Subseafloor ecosystems constitute a significant portion of Earth’s biosphere. The estimated total number of cells in marine sediment (~1029)1 rivals estimated totals in the ocean and in soil2. Although total cell abundance is unknown for the igneous aquifer that underlies marine sediment, microbes are routinely recovered from this aquifer3, chemical traces of their activity are pervasive in its altered rock4,5 and formation fluid6, and physical textures suggestive of microbial alteration are common7. The total volume of subseafloor habitats is immense; the volumes of marine sediment and igneous basement cooler than 122 °C [the currently accepted high-temperature limit to life8] are respectively 2.6 × 108 km39 and 1.0 × 109 km3 (Methods). The volumes of potentially habitable void space within the sediment9 and basement, respectively, are about 8.5 × 107 km3 and 8.0 × 107 km3 (each equal to about 6% of ocean volume [1.3 × 109 km3]10).

Mean respiration per cell is very low in subseafloor sediment11,12. Although their total abundance is high, microbial cells in marine sediment are generally much smaller than microbial cells in the surface world. Consequently, the total biomass of marine sedimentary life is <1% of Earth’s total biomass1. Per-cell catabolic rates and total biomass are presently unknown for microbial communities in the igneous aquifer beneath the sediment.

Because subseafloor communities reside at the interface between the biologically active surface world and the large geological reservoirs of biologically important chemicals, subseafloor microbial activities play a fundamental role in Earth’s biogeochemical cycles. This role is particularly visible in the organic-fueled activity of marine sedimentary life. Burial of reducing power (electron donors) in the form of organic matter and pyrite (FeS2) is the principal driver of Earth’s surface oxidation over geologic time13. Subseafloor life is the last biological filter through which organic matter passes on its way to burial and subduction. Among Earth’s surface and near-surface environments, marine sediment is the largest reservoir of carbon (including both the reduced carbon in organic matter and the oxidized carbon in sedimentary carbonates)14 and the third largest reservoir of nitrogen (after the atmosphere and continental crust)15.

The general geography of subseafloor catabolic activity is well constrained. Anaerobic activity dominates in the fast-accumulating sediment of continental margins and oceanic upwelling regions, where subseafloor respiration greatly outpaces the flux of dissolved oxygen (O2) into the sediment from the ocean. Aerobic activity dominates in the very slowly accumulating sediment of the deep open ocean, where subseafloor respiration rates are very low and the O2 flux keeps pace with respiration throughout the sediment column16 (Fig. 1). Rates of organic-fueled activity are highest in young, near-seafloor sediment and decline rapidly with increasing depth below seafloor17 (Fig. 2). Similarly, oxidation of igneous ocean crust is fastest for young crust (less than circa 10 Ma) and relatively slow thereafter18. Crustal oxidation is generally greatest where permeability is highest and seawater flow through the igneous aquifer is consequently fastest19. This is typically in discrete zones within the upper 200–500 m of the igneous sequence20 (Fig. 3).

Fig. 1
figure 1

Global distributions of dissolved O2 and SO42− in subseafloor sediment. In yellow regions, dissolved SO42− is expected to disappear within the sediment and HCO3 is expected to be the predominant net electron acceptor at greater depths. In white regions, dissolved SO42− is expected to penetrate throughout the sediment, from seafloor to igneous basement. Dark blue and light blue regions respectively represent the minimum and maximum areas over which dissolved O2 is expected to penetrate throughout the sediment from seafloor to basement16. SO42− reduction is expected to dominate net subseafloor organic oxidation in the yellow and white regions. O2 reduction is expected to dominate net subseafloor organic oxidation in the blue regions. Orange dots mark drill sites where SO42− disappears within the first 500 meters of sediment. White dots mark drill sites where SO42− penetrates the entire sediment sequence. Dark blue dots mark drill sites where dissolved O2 and dissolved SO42− penetrate the entire sediment sequence. Medium blue dots mark coring sites where O2 and SO42− penetrate to the bottom of the cores (meters to tens of meters) and may penetrate to basement. X symbols mark sites where the full extent of SO42− penetration is unknown, because SO42− is present in all measured samples, but the sedimentary sequence continues far below the last measurement. More information regarding the generation of this map is available in the Methods section

Fig. 2
figure 2

Profiles of net respiration rates at representative sites. a Profile of net respiration rate (O2 reduction rate) as a function of sediment depth at oxic IODP Site U1370 (dark blue zone in Fig. 1). b Profile of net respiration rate (SO42− reduction rate) as a function of sediment depth at anoxic ODP Site 1226, where dissolved SO42− is present throughout the sediment column (white zone in Fig. 1). c Profile of net respiration rate (SO42− reduction rate) as a function of sediment depth at anoxic ODP Site 984, where dissolved SO42− is depleted at a relatively shallow sediment depth (yellow zone in Fig. 1). Circles mark chemical concentration data color-coded to the Fig. 1 geographic zones in which the sites occur (dark blue, white and yellow, respectively). Red lines mark net respiration rates and pink bars mark uncertainty (first standard deviation). Crosshatched zones in A and B mark igneous basement

Fig. 3
figure 3

Profile of Fe oxidation ratio [Fe(III)/Fetotal] as function of basement depth. Fetotal = Fe(III) + Fe(II). Higher ratios of Fe(III)/Fetotal show that a larger fraction of the total Fe is oxidized. Filled circles mark data from DSDP Hole 504B. Open circles mark data from ODP Hole 896 A. All data from Alt et al.20. Figure modified from Alt et al.20

Given the ubiquity of subseafloor life and its central role in chemical exchange between the surface and subsurface worlds, this review has three major objectives. The first objective is to synthesize present understanding of subseafloor microbes, their catabolic activities, and their global consequences. The second objective is to advance understanding of the factor(s) that ultimately limit rates of subseafloor activities. The third objective is to identify major gaps in scientific understanding of subseafloor communities and activities, and recommend approaches to filling those gaps.

