Despite the success of previous ODP legs, which set out to sample exposures of the lower oceanic crust and upper mantle, many questions regarding the compositional heterogeneity, stratigraphy, and petrogenesis of the ocean crust remain unanswered. Leg 176 will return to Site 735 on the Southwest Indian Ridge to deepen Hole 735B to at least 2 km to in order to characterize the nature of the lower oceanic crust at a slow-spreading ridge. Hole 735B was originally drilled to 500 m below the seafloor (mbsf), into gabbroic rocks exposed on a wave cut terrace atop the transverse ridge flanking the Atlantis II Fracture Zone (Fig. 1). The primary objective of this leg is to investigate the magmatic, metamorphic, and tectonic processes that attend seafloor spreading in the lower oceanic crust. Based on the phenomenal recovery and superb operational conditions encountered when Hole 735B was cored during Leg 118, we expect that deepening the hole will provide core that will directly address current models of the seismic structure of the oceanic crust. These operations will also provide information on the role offractures and fluid penetration into the lower oceanic crust and perhaps the sub-oceanic mantle, and define gradients in metamorphic facies, if they exist. From a tectonic perspective, deepening Hole 735B will establish the spacing and morphology of major ductile-shear zones recognized in the recovered core from Leg 118. Layered cumulate gabbros, as might be expected according to the ophiolite model of ocean crust stratigraphy, were not included in the nearly complete section recovered during Leg 118. Only continued drilling will establish if these lithotypes are indeed absent in this environment, discounting the hypothesis of a long lived magma chamber. Additionally, recognizing that dredging along the fracture zone has recovered mantle lithologies, and that remote sensing studies have established the presence of a seismic transition with potential correlation to the crust/mantle boundary, we have the opportunity to sample, for the first time, the transition between rocks produced by crystallization of magma extracted from the Earth’s mantle and the residues from which these liquids were derived. The overriding goal of this project is to recover sufficient gabbroic rock to ensure representation of all major lithologies to develop a consistent model of the igneous, metamorphic, and tectonic processes that occurred during the formation and subsequent exhumation of this section through the ocean crust.
In the event that drilling conditions preclude deepening Hole 735B, we propose to drill a transect of 500-m-deep holes away from Hole 735B along a lithospheric flow line. Because 800-m spacing of these holes equates to 100,000-yr age increments along the flow line, we will be able to test lateral and temporal variability of processes in the lower oceanic crust. This proposed transect of at least three holes will span an interval of 400,000 years of crustal generation. Owing to basement rotation perpendicular to the flow line, we should sample successively deeper parts of the section, which will reflect multiple magmatic and amagmatic episodes.
After an initial debate in the late 1950s and early 1960s, a consensus was achieved for a fairly straightforward layer-cake geologic model for the stratigraphy of the ocean crust. This model was based on the match of density and P- and S-wave velocities of gabbro, diabase, and tholeiitic basalt dredged from the seafloor and found in ophiolite complexes (fossil sections of ocean crust tectonically emplaced in island arcs and continental margins) to the seismic character of the ocean crust (e.g., Christensen, 1972). An overall stratigraphy was then compiled extrapolating the observed simple-layered seismic structure to the vertical lithologic stratigraphy seen in tectonically disrupted ophiolites.
Based on this information, layer 3 was expected to consist of a uniform layer of magmatic cumulates deposited on the floor and walls of a large continuous magma chamber, overlain by evolved ferrogabbros along its roof. In reality, this internal stratigraphy was more dependent on the documented stratigraphy of the large layered intrusions found on the continents than it was on the internal stratigraphy of gabbros in ophiolites. These large layered intrusions have dominated the thinking of the petrologic community with their systematic progression in chemistry from top to bottom and their historic role in establishing the key role of fractional crystallization in the evolution of magmas. It has been a natural impulse to impose their stratigraphy on that of lower crust in both ophiolites and the present-day ocean basins, thereby, evolving the extremely attractive paradigm of the "infinite onion". The "infinite onion" example consists of a large continuous magma chamber underlying the global ocean ridge system, disrupted only by the largest of ocean fracture zones, from which layers of ocean crust continuously grow at top, sides, and bottom to form a uniform coarse gabbroic layer comprising two thirds of the ocean crust (Cann, 1970).
