Research Article

Sulfur Isotopic Composition of Cenozoic Seawater Sulfate

See allHide authors and affiliations

Science  20 Nov 1998:
Vol. 282, Issue 5393, pp. 1459-1462
DOI: 10.1126/science.282.5393.1459


A continuous seawater sulfate sulfur isotope curve for the Cenozoic with a resolution of ∼1 million years was generated using marine barite. The sulfur isotopic composition decreased from 19 to 17 per mil between 65 and 55 million years ago, increased abruptly from 17 to 22 per mil between 55 and 45 million years ago, remained nearly constant from 35 to ∼2 million years ago, and has decreased by 0.8 per mil during the past 2 million years. A comparison between seawater sulfate and marine carbonate carbon isotope records reveals no clear systematic coupling between the sulfur and carbon cycles over one to several millions of years, indicating that changes in the burial rate of pyrite sulfur and organic carbon did not singularly control the atmospheric oxygen content over short time intervals in the Cenozoic. This finding has implications for the modeling of controls on atmospheric oxygen concentration.

Global changes in climate and atmospheric chemistry are intimately related to the sedimentary S cycle (1, 2). Seawater sulfate represents one of the main reservoirs of dissolved S (3). Evidence for large-scale transfers of S between different sedimentary reservoirs is provided by the evaporite-based isotope record of oceanic sulfate (4–6). Here, we present a high-resolution, continuous seawater S isotope age curve (Fig. 1) that could help to elucidate the factors affecting the cycles of S and C and atmospheric oxygen content over geologic time.

Figure 1

(A) S isotopic composition of seawater sulfate over the past 65 million years. Marine barite samples are from the following locations: Holocene core-top sediment samples are from the Pleiades expedition (1° to 6°N, 105° to 138°W), Eurydice expedition (1° to 5°N, 155° to 165°E), Indomed expedition (10° to 30°N, 13° to 33°W) and Joint Global Ocean Fluxes (JGOFS) EqPac expedition (12°S to 9°N, 140°W). PC72 is a piston core obtained during the JGOFS EqPac expedition (0°N, 139° W). DSDP core samples are from Leg 85 site 572 (1°26′N, 113°50′W, 3893 m, 0 to 15 Ma), Leg 85 Site 574 (4°12′N, 113°19′W, 4561 m, 0 to 37 Ma), Leg 85 Site 575 (5°51′N, 135°02′W, 4536 m, 0 to 22 Ma), Leg 86 Site 577 (32°26′N, 157°43′W, 2675 m, 0 to 65 Ma), Leg 32 Site 305 (32°26′N, 157°50′E, 2675 m, 27 to 65 Ma), and Leg 41 Site 366 (3°26′N, 9°50′W, 2853 m, 25 to 65 Ma). (B) S isotope curve over the past 65 My, obtained from evaporites [data from (6)]. All the S isotopic ratios are reported in per mil relative to the Canyon Diablo Troilite standard and are plotted with respect to the chronology in (41). The sulfur isotope data are available on request from the corresponding author.

The concentration of dissolved sulfate in the ocean and its δ34S value are controlled by (i) a balance between the flux of S to the oceans from continental weathering and its isotopic value [δ34S of this source is variable but is typically lower than seawater, 0 to 10 per mil (1)], and (ii) the deposition rate of S-bearing sediments, including evaporite sulfates with δ34S values approximately equal to that of seawater, and pyrite sulfide with a large isotopic fractionation signature [for example, modern δ34S values of pyrites average about –20 per mil (7)]. Volcanism and hydrothermal activity are also sources of reduced sulfur to the ocean [δ34S values of these sources vary from about 0 to 3.5 per mil (1, 8, 9)]. Therefore, considerable S isotopic variations can be generated. For example, during periods of enhanced pyrite formation and burial, the light S isotope is preferentially removed from the ocean and the δ34S value of seawater sulfate increases. These and other fluctuations are recorded in contemporaneous evaporites and marine barite.

Suitability of barite for recording seawater sulfate isotopic composition. Records of the S isotopic composition of seawater sulfate have been obtained mostly from isotope data on marine evaporite sulfates [for reviews see (5, 6, 10)], but the geological record of marine evaporites is episodic, with gaps of millions of years (5). Evaporites are also susceptible to diagenesis, their age control is often problematic (3), and a purely marine origin of some is questionable (11). In another attempt, Burdett et al. (12) used the S isotope composition of carbonate-associated sulfate in tests of planktonic foraminifers. However, the nature of SO42−incorporation in carbonates is not well understood (13), and its behavior during carbonate diagenesis is unknown. Because carbonates are susceptible to diagenetic alteration (14), and because biological vital effects during the incorporation of sulfate cannot be ruled out, the use of marine carbonates for establishing a record of S isotopes of seawater sulfate is also problematic.

