An integrated sulfur isotope model for Namibian shelf sediments

In this study the sulfur cycle in the organic-rich mud belt underlying the highly productive upwelling waters of the Namibian shelf is quantified using a 1D reaction-transport model. The model calculates vertical concentration and reaction rate profiles in the top 500 cm of sediment which are compared to a comprehensive dataset which includes carbon, sulfur, nitrogen and iron compounds as well as sulfate reduction (SR) rates and stable sulfur isotopes (S-32, S-34). The sulfur dynamics in the well-mixed surface sediments are strongly influenced by the activity of the large sulfur bacteria Thiomargarita namibiensis which oxidize sulfide (H2S) to sulfate (SO42-) Using sea water nitrate (NO3-) as the terminal electron acceptor. Microbial sulfide oxidation (SOx) is highly efficient, and the model predicts intense cycling between SO42- and H2S driven by coupled SR and SOx at rates exceeding 6.0 mol S m(-2) y(-1). More than 96% of the SR is supported by SOx, and only 2-3% of the SO42- pool diffuses directly into the sediment from the sea water. A fraction of the SO42- produced by Thiomargarita is drawn down deeper into the sediment where it is used to oxidize methane anaerobically, thus preventing high methane concentrations close to the sediment surface. Only a small fraction of total H2S production is trapped as sedimentary sulfide, mainly pyrite (FeS2) and organic sulfur (S-org) (similar to 0.3 wt.%), with a sulfur burial efficiency which is amongst the lowest values reported for marine sediments (<1%). Yet, despite intense SR, FeS2 and S-org show an isotope composition of similar to 5 parts per thousand at 500 cm depth. These heavy values were simulated by assuming that a fraction of the solid phase sulfur exchanges isotopes with the dissolved sulfide pool. An enrichment in H2S of S-34 towards the sediment-water interface suggests that Thiomargarita preferentially remove (H2S)-S-32 from the pore water. A fractionation of 20-30 parts per thousand. was estimated for Sox (epsilon(SOx)) with the model, along with a maximum fractionation for SR (epsilon(SR-max)) of 100 parts per thousand. These values are far higher than previous laboratory-based estimates for these processes. Mass balance calculations indicate negligible disproportionation of autochthonous elemental sulfur; an explanation routinely cited in the literature to account for the large fractionations in SR. Instead, the model indicates that repeated multi-stepped sulfide oxidation and intracellular disproportionation by Thiomargarita could, in principle, allow the measured isotope data to be simulated using much lower fractionations for epsilon(SOx) (5 parts per thousand) and epsilon(SR) (78 parts per thousand.). (c) 2008 Elsevier Ltd. All rights reserved.