Diversities of subseafloor organisms and activities

Subseafloor sediment and the underlying basaltic aquifer are inhabited by many lineages of Bacteria21,22,23, Archaea22,23, and a subset of the Eukarya (fungi)24,25. Transcriptional data indicate that members of all three domains are active in subseafloor sediment25. Viruses are also abundant in both the sediment26 and the basaltic aquifer27.

Due to low per-cell energy fluxes28, selection is fierce in subseafloor sediment, leading both microbial diversity29 and cell concentration1 to decline exponentially with sediment depth and age. Broadly defined taxonomic groups (Chloroflexi, Deltaproteobacteria, Firmicutes) are present in relatively cool subseafloor sediment throughout much of the ocean22,23,29,30,31,32,33,34, suggesting that the consequences of this selection are relatively consistent at coarse taxonomic levels.

Geographic distributions of taxonomic richness and microbial abundance are largely unknown for the subseafloor igneous aquifer. However, different habitats within this environment are known to harbor distinctly different communities. For example, hot (65 °C) anoxic fluid in 3.5-Ma basalt of the Juan de Fuca Ridge (northeastern Pacific) contains abundant thermophilic anaerobes3,35, while relatively cold (3–4 °C) oxic fluid of the North Pond aquifer (central North Atlantic) is dominated by diverse Proteobacteria36,37. Communities from both aquifers are dominated by bacteria, with only trace concentrations of archaea35,37.

Subseafloor organisms glean energy from a broad range of redox reactions (e.g., disproportionation reactions, such as fermentation, and respiration reactions). In both the sediment and the igneous basement, electron acceptors (oxidants) for respiration include dissolved chemicals (O2, nitrate [NO3], sulfate [SO42−]) carried by diffusion or flow into the subseafloor, oxidized elements in minerals, and hydrogen peroxide (H2O2) from water radiolysis. In sediment, electron donors (reductants) include organic matter and reduced minerals deposited from the overlying world28, and hydrogen (H2) from in situ radiolysis of water38. In the igneous crust, electron donors include reduced elements in the rock (which originated in the underlying mantle)18, dissolved organic matter that enters with water from the overlying ocean6, and H2 from in situ water radiolysis39. Although in situ radiolysis may support subseafloor respiration in both the sediment and the underlying igneous aquifer, it may not have a net effect on subseafloor chemistry or global biogeochemical cycles, because it simultaneously creates reductant (H2) and oxidant (H2O2 and oxidized mineral species) in stoichiometric balance.

The traditional model of subsurface respiration assumes successive zones of reduction of O2, NO3, oxidized manganese (Mn[IV]), oxidized iron (Fe[III]), SO42−, and carbon dioxide (CO2) with increasing distance from the oxic surface world. In marine sediment, this sequence is vertical, with increasing depth below seafloor40,41 (Box 1). In aquifers with active fluid flow, the sequence depends on distance from the location where the fluid enters the aquifer from the surface42, e.g., the location where seawater enters the subseafloor igneous basement.

Most metabolic activities known to occur in subseafloor ecosystems were first discovered in surface or near-surface ecosystems. However, a few were first discovered in subseafloor sediment, including anaerobic oxidation of methane43, sulfate-reducing ammonium oxidation44, and microbial production of ethane and propane45.

The specific taxa responsible for individual metabolic activities are not yet known for most subseafloor environments46. The taxa responsible for specific activities may vary considerably in space and time; a recent metagenomic study of subseafloor planktonic communities in a cold oxic basalt aquifer found a high degree of metabolic functional redundancy over two years of sample collection, despite large shifts in community composition37.

Global consequences of subseafloor biological activities

Subseafloor microbial activity has at least five major consequences for global biogeochemical cycles (Fig. 4). First, it plays an important role in the oxidation-reduction (redox) evolution of Earth’s surface and near-surface environments47. Burial of organic matter is the principal pathway for removing reduced material from the surface world and subsurface microbial activity limits the rate of organic burial. In other words, subsurface microbial activity is the final throttle on the primary driver of Earth’s surface oxidation. Second, pyrite (FeS2) produced as a result of microbial SO42− reduction in marine sediment is the principal sink for sulfur from the world ocean48. Third, this microbially driven pyrite production is a principal source of alkalinity to the world ocean11. These consequences of pyrite precipitation are significant for two reasons: sulfate and bicarbonate (the principal contributor to alkalinity) are the second and third most abundant anions in seawater (after chloride)49, and alkalinity influences the distribution of CO2 between the atmosphere and ocean by determining the speciation of dissolved inorganic carbon. Fourth, subseafloor NO3 reduction decreases the amount of biologically fixed nitrogen available for marine biomass production (which is often N-limited)50. Fifth, subseafloor microbial activities play a significant role in creating and/or destroying resources of economic interest, including marine deposits of hydrocarbons51, phosphate52, dolomite53,54, and barite55.

Fig. 4
figure 4

Major net chemical fluxes due to microbial activities in subseafloor environments. Subseafloor microbial activities control burial fluxes of organic carbon (Corg), organic nitrogen (Norg), and reduced sulfur (Sred) (typically buried as iron sulfide), as well as production fluxes of alkalinity (ALK) and N2. In doing so, they control the rate at which reducing power is buried. Microbes in igneous basement contribute to burial of oxidizing power (via reduction of O2 and NO3)

Impact of subseafloor activity on the global redox budget

Oxidation-reduction (redox) reactions transfer electrons from one chemical to another. This transfer constitutes one of the most significant processes on Earth’s surface. The primary driver of redox activity on Earth’s surface is oxygenic photosynthesis, which converts water and carbon dioxide into oxidized and reduced compounds (O2 and organic matter, respectively)13. Organic-fueled respiration reverses this process by oxidizing the organic matter.

