Depth of dolomite formation
The 87Sr/86Sr ratios of all of the studied dolomite plots close to the profile for Quaternary seawater composition (Fig. 8) indicate that all of the samples have an early diagenetic origin and have not precipitated from hydrothermal fluids. Additional evidence from petrographic relationships indicates primary precipitation of dolomite in the sedimentary sequence of the upper slope and outer shelf sites with cementation and no replacement of precursor carbonate. The growth of framboidal pyrite post-dates the precipitation of the dolomite, providing additional support for early dolomite formation. At all sites, HS− concentrations decrease below the SMI, often reaching 0. Below this depth, Fe2+ is remobilized and available to nucleate as framboidal crystals. Dissolved iron has been measured in the methanogenic zone by shipboard analysis but is below the detection limit in the sulphidic zone (D'Hondt et al., 2003). Early formation of dolomite layers has been reported from other deep-sea sites, such as the Romanche Fracture Zone in the equatorial Atlantic (Bernoulli et al., 2004) and in the Oligocene to Miocene deep-sea fan succession of the Gonfolite Lombarda Group, northern Italy (Bernoulli & Gunzenhauser, 2001).
Strontium isotope and carbon isotope data indicate a general disequilibrium between dolomite and modern porewater. Also, δ18O values around 4‰ are within the range expected for dolomite precipitated from seawater under low-temperature conditions (Vasconcelos et al., 2005). Some scatter in the data is probably due to varying seawater composition and palaeotemperatures throughout the Quaternary. These disequilibria suggest that the dolomite formation occurred mainly in the past, and the depth of present dolomite formation is reflected primarily in the porewater chemistry. Since the 87Sr/86Sr values in the porewater show a linear gradient, due to diffusive mixing between brine and seawater, the vertical projection of the values measured in the dolomites on the mixing line (regression line at Site 1229) constrain dolomite precipitation to the uppermost 30 mbsf within the sequence at the Peru Margin shelf Site 1229/681 (Fig. 8C). Also, Sites 1227/684 (Fig. 8A), and 1228/680 (Fig. 8B) show similar trends, with a few more radiogenic values at Site 1228/680. Even if these values are due to leaching of volcanic ash layers of a different composition (see Hart & Miller, 2006), the disequilibrium between 87Sr/86SrDOL and 87Sr/86SrW still indicates that the dolomite layers form at shallow depth. Layers that occur at greater depths than 30 mbsf are not forming now but formed in the past, when they where located nearer to the sediment/water interface. Thus, the δ13CDOL values can be used as a proxy for the activity of the ‘deep biosphere’ in the past (see below).
Porewater Mg and Ca profiles show the most pronounced concentration minima around 30 mbsf and indicate that present dolomite formation probably occurs at this depth (e.g. Site 1229; Fig. 11C). Therefore, downward diffusion from seawater, as well as upward diffusion from the brine delivers Mg and Ca for dolomite formation. To form a dolomite layer of 3 to 5 cm thickness, precipitation must be focused at a particular site with Mg and Ca diffusing to that site for a certain amount of time. In fact, with the Mg-gradients commonly observed at the studied sites, a dolomite layer in the range of 2 to 3 mm per 10 000 years would be precipitated. These are minimum rates not taking into account the episodicity of the dolomite formation, but it is consistent with the amount of dolomite formed throughout the Quaternary. The minima of the Ca and Mg correlate with maxima in alkalinity, where alkalinity is highly increased by sulphate reduction. Elevated alkalinity in the uppermost 5 mbsf is probably due to sulphate-reducing activity in freshly deposited organic carbon-rich layers, whereas the high alkalinity at 30 mbsf is due to increased microbial activity at the SMI. At Site 1227, alkalinity at 5 mbsf is higher than at the SMI and correlates with the Mg and Ca profiles, which show a pronounced minimum at this depth. It is also noted that at Site 1227 there is evidence that the modern porewater chemistry is not in equilibrium with the current redox zonation and that the SMI has shifted downwards in the recent past. In summary, based on the interpretation of the data, active dolomite formation is restricted to the most biogeochemically active horizons, which is often coincident with the SMI.