Two decades of geophysical testing of this model and study of the stratigraphic structure and chemistry of ophiolites have thrown this simple paradigm into question. Interpretations of seismic structure along ridges have become increasingly complex, while the provenance of ophiolites has become increasingly ambiguous and generally believed to be atypical of ocean crust. Recent work suggests that the ocean crust has a complex, three-dimensional structure that is highly dependent on magma supply and spreading rates without large steady-state magma chambers (e.g., Whitehead et al., 1984; Detrick et al., 1990; Sinton and Detrick, 1992; Barth, 1994; Carbotte and MacDonald, 1994). Compilations of dredge results and seismic data have suggested that a continuous gabbroic layer does not exist at slow-spreading ridges (Mutter et al., 1985; NAT Study Group, 1985; McCarthy et al., 1988; Dick, 1989; Cannat et al., 1992; Tucholke, unpubl. data), and that its internal stratigraphy is governed by dynamic processes of alteration and tectonism as much as by igneous processes. The exceptional abundance of serpentinized peridotite in dredge hauls from the walls of rift valleys in fracture zones and in the rift mountains away from fracture zones (Aumento et al., 1971; Rona et al., 1987; Dick, 1989; Cannat et al., 1992; Tucholke, unpubl. data) is also raising the serious possibility that serpentinite is a major component of seismic layer 3, as originally suggested by Hess (1962).
This re-evaluation of the geometric relationships between lower oceanic crust and the upper mantle has led to the development of three general models for the lithologic architecture of slow-spreading ocean crust (Fig. 2). One model (Fig. 2A) proposes that relatively small gabbroic bodies discordantly cut each other and the upper mantle, with the proportion of gabbroic bodies to mantle lithologies diminishing with depth (see Swift and Stephen, 1992; Cannat, 1993; Sleep and Barth, 1994). This model suggests that the primitive troctolites recovered from the bottom of Hole 735B represent deep level gabbroic rocks, and continued drilling should rapidly intersect upper mantle lithologies. A second interpretation (Fig. 2B) is that the lower crust is made up of an assembly of small discordant gabbroic bodies that transit abruptly into the upper mantle following the concept proposed by Cann (1970) and Nisbet and Fowler (1978) and recently amplified by Smith and Cann (1993). Finally, a third model based on ophiolite studies (Fig. 2C) suggests that the discordant gabbroic bodies occur only in the upper crust, and then grade downward into large layered gabbro intrusions, which overlie and have a sharp magmatic contact with upper mantle peridotite (Pallister and Hopson, 1981; Smewing, 1981). While recent wide-angle seismic profiles over and around the Atlantis II Bank image a distinct Moho with a 4-5 km thick overlying layer (Müller et al., 1995), these data do not unequivocally rule out any of the three models. This seismic boundary may represent a magmatic transition from gabbroic rocks to peridotite; alternatively we could be seeing a transition from hydrous serpentinized peridotite and fresher peridotite. The gabbroic rocks recovered from the 500-m penetration of Hole 735B are invaluable in detailing the compositional heterogeneity of the deep ocean crust; however, they provide insufficient information to determine which model is correct. Consequently, the primary objective of deepening Hole 735B is to penetrate deeply enough into the ocean crust to evaluate which model is most applicable to slow-spreading oceanic crust. An obvious potential highlight of this drilling program is that we may meet one of the longstanding ambitions of several ODP thematic panels and a principal objective of the Mantle/Crust Interactions Working Group at COSOD II-sampling the transition from the oceanic crust to the upper mantle.