Over Earth history, the ocean and atmosphere have been progressively oxidized by continuous burial of reduced biological products (principally organic matter and pyrite)13. This burial results from the failure of subsurface microbes to oxidize all of the organic matter within sediment and from the microbially mediated transfer of reducing power from organic matter to pyrite. Major manifestations of this long-term oxidation include the precipitation of banded iron oxide deposits during the Archean and early Paleoproterozoic56 and the appearance of abundant free O2 in the ocean and atmosphere57. Consequences of this long-term oxidation include Earth’s astonishing diversity of oxygen-reliant organisms58, metabolites59, and oxidized minerals60.

The rate of organic carbon production (net primary production) via photosynthesis is ~4.5 × 1015 moles C yr−1 in the ocean61 and ~4.7 × 1015 moles C yr−1 on land62. Organic carbon flux to the seafloor is ~2 × 1014 moles C yr−161,63. Assuming a typical marine organic ratio of nitrogen to carbon in the seafloor organic flux (0.15)64, organic nitrogen flux to the seafloor is ~0.3 × 1014 moles N yr−1. Most of the organic-matter flux is ultimately oxidized, sustaining life at the seafloor and within the sediment. However, a significant fraction of the organic flux to the seafloor is permanently buried, rather than respired; the net organic C burial rate in marine sediment is in the range of 0.2 × 1014 moles C yr−161 to 0.7 × 1014 moles C yr−163 (Table 1). Most of the flux to the seafloor (70–85%) and the long-term burial (~90%) occurs in coast and shelf environments, in water depths shallower than 200 meters61,63.

Table 1 Major net chemical fluxes due to microbial activities in subseafloor environments. Organic carbon burial flux estimates from Sarmiento and Gruber35 and Dunne et al.37. Estimates of NO3 flux to sediment and N2 flux from sediment from Gruber50. All other fluxes calculated as described in the text. As described in the text, the extent to which microbial activities are responsible for the net oxidizing flux into the igneous basement is not yet known

By oxidizing organic matter in marine sediment, microbial communities reduce the mean oxidation state of the ocean and atmosphere. From a global redox perspective, this effect is a subseafloor manifestation of the same processes that consume organic matter in the surface world. Net microbial respiration in marine sediment generally consumes both oxidants (electron acceptors) and reductants (electron donors) that originated in the surface photosynthetic world. The principal net oxidants include dissolved species (O2, NO3, SO42−) that diffuse into sediment from the overlying ocean, and solid-phase oxidized metals (e.g., Fe[III] and Mn[IV] in minerals). The principal net reductant is buried organic matter. The principal products of organic-fueled respiration in marine sediment are dissolved inorganic carbon (DIC = HCO3 + CO32− + dissolved CO2) and dissolved nitrogen species. This DIC generally has one of three fates: some returns to the ocean via diffusion and/or advection as DIC, some is incorporated into carbonate minerals (e.g., CaCO3) and remains buried, and some is buried as DIC where sediment accumulates faster than chemicals can diffusively return to the ocean. Nitrogen in buried organic matter is released by organic degradation as ammonium (NH4+) and amines50. Much of this reduced nitrogen is oxidized to NO3 and returns to the ocean. Some of this NO3 is transformed to N2 (denitrified) during its migration to the ocean. Some of the reduced nitrogen is buried as dissolved NH4+, incorporated into mineral phases, or transformed to N2 by anaerobic processes44.

For this review, we quantify sedimentary redox fluxes in terms of the number of electrons transferred during reduction from the chemical species stable in contact with Earth’s oxygenated atmosphere (e.g., CO2, SO42−, NO3) to the reduced chemical species that persist in subseafloor sediment (e.g., organic matter, FeS2) (Methods). We similarly quantify basement redox fluxes in terms of the electron numbers required to oxidize the reduced chemicals in basalt. Previous studies have presented redox fluxes in terms of O2 equivalents47 or H2 equivalents;13 one O2 equivalent corresponds to four electron equivalents and one H2 equivalent corresponds to two electron equivalents.

The total electron-equivalent burial rate in marine sediment, including reduced C, N and S, is 1 × 1014 to 4 × 1014 moles e yr−1 (Table 1). The marine sedimentary organic carbon burial rate of 1.6 × 1013 to 6.5 × 1013 moles C yr−161,63 equals 6.4 × 1013 to 2.6 × 1014 moles e yr−1. Assuming an N/C ratio equal to typical marine organic matter (0.15)64, the marine sedimentary organic nitrogen burial rate is 2.4 × 1012 to 9.8 × 1012 moles N yr−1 (Table 1). Electron-equivalent burial as organic nitrogen (1.9 × 1013 to 7.8 × 1013 moles e yr−1) is ~1/3 the rate of electron-equivalent burial as organic carbon.

Marine sedimentary communities also modulate redox pathways by transferring electrons from one buried chemical species to another, most notably by transferring electrons from organic matter to iron and sulfur, leading to FeS2 formation. In this case, reducing equivalents (electron equivalents) that were originally buried with organic matter continue to be buried with the sulfur and iron to which they were transferred. This transfer of electrons is significant. Given an average burial ratio of organic carbon to reduced sulfur in marine sediment (7.5 mole C/mole S)65 and the organic burial rates mentioned above, the rate of reduced sulfur burial, principally as FeS2, is 2 × 1012 to 9 × 1012 moles S yr−1 (Table 1). The burial rate of electron equivalents as reduced sulfur (1.4 × 1013 to 6.3 × 1013 moles e yr−1) is approximately 1/4 the rate of electron-equivalent burial as organic carbon and slightly lower than the rate of electron-equivalent burial as organic nitrogen.