Figure 11. Correlation of sulphate and methane concentration with alkalinity, Mg and Ca concentrations and bacterial cell counts at Site 1229 (D'Hondt et al., 2003), showing high cell concentrations at the upper and lower SMIs. At these boundaries increased microbial activity occurs, which causes maximum alkalinity at these depths with the potential to induce dolomite precipitation.
Download figure to PowerPoint
The microbial factor in dolomite precipitation
In the classic concept, the depth and the diagenetic zone of formation are controlled by the sedimentation rate, which limits the downward diffusion of SO from seawater (Kelts & McKenzie, 1984; Baker & Burns, 1985; Burns & Baker, 1987). Depending on the depth of diffusion, dolomite is formed either in the sulphate reduction zone or in the methanogenic zone and shows the δ13C values typical for the zone of formation (Claypool & Kaplan, 1974). In this ‘organic dolomite model’, the depth of major organic matter degradation is, therefore, determining where dolomite precipitates independent of the biological activity at the site of precipitation.
In general, our interpretation is in agreement with this model, however, it is suggested that the formation of distinctive layers occurs at strictly focused sites within the sedimentary section, along geochemical interfaces, where the microbial activity is the driving force for dolomite precipitation. The observation of ‘microbial hotspots’ coincident with the SMI (Fig. 11), maxima in alkalinity and minima in Mg and Ca concentration support this hypothesis and would explain the focused growth of multiple, less than 5 cm thick, dolomite layers within a more than 100 m thick sedimentary succession with diffusion of Mg and Ca over long distances towards that particular site independent of the lithology at the particular horizon. Based on the different models used to explain dolomite formation, the biogeochemical activity found at the SMI would, in any case, favour dolomite precipitation. Alkalinity is strongly increased at the SMI, which increases the supersaturation of dolomite at this particular horizon. Often, the SMI shows maxima in alkalinity or a sharp change in slope (e.g. Site 1230), which indicates net production of alkalinity. As a secondary effect, by the consumption of Mg and Ca in stoichiometric proportions during dolomite precipitation, the Mg/Ca ratio is increased. Kinetically, dolomite precipitation may be favoured by eliminating the inhibiting effect of SO, as suggested by Baker & Kastner (1981). However, this effect has not been demonstrated for low-temperature conditions. Most importantly, the SMI shows extraordinarily high cell concentrations (Fig. 10; D'Hondt et al., 2003, 2004; Parkes et al., 2005) and can be considered as a microbial ‘hotspot’ within a deep sub-seafloor biosphere. The presence of living microbes is required for low-temperature dolomite precipitation, as shown by van Lith et al. (2003), by providing the appropriate physico-chemical conditions to overcome the kinetic inhibition. The mechanism of this mediation process is not clearly understood and needs to be investigated. Possibly, the formation of extracellular polymeric substances (EPS) along geochemical interfaces may play an important role, but remains to be observed in deep-sea sediments. A similar interpretation was based on observations from a Bahamian stromatolite, where calcite crystals formed in a single EPS layer attributed to high sulphate reducing activity (Visscher et al., 2000).
The type of microbial activity at the present SMI is indicated by δ13CDIC, which often reaches strongly negative values due to AMO. This is clearly the case at Site 1230, where δ13CDOL is lower than −30‰. This is consistent with high production of CH4 and the presence of gas hydrates at greater depth. Nevertheless, δ13CDIC is not as negative as the δ13CDOL, which indicates that AMO may be episodic (see discussion below). Moreover, δ13CDIC and δ13CDOL at the shelf sites do not show any evidence for significant AMO. Also, CH4 production is low at the shelf sites, with μm concentrations of CH4, which is insufficient to account for most of the sulphate reduction through AMO. DNA studies by Parkes et al. (2005), Schippers & Neretin (2006) and Inagaki et al. (2006) show generally low amounts of Archaeal 16S rRNA genes and other groups, such as green non-sulphur bacteria, seem to be dominant at the SMI. However, based on the strongly varying δ13CDOL values, which reach in different depths much higher values than modern δ13CDIC, methanogenesis was much higher in the past, and the high bacterial cell numbers observed at the SMI may be mainly dormant remains of a much more active SMI in the past. The dolomite could have incorporated a broad range of δ13C values under non-steady-state conditions.
