Journal of Volcanology and Geothermal Research 110 ?2001) 191±233 www.elsevier.com/locate/jvolgeores Formation of the lower ocean crust and the crystallization of gabbroic cumulates at a very slowly spreading ridge James H. Natlanda,*, Henry J.B. Dickb aRosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, FL 33149, USA bWoods hole Oceanographic Institution, Woods hole, MA 02543, USA Revised 25 June 2000 Abstract Ocean Drilling Program Hole 735B was extended to 1508 m below the sea ¯oor during Leg 176, atop a shallow banknear Atlantis II Fracture Zone on the very slowly spreading Southwest Indian Ridge. All the drilling was in gabbro, and recovery averaged nearly 87%. The drill penetrated a series of stacked plutons consisting mostly of olivine gabbro, but some with troctolite. Each pluton is some 200±500 m thick, each has its own internally coherent stratigraphy, and each apparently represents an individual event of signi®cant magma in¯ation and addition to the crust. The entire column was extensively deformed along inclined zones of distributed shear before it was completely frozen, this marking the onset of unroo®ng of the rocks and their ascent to high rift mountains. The deformation mobilized late-stage melts into ¯ow patterns which led to concentration of ilmenite and magnetite in hundreds of seams of oxide gabbro along or near zones of strong crystal±plastic deformation, the highly differentiated melts overall being concentrated by buoyancy forces toward the top of the section, especially in one zone nearly 70 m thick. However, upward ¯ow was ultimately blocked or de¯ected by zones of impermeable rockresulting either from downward freezing or grain-size reduction during shear. A melt lens probably did not form at the base of sheeted dikes, as it does at the East Paci®c Rise. Despite this, the rocks are cumulates, and most are adcumulates, with very low residual melt porosities. Cumulate theory based on stratiform, layered intrusions does not entirely apply to these rocks. Instead, all gabbros, including the oxide gabbros, crystallized in a dense crystal mush in patterns dominated by fractures, channelized ¯ow, and intergranular porous ¯ow. Most gabbros are not layered; weakmodal layering of uncertain origin is present in ,2% of the rocks. Porosity reduction leading to formation of adcumulates at all stages was extremely ef®cient. This occurred in the course of synkinematic differentiation, also called differentiation by deformation, and involved compaction under conditions of lithostatic loading and shear, emplacement of crystal mushes along inclined, possibly curving faults, dissolution and reprecipitation of minerals along grain boundaries, and pressure solution, which was the ®nal agent in porosity reduction. As the blockwas lifted from beneath the rift-valley ¯oor, crystal±plastic deformation gave way to brittle fracture, and the now dominantly subsolidus metamorphism shifted from high-temperature assemblages characterized by amphibole near the top of the section, to low-temperature assemblages with smectite±chlorite and zeolite near the bottom. The rocks acquired their stable magnetization during formation of amphiboles and secondary magnetite during this metamorphism. The entire body of rockis reversely stably magnetized at a consistent inclination, and it is rotated to the south, perhaps along a curving detachment surface, away from the ridge segment where it formed, by about 208. It has an intensity of magnetization suf®cient to account for the magnetic anomaly observed over the site. Drilling did not reach ultrama®c rock, but dredging indicates that peridotite is probably within a few hundred meters of the * Corresponding author. E-mail address: [email protected] ?J.H. Natland). 0377-0273/01/$ - see front matter q 2001 Elsevier Science B.V. All rights reserved. PII: S0377-0273?01)00211-6 192 J.H. Natland, H.J.B. Dick / Journal of Volcanology and Geothermal Research 110 52001) 191±233 bottom of the hole. Seismic Moho, however, is placed at 5 km beneath the summit of Atlantis Bank. Much of the rock between Moho and the bottom of Hole 735B must therefore be partially serpentinized peridotite. q 2001 Elsevier Science B.V. All rights reserved. Keywords: ocean crust; gabbros; petrology; cumulates; Southwest Indian Ridge 1. Introduction discovered. Consider the following sequence. The simplest intrusion is that of a single injection of A long-standing objective of the Ocean Drilling magma into some portion of the crust. Small examples Program has been to drill entirely through the ocean are thin dikes or sills, which crystallize from crust into the upper mantle, to see how the ocean crust quenched surfaces inward, without much internal is formed. Alternatively, the lower ocean crust can be magma movement ?e.g. Marsh, 1989). A larger sampled where faulting or erosion has removed body, but which is still nevertheless considered to basalts and dikes, which are