Oxidation of the fractured igneous basement that underlies the sediment also plays a role in global redox cycling, by reducing the oxidation state of the ocean and atmosphere47. As seawater hydrothermally circulates through this aquifer, electron acceptors from the ocean, principally O2 and NO3, oxidize reduced elements, principally iron and sulfur, in igneous minerals18. This circulation introduces electron acceptors to the upper crust throughout much of the ocean; dissolved O2 and NO3 is present in Eocene and Miocene basalt of the eastern Pacific21, Pleistocene to Cretaceous basalt of the central South Pacific16, and Miocene basalt of the North Atlantic66. The resulting oxidation of 1.7 ± 1.2 ×  1012 moles Fe yr−1 and 1.1 ± 0.7 × 1011 moles S yr−118 reduces the ocean and atmosphere by 2.5 ± 1.3 × 1012 moles eyr−1 (Table 1). The extent to which microbial respiration drives this oxidation is unknown, because dissolved O2 and NO3 abiologically oxidize reduced iron and sulfur67.

Multiple lines of evidence suggest that microbes harvest at least some of the energy from the mineral-oxidation reactions in the igneous basement. Microbial communities are routinely sampled from the basement aquifer3,36, members of these communities readily colonize fresh mineral surfaces68,69, and chemical data indicate that microbes play a role in late stages of rock alteration within the aquifer (e.g., SO42− reduction, methanogenesis)4,5. However, the extent to which subseafloor communities utilize the energy from these reactions is unknown. More importantly for this review, the net effect of that harvest on the global redox cycle is also unknown. For example, it is not yet clear if microbial communities enhance the total rate of basement oxidation by increasing the extent to which these reactions penetrate into the rock.

Impact on the global sulfur cycle

As the primary sink of sulfur from the ocean48 and a major source of alkalinity to the ocean11, SO42− reduction coupled to FeS2 precipitation in marine sediment directly affects ocean chemistry and indirectly affects atmospheric chemistry and climate. Sulfate is the dominant electron acceptor in the anoxic sediment that blankets ~60% of the seafloor. However, only a fraction of the sulfur reduced globally in marine sediment is precipitated as FeS2. Sulfur is reduced at the shallowest depths and highest rates in coastal environments, where up to 95% of the resulting sulfide diffuses upward and is re-oxidized70. In open-ocean environments, such as the equatorial Pacific, SO42− is reduced at greater sediment depths and iron concentration is great enough that ~100% of the reduced sulfur precipitates as FeS271.

The natural (pre-anthropogenic) flux of sulfur to the ocean is almost entirely via rivers. A recent estimate of this flux is ~2 × 1012 moles S yr−148. This estimate ignores rapid cycling of sulfur between the ocean and the immediately overlying atmosphere. It also ignores rapid microbial cycling of sulfur reduction and re-oxidation in shallow marine sediment and the water column of oxygen-minimum zones. Again ignoring these rapid cycles, the flux of sulfur from the ocean is almost entirely via metal sulfide precipitation (primarily FeS2 in marine sediment)13,48. As discussed above, we estimate the rate of reduced sulfur burial in marine sediment to be 2 × 1012 to 9 × 1012 moles S yr−1 (Table 1).

Impact on alkalinity and atmospheric CO2

Precipitation of metal sulfide following SO42− reduction removes the anionic charge of SO42− from the ocean. This charge is balanced by production of alkalinity11 (alkalinity is equal to the sum of positive charge equivalents minus the equivalents of negative charge from the conjugate bases of strong acids [Cl and SO42−])49. Consequently, precipitation in marine sediment of 2 × 1012 to 9 × 1012 moles reduced S yr−1 generates 4 × 1012 to 1.8 × 1013 moles of alkalinity equivalent yr−1 (Table 1). Given these estimates and a riverine primary alkalinity flux to the ocean of 1.3 × 1013 mole equivalents yr−172, microbially induced precipitation of sedimentary FeS2 generates 24–58% of the annual primary alkalinity flux to the ocean. CO2 partial pressure (pCO2) is highly sensitive to the value of alkalinity; for present-day ocean chemistry, a 25% increase in alkalinity will decrease atmospheric pCO2 by more than an order of magnitude (Methods).

Impact on the global nitrogen cycle

Ongoing burial of organic nitrogen is the primary sink of nitrogen from the atmosphere and ocean. Furthermore, microbial NO3 reduction (denitrification) linked to organic matter oxidation in marine sediment is the principal sink of fixed nitrogen from the ocean; the rate of sedimentary NO3 reduction is about a factor of two greater than the rate of denitrification in the water column and an order of magnitude larger than the rate of organic nitrogen burial50. In the absence of sedimentary denitrification, most marine ecosystems would not be nitrogen-limited. Most of this denitrification occurs in coast and shelf sediment, where the flux of organic matter to the seafloor is highest.

Impact on geological resources

Subseafloor microbial activities also play significant roles in creating and destroying resources of economic interest. The importance of subseafloor microbial activities is most fully delineated for destruction of subseafloor hydrocarbon resources51. More than 90% of the methane produced in marine sediment is consumed by anaerobic communities within the sediment73. Biodegradation is also extensive in relatively cool (50 °C) marine oil reservoirs, which have typically lost up to ~50% of their mass of C6+ hydrocarbons51. Biodegradation decreases greatly as a function of reservoir temperature. Evidence of oil biodegradation is generally absent from reservoirs warmer than 80 °C, suggesting that the thermal limit to life in the reservoirs is close to that temperature51. Although microbial roles are less thoroughly studied for production and destruction of other economic resources, rates and subsurface locations of organic oxidation, sulfate reduction and methane production are known to play central roles in precipitating sedimentary deposits of phosphate52, barite55, dolomite53,54, and gas hydrates. Microbial activities may also play central roles in formation of marine metal deposits, including hydrothermal sulfides74.

