pg ml-1) 15 ± 8 16 ± 8 30 ± 15 30 ± 15 3 ± 1.5 × 105
Archaeal DNA (pg ml-1) ∼5 × 10-4 <d.l. ∼5 × 10-4 ∼5 × 10-4 206 ± 100
Cell density (cells ml-1) 5.1 ± 0.5 × 104 NA NA 3.3 ± 0.3 × 104 NA
  • View inline* The concentrations derived from charged balance were 10.2 mM for sample 1, 27.6 mM for sample 3 and 50.6 mM for sample 4.

  • View inline The concentrations of dissolved gases were reported as concentrations corrected for diffusive loss (8). Diffusive correction was not applied to sample 4 because of the lack of noble gas analysis.

  • View inline The positive O2 content for sample 1 may be derived from incomplete isolation of fracture water from the mining environment caused by the extremely high water pressure and flow rate.

  • View inline§ The carbon isotopic value was referenced to Pee Dee belemnite (PDB).

  • View inline The uncertainties for aqueous and gas chemistry are ±10% and ±20%, respectively. The uncertainties for isotopic measurements are ±0.5‰ for δ13C-hydrocarbon and ±5‰ for δ2H-hydrocarbon and δ2H-H2.

  • Fracture water samples yielded a uniform community structure dominated by a single phylotype (MP104-0916-b1) that constituted >88% of the clones in the 16S rDNA libraries we generated but did not appear in the water used for mining (Fig. 1A). This phylotype was related (96% similarity) to an uncultured clone recovered from thermal fluids of oceanic crust (11) or to Desulfotomaculum kuznetsovii (91% similarity) (12), a sulfate reducer growing at moderately thermophilic conditions. Of the other minor bacterial and archaeal phylotypes associated with the fracture water, MP104-1109-a19 resembled (98% similarity) environmental clones (the SAGMEG-2 group) obtained from fracture water in an adjacent mine (13) (Fig. 1B). Other phylotypes were closely related (98 to 99% similarity) to environmental clones recovered from surface environments (such as soils and sludge) or to various mesophiles distributed within Proteobacteria, Firmicutes, and Methanobacteria (Fig. 1). These minor phylotypes did not resemble those from the mining water [see supporting online material (SOM)].

    Fig. 1.

    Phylogenetic trees for bacteria (A) and archaea (B) based on 16S rDNA sequences recovered from fracture and mining water in the Mponeng gold mine, South Africa. The fraction number in parentheses represents the number of clones for each phylotype versus the total number of clones analyzed in each sample. The scale bar is equivalent to the sequence variation of 10%.

    High-density 16S rDNA microarray analysis was also used as a more sensitive approach to assess microbial diversity. Array hybridization results supported the finding that microbial diversity was much less (table S2) and the Firmicutes were of greater relative abundance (table S3) in the fracture water than in the mining water. Some overlap between the sequences found in the fracture water and those in the mining water is expected because mining water is a mixture of recycled fracture water and surface water.

    The geochemistry of the fracture water was characterized to identify its origin and to assess the principal metabolic pathways (Table 1 and tables S1 and S4). Over the observation period, the temperature decreased from >60° to 52°C, while the pH was constant and the redox potential (Eh) increased from –330 to –263 mV. The Cl, Br, and SO 2–4 concentrations increased significantly, whereas other anion concentrations fluctuated or decreased slightly. Reduced gases (H2 and CH4) were abundant (>1 mM), and their concentrations increased as water flow rates declined and gas flow rates remained constant. Formate and acetate concentrations ranged from 7 to 9 μM and 22 to 36 μM, respectively. The δ2H and δ13C of the abundant dissolved C1–4 hydrocarbons indicated an abiogenic origin, based on their similarity to abiogenic hydrocarbon gases identified at other Precambrian Shield sites (9). The δ2H and δ18O values of the fracture water, –29.1 per mil (°) and –6.9° VSMOW (Vienna standard mean ocean water), respectively, plotted slightly above the global and local meteoric line (14). The δ2H-H2 values ranged between –680 and –690° VSMOW, which when compared with the δ2H-H2O value yielded isotopic equilibrium temperatures of 62°C for the first sample, declining to 49°C for the last sample. Based on the thermal conductivity model (15) and the heat flow data (16), the initial temperature equates to that at 4.2 kmbls and the final temperature to that at 2.9 kmbls.

