DISCUSSION

During Leg 201, two different types of dolomite with different morphologies and depths of precipitation were found in the Peruvian shelf sediments and in the trench. All recovered dolomites were formed in lower neritic to abyssal sediments and not shallow-marine evaporative conditions, as shown by paleobathymetry determined from benthic foraminifers (Resig, 1990). Older dolomites were recovered by dredging (Kulm et al., 1984) or deeper drilling during Leg 112 (Thornburg and Suess, 1990), and some of them were shown to be formed in evaporitic basins in the Miocene, when the brine also formed.

At all sites drilled on the Peru margin during Leg 201, dolomite layers were recovered in sediments formed under upwelling conditions. They are 2–5 cm thick, laminated, and densely lithified. Friable dolomite is rare, but it does occur around hard layers and rarely as single laminae. It is never homogeneously distributed in the sediment column. The fine-grained and homogeneous dolomicrosparite consists of micron- to decimicron-size euhedral rhombs. Only small amounts of possible precursor carbonate are present in the sediment, and dolomite forms as a primary precipitate. Foraminiferal tests and diatom frustules are well preserved and not replaced but cemented by the growing diagenetic dolomite. Foraminifers show growth of fibrous calcite cement at different sites. This cement formed before dolomite growth and also shows evidence of dolomitic replacement. Often, dolomite rhombs appear as replicas of the porous surface of the foraminiferal tests and the diatom frustule. However, the pores of a strongly dissolved siliceous frustule at Site 1228 are not completely filled with dolomite. The formation of framboidal pyrite postdates dolomite formation. This indicates that the dolomitization process was terminated early. The youngest dolomite layers often appear between 10 and 30 mbsf; most are probably formed in this depth range, where the uncompacted diatomaceous sediments still have a high porosity. This observation is in agreement with observations from other environments, such as the Miocene dolomite-cemented diatomite layer in the Southern Alpine Gonfolite Lombarda Group (Bernoulli and Gunzenhauser, 2001). Dolomite layers show similar size and shape within each site, but they are different between the sites. Cementation style is also characteristic for each site. This indicates that the discrete dolomite layers probably formed under the same conditions at each site. Results from XRD and EMPA show that all Peru margin dolomites have a well-ordered stoichiometry with a slight enrichment in calcium. Dolomite layers are not always correlated with diatomaceous horizons, thus not directly depend on lithology. However, on a broad scale, dolomite only occurs in sediments associated with the upwelling zone. Because they occur as distinct layers, dolostones in the Peru margin probably formed along a sharp geochemical interface relatively early in the sedimentary column. An early precipitation (i.e., in the uppermost 30 mbsf) is also suggested by Sr isotopic composition similar to seawater values and the disequilibrium of ¹³C of the dolomite with ¹³C in dissolved inorganic carbon of modern pore water (M.L. Musgrove and D. Schrag, pers. comm., 2004; and P. Meister, unpubl. data). A possible interface would be the boundary between the sulfate-reducing and methanogenic diagenetic zones. At this interface, sulfate ions, proposed as a kinetic inhibitor (Baker and Kastner, 1981), are removed and alkalinity reaches peak values (D'Hondt, Jørgensen, Miller, et al., 2003), whereas Mg and Ca concentrations simultaneously decrease at these depths (Suess, von Huene, et al., 1988). Also, these interfaces (in particular at Site 1229) seem to be hot spots of microbial activity (D'Hondt, Jørgensen, Miller, et al., 2003) and, therefore, may play an important role in dolomite formation. Carbon-13 values of the dolomite layers are strongly variable, but fall into the range of values measured in the dissolved inorganic carbon at the sulfate/methane interface.

Dolomite layers found below 200 mbsf in the Peru Trench are strongly brecciated and consist of centimeter-scale clasts with planar subhedral dolomite grains overgrowing the original sediment. Remnants of marine calcitic organisms are almost completely replaced by the coarse anhedral dolomite. Xenotopic textures are nicely developed in the matrix as a result from simultaneous growth of neighboring crystals along compromise boundaries. Simultaneous growth is only possible if the dolomitizing fluid is strongly supersaturated with respect to dolomite. Cathode luminescence revealed several generations of dolomitic cement formed in the open spaces between the clasts contemporaneously with the brittle deformation. Open pore spaces are partially filled with soft sediment and partially filled with coarse sparitic dolomite and a central cavity. The last phase of diagenesis ended with the precipitation of fine-grained pyrite.

Dolomite mineralization is strongly related to brecciated layers, which were observed also in several deeper layers during Leg 112. The brittle deformation has been related to shear zones within the accretionary prism, which only occur at depths deeper than 200 mbsf. Therefore, the synkinematic precipitation of dolomite occurred not early in the sedimentary column but more deeply buried, in compacted sediment. The stiffness of the compacted sediment, and perhaps also the presence of gas hydrates, are responsible for the brittle behavior and preferential deformation along discrete fault zones, where hard lithified dolomite was brecciated. Most likely these shear zones provide fluids from deeper parts of the accretionary prism that enhance precipitation of dolomite. Last-stage precipitation of high amounts of pyrite supports the presence of a sulfidic species in the pore water. So far, neither the microbial model (mediation through enhanced microbial activity along the fluid conduits) nor any abiogenic model (e.g., rapid fluid decompression) can be excluded for the formation of the dolomite breccias.

The fact that dolomite occurs as hard lithified layers and not continuously as disseminated rhombs, as in the Monterey Formation, which experienced higher temperature diagenesis (Isaacs, 1984), indicates that dolomite precipitation is not a continuous process but probably forms at distinct geochemical interfaces. This also explains why dolomite layers form in different horizons with different sediment composition and porosity but on a large scale related to organic carbon–rich upwelling sediment. An analog example could be barite layers that formed at the boundary between the methanogenic and the sulfate-reduction zone (Site 1227) (D'Hondt, Jørgensen, Miller, et al., 2003) and show morphologies very similar to the dolomite layers. For dolomite formation, Mg and Ca concentrations in the pore water (Suess, von Huene, et al., 1988) are not the controlling factors, as these ions become depleted at the depth of dolomite formation, but the consumption of sulfate and an increase of alkalinity are important controls. Dolomite precipitation at the methane/sulfate interface, where sulfate is completely consumed by sulfate-reducing bacteria, would be consistent with the sulfate inhibition model. Crystal morphologies typical for microbially mediated dolomite, as observed in culture experiments and in the natural environment (Lagoa Vermelha), were not found in the deep-sea sediments. This study, however, provides evidence that dolomite formation occurs at depths, where microbial cell concentration and activity (Shipboard Scientific Party, 2003) are high, suggesting the relevance of the microbial dolomite model in this deep-sea environment. Microbial activity is enhanced at biogeochemical interfaces where high gradients in ion concentration can be used as an energy source. At the same time, these interfaces are maintained by the bacteria and represent zones with ion concentrations suitable for dolomite precipitation.