The limits of subseafloor activity

Since active subseafloor communities are present throughout the world ocean, a paramount biogeochemical mystery has been the failure of those communities to completely consume the electron donors present in subseafloor environments. Even in oxic marine sediment, traces of organic matter may survive for 100 Myrs or more16. Reduced iron and reduced sulfur persist in subseafloor basalt for more than 100 Myrs18. This failure plays a significant role in the redox evolution of Earth’s atmosphere and ocean, because, as described above, sequestration of organic matter in marine sediment is a principal sink for reducing power from the surface world. Furthermore, failure of microbial communities to oxidize all of the iron and sulfur in the subseafloor igneous aquifer leaves oxidizing power in the ocean that might have otherwise been lost to mineral oxidation. The failure to consume all of the buried organic matter also decreases the influence of subseafloor respiration on global cycles of sulfur, nitrogen and alkalinity. The persistence of organic matter in marine sediment is also crucial for thermogenic formation of major hydrocarbon resources, because oil and gas cannot form in the absence of their feedstock.

Not all sedimentary organic matter is consumed

Organic oxidation rate is generally highest at the seafloor and decreases with sediment depth, declining rapidly near the seafloor and increasingly slowly with greater distance from the seafloor. This relationship is typically modeled as a power-law function17,75,76. The basic reaction series can be simplified as follows:

$${\mathrm{6CH}}_{\mathrm{2}}{\mathrm{O}}_{{\mathrm{POM}}}--- > \left( {{\mathrm{CH}}_{\mathrm{2}}{\mathrm{O}}} \right)_{{\mathrm{6}}\;{\mathrm{(dissolved)}}}--- > 2\left( {{\mathrm{CH}}_{\mathrm{2}}{\mathrm{O}}} \right)_3 \\ \hskip 10pt+\,{\mathrm{3SO}}_{\mathrm{4}}^{2 - }--- > {\mathrm{6CO}}_2 + {\mathrm{3H}}_{\mathrm{2}}{\mathrm{S}}$$

Standard models of organic persistence in marine sediment assume that the first step is the rate-limiting step75. However, if organisms are operating at their minimum harvestable reaction affinity77, catabolic reactions cannot proceed faster than reactants appear or reaction products disappear at the reaction location78. In this case, the rate-limiting step may occur later in the reaction series, such as the final respiration step of the above sequence, which will be limited by the rates at which SO42− is introduced and the rate(s) at which CO2 and H2S are removed (Box 2).

This dependence of organic oxidation rate on chemical fluxes and minimum reaction affinity will naturally lead to the typical depth-dependence of organic oxidation rate (Box 2). Because the time required for chemical diffusion is proportional to the square of diffusive distance79, where advection is insignificant (below the depth of bio-irrigation), organic oxidation rate must decline exponentially with distance from sources of catabolic reactants and sinks of catabolic products (Box 2). Where electron acceptors and catabolic products respectively diffuse from and to the ocean above, the rate of organic oxidation will decline exponentially with depth beneath the seafloor.

Most quantitative models of the decline in bulk organic oxidation rate with sediment depth assume that organic persistence results from organic molecular structure, with selective preservation of the least reactive (most carbon-rich) compounds75,76. These models are variations on Berner’s 1964 G Model, which assumed organic oxidation rate to be proportional to organic matter concentration76. More recent models assume that bulk organic oxidation rate declines with sediment depth and age because increasingly recalcitrant pools of organic matter are successively depleted17. This assumption is consistent with laboratory experiments that show sequential utilization of different sedimentary organic carbon pools by aerobic heterotrophs80. Despite this consistency, reliance on relative degradability to explain organic oxidation rates is problematic for the following reasons. First, the molecular composition of these organic pools is never identified by studies that assume them to model organic oxidation rates17,76. Second, the degradability of an organic-matter pool, such as kerogen (the fraction of organic matter insoluble to organic solvents) in deeply buried sediment, is not inherent to the organic molecules that compose that pool, but depends on the environment in which the organic pool is present81. For example, 365-million-year-old organic matter is quickly degraded by microbes when exposed by erosion82. Third, preferential degradation or persistence of different classes of organic compounds may be better considered as a function of community metabolic capabilities than as an intrinsic property of organic molecules83. Fourth, preferential persistence of the most carbon-rich compounds is difficult to reconcile with preservation of relatively low C/N ratios for tens of Myrs84,85, and with subseafloor respiration of C/N ratios close to the average marine (Redfield) value for tens of Myrs16,29,86.

Other studies have proposed that organic matter persists because it is adsorbed to mineral grains85,87,88. This hypothesis is consistent with a generally close association between high bulk organic concentrations and abundance of fine-grained minerals65. However, reliance on organic adsorption to minerals as the primary explanation of organic persistence in marine sediment is difficult to sustain for the following reasons. First, it is difficult to reconcile the molecular integrity of some compounds, such as certain fatty acids and pigments, with the solubilization that must precede adsorption89. Second, discontinuous distribution of organic matter renders binding to mineral grains an insufficient explanation of organic persistence in continental-margin sediment65,90. Third, much of the adsorption to minerals is easily reversed and ultimately cannot preserve organic matter91.

These different explanations have fundamentally different implications for the relationship between organic oxidation rates and organic-matter persistence in deep old sediment. To the extent that organic oxidation rate depends on minimum reaction affinity and diffusion rates of dissolved catabolic reactants and products (Box 2), organic matter can persist simply because oxidation rate declines with sediment depth. In contrast, selective-preservation hypotheses assume that organic oxidation rate declines because residual organic molecules are chemically recalcitrant. Finally, the mineral adsorption hypothesis could be used to predict that organic oxidation rate declines because residual organic matter is strongly adsorbed to minerals.

Demonstration of increased organic recalcitrance or increased mineral adsorption with sediment age is necessary to respectively support the hypotheses that organic recalcitrance or mineral adsorption drives the decrease in organic oxidation rate with increasing sediment depth and age. However, neither test is sufficient to prove either hypothesis, because organic recalcitrance and/or mineral adsorption might increase with sediment age even if they do not drive the decrease in organic oxidation rate with sediment depth and age.

