Global rates of marine sulfate reduction and implications for sub–sea-floor metabolic activities

See allHide authors and affiliations

Science  23 May 2014:
Vol. 344, Issue 6186, pp. 889-891
DOI: 10.1126/science.1249213


Sulfate reduction is a globally important redox process in marine sediments, yet global rates are poorly quantified. We developed an artificial neural network trained with 199 sulfate profiles, constrained with geomorphological and geochemical maps to estimate global sulfate-reduction rate distributions. Globally, 11.3 teramoles of sulfate are reduced yearly (~15% of previous estimates), accounting for the oxidation of 12 to 29% of the organic carbon flux to the sea floor. Combined with global cell distributions in marine sediments, these results indicate a strong contrast in sub–sea-floor prokaryote habitats: In continental margins, global cell numbers in sulfate-depleted sediment exceed those in the overlying sulfate-bearing sediment by one order of magnitude, whereas in the abyss, most life occurs in oxic and/or sulfate-reducing sediments.

Mapping sub–sea-floor communities

The sea floor is teeming with microbes, whose sheer numbers produce a major effect on the global biogeochemical cycles of carbon, sulfur, and other important nutrients. Bowles et al. constructed a map showing how deeply sulfates penetrate marine sediments worldwide and how quickly that sulfate is chemically reduced by microbes in the sub–sea-floor. Globally, almost a third of the organic carbon that reaches the sea floor is consumed during sulfate reduction, and the vast majority of microbial cells in the sub–sea-floor at continental margins get their energy through the biochemical processes of fermentation and methanogenesis.

Science, this issue p. 889.

Sulfate reduction is a ubiquitous microbial process in oceanic sediments and an important pathway for carbon oxidation and redox cycling (1, 2). Nevertheless, the currently estimated global sulfate-reduction rate (SRR) of 75 teramoles (Tmol) of sulfate per year (1) is based on coarse spatial averaging and is not consistent with the up-to-date assessment of the global organic matter flux to marine sediments of 79 to 192 Tmol carbon per year (3, 4). Net sulfate reduction follows a two carbon–to–one sulfur stoichiometric ratio [Embedded Image; for example, see (5)]. Therefore, the current estimates for the rate of subsurface sulfate reduction and for the organic carbon flux to the sediment suggest that either insufficient organic carbon reaches the sediment to account for sulfate reduction or that most (78%) organic matter is channeled toward sulfate reduction. Nevertheless, the organic carbon reaching the sediment must also foment other prominent redox reactions such as carbon respiration (4, 6). Moreover, a sizable portion of sedimentary organic matter successfully survives early diagenesis and is buried (7). This discrepancy in global geochemical cycles gives impetus for an amended global view on sulfate reduction, which can be merged with recently revised global prokaryotic abundances to properly assess the activity of sulfate-reducing microorganisms at a global scale (5, 811).

We used currently available sulfate-concentration profiles from multiple scientific ocean drilling programs (12) to estimate global net SRRs (Fig. 1A). These profiles were best described by assuming that sulfate concentrations exponentially decrease with depth. A total of 199 sulfate profiles (Fig. 1B) with a mean error square value <4 mM2 based on a least-squares regression were selected for the global SRR analysis. We then used depth-decay constants (b) extracted from these profiles to train an artificial neural network (ANN) using high-resolution (1 × 1 degree) satellite observations and water-column chemistry maps (e.g., surface water chlorophyll A, particulate organic carbon, and bottom water O2) (1214) (table S1).

Fig. 1 Global SRRs in marine sediments.

(A) Global distribution of depth-integrated SRRs (millimoles per square centimeter per year) based on predictions of the exponential depth-decay constant (b) from an ANN in a 1° by 1° resolution. Black point symbols represent 199 Deep Sea Drilling Project/Ocean Drilling Program/Integrated Ocean Drilling Program sites with sulfate profiles described by an exponential fit ultimately used to train, validate, and test the ANN. (B) The correspondence of predicted b values from the ANN and the actual fit values of b for all profiles (correlation coefficient R = 0.88).

The ANN predicts depth-decay constants ranging 16 orders of magnitude from 5.8 × 10−13 to 3.2 × 102 m−1, while depth-integrated SRRs calculated by a steady-state diffusion, advection, and reaction function (eq. S8) (12) ranged from 5.8 × 10−12 to 8.2 mmol cm−2 year−1. The highest SRRs were predicted in shelf environments, and the lowest SRRs were calculated in the nutrient-poor oceanic gyres (Fig. 1A and Table 1). These general trends are corroborated by a previous prediction of global, depth-integrated SRRs derived from and mainly reflecting the distributions of primary productivity (2). Furthermore, these trends are consistent with observations in previous studies of global sulfate profiles from oceanic deep drilling programs (5), as well as global compilations of radiotracer gross SRR measurements (15). Although the ANN is trained and validated solely with deep-sea drilling data (Fig. 1B) (12), it replicates exponential sulfate depth-decay coefficients from several published short cores (<15 m) (fig. S4) taken below the shelf break and captures deep-sea locations characterized by high regional SRRs coupled to methane oxidation [e.g., 0.05 to 0.4 mmol cm−2 year−1 in the Arabian Sea (16)].