    Noble gas analyses yielded elevated radiogenic and fissiogenic concentrations of 4He, 40Ar, 134Xe, and 136Xe (table S1) but exhibited no systematic trend during the observation period. The bulk model age of the fracture water ranged from 15.8 ± 7.9 million years to 25.0 ± 3.8 million years (Table 1). This model age represents either the true subsurface residence time or the mixing between a hydrothermal fluid that is ancient (0.8 to 2.5 billion years old); saline; and H2-, C1–4 hydrocarbon–, and SO 2–4-rich; and paleometeoric water that is ∼3 to 4 million years old; moderately saline; and H2-, C1–4 hydrocarbon– and SO 2–4-poor (SOM).

    The exceptional dominance of the Firmicutes clones; δ2H and δ18O values of the fracture water that are distinct from those of the mining water (17); and increasing H2, CH4, He, Cl, Br, and SO 2–4 concentrations during the observation period suggest that the fracture environment favors the survival of these Firmicutes-related microorganisms and that mining operations had minimal impact on this environment during depressurization and dewatering. Desulfotomaculum-related environmental clones and isolates have been detected in other subsurface environments, including an oil reservoir (18), deep sedimentary strata (19, 20), an ore deposit (21), and saline water emanating from a 3.2-kmbls borehole located 13 km east of this borehole (22). The evident success of these microorganisms may not be surprising because their ability to form endospores would facilitate their survival during periods of low water activity, nutrient deprivation, and suboptimal temperature (23). Whether the less abundant members of the clone libraries represent indigenous microorganisms that are capable of acquiring energy through metabolisms that are distinct from those of the strains they most resemble phylogenetically, or alternatively represent moribund relics of shallower, less saline paleometeoric water that has mixed with this fracture water, is not known.

    Sulfur isotopic analyses yielded δ34S values of HS ranging from 11 to 13° Vienna Canon Diablo meteorite (VCDT) and those of SO 2–4 from 19 to 26° VCDT (table S4). The depleted δ34S value of HS relative to that of SO 2–4 (–7.7° for sample 4 and –12.4° for sample 7) is consistent with microbial sulfate reduction. The fact that SO 2–4 concentrations in the fracture water were higher than that of the overlying dolomitic aquifer (13), and that SO 2–4 has been reported in analyses of fluid inclusions from hydrothermal quartz in the VCR (24) and the reservoir mixing-fractionation model (SOM), all suggest that the SO 2–4 originated from the ancient hydrothermal fluid, not the paleometeoric end member. The δ34S values of fracture pyrite (0 to 2° VCDT) were less than that of the coexisting barite (10.1° VCDT) (table S4) and were consistent with those of hydrothermal 2.0-billion-year-old pyrite from the VCR in the same mine (25). The Δ33S values for all samples clustered around 0° (table S4). To reproduce the Δ33S of total dissolved S species (–0.022°) for sample 7, a mixing of SO 2–4 derived from pyrite oxidation and barite dissolution with a Rayleigh isotopic fractionation by microbial sulfate reduction was required (SOM). Such a mixture, composed of 30% SO 2–4 derived from the oxidation of pyrite by radiolytically produced oxidants (10) and 70% SO 2–4 derived from the dissolution of barite, produced an initial δ34S value of 8.13° VCDT and a Δ33S value of –0.0087° (SOM). The isotopic evolution from the initial 8.13° VCDT to the present observation (18.63° VCDT for total dissolved S species) requires a Rayleigh fractionation process in which 70% of the initial SO 2–4 was removed as microbially precipitated pyrite (SOM). By this model, the total HS formed by microbial sulfate reduction would be 4.35 mM [1.52 mM × 70%/30% + 0.80 mM HS (SOM)].

    The Gibbs free energy for microbial redox reactions was calculated to provide additional constraints on the dominant respiratory pathway occurring in this fracture water. Sulfate reduction dominated the free-energy yields for a wide range of electron donor and acceptor combinations (Table 2). The freeenergy yields for these reactions were much greater than the minimum requirement for synthesis of 1/3 of an adenosine triphosphate molecule by pumping one proton across the cell membrane (∼–20 kJ) (26). Among electron donors available for sulfate reduction and methanogenesis, H2 yielded more free energy than acetate and formate, and H2-utilizing sulfate reduction yielded the highest free energy and energy flux (Table 2), suggesting that CO2-utilizing methanogens and acetogens cannot sustain as high a population density as sulfate reducers and therefore would be minor constituents of the community, as is observed in the clone libraries. The prominence of the free-energy flux for H2-utilizing sulfate reduction over other metabolic reactions is consistent with the depletion of δ34S values of HS relative to SO 2–4 (27) and the physiological characteristics inferred from the dominant phylotype.

    Table 2.

    Gibbs free energy (ΔG), substrate consumption rate and steady-state free-energy flux for various microbial redox reactions.