Previous Investigations at Site 735
During Leg 118, a large intact 500-m section of gabbros was recovered from Site 735. These gabbros were unroofed and uplifted on the transverse ridge flanking the Atlantis II Fracture Zone. The complex internal structure and stratigraphy of the recovered section provided a first look at the processes of crustal accretion and on-going tectonism, alteration, and ephemeral magmatism at a slow-spreading ocean ridge. The section was not formed in a large steady-state magma chamber, but by continuous intrusion and reintrusion of numerous small, rapidly crystallized bodies of magma. There is little evidence of the process of magmatic sedimentation that is so important in layered intrusions. Instead, new batches of magma are intruded into and initially supercooled by a lower ocean crust that consists of wholly crystalline rock and semisolidified crystal mush. This leads to undercooling and rapid initial crystallization of new magmas to form a highly viscous or rigid crystal mush that prevents the formation of magmatic sediments. Initial crystallization is followed by a longer, and petrologically more important, period of intercumulus melt evolution in a highly viscous crystal mush or rigid melt crystal aggregate.
As a consequence, long-lived magma chambers or melt lenses were virtually absent throughout most of crustal formation beneath the Southwest Indian Ridge. Thus, melts in the highly viscous or rigid intrusions were largely uneruptable throughout most of their crystallization. This explains the near absence of highly evolved magmas such as ferrobasalts along the Southwest Indian Ridge (Dick, 1989), as opposed to fast-spreading ridges where they are common and a long-lived melt lens is believed to underlie the ridge axis (e.g., Sinton and Detrick, 1992).
Wall-rock assimilation occurring while small batches of melt work their way up through the partially solidified lower crust appears to have played a major role in the chemical evolution of the section and, therefore, in the chemistry of the erupted basalt. This process has been largely unevaluated for basalt petrogenesis to date, which has thrown into question the simple models for the formation of mid-ocean-ridge basalt (MORB) drawn from experimental studies. These studies assume equilibrium crystallization and melting processes throughout magma genesis.
An unanticipated major feature of the drilled section is the evidence for deformation and ductile faulting of the still partially molten gabbros. This deformation apparently occurred over a narrow window, late in the cooling history of the gabbros (probably at 70-90% crystallization) when they became sufficiently rigid to support a shear stress. This produced numerous small and large shear zones, creating zones of enhanced permeability into which the late intercumulus melt moved. This synkinematic igneous differentiation of intercumulus melts into the shear zones transformed the gabbro there into oxide-rich ferrogabbros. The net effect of these magmatic and tectonic processes was to produce a complex igneous stratigraphy with undeformed oxide-free olivine gabbros and microgabbros criss-crossed by bands of sheared ferrogabbro. Synkinematic differentiation is probably ubiquitous in lower ocean crust formed at slow-spreading ocean ridges, and should be recorded in ophiolite suites formed in similar tectonic regimes.
At Site 735, ductile deformation and shearing continued into the subsolidus regime, causing recrystallization of the primary igneous assemblage under granulite facies conditions, and the formation of amphibole-rich shear zones (Stakes et al., 1991; Dick et al., 1991a; Cannat et al., 1991). Here again, formation of ductile shear zones localized late fluid flow with the most intense alteration occurring in the ductile faults (Dick et al., 1991a). Undeformed sections of gabbro also underwent enhanced alteration at this time principally by replacement of pyroxene and olivine by amphibole.
A consequence of simultaneous extension and alteration has been far more extensive alteration at high temperatures than found in layered intrusions that were intruded and cooled in a static environment. An abrupt change in alteration conditions of the Hole 735B gabbros, however, occurred in the middle amphibolite facies with the cessation of shearing and ductile deformation. Mineral vein assemblages changed from amphibole-rich to diopside-rich, reflecting different fluid chemistry. Continued alteration and cooling to low temperature occurred under static conditions similar to those found for large layered intrusions. These changes likely occurred due to an inward jump of the master faults defining the rift valley walls, thus transferring the section out of the zone of extension and lithospheric necking beneath the rift valley into a zone of simple block uplift in the adjoining rift mountains. Ongoing hydrothermal circulation, no longer enhanced by stresses related to extension, was greatly reduced, driven only by thermal-dilation cracking as the section cooled to ambient temperature.