Table 1 Weighted average depth-integrated SRR for different water depths.

The total area covered here is ~349 by 106 km2 or ~97% of the total ocean.

View this table:

From an environment-specific perspective, the ANN predicts a fourth of the previously estimated area-weighted and depth-integrated SRR for shelf sites: 0.097 mmol cm−2 year−1 (15) versus 0.025 mmol cm−2 year−1 (Table 1). All average depth-integrated rates determined here are considerably lower than previously reported (15) but appear to agree better with regionally averaged sulfate penetration depths. For example, assuming the previous upper continental slope average areal rate of 7.4 × 10−2 mmol cm−2 year−1 (15), a tortuosity-corrected diffusion coefficient of 120 cm2 year−1, and a porosity of 0.8, the average sediment depth where sulfate depletes to 0.3 mM in sediments would be only 35 cm or 1.74 m, assuming a linear or an exponential profile, respectively. The calculations here yield an average depth-integrated rate for the upper slope of 1.7 × 10−2 mmol cm−2 year−1. This rate would produce 0.3 mM sulfate concentrations at a depth of 1.2 or 20 m below sea floor (mbsf) for these same cases, respectively. The main reason for this difference is that the previously compiled global SRRs did not account for large sea-floor surface areas (e.g., ~70% of the continental shelf) consisting of organic-poor relict sands (17) and, thus, were averaged with a bias toward high-activity, organic-rich sites (8, 18). This and other geochemical and depositional heterogeneities observed in coastal sediments may also explain some of the larger deviations between fitted and ANN-predicted b values for some shallow-water cores (Fig. 1B). For instance, order-of-magnitude variations in SRRs have been estimated within a single muddy basin [e.g., within 3000 km2 in Arkona Basin (19)] and likewise for small gassy basins [e.g., 8 km2 in Aarhus Bay (20)].

The ANN-based global net SRR estimate (11.3 Tmol sulfate year−1) (Table 1) is roughly 15% of previous estimates for gross SRR (1). Although calculating global net as opposed to gross SRR could explain this divergence, it is highly unlikely that gross SRRs account for more than 78% of the global organic carbon flux to the sea floor. In contrast, the 11.3 Tmol sulfate year−1 predicted by the ANN is equivalent to 22.6 Tmol organic carbon year−1 [assuming a 2 C–to–1 S stoichiometry (5, 12)] or a more realistic 11 to 29% of the estimated global organic carbon flux to the sea floor.

This substantial diminution in global SRRs inherently affects previous conceptions of global microbial process distributions in sub–sea-floor sediments. Subsurface microorganisms largely depend on harvesting energy from the organic matter reaching the sea floor. This amount is minor in comparison to the carbon supplied to seawater prokaryotes via photosynthesis (4.3 Pmol C year−1) (21). In spite of this sharp contrast in carbon availability, the marine subsurface total prokaryotic biomass is approximately equal to that of seawater (8).

Coupling the ANN-derived global SRR maps to global sub–sea-floor biomass maps (8) allows for the calculation of potential cell-specific rates, which can help us to further elucidate the activities of sulfate-reducing microorganisms across various global sedimentary environments. These microorganisms thrive in anoxic surficial sediments where sulfate and labile organic substrates coincide. Within inner-shelf sediments (<50 m water depth), which typically receive the highest inputs of labile organic matter, area-weighted SRR averages (Fig. 2, A and B) exhibited the highest rates (1.5 nmol cm−3 day−1), leading to submicromolar sulfate concentrations by 6 mbsf. Furthermore, inner-shelf sediments comprise the highest prokaryotic cell abundances (Fig. 2C) and cell-specific rates (Fig. 2D) around 0.1 fmol cell−1 day−1. These data are in strong contrast to those of deep-water environments, which receive considerably less organic carbon. Peak SRRs in abyss sediments (>3500 m water depth) are a fraction of the shallow-water counterparts, at 0.03 pmol cm−3 day−1. Furthermore, cell abundances are generally lower, with the cell-specific rates reaching a maximum of 9 × 10−4 fmol cell−1 day−1.

