Geochemistry of mafic clasts of ODP Holes 125-778A and 125-779A

Clasts of metamorphosed mafic igneous rock of diverse composition were recovered in two drill sites on a serpentine mud volcano in the outer Mariana forearc during Ocean Drilling Program Leg 125. These clasts are xenolithic fragments that have been entrained in the rising serpentine mud, and make up less that 9% of the total rock recovered at Sites 778 and 779. Most samples are metabasalt or metadiabase, although one clast of possible boninite and one cumulate gabbro were recovered. On the basis of trace element signatures, samples are interpreted to represent both arc-derived and mid-ocean ridge-derived compositions. Rocks with extremely low TiO2 (<0.3 wt%) and Zr (<30 ppm) are similar to boninite series rocks. Samples with low TiO2 (<0.9 wt%) and Zr (<50 ppm) and extreme potassium enrichment (K2O/Na2O >3.9) may represent island arc rocks similar to shoshonites. However, the K2O/Na2O ratios are much higher than those reported for shoshonites from modem or ancient arcs and may be the result of metamorphism. Samples with moderate TiO2 (1.4 to 1.5 wt%) and Zr (72 to 85 ppm) are similar to rocks from mid-ocean ridges. A few samples have TiO2 and Zr intermediate between island arc and mid-ocean ridge basalt-like rocks. Two samples have high iron (Fe2O3* = >12.8 to 18.5 wt%) (Fe2O3* = total iron calculated as Fe2O3) and TiO2 (>2.3 wt%) and resemble FeTi basalt recovered from mid-ocean ridges. Metamorphism in most samples ranges from low-temperature zeolite, typical of ocean floor weathering, to prehnite-pumpellyite facies and perhaps lower greenschist. Blue amphibole and lawsonite minerals are present in several samples. One diabase clast (Sample 9) exhibits Ca enrichment, similar to rodingite metamorphism, typical of mafic blocks in serpentinized masses. The presence of both low-grade (clays and zeolites) and higher grade (lawsonite) metamorphism indicates retrograde processes in these clasts. These clasts are fragments of the forearc crust and possibly of the subducting plate that have been entrained in the rising serpentine and may represent the deepest mafic rocks ever recovered from the Mariana forearc. The variable compositions and degree of metamorphism of these clasts requires at least two tectonic origins. The recovery of clasts with mid-ocean ridge and arc chemical affinities in a single drill hole requires these clasts to have been "mixed" on a small scale either (1) in the forearc crustal sequence, or (2) after inclusion in the rising serpentine mud. The source of the MORB-like samples and an explanation for the presence of both MORB-like and arc-like rocks in close proximity is critical to any model of the evolution of the Mariana forearc. The source of the MORB-like samples likely will be one (or more) of the following: (1) accretion of Pacific plate lithosphere, (2) remnants of original forearc crust (trapped plate), (3) volcanism in the supra-subduction zone (arc or forearc) environment, or (4) derivation from the subducting slab by faulting along the dÈcollement.

Mineralogy of serpentine muds of ODP Leg 125 holes (Table 1)

Large serpentinite seamounts are common in the forearc regions between the trench axis and the active volcanic fronts of the Mariana and Izu-Bonin intraoceanic arcs. The seamounts apparently form both as mud volcanoes, composed of unconsolidated serpentine mud flows that have entrained metamorphosed ultramafic and mafic rocks, and as horst blocks, possibly diapirically emplaced, of serpentinized ultramafics partially draped with unconsolidated serpentine slump deposits and mud flows. The clayand silt-sized serpentine recovered from three sites on Conical Seamount on the Mariana forearc region and from two sites on Torishima Forearc Seamount on the Izu-Bonin forearc region is composed predominantly of chrysotile, brucite, chlorite, and clays. A variety of accessory minerals attest to the presence of unusual pore fluids in some of the samples. Aragonite, unstable at the depths at which the serpentine deposits were drilled, is present in many of the surficial cores from Conical Seamount. Sjogrenite minerals, commonly found as weathering products of serpentine resulting from interaction with groundwater, are found in most of the samples. The presence of aragonite and carbonate-hydroxide hydrate minerals argues for interaction of the serpentine deposits with fluids other than seawater. There are numerous examples of sedimentary serpentinite deposits exposed on land that are very similar to the deposits recovered from the serpentine seamounts drilled on ODP Leg 125. We suggest that Conical Seamount may be a type locality for the study of in situ formation of many of these sedimentary serpentinite bodies. Further, we suggest that both the deposits drilled on Conical Seamount and on Torishima Forearc Seamount demonstrate that serpentinization can continue in situ within the seamounts through interaction of the serpentine deposits with both seawater and subduction-related fluids.

Ultimate and failure strengths of normal muds and serpentinite muds of ODP Leg 125 samples

Serpentinite seamounts in the Mariana forearc have been explained as diapirs rising from the Benioff zone. This hypothesis predicts that the serpentinites should have low strengths as well as low densities relative to the surrounding rocks. Drilling during Leg 125 showed that the materials forming Conical Seamount in the Mariana forearc and Torishima Forearc Seamount in the Izu-Bonin forearc are water-charged serpentinite muds of density <2 g/cm**3. Wykeham-Farrance torsion-vane tests showed that they are plastic solids with a rheology that bears many similarities to the idealized Cam clay soil model and is well described by critical-state soil mechanics. The serpentinite muds have ultimate strengths of 1.3 to 273.7 kPa and yield strengths of approximately 1.0 to 50 kPa. These muds thus are orders of magnitude weaker than salt and are, in fact, comparable in density and strength to common deep-sea clay muds. Such weak and low-density materials easily become diapiric. Serpentinite muds from the summit of Conical Seamount are weaker and more ductile than those on its flanks or on Torishima Forearc Seamount. Moreover, the summit muds do not contain the pronounced pure- and simple-shear fabrics that characterize those on the seamount flanks. The seamounts are morphologically similar to shield volcanoes, and anastomosing serpentinite debris flows descending from their summits are similar in map view to pahoehoe flows. These morphologic features, together with the physical properties of the muds and their similarities to other oceanic muds and the geochemistry of the entrained waters, suggest that many forearc serpentinite seamounts are gigantic (10-20 km wide, 1.5-2.0 km high) mud volcanoes that formed by the eruption of highly liquid serpentinite muds. Torishima Forearc Seamount, which is blanketed by more ìnormalî pelagic/volcaniclastic sediment, has probably been inactive since the Miocene. Conical Seamount, which seems to consist entirely of serpentinite mud and is venting fresh water of unusual chemistry from its summit, is presently active. Muds from the flanks of Conical Seamount are stronger and more brittle than those from the summit site, and muds from Torishima Forearc Seamount are stronger yet; this suggests that the serpentinite debris flows are compacted and dewatered as they mature. The shear fabrics probably result from downslope creep and flow, but may also be inherited.

(Table 1) Osmium concentrations and isotopic ratios for ODP Leg 125 peridotites

Mantle peridotites drilled from the Izu-Bonin-Mariana forearc have unradiogenic 187Os/188Os ratios (0.1193 to 0.1273), which give Proterozoic model ages of 820 to 1230 million years ago. If these peridotites are residues from magmatism during the initiation of subduction 40 to 48 million years ago, then the mantle that melted was much more depleted in incompatible elements than the source of mid-ocean ridge basalts (MORB). This result indicates that osmium isotopes record information about ancient melting events in the convecting upper mantle not recorded by incompatible lithophile isotope tracers. Subduction zones may be a graveyard for ancient depleted mantle material, and portions of the convecting upper mantle may be less radiogenic in osmium isotopes than previously recognized.