We used marine barite, a readily available ubiquitous minor phase in deep-sea sediments (15, 16), for the construction of the seawater sulfate S isotope record. It continuously forms in the water column; the associated sediment ages can be determined using bio- and magnetostratigraphy; barite is not particularly susceptible to diagenetic alteration in oxic sediments (16, 17); S is one of its major constituents; and at Earth-surface temperatures, the fractionation factor of S between seawater and barite is small (<0.4 per mil). Furthermore, barite has been shown to reliably record the seawater Sr isotopic composition (16). For these reasons, marine barite is a suitable mineral for recording the history of seawater S geochemistry.

We separated marine barite from more than 20 Holocene core-top sediments from the Atlantic, Pacific, and Indian oceans, as well as from six Deep Sea Drilling Project (DSDP) cores spanning the last 65 million years (My) of Earth's history (see Fig. 1 for sample locations). Separation and purification of barite were carried out using a sequential leaching method modified from (16) to ensure that no sulfide S reprecipitated as barite during the procedure (18). The barite S was converted to silver sulfide and fluorinated to SF6 (19,20).

The mean δ34S value for all core-top samples of barite was 21.18 per mil relative to the Canyon Diablo Troilite standard (Fig. 1). The δ33S value is 10.82 per mil, and δ36S is 39.45 per mil. The precision was ±0.20 per mil. This δ34S value compares well with the modern seawater δ34S value of 20.99 ± 0.25 per mil determined by (21). This comparison indicates that marine barite records the contemporaneous S isotopic composition of seawater. Moreover, data from older marine barite samples from different cores agree with each other at any given time (Fig. 1), suggesting that barite in these samples reliably records and preserves the S isotopic composition of seawater. Over the 65-My time interval, the crystal habit of the barite does not change, and the Sr isotopic ratios of the barites (16) are within the range of the published Sr isotopic record for carbonates of corresponding age. Both observations indicate that the marine barites used in our study have not significantly recrystallized over the past 65 My.

Systematics of the S isotope composition of seawater sulfate. The major features of the original evaporite-based seawater sulfate S isotope curve for the Cenozoic are seen in the new marine barite data (Fig. 1). However, the new data set, obtained at a time resolution of ∼1 My, shows considerably more detail, filling significant gaps and revealing newly recognized features. The oceanic residence time of dissolved sulfate is ∼20 My (22). Our data set indicates that substantialvariations in δ34S have taken place on shorter time scales, implying that the oceans have not been at isotopic steady state with respect to sulfur. This result should be taken into account in future efforts to model the linkage between the S and C cycles and atmospheric O2(1).

The low values of δ34S between 65 and 55 million years ago (Ma) indicate that during this time interval, either pyrite deposition rates were lower or there was increased input of isotopically light S to the oceans. The minimum at ∼55 Ma, the Paleocene-Eocene boundary, coincides with an important benthic foraminiferal global extinction; large, negative, short-term C isotopic excursion (Fig. 2A); a maximum in deep ocean temperature (∼18°C); and intense volcanism (23). The largest shift in the S isotope ratio during the Cenozoic was in the early to mid-Eocene from 55 to 45 Ma. The pronounced shift from 17 to 22 per mil over a period of <10 My indicates that large changes in the external forcing mechanisms, such as tectonic uplift and weathering, volcanism, or oceanic vertical mixing and productivity (22, 24), influenced the fluxes and isotopic compositions of S input or burial, or both. Possible explanations for this increase include (i) a reduction of the river input of S to the ocean (by ∼30% over 10 My) without changing the S output and sulfate/sulfide ratio, thus causing a decrease of ∼20% in the seawater S concentration; (ii) a pronounced decrease in hydrothermal activity or volcanism; and (iii) a large increase in the rate of pyrite formation and burial relative to that of sulfates. Evidence for significant decreases in volcanic activity as volcanism in the North Atlantic and Caribbean decreased sharply at ∼55 Ma (23) exists for this time interval. No direct evidence for increased input fluxes from the continents starting around 55 Ma exists, but the initial collision between India and Asia occurred then (25). Could the S curve be a manifestation of early uplift and weathering caused by the collision, as may be suggested by the cooling trend between 55 and 40 Ma (26) and supported by the187Os/186Os shift toward more radiogenic values (27)? Intensified weathering and increased input of nutrients to the ocean would enhance marine biological productivity and sedimentation rates. These effects would promote increased pyrite formation and burial, which is limited by the availability of marine organic carbon (22, 28) and would lead to higher seawater δ34S values. A combination of the second and third possibilities is the most likely explanation for this isotopic increase.

Figure 2

(A) Cenozoic seawater sulfate S isotope record (this work) and bulk deep-sea sediment carbonate δ13C record (30) averaged over 1-My increments with respect to the chronology in (41). (B) A plot of the S isotopic composition in marine barite versus the C isotopic composition of marine carbonate during the Cenozoic. The data are grouped according to age.