Persistence of electron donors in subseafloor sediment is not limited to solid-phase organic matter. For example, the turnover times of dissolved fatty acids in sulfate-reducing subseafloor sedimentary ecosystems ranges from approximately tens to hundreds of years (Methods). These relatively long turnover times of dissolved electron donors are not explainable by chemical recalcitrance or adsorption to minerals. However, they are a natural consequence of microbial thermodynamics coupled to subseafloor timescales of chemical diffusion (Box 2).

The causal relationship between organic oxidation rates and organic-matter persistence may differ from one marine sedimentary environment to another. Diffusion of chemical species may limit organic oxidation rates in fast-accumulating anoxic sediment, where organic oxidation reactions operate at minimum harvestable affinities of reaction. However, in sediment where electron acceptors are abundant and reaction products are scarce, organic persistence may result from mineral shielding or molecular inaccessibility of the organic matter. Such circumstances might apply in the slowly accumulating organic-poor oxic sediment that blankets much of the abyssal ocean16,85.

The igneous subseafloor is not fully oxidized

The factors that limit oxidation rates in the igneous subseafloor are rarely addressed explicitly. The permeability of un-fractured basalt is extremely low and rock alteration is closely associated with fractured surfaces92. Given these points, persistence of reduced iron and sulfur in subseafloor basalt can most simply be ascribed to physical inaccessibility of the reduced elements within the un-fractured rock. This ascription is a variation on the diffusive-timescale explanation that we present for sedimentary organic oxidation above and in Box 2; however, in the igneous basement, permeability is so low that chemical diffusion of electron acceptors into un-fractured rock is extremely slow, even for sub-millimeter distances.

The persistence of dissolved organic matter (DOM) in the igneous aquifer is more enigmatic. DOM enters the aquifer with water from the overlying ocean6. Because DOM is in the water, not the rock, it is in contact with dissolved electron acceptors. Although some DOM is consumed within the aquifer, DOM is present along the entire flow path. In the cool oxygenated aquifer of North Pond (North Atlantic), DOM has radiocarbon model ages of several thousand years6. These model ages are significantly older than the residence time of water in the aquifer or the residence time of deep water in the North Atlantic, suggesting ongoing supply of aged organic matter to the deep ocean and the aquifer.

Microbial activities and the subseafloor limit to life

The preceding paragraphs and Box 2 focus on limits to subseafloor activity in environments where organisms are present and metabolically active. In subseafloor environments, fluxes of bioavailable energy may be too low to maintain community function under physical conditions that differ notably from physical limits to life in other environments. For example, the high-temperature limit to oil and gas bioalteration (80–90 °C)51 is considerably lower than the high-temperature limit to life in energy-rich and nutrient-rich autoclave conditions (122 °C)8. This difference illustrates that the high-temperature limit to life may vary from one environment to another as a function of bioavailable energy flux or nutrients9,51.

Factors other than temperature may also sterilize subseafloor systems. More precisely, changes in chemical conditions might sterilize subseafloor environments after multi-Myr intervals of selection for previous chemical conditions. For example, the oxic condition of abyssal clay sequences for tens of Myrs may drive obligate anaerobes in the clay to extinction, to be followed by local extinction of aerobes if sedimentation rates increase and dissolved O2 disappears from the clay.

Future directions

As indicated in the preceding sections, many key issues remain to be resolved. Global rates of subseafloor microbial activities and their consequences are only roughly quantified. For example, recent estimates of organic carbon burial in marine sediment differ by ~4 × 61,63 and recent estimates of global SO42− reduction rate in marine sediment differ by ~7.5× 93,94. Our estimates of FeS2 precipitation rate, alkalinity production rate and organic nitrogen burial rate in marine sediment are based on estimates of organic carbon burial rate and also vary by ~4×. This lack of precision is exacerbated by incomplete understanding of microbial processes in the subseafloor. For example, the extent to which microbes participate in redox alteration of inorganic phases in subseafloor basalt and sediment remains to be determined. More precisely, the extent to which microbes harvest the energies of mineral-altering redox processes that can proceed abiologically is not known. Furthermore, the extent to which microbes actively enhance mineral alteration, e.g., by mining more deeply or more rapidly into minerals and/or rock than done by abiotic alteration, is also unknown.

The factors that limit rates of subseafloor catabolic activities are crucial to the operation of Earth’s biogeochemical cycles but not fully understood. Consideration of reaction affinities and dissolved chemical transport suggests that diffusion rates of catabolic reactants and products play an important and previously under-recognized role in limiting subseafloor catabolic rates and, thereby, sustaining the persistence of electron donors in subseafloor environments.

The specific taxa responsible for individual activities are not known for most subseafloor environments46. The metabolic networks that sustain their communities for millions of years are largely unexplored. The nature and extent of communication between cells in subseafloor communities is unexamined. At the most fundamental biological level, the genomic properties and patterns of gene expression that enable lineages of subseafloor microbes to survive at extraordinarily low rates of activity for millions of years remain unknown28.

Better quantification of subseafloor microbial activity rates and their consequences will require integration and quantitative analysis of data from diverse studies. For example, more accurately quantifying global burial rates of organic carbon, organic nitrogen and sulfide in marine sediment will require synthesis of geochemical data from shallow and deep sediment of coastal environments, continental shelves and the open ocean. And precise quantification of sulfide burial’s effect on ocean alkalinity and atmospheric chemistry will require quantification of both sulfide burial and sulfate flux to the ocean.

Deeper understanding of subseafloor activities and their consequences will require Earth and life scientists to deliberately integrate techniques from diverse fields (e.g., molecular biology, aqueous geochemistry, sedimentology, petrology and mineralogy) to test key hypotheses. For example, fuller understanding of how dissolved chemical transport and reaction affinities limit rates of subseafloor activities – and ultimately control global biogeochemical fluxes – will require integrated study of in situ abilities to undertake specific catabolic reactions, in situ reaction rates, and in situ energies of reaction.