Fig. 2 Subsurface profiles of area-weighted parameters in various oceanic depth zones.

(A) Sulfate profiles (millimolar), (B) SRRs (femtomoles per cubic centimeter per day), (C) cellular abundances (cells per cubic centimeter) (8), and (D) cell-specific rates (femtomoles per cell per day). Thick lines represent a 10% sulfate-reducing microorganism contribution to the total population (baseline scenario discussed in the text), whereas the shaded region for each line in (D) represents the range of 1 to 30% contribution of sulfate reducer to the total population.

Simultaneous measurements of SRRs and sulfate-reducing microorganism abundances are rare, but the existing data are consistent with our model (2226). The majority of these data exist for relatively shallow-water, high-productivity sites (e.g., Aarhus Bay), with surficial cell-specific SRRs around 0.1 fmol cell−1 day−1 and reaching 1.0 × 10−3 fmol cell−1 day−1 by ~1 mbsf (11). These data are within the range of our areal weighted average for 0.1 mbsf of 0.1 fmol cell−1 day−1 (<50 m water depth) (Fig. 2). In our modeled shallow-water environments (i.e., inner and outer shelves), high SRRs lead to peak cell-specific rate values near the sediment-water interface. These cell-specific rates taper quickly to zero as sulfate becomes exhausted. Our results show that peak values for the slope also occur near the sediment-water interface (with a slight increase with sediment depth) and gradually decrease as sulfate approaches zero. The abyss, however, does not reach a peak cell-specific rate within the top 80 mbsf, and values remain one order of magnitude lower than the peak values for the other environments. Notably, for the assumed fractions of sulfate reducers within the total microbial community (1 to 30%), the general trends persist for cell-specific SRRs in different environments (Fig. 2D).

Results of the global survey show a distinct trend between environments in the continental margin and the abyss. The abyss (>3500 m water depth) is typically characterized by organic-poor sediments that allow for deep sulfate penetration. This prevalence of sulfate at great sedimentary depths indicates that most cells within the habitable deep sedimentary biosphere (down to 4000 mbsf or the specific basement depth) are found in either oxic or sulfate-reducing settings (Table 2). Nevertheless, within the other environments on the continental margins, sulfate is removed at comparatively shallow sediment depths (<100 m) (Table 2). The sulfate-methane transition zone (SMTZ) is a distinct geochemical horizon that represents an important transition from sulfate-reducing (overlying the SMTZ) to methanogenic sediments (underlying the SMTZ) (1, 19, 20). Limited data from deep-sea cores at these sites suggest that acetate and hydrogen can be abundant and thus serve as substrates for a vast methanogenic subsurface (2729).

Table 2 Global analysis of SRRs with respect to published organic carbon fluxes to the sea floor for various water depth environments.

NR, concentration not reached. The global values in bold represent totals, whereas those in italics represent area weighted averages. SR, sulfate reduction.

View this table:

Collectively, these observations indicate that although the lack of reduced substrate limits sulfate reduction in deep-sea sediments, the continental margins harbor an expansive biosphere below the SMTZ where traditional, energy-rich electron acceptors are exhausted. Thus, this fraction of the microbial biosphere is largely fermentative and methanogenic (Table 2 and figs. S5 and S6). Roughly estimating the SMTZ at the depth at which sulfate depletion reaches 0.1 mM sulfate, habitable sediments located below the SMTZ would make up a total global subsurface volume of 108 km3 (32% of total), hosting ~50% of the sub–sea-floor biomass (12). However, ~90% of cells in the subsurface at the continental margins (<3500 m water depth) would be situated below the SMTZ (12).

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S6

Table S1 to S4

References (3044)

References and Notes

  1. See supplementary materials on Science Online.
  2. Acknowledgments: All data used in this study are publically available via Janus and Pangea ( We thank Deep Sea Drilling Project/Ocean Drilling Program/Integrated Ocean Drilling Program, Janus, and Pangea for compiling and providing these data sets. We thank A. Boetius, J. J. Middelburg, and S. Joye for helpful comments during the development of this work. Primary support for this work was provided by the Research Center/Cluster of Excellence “The Ocean in the Earth System” (MARUM) funded by the Deutsche Forschungsgemeinschaft in the framework of a Postdoctoral Fellowship awarded to M.W.B. Additional funding was provided by the European Research Council (ERC) under the European Union’s Seventh Framework Programme–“Ideas” Specific Programme, ERC grant agreement no. 247153, the Helmholtz Association (AWI Bremerhaven), and Utrecht University through its strategic theme Sustainability, sub-theme Water, Climate, and Ecosystems.
View Abstract

Navigate This Article