Fe(III) Input
U1474 U1475
U1478
U1482 U1483
U1485 U1486
U1488 KN223-02
KN223-16
40 45 50 55 60 65 70 75 80 85 90
1 10 100 1000 10000 100000
Estimated Net Sulfate Reduction Rate (μmol/m3/yr)
Estimated34ε(‰)
105
104
103
102
101
0 10 20 30 40 50 60 70
EstimatedNetSulfateReductionRate(μmol/m3/yr)
Estimated Sedimentation Rate (cm/kyr)
A B
Figure V.28: Relationships among mean nSRR, sedimentation rate, and34ε.
(A) Scatter plot of I-CANDI model estimates of34ε (vertical axis) versus net sulfate reduction rate (nSRR; horizontal axis) for each of the sites modeled in this study. Note that nSRR is plotted on a logarithmic scale. (B) Scatter plot of estimated nSRR versus estimated sedimentation rate for each of the sites in this study.
256]) have observed H2S oxidation that creates SO42−depleted in34S. Such ox- idation could account for the Δδ34S reversal if SO42−derived from oxidation is the dominant component of the ambient SO42− pool. Small amounts of solid phase S oxidation or of blank contamination could also skew the SO42–δ34S data toward lower values under these conditions.
Although we do not have H2S δ34S data from any sites beyond Sites U1482 and U1483, we consistently observe a return of SO42– δ34S to seawater-like values (∼+21h) below the SMT at all sites where we have data. This consistency in SO42–δ34S below the SMT suggests that a common process is responsible. No- tably, the amount of SO42−in these post-SMT samples is vanishingly small in our data as well as in other studies [255]. Based on these considerations, we favor blank contamination as the source of this Δδ34S reversal. We are currently un- able to distinguish whether such contamination occurred during shipboard pro- cesses (e.g. small amounts of contamination with seawater during coring plus some sulfide oxidation) or during shore-based laboratory processing. Analysis of pore water species that are differentially sensitive to each of these contami- nation sources is needed to resolve this uncertainty.
V.6.7 Implications for the global marine sulfur cycle
Our results demonstrate that high34ε is an ubiquitous feature of sulfur cycling in modern deep marine sediments. Although the physical parameters of de- positional environments [127, 246] and the amount of iron and labile organic matter input may modulate the δ34S of pyrite that is buried in these sediments, past observations [4, 37, 38, 115, 158, 188, 197, 220, 222, 266, 293, 329] and our modeling show that the depletion of this deep ocean pyrite in 34S relative to pyrite in shelf sediments is robust.
Previous efforts to characterize global nSRRs (e.g. [43]) have often ignored sites that are not well characterized by an exponential decrease in[SO42−]with depth. Such filtering of sites eliminates inclusion of many continental slope sites with relatively linear[SO42−]profiles (e.g. Sites U1482 and U1483 in this study) or with upward fluxes of SO42−from sources at depth (e.g., Sites U1486 and U1488). Our modeling shows that such exclusion probably imparts a sub- stantial bias in global nSRR estimates; both of these varieties of sites display elevated nSRRs compared to sites with exponential[SO42−]decreases (e.g. Sites U1474 and U1475) in our study. Preferential sampling of high nSRR sites and other sources of bias prevent us from concluding that the global nSRR estimate of Bowles et al. (2014) [43] is too low; in fact, their estimate of 11.3 x 1012molyr is already four times higher than a recent estimate of the pre-anthropogenic riverine SO42– input flux [52]. However, our analysis does show that a more serious accounting of these sites may be necessary to bring estimates of the net input and output fluxes in the modern marine S cycle into better agreement.
We can roughly constrain the size of the bias that may be present in current global nSRR estimates based on the exclusion of the sites with a SO42– source at depth. Although global fluxes of S to and from the crust due to hydrother- mal activity remain poorly constrained, prior studies [2, 3, 358] have estimated the flux of S out of seawater through circulating fluids to be in the range of 0.3 to 2.5 Tmolyr . Most of this flux is removed via the precipitation of anhydrite that later dissolves under cooler off-axis temperatures [2, 3]. For comparison, the amount of SO42− reduction estimated by Bowles et al. (2014) [43] to occur in sites below 2000 m water depth is about 2.9 Tmolyr ; most spreading centers are within this depth range. If 100% of the anhydrite that dissolves from cooling crust is removed via SO42− reduction in the sediments and does not re-enter seawater, this additional SO42−flux would increase the amount of SO42−reduc-
tion occurring in these deep environments by 10 to 86%. These numbers are admittedly an upper limit on the contribution of SO42− from bottom fluids to global SO42−reduction if the Bowles et al. (2014) [43] model is accurate, but the contribution could be even more significant if the model overestimates global SO42−reduction rates (see Chapter VI).
Finally, Figure V.27 presents a useful framework for interpreting the influence of past Earth system changes on the S isotopic composition of pyrite buried glob- ally. For example, Jones and Fike (2013) [160] observed an increase in pyrite δ34S in end-Ordovician sediments without any corresponding increase in the δ34S of seawater SO42−proxies. This decrease in Δδ34S was attributed to a decrease in
34ε, but our analysis suggests that a decrease in ambient water temperatures [101] and higher bottom water oxygen concentrations could also contribute to this change. Halevy et al. (submitted) [127] have also recently argued that a decrease in seawater [SO42−] and the corresponding diffusive flux of SO42– to sediments can explain the lower Δδ34S values observed in Archean and Protero- zoic sediments. While this could be true, our model indications that lower bot- tom water O2 concentrations (e.g. [58, 94, 167]), lower RRPOCs (e.g. [32]), higher temperatures (e.g. [181]), and lower sedimentation rates (e.g. [130]) all force pyrite δ34S towardslower values (i.e., higher Δδ34S) complicate this pic- ture. More extensive sensitivity studies with these variables — as well as Fe(III) mineral fractions of differing reactivities [47] — are needed to fully evaluate the net effect of these competing influences on Δδ34S through time.
V.7 Conclusions
Here, we have undertaken a study of sedimentary S cycling in deep ocean sed- iments through measurements and modeling of S geochemical profiles at 11 sites cored on IODP Expedition 361, IODP Expedition 363, andR.V. Knorrcruise KN223. We find high diversity in the character of the[SO42−]and δ34S profiles across the different sites. However, all are united by large (> 45h)34ε estimates based on closed system and open system modeling of their profiles. The mod- eled δ34S of the pyrite buried at each of these sites is quite depleted in34S relative to pyrite buried in shallow shelf sediments, but exhibits a complicated depen- dence on POC rain rate, Fe(III) rain rate, and oxygen penetration depth. No- tably, nSRRs are higher among the sites with quasi-linear[SO42−]profiles and with[SO42−]profiles indicating a deep source of SO42–than at sites with a tradi-
tional exponential decrease in[SO42−]. Ignorance of such sites in estimates of the global nSRR likely biases these estimates, and accounting for the sites could help future estimates resolve isotope mass balance within the modern marine S cycle. In addition, better understanding of the effects of non-S species on the δ34S of pyrite preserved in marine sediments is needed. These considerations motivate continued investigation into the controls on pyrite δ34S in marine sed- iments.
C h a p t e r VI