Hypothesis-driven laboratory experiments and direct study of natural communities will both be crucial to advance understanding of the physiological features, metabolic strategies, and genomic properties that enable lineages of subseafloor microbes to persist at extraordinarily low rates of activity for millions of years. Laboratory experiments are necessary to elucidate the features and strategies that enable microbes to thrive or survive under specific subseafloor conditions, at least on laboratory timescales, and the genes that code for those features and strategies. Direct studies of natural subseafloor communities are needed to delineate the conditions in which they live, to characterize microbial interactions in subseafloor communities, and to identify features and strategies that enable microbial lineages and communities to thrive or survive under subseafloor conditions for many millions of years.

With both laboratory experiments and natural-sample studies, a broad range of analytical approaches is crucial to understand the organisms and communities. Genomic studies are necessary to know what capabilities are available. Transcriptomic studies are needed to understand which capabilities are expressed. Biogeochemical studies are required to determine in situ reaction rates, in situ energies of reaction, and consequences of microbial activities. Exacting studies of organic composition and mineral composition are necessary to identify the substrates available to subseafloor organisms and the consequences of microbial activities. Natural-community studies often benefit from explicitly placing them in a temporal context, e.g., by repeatedly sampling a subseafloor flow path over time37, by sampling crust of very different ages, or by treating data from younger samples in the same sediment column as potentially representative of earlier stages in the history of a subseafloor community29,95,96 or habitat. Such temporal context is crucial for understanding how subseafloor communities are assembled. It is also critical for understanding how their biogeochemical roles and consequences vary throughout the subseafloor in space and time.

Individual studies need not integrate across many categories of study. Appropriate laboratory experiments can bring great value whether they are tied to specific field sampling programs or not. Similarly, well-posed natural-sample studies can effectively address key problems (e.g., quantification of specific biogeochemical fluxes or identification of globally consistent genomic and/or transcriptomic patterns) without close integration to other categories of natural-sample study or to laboratory experiments. Ultimately, however, each category of study derives greater strength by integration with one or more of the other categories. For example, laboratory studies inform the understanding of data from subseafloor samples, and subseafloor data inform the hypotheses that drive the laboratory experiments. Understanding of natural subseafloor communities and their consequences also benefits from synthesizing results of different approaches to natural samples. For example, biogeochemical results might explain why a specific capability that is present is not expressed, and genomic studies might explain why a reaction that is energetically feasible does not occur.

As shown by these examples, close integration of biological, chemical and physical approaches will ultimately be necessary to further advance understanding of subseafloor life and its biogeochemical consequences. This integration will require both Earth and life scientists to begin thinking of many geological and geochemical features (such as organic recalcitrance and rates of subseafloor redox processes) as ecosystem properties, rather than a priori environmental features.


Calculation of habitable basement volume

The total volume (rock + water) of submarine igneous basement cooler than 122 °C is 3.0 × 108 km3. We calculated this total volume using the method of Heberling et al.97; however, instead of globally averaged theoretical and observed heat flow values98 and constant sedimentation rate of 3 m Myr−1, we used global grids of heat flow99 and sediment thickness100 to calculate global maps of depth to the 122 °C isotherm with a grid resolution of 0.1 × 0.1 degrees. We then spatially integrated these depths to calculate a global volume.

Assuming 10% in upper 500 m of basement and a linear decrease in porosity to zero at 1000 m into basement101, habitable pore space in the basement is 2.2 × 107 km3 (about 2% of ocean volume).

Determination of dissolved O2 and SO4 2− distributions

We used interstitial water concentrations of dissolved O2 and SO42− from ODP and IODP drill sites102 and a two-step approach to create the global map of electron acceptor distributions in sediment (Fig. 1). The first step was to plot the chemical species as a function of sediment depth and note whether their concentration goes to zero within the sediment or is non-zero all the way to basement. Of 380 ODP and IODP drill sites with SO42− data, SO42− concentrations go to zero within the sediment at 150 sites, 70 sites had non-zero SO42− concentrations from seafloor to basement, and 160 sites were undetermined because the sites were not drilled to basement or SO42− measurements were stopped at depths far above basement. The dissolved O2 results were previously reported in D’Hondt et al. (2015)16.

The second step identified relationships between the presence and absence of the various chemical species as a function of sedimentation rate and total sediment thickness. This method was previously used for O2 in sediment;16 we extended it to include the SO42− data (Fig. 5). Our results indicate that SO42− goes to zero in the sediment column when sediment thickness exceeds about 500 m or sedimentation rate exceeds 35 m Myr−1. In comparison, dissolved O2 concentrations go to zero in the sediment where sediment thickness exceeds about 150 m or sedimentation rate exceeds 15 m Myr−1. These ‘boundaries’ for persistence or absence of SO42− to igneous basement were combined with similar boundaries for dissolved O2 (e.g., D’Hondt et al., 2015) to generate Fig. 1 from global sediment-thickness maps100 and depth-averaged sedimentation rates. The depth-averaged sedimentation rates were derived from the sediment thickness maps and global maps of ocean basement age103.

Fig. 5
figure 5

Plot of sediment thickness and sedimentation rate at ODP and IODP sites. Only sites where dissolved SO42− has been measured in sedimentary porewater are shown. Orange dots indicate sites where SO42− concentrations go to zero within the upper 500 m of sediment. White dots indicate sites where SO42− is present from seafloor to basement. The yellow, dark blue and light blue fields identify the sediment thickness and sedimentation rate boundaries used to generate Fig. 1. The blue fields are based on previous results for dissolved O2 reported by D’Hondt et al.16. Dissolved O2 may or may not be present throughout the sediment at sites marked by white dots in the blue fields; no O2 measurements were made at these sites. White dots in the yellow areas include sites where flow introduces SO42− through rubble or from beneath the seafloor and sites where SO42− diffuses upward from underlying brine. Orange dots in the white field generally mark sites where dissolved chemical concentrations are not at diffusive steady state (e.g., because sedimentation rate has been unusually high in the recent geologic past)