    Sample no.134
    ΔG (kJ mol-1)Substrate rateView inline(μM year-1)Energy fluxView inline(kJ cell-1 s-1)ΔG (kJ mol-1)Substrate rateView inline(μM year-)Energy fluxView inline(kJ cell-1 s-1)ΔG (kJ mol-1)Substrate rateView inline(μM year-1)Energy fluxView inline(kJ cell-1 s-1)
    H2-sulfate reductionView inlineView inline -148 5.9 × 106 -9.2 × 10-13 -155 8.6 × 106 -1.4 × 10-12 -146 1.8 × 107 -2.7 × 10-12
    Acetate-sulfate reductionView inlineView inline -86 1.9 × 105 -1.7 × 10-14 -94 1.5 × 105 -1.5 × 10-14 -82 2.4 × 105 -2.1 × 10-14
    H2-methanogenesisView inline -94 2.1 × 105 -2.1 × 10-14 -95 2.9 × 104 -2.9 × 10-15 -93 3.8 × 105 -3.7 × 10-14
    Acetate-methanogenesisView inline -33 1.9 × 105 -6.6 × 10-15 -34 1.5 × 105 -5.5 × 10-15 -28 2.4 × 105 -7.1 × 10-15
    Formate-methanogenesisView inline -74 1.8 × 104 -1.4 × 10-15 -86 1.5 × 104 -1.3 × 10-15 -42 1.8 × 103 -8.1 × 10-17
    H2-acetogenesisView inline -62 1.4 × 105 -9.3 × 10-15 -61 2.0 × 104 -1.2 × 10-15 -64 2.5 × 105 -1.7 × 10-14
  • View inline* The calculation for maximum substrate consumption rate and steady-state free-energy flux is shown in SOM.

  • View inline The HS- concentration used in the calculation for sulfate reduction for samples 1 and 3 were assumed to be 1.3 mM. The free energy for sulfate reduction only varied less than 5% when HS- concentration was changed between 1.1 to 1.5 mM.

  • View inline The reactions for free energy calculations were as follows: H2-sulfate reduction: 4 H2 + H+ + SO42- → HS- + 4 H2O; acetate-sulfate reduction: CH3COO- + SO42- → 2 HCO3- + HS-; H2-methanogenesis: 4 H2 + H+ + HCO3- → CH4 + 3 H2O; acetate-methanogenesis: CH3COO- + H2O → CH4 + HCO3-; formate-methanogenesis: 4 HCOO- + H+ + H2O → CH4 + 3 HCO3-; and H2-acetogenesis: 4 H2 + H+ + 2 HCO3- → CH3COO- + 4 H2O. The uncertainties are ±10%.

  • The in situ rate of microbial sulfate reduction was estimated to range from 0.22 to 1.45 nM year–1, or from 5.5 × 10–18 to 3.6 × 10–17 moles per cell year–1 (assuming that all observed cells, 4 × 107 cells liter–1, actively reduce SO 2–4), based on the potential microbially produced sulfide [4.35 mM (SOM)] and fracture water residence time of ∼3 or ∼20 million years inferred from noble gas analysis. Such a rate is comparable to rates reported in subseafloor sediments (28) and to the estimated rate of radiolytic H2 generation (10), but far less (4 to 5 orders of magnitude) than the maintenance energy demand of mesophilic sulfate reducers as determined in laboratory experiments (29) (SOM). The estimated in situ sulfate reduction rate, when combined with the experimental maintenance energy demand [48 mol (g dry weight cell)–1 year–1] (29) and assuming 20 fg per cell, would support only 200 to 1600 cells liter–1, which is far below the observed ∼4 × 107 cells liter–1. If cells were constantly growing and dying, then the experimental sulfate reducer yield of 12.2 g of dry weight cell mol–1 (29), when combined with the in situ sulfate reduction rate, would correspond to cell turnover times of 45 to 300 years. The isotopic estimate of the long-term in situ microbial activity, however, is 109 to 1010 times less than the maximum substrate consumption rate (Table 2), which would seem to indicate that as-yet-unidentified factors play a role in restraining the microbial respiration.

    The hot, reducing, gaseous water emanating from a fracture at 2.8 to 4.2 kmbls harbored a microbial community dominated by a single Firmicutes phylotype. The Firmicutes probably penetrated the Mponeng fracture zone at current depths during infiltration of paleometeoric water between 3 and 25 million years ago and since then have relied on nonphotosynthetically derived H2 and SO 2–4 converted from Archaean/Proterozoic pyrite/barite. Nutrient concentrations have remained much higher than observed in shallower crustal environments, suggesting that the deep crustal biosphere may be energy-rich, is not approaching entropic death, and is capable of sustaining microbial communities indefinitely by geological processes.

    Supporting Online Material

    SOM Text

    Figs. S1 and S2

    Tables S1 to S4


    References and Notes

    View Abstract

    Navigate This Article