STRUCTURE AND TECTONICS

One of the main aims of Leg 209 was to measure ductile deformation fabrics in residual mantle peridotites in order to constrain the mode of mantle upwelling and corner flow beneath slow-spreading ridges. We hoped to use mineral shape fabrics, as well as lattice-preferred orientation of olivine, to determine whether ductile deformation was radial, consistent with three-dimensionally focused, buoyancy-driven upwelling, or orthogonal to the rift axis, consistent with two-dimensional, plate-driven upwelling. Our expectations were based on extensive observations of mineral shape fabrics in residual mantle peridotites; in particular, field and laboratory measurements of spinel and pyroxene lineation, together with laboratory observations of olivine elongation and subgrain orientation, have been used to map ductile deformation trajectories in the Oman ophiolite and other massifs (e.g., Nicolas et al., 1972; Nicolas and Violette, 1982).

Summary of Shipboard Observations

As noted above, the third important observation emphasized in our synthesis paper (Kelemen et al., submitted [N1]) was that most peridotites lacked textural evidence for subsolidus plastic deformation. Instead, shipboard observations indicated that in most peridotite both spinel and pyroxene form equant grains or irregular grains interstitial to olivine crystals (Shipboard Scientific Party, 2004b: figs. F6, F7, F9, F44; 2004c: figs. F19, F91; 2004d: fig. F38; 2004e: figs. F16, F42; 2004g: figs. F10-F15, and especially figs. F22 and F36). Spinel, in particular, forms skeletal grains whose extensions are commonly <100 µm wide but extend along olivine grain boundaries over millimeters in two and three dimensions. In some cases, these spinels are intergrown with pyroxene and could have been armored by "strong" pyroxene porphyroclasts during ductile deformation of olivine. However, in other samples the spinel was present along pyroxene-free olivine grain boundaries. Similarly, at every site we recovered peridotites in which pyroxenes are poikilitic or interstitial to olivine.

The shipboard structural geology team used a semiquantitative scale of crystal-plastic deformation intensity in peridotites, ranging from a lack of any crystal-plastic shape fabric (0, also called "protogranular"), through several stages of foliation and porphyroclast development, to mylonitic (4) and ultramylonitic (5) fabrics. Recovered peridotites from Site 1268 had an average deformation intensity of ~0.9, those from 1270 had an average intensity of ~1.2, from 1271, ~0.5, from 1272, ~0.2, and from 1274, ~0.3. Impregnated peridotites at Site 1275 had an average ductile deformation intensity <0.1. Omitting mylonitic rocks (crystal fabric intensity > 3), these averages become ~0.5 for Site 1268 and ~1 for Site 1270. Averages are not changed significantly by omitting mylonitic intervals for all other sites. In contrast, peridotites from ODP Site 895 at Hess Deep, deformed beneath the East Pacific Rise (e.g., Mével, Gillis, Allan, and Meyer, 1996), and from the intermediate- to fast-spreading Oman ophiolite (e.g., Nicolas et al., 2000; Nicolas and Violette, 1982) generally have well-developed spinel shape fabrics and discernable orthopyroxene shape fabrics, corresponding to a crystal-plastic deformation intensity of 2–3 on the shipboard scale used during Leg 209.

The presence of interstitial spinel without pyroxene, and interstitial pyroxene without spinel, virtually rules out subsolidus exsolution of spinel from pyroxene as the cause for the interstitial textures. Instead, we infer that the interstitial textures formed during melting, melt migration, igneous dissolution, and/or precipitation from melt migrating by porous flow along crystal grain boundaries. Given that these rocks, and mixtures of these peridotites with small amounts of primitive basaltic melt, have solidus temperatures >1200°C at 0.4–0.7 GPa (e.g., Hirschmann, 2000), we infer that most peridotites recovered by drilling during Leg 209 have not recorded measurable shear strain at temperatures <1200°C. Given the estimates for the thickness of the thermal boundary layer summarized in "Thermobarometry" above, this inference indicates that most peridotites recovered during Leg 209 were not penetratively deformed at depths less than ~20 km below the seafloor.

High-temperature (>900°C) mylonitic shear zones with recrystallized olivine, pyroxene, and/