The dynamic deep biosphere
Different dolomite samples recovered in the sediments of the Peru margin during ODP Legs 112 and 201 show evidence of formation in the past at a shallower depth than their present position and, thus, document an evolution through most of the Pleistocene, which was influenced by a highly dynamic deep biosphere. The variations in δ13CDOL throughout the sedimentary sequences and the disequilibrium between the modern δ13CDIC and δ13CDOL imply shifts in the depth of the diagenetic zones, as well as dramatic variations of the rates of microbial activity through time. The positive δ13CDIC values in the methanogenic zone are mixed at the SMI with the negative values from the sulphate reduction zone and, after a shift in the redox zonation, a certain amount of time is required for the δ13CDIC value to come to a new equilibrium. By this mechanism, a range of different δ13CDOL values is possible at the SMI. Indeed, disequilibrium of the δ13CDIC values with modern redox zonation is observed at Site 1227 (Fig. 7A), which shows that even positive δ13CDIC values at the SMI are possible. At the shelf sites, the δ13CDIC values at the SMI do not indicate a high contribution of AMO and, thus, extremely negative δ13CDOL values are not expected. In contrast, strong AMO at the SMI in the trench Site 1230/685 is consistent with the observed extremely negative δ13CDOL.
Based on the numeric model (Fig. 10), it is proposed that these changes are dominantly controlled by the activity of microbial sulphate reduction at specific intervals. S-shaped sulphate concentration profiles, measured at different sites, indicate active consumption at certain stratigraphic horizons at shallow depth. Numerical modelling shows that deposition of a 2·5 m thick organic carbon-rich layer, in which the sulphate reduction rate may reach 50 mmol m−3 a−1, leads to a rapid removal of sulphate and a new SMI may be formed. The porewater sulphate model applied to Site 1229 demonstrates that, triggered by sulphate reducing activity in horizons with different organic matter content, the SMI may potentially migrate upwards from 30 to 5 mbsf and allow for dolomite precipitation at very shallow depths. For example, a shallow occurrence of dolomite was observed at Blake Ridge (Rodriguez et al., 2000). Consumption of SO may be horizontally restricted if the permeability is reduced, such as in clay-rich sediments. In such cases, more lense-shaped, nodular concretions would be expected, as observed in the Miocene Drakes Bay Formation, California (Burns et al., 1988), whereas in porous diatom ooze, horizontal layers may form.
Dynamic activity of the deep biosphere may be triggered by different mechanisms. At the shelf sites, glacial–interglacial sea-level changes superimposed on tectonic uplift and subsidence may have a major effect on the diagenetic system by switching on and off the upwelling cell in the basin, which is reflected in strongly varying TOC concentrations in different intervals (Wefer et al., 1990). The frequency of dolomite layers throughout the sedimentary column of Site 1229 (Fig. 3) seems to correlate roughly with TOC concentration, as well as colour reflectance values (D'Hondt et al., 2003) and the reconstructed bathymetry curve (determined by benthic foraminiferal assemblages, Resig, 1990). Thus, the SMI may have moved downwards during glacial lowstand, when microbial activity was low, and upwards during interglacial highstands, when microbial activity was high. This mechanism may explain the indirect coupling of dolomite formation with orbital cycles, as suggested by Compton (1988) based on the spacing of dolomite layers in the Monterey Formation. The formation of the layers was episodic and a massive layer can only form, when the site of precipitation is focused for a sufficient length of time.
Another mechanism was probably active at the trench site, where an extremely negative δ13CDOL value less than −30‰ was measured in a dolomite layer at 5 mbsf. This indicates intense AMO in the past, but the modern porewater δ13CDIC value only approaches −13‰. This increase in sulphate reduction and methane oxidation may have been triggered by decomposition of gas hydrates, possibly due to decompression during glacial sea-level lowstand. In an extreme case, the SMI might reach the sediment surface, causing seepage of methane into the seawater. Seep carbonates are common at both active and passive continental margins and show high variability of methane seepage through time (e.g. Thornburg & Suess, 1990). Variation in AMO was also proposed from the New Jersey continental shelf (Malone et al., 2002), based on the disequilibrium between δ13CCarbonate and δ13CDIC.