dif®cult to drill. The be the result of one injection of magma, is the high transverse ridges of major fracture zones in the Skaergaard intrusion, in which magma convected Atlantic and Indian oceans have long been known as and crystallization processes were diverse ?e.g. places where gabbroic and serpentinized ultrama®c McBirney, 1996; Irvine et al., 1998). The next most rockare consistently exposed to the dredge, and complicated might be that of a steadily supplied, which therefore afford the drill a short path to the repetitively injected, and frequently tapped, central crust±mantle transition. One such location is atop a volcano, perhaps like Kilauea, as viewed from the ¯at platform at Atlantis Bank, the eastern transverse top, but more pertinently one that produces cumulates ridge on Atlantis II Fracture Zone, Southwest Indian with a repetitive sequence in both rhythmic and cryp- Ridge ?Fig. 1). The bankwas swath mapped in 1986 tic mineralogical variation, when considered from the ?Dicket al., 1991a) and then drilled in 1987. That interior. The best-studied example is probably Rhum, section, obtained in only 720 m of water at ODP where cyclic cumulates marked by such repetitions Hole 735B during Leg 118 ?Robinson, Von Herzen were originally believed to have precipitated from a et al., 1989) comprises 504 m of abyssal gabbro, and large, circulating, and regularly injected magma the hole, which was equipped with a re-entry cone chamber ?Brown, 1956; Wager and Brown, 1967). atop a hard-rockbase, was left open for future drilling. This picture has been modi®ed by more recent ®eld The renewed drilling tookplace during October± study ?e.g. BeÂdard et al., 1988), but Rhum, perhaps December of 1997 and extended the hole to 1508 m more than any other place, provided a conceptual below the sea ¯oor ?mbsf) before a sudden and unex- frameworkfor considering evolution of basaltic pected drill-string failure in high seas blocked further liquids by means of crystal fractionation in recurrently drilling. The recovery rate of cored rockwas 86.6%, replenished and tapped, fractionating ?RTF) magma virtually identical to the rate during Leg 118 ?Dick, chambers ?e.g. O'Hara, 1977; Usselman and Hodge, Natland, Miller et al., 1999). The section, almost 1978; O'Hara and Mathews, 1981). entirely in gabbro, is ideal to contribute to models Some students of ophiolites postulated that variants of crustal accretion at spreading ridges, and to of this type of volcano occur in the ocean crust ?e.g. compare to ophiolites, layered intrusions, and other Greenbaum, 1972; Pallister and Hopson, 1981; kinds of ocean crust. Smewing et al., 1984) and this idea was taken up In the past, analogies of the unseen gabbroic particularly for fast-spreading ridges ?Cann, 1974; portion of the ocean crust to the products of very Rosendahl, 1976; Natland, 1980; Wilson et al., different types of magma chambers have been offered 1988). Besides having repetitive supply and eruption ?e.g. Cann, 1974; Rosendahl, 1976; Bryan and Moore, of magma, these volcanoes are also regularly and 1977; Nisbet and Fowler, 1978; Sinton and Detrick, continuously rifted by sea-¯oor spreading, being 1992). The gabbros of Hole 735B represent one of the pulled apart from top to bottom right down the middle, most complex types of magma chamber thus far with each dilation of the crust volumetrically J.H. Natland, H.J.B. Dick / Journal of Volcanology and Geothermal Research 110 52001) 191±233 193 balancing each injection of magma. The crust has high-temperature deformation, and magmatic little relief, and is virtually constant in thickness. differentiation acted in concert with formation of However, as we now understand it, there is no large metamorphic textures in the selfsame rocks. magma body beneath the axis of the East Paci®c Rise In 1987, at the time of Leg 118, no matter how the ?Sinton and Detrick, 1992), as indeed, there was none crust at slowly spreading ridges was thought to form, at Rhum ?cf. Emeleus et al., 1996). Instead, drilling no one conceived of this. No accounts in the literature shows that at the East Paci®c Rise, high-level prepared us for the intimate connection between cumulates, at least, form in a narrow, generally verti- magmatism and shear that these rocks revealed. cal or dike-like, channelized network within a crystal They challenged our ability even to name them, let mush ?Natland and Dick, 1996). Some melt collects at alone breakthem down into a rational igneous strati- a thin and narrow sill-like lens at the base of the dikes graphy. In retrospect, we can identify some places on ?e.g. Kent et al., 1993) and also in another at the land where counterparts may exist, and study of both mantle transition ?Crawford et al., 1999). This is not ophiolites and layered intrusions has produced a few a simple volcano, yet it is unquestionably a common parallels and clari®ed some of what was puzzling in type on the Earth's surface.