The complex section of rock drilled at Site 735 formed beneath the very slow-spreading Southwest Indian Ridge (0.8 cm/yr half rate) and represents the slow end of the spectrum for crust formation at major ocean ridges far from hot spots. Such ridges have the lowest rates of ocean ridge magma supply, and crustal accretion is most heavily influenced by deformation and alteration. At the opposite end of the spreading rate spectrum (7-9 cm/yr), where the majority of the seafloor has formed, the crustal stratigraphy is likely different. Judging from the results of Hole 735B, the critical brittle-ductile transition has migrated up and down through the lower crust due to the waxing and waning of magmatism beneath the Southwest Indian Ridge. In contrast, this transition may be more stable near the sheeted dike gabbro transition at faster spreading ridges such as the East Pacific Rise, reflecting a near steady-state magma chamber or crystal mush zone. This should produce an internal stratigraphy for the lower crust quite different than that described here.
The rather general conclusions drawn to date from Hole 735B, so different from what was anticipated, are based on study of only a small part of what is likely to be a 2- to 4-km-thick section, and thus may represent only part of a more complex overall stratigraphy. By deepening Hole 735B and eventually drilling an offset section of holes along a lithospheric flow line, we will obtain a representative section of the lower ocean crust at one of the two critical ends of the spreading spectrum that, together with seafloor mapping, will permit a true three-dimensional view of the ocean crust. The suite of holes will also provide a natural laboratory for downhole geophysical experiments, where hole-to-hole magnetotelluric and permeability experiments can be conducted, and that will provide a downhole seismic laboratory from which the nature of layer 3 may be directly tested.
Tectonic Setting
Site 735 is located in the rift mountains of the Southwest Indian Ridge, 18 km east of the present-day axis of the Atlantis II Transform Fault (Fig. 1). The Southwest Indian Ridge has existed since the initial breakup of Gondwanaland in the Mesozoic (e.g., Norton and Sclater, 1979). Shortly before 80 Ma, plate readjustment in the Indian Ocean connected the newly formed Central Indian Ridge to the Southwest Indian Ridge and the Southeast Indian Ridge to form the Indian Ocean Triple Junction (Fisher and Sclater, 1983). Steady migration of the triple junction to the northeast has created a succession of new ridge segments and fracture zones including the Atlantis II. Thus, the Atlantis II Fracture Zone and the adjacent ocean crust is entirely oceanic in origin, free from complications due to continental breakup as postulated for some equatorial fracture zones along the Mid-Atlantic Ridge (e.g., Bonatti and Honnorez, 1976).
Over the last 34 m.y., the spreading rate along the Southwest Indian Ridge has been relatively constant, near 0.8 cm/yr, at the very slow end of the spreading-rate spectrum (Fisher and Sclater, 1983). All the characteristic features of slow-spreading ridges, including rough topography, deep rift valleys, and abundant exposures of plutonic and mantle rocks, are present on the Southwest Indian Ridge (Dick, 1989). Significantly, two thirds of the rocks dredged from the walls of the active transform valleys are altered mantle peridotites, whereas most of the remainder are weathered pillow basalts. This exceptional abundance of peridotite, compared to dredge collections of similar size from the North Atlantic, indicates an unusually thin crustal section in the vicinity of Southwest Indian Ridge transforms. Moreover, the paucity of dredged gabbro along the Southwest Indian Ridge suggests that magma chambers were small or absent near fracture zones.
The thin crust adjacent to fracture zones is thought to reflect segmented magmatism along the Southwest Indian Ridge that produces rapid along-strike changes in the structure and stratigraphy of the lower ocean crust (e.g., Whitehead et al., 1984; Francheteau and Ballard, 1983; Crane, 1985; Schouten et al., 1985; MacDonald et al., 1986). This model views the Southwest Indian Ridge as a series of regularly spaced, long-lived shield volcanoes and underlying magmatic centers, which undergo continuous extension to form the ocean crust (Dick, 1989). Site 735 is situated some 18 km from the Atlantis II Transform Fault, and was accordingly situated near the mid-point of a hypothetical magmatic center beneath the Southwest Indian Ridge at 11.5 Ma (Dick et al., 1991b).