In the Oligocene to Pliocene, a 35-My period during which Earth experienced significant tectonic and climate changes, including the initiation and intensification of Antarctic glaciation, closure of the Mediterranean basin and of the Central American Isthmus, and intensified uplift and denudation in the Himalayas (25, 29,30), the oceanic sulfate δ34S value varied by <2 per mil (Fig. 1). The only pronounced feature of the isotopic record in the last 35 My is the small minimum in the Oligocene (at ∼30 Ma), also observed in the evaporite record (6), which most likely represents a decrease in pyrite burial rates. The pronounced δ34S decrease, of ∼0.8 per mil in the last ∼2 My, may also indicate a decrease in the pyrite burial rate, possibly because of a higher degree of ventilation and oxidation of deep water as it forms at cold high latitudes. This and the Oligocene shifts are also manifested in the C isotopic record (30) (Fig. 2A). An increase in the relative input of lighter S isotopes as a result of increased weathering, as indicated by the seawater Sr isotopic curve, could also cause this shift.

S and C isotope coupling and implications for atmospheric O2 . Burial and oxidation of organic carbon and sulfide S are considered the dominant controls on the mass of atmospheric oxygen (1, 2, 9, 22, 24, 28, 31). Geological evidence suggests that the mass of atmospheric oxygen has not decreased or increased by more than a factor of ∼2 of the present amount (∼38 × 1018 moles) since the Cambrian (32). Accordingly, major shifts in the redox state of S in the environment must have been balanced by complementary redox-state shifts of C, resulting in the expectation of a negative correlation between δ34Ssulfateand δ13Ccarbonate at any given time. The δ34Ssulfate and δ13Ccarbonate records provide estimates for the relative partitioning of C and S between their respective oxidized and reduced reservoirs through time. These records have been extensively used in models to assess the history of atmospheric Po2 as well as the histories of C and S burial and weathering fluxes. When these isotopic data are averaged over time periods of ≥50 My, for the Phanerozoic, a negative correlation between δ34Ssulfate and δ13Ccarbonate is observed (33). A comparison between the detailed δ13C curve of marine carbonates, averaged over 1-My intervals (30), and the 1-My-resolution S isotope curve presented here (Fig. 2A) reveals no such correlation nor other significant correlations (Fig. 2B), suggesting that the changes in organic C and pyrite S burial rates do not balance each other over short time scales of a few million years or less. Using these two isotopic data sets in a model such as described by Kump and Garrels (24), in which the atmospheric oxygen level is driven solely by sulfide S and organic C burials, yields fluctuations in atmospheric oxygen concentration during the Cenozoic of >20%. Either such changes in atmospheric oxygen concentration have occurred or the stability of atmospheric oxygen concentration has been maintained, but not solely by the coupling between the C and S cycles, especially on time scales of 1 My to several million years. A deeper understanding of all the processes that influence the oxygen budget is needed before atmospheric oxygen fluctuations can be modeled reliably using the S and C isotope data of seawater sulfate and carbonate. These processes include the recently suggested involvement of the phosphorus cycle (34), hydrothermal and volcanic activities (9, 35), and iron oxidation (36). More elaborate models with negative feedback mechanisms that provide limits on atmospheric Po2 excursions are needed to produce a realistic history of atmospheric O2.

The observation that the δ34Ssulfate and δ13Ccarbonate records do not correlate on time scales of 1 My to several million years (Fig. 2) is a consequence of the great difference in the residence times and oceanic reservoir sizes of C (105 years and 2 × 1018 moles, respectively) and S (2 × 107 years and 39 × 1018 moles, respectively) (5, 22, 37). Thus, the time required for the oceanic C and S concentrations and isotope ratios to reach a new steady state in response to input or output flux perturbations differs substantially; consequently, the oceanic C and S systems may be decoupled by millions of years (22, 38). Another factor that would preclude the suggested inverse relation between marine δ34Ssulfate and δ13Ccarbonate is the observation that in normal marine settings, the amount of metabolizable organic matter is the most important limiting factor for bacterial sulfate reduction processes (39). This dependence results in a positive correlation between the burial rates of organic carbon and pyrite sulfur, as documented in both modern and ancient marine environments (39, 40). Because organic C and pyritic S are enriched in the lighter isotopes, their increased burial rates would shift the oceanic δ34Ssulfate and δ13Ccarbonate to higher values and would lead to a positive correlation between these two parameters. The increased atmospheric O2 concentration resulting from increased organic carbon burial will be amplified by the S cycle (9,22).

  • * To whom correspondence should be addressed. E-mail: apaytan{at}


View Abstract

Stay Connected to Science

Navigate This Article