Reaction-rate calculations

To quantify net rates of O2 reduction (IODP Site U1370) and SO42− reduction (ODP Sites 1226 and 984) from dissolved chemical data (Fig. 2a–c), we used a modified version of the Matlab-based numerical procedures of Wang et al.71. We modified the approach of Wang et al. by using an Akima spline, instead of a 5-point running mean, in order to generate a best-fit line to the chemical concentration data. We determined standard deviations through use of a Monte Carlo simulation (n = 50). For these calculations, diffusivities are from Schulz104, corrected for in situ temperature at Sites 1226 and U1370 (temperature was assumed constant at Site 984, where in situ temperature was not measured). Porosities and in situ temperature data are from shipboard measurements of the respective drilling expeditions105,106,107 (available at Sedimentation rates are from the shipboard age models generated for each drill site105,106,107.

Calculation of electron-equivalent burial rates

We calculate electron-equivalent burial rates by multiplying the difference between the dominant oxidation state of the chemical element (C, N or S) at Earth’s surface and its oxidation state in marine sediment. The dominant surface-Earth oxidation states for C, N, and S are IV, 0 and VI, respectively. The oxidation states in marine sediment are taken as 0, -III and -I for organic C, organic N and S in FeS2.

Calculation of alkalinity effect on CO2 partial pressure

Increasing alkalinity from present ocean values will cause net conversion of dissolved CO2 to predominantly HCO3 and to a lesser extent CO32− (dissolved CO2 = H2CO3 + CO2). We calculated the exact change in equilibrium CO2 partial pressure for a given change in alkalinity using the thermodynamic constants for the aqueous CO2 system49.

Dependence of organic degradation on diffusive distance

Our idealized model of the depth dependence of organic degradation rate assumes that the rate of a catabolic reaction sequence (i.e., production of a sugar monomer from particulate organic matter, followed by fermentation and respiration), occurs at or near steady state at a given sediment depth and that the energy (−ΔG) yielded by each biologically conserved reaction (fermentation or respiration) must equal or exceed the minimum amount that can be biologically conserved (ΔGminimum)77,108,109. That is, the affinity (A = –ΔG) for each catabolic reaction must equal or exceed the biological minimum affinity (Aminimum). This assumption is supported by observations of in situ energies of reactions in marine sediment44,77,110,111. Here, we use SO42− reduction as an example. In this approach, the catabolic production rate of DIC (of which HCO3 is the major species) depends on the HCO3 concentration, 1 – c/cA-min, to keep A ≥ Aminimum, where cA-min is the HCO3 concentration above which A drops below Aminimum and the reaction sequence does not proceed. This dependence of rate on metabolite concentrations was previously applied by Boudart (1978)112 to generic reaction sequences and adopted by Jin and Bethke (2009)78, and Dale et al. (2008)113. We extend this approach by explicitly identifying the effect of diffusive distance on reaction rate, via its effect on metabolite concentrations. We choose HCO3 for this demonstration because, in sulfate-reducing sediment where sulfide is scavenged by iron, HCO3 exhibits the largest relative change in concentration.

P(z), the production of DIC at depth z is then given by

$$P\left( z \right) = P_{\mathrm{max}}\left( {1 - c\left( z \right)} \right]/\left. {c_{A - {\mathrm{min}}}} \right)$$

where c(z) is the HCO3 concentration at depth z, and Pmax is the rate when c = 0 and the local mass balance is

$$\frac{{\partial c\left( {z,t} \right)}}{{\partial t}} = 0 = D\frac{{\partial ^2c\left( {z,t} \right)}}{{\partial z^2}} + P\left( z \right)$$
$$0 = D\frac{{\partial ^2c(z,t)}}{{\partial z^2}} + P_{\mathrm{max}}\left(1 - \frac{{c\left( z \right)}}{{c\left( z \right)_{A - {\mathrm{min}}}}}\right)$$

where D is the HCO3 diffusion constant. A solution to Eq. (3) is

$$c\left( z \right) = \left( {c\left( {z = 0} \right)-c_{A - {\mathrm{min}}}} \right) \ast \left( {{\mathrm{exp}}\left( { - P_{\mathrm{max}}} \right./\left( {D \ast c_{A - {\mathrm{min}}}} \right.} \right) + c_{A - {\mathrm{min}}}$$

and the depth-dependent metabolic rate is then

$$P\left( z \right) = P_{\mathrm{max}} \ast {\mathrm{exp}}\left( {\left( { - P_{\mathrm{max}}/\left( {Dc_{A - {\mathrm{min}}}} \right)^{0.5}z} \right)} \right)$$

Where Pmax = DIC production rate when cDIC = 0.

Turnover-time calculations

The turnover times of low-molecular-weight fatty acids are calculated as the concentration of lactate, acetate or formate divided by the rate of SO42− reduction corrected for stoichiometry (respectively 1.0, 0.67, and 0.5 times the rate of SO42− reduction for acetate, lactate, and formate) at IODP Site 1226, a site where SO42− is reduced but not entirely depleted. The different ratios to SO42− reduction are used to account for the different oxidation states and numbers of carbon atoms in each molecule, assuming the initial particulate organic carbon (POC) had oxidation state 0. Because we do not know the fraction of total SO42− reduction rate that goes to oxidize each of these low-molecular-weight acids, we calculated a potential range of turnover times based on each one separately. At this site, concentrations of these low-molecular-weight fatty acids are all in the range of 1 µM111 and the subseafloor SO42− reduction rate declines from 3 × 10−8 moles liter−1 yr−1 to ~0.2 × 10−8 moles liter−1 yr−1 with increasing depth below seafloor71. Given the relatively constant low-molecular-weight fatty acid concentrations, the decline in SO42− reduction rate leads to turnover times increasing from the low tens to low hundreds of years with increasing sediment depth.