Geology of the Atlantis II Fracture Zone
Hole 735B is located on a shallow bank, informally named Atlantis Bank, on the crest of a 5 km-high mountain range constituting the eastern wall of the Atlantis II Transform valley. This transverse ridge is similar to many other flanking fracture zones on the Southwest Indian Ridge (e.g., Engel and Fisher, 1975; Sclater et al., 1978, Fisher et al., 1986; Dick, 1989) where abundant plutonic rocks, particularly peridotite, are uplifted to a shallow level and exposed. The bank consists of a platform, roughly 9 km long in a north-south direction and 4 km wide, which is the shallowest of a series of uplifted blocks and connecting saddles that form a long, linear ridge parallel to the transform. The top of the platform is flat, with only about 100 m relief over 20 km2. A video survey of a 200 x 200 m area in the vicinity of the hole showed a smooth flat wave-cut platform exposing foliated and massive jointed gabbro locally covered by sediment drift. The platform probably formed by erosion of an island similar to St. Paul's Rocks in the central Atlantic, and then subsided to its present depth from normal lithospheric cooling (Dick et al., 1991b). A similar wave-cut platform occurs on the ridge flanking the DuToit Fracture Zone (Fisher et al., 1986).
The foliation seen in the video survey and at the top of the drill core appears to strike east-west, parallel to the ridge axis and orthogonal to the fracture zone. The orientation of similarly foliated peridotites exposed on St. Paul's Rocks has been measured and is also parallel to the Mid-Atlantic Ridge and orthogonal to St. Paul's Fracture Zone (Melson and Thompson, 1971). This foliation, projected along strike across the Atlantic Bank platform, intersects a long ridge coming up the wall of the fracture zone, which is oriented obliquely west-northwest to the transform. Ridges produced by land-slips and debris flows normally are oriented orthogonal to the fracture zone. The suspicion then is that this oblique ridge and a similar one 2 km to the north represent the trace of the thick zone of foliated gabbros down the wall of the transform. Given the once shallow water depth, the canyon between the two ridges may be erosional and the foliated gneissic amphibolites resistant remnants. A three-point solution for the dip of the shear zone, based on the trend of this ridge, and an east-west strike gives a dip of approximately 40°, which is close to that observed in the drilled amphibolites.
This shear zone represents a ductile-fault, and thus does not represent a simple stratigraphic discontinuity. The rocks at the top of the shear zone are gabbronorites that pass gradually into a zone of olivine gabbro toward its base. The shear sense determined from drill cores is normal. Thus, it would appear that the rocks to the north of the drill site are down-thrown an unspecified amount. Any offset drill sites to the north would start higher in the stratigraphic section.
The site is located between magnetic anomalies 5 and 5a, approximately 93 km south of the present-day axis of the Southwest Indian Ridge and 18.4 km from the inferred axis of transform faulting on the floor of the Atlantis II Fracture Zone (Dick et al., 1991b). Given the position of the site, the relatively constant spreading direction over the last 11 m.y., and the ridge-parallel strike of the local foliation, the Atlantis Bank gabbros must have crystallized and deformed beneath the median valley of the Southwest Indian Ridge, 15 to 19 km from the ridge-transform intersection, around 11.5 Ma. A single Pb-zircon of 11.3 Ma, as reported by Stakes et al. (1991) for a trondhjemite in amphibolite near the top of the hole, confirms the age determined by plate reconstruction.
The gabbros were subsequently uplifted in a large horst from beneath the rift valley 5 to 6 km up into the transverse ridge (Dick et al., 1991b). The single uniform magnetic inclination throughout the section demonstrates that there has been no late tectonic disruption of the section, although the relatively steep inclination suggests block rotation of up to 18° (Pariso et al., 1991). Thus, unlike some rocks dredged from fracture-zone walls, those drilled in Hole 735B formed beneath the rift-valley floor away from the transform fault. Petrologically, these rocks likely represent a typical igneous section of Southwest Indian Ridge ocean crust with an intact metamorphic and tectonic stratigraphy. The rocks, thereby, record brittle-ductile deformation and alteration at high temperatures beneath the rift valley, as well as subsequent unroofing and emplacement on ridge-parallel faults.
The unroofing and exposure of the Hole 735B section relates to the present-day asymmetric distribution of plutonic and volcanic rocks north and south of the ridge axis near the fracture zone, as well as to the striking physiographic contrast between crust spreading in opposite directions at the ridge-transform intersection (Fig. 3) (Dick et al., 1991b). These features suggest that a crustal weld periodically formed between the shallow levels of the ocean crust and the old, cold lithospheric plate at the ridge-transform intersection. This weld caused the shallow levels of the newly formed ocean crust to spread with the older plate away from the active transform, thereby causing the creation of long-lived detachment faults. Beneath the faults, the deep-ocean crust that was spreading parallel to the transform was unroofed and emplaced up into the rift mountains to form a transverse ridge. A similar model was proposed by Dick et al. (1981) to explain the asymmetric physiography and distribution of plutonic and volcanic rocks at the Kane Fracture Zone in the North Atlantic. At the Kane Fracture Zone, the surface of the detachment fault has actually been observed by submersible (Dick et al., 1981; Mevel et al., 1991). Detachment faults similar to the one proposed to explain unroofing of the lower crust at the Atlantic Fracture Zone have been suggested to occur periodically within rift valleys by fault capture during amagmatic periods (Harper, 1985; Karson, 1991). Thus, the structures and fabrics seen in core from Hole 735B are likely to be representative of the kinds of fabrics generally found in lower crustal sections formed at slow-spreading ridges. It is true, however, that due to the proximity to the transform the extent of the ductile shear may be greater than elsewhere beneath the rift valley.
The principal objective of Leg 176 will be to deepen Hole 735B to a depth of at least 2 km (Fig. 4) to determine the nature of the magmatic, metamorphic, and tectonic processes in the lower oceanic crust. Leg 176 will:
Supplementary objectives for the leg if deepening Hole 735B is not achieved:
It is reasonable to assume that drilling conditions will remain roughly constant from the hole's present depth until thermal problems occur, or when serpentinized peridotite is encountered. While the depth at which the latter will occur is difficult to predict, it is reasonable to expect thermal problems at about 2000 m sub-basement, based on experience at Hole 504B, where the present temperature is now close to 190°C.
Based on the hypothesis that brittle fracture and brecciation decrease with depth in plutonic layer 3 due to the steep geotherm beneath an ocean ridge, and that fine-grained rocks are unlikely to be encountered lower in the section, it is reasonable to suspect that the overall penetration rate would remain close to that for Leg 118. Five hundred meters were drilled at Hole 735B in 19 days, including setting a guide base and starting with the mud motor rather than the top drive, and using minimum bit weight and extreme operational conservatism. Improved bit design subsequent to Leg 118 and the observed lack of bit wear during Leg 118, suggests that operations during Leg 176 should extend bit life, which should make up for increased trip time as the hole is deepened.
The priorities for 735B drilling defined by PCOM are as follows:
2) Log the deepened hole
3) Conduct both Packer and VSP experiments in the deepened hole
4) In the event difficulties are encountered while drilling, the following priorities should be maintained:
5) Efforts should focus on the wave-cut terrace on which Hole 735B is located.
Establishing the depth/seismic tie by means of a VSP and synthetic seismograms is essential to identifying deep crustal reflectors. During the first phase of drilling Hole 735B during Leg 118, geochemical data from the geochemical logging tool (GLT), compressional- and shear-wave velocity and amplitude logs, the borehole televiewer (BHTV), and magnetic susceptibility logs were especially useful in delineating structural and stratigraphic features of magmatic layering and fractures. In anisotropic formations such as those at Site 735, the shear-wave velocity and amplitude measured at different azimuths in the borehole may indicate fracture orientation and the regional paleostress direction. The Formation MicroScanner (FMS) high-resolution images will also vastly improve the determination of the fracture and alteration zone distribution in the crust and should be given high priority. FMS logs will also be useful in re-orientation of the structural markers in the cone. Magnetic susceptibility was useful in identifying metallic oxides that are quite abundant in the upper 500 m of the hole. Log resistivities were as high as 40,000 ohm-m in the upper 500 m of the hole, and the dual laterolog (DLL) is recommended in such high-resistivity environments. In summary, the Quad, GLT (providing low core recovery necessitates it), dipole sonic, DLL, FMS, and BHTV tool strings, should be run. In addition, a vertical-incidence VSP and a high-temperature magnetometer are recommended.
To Leg 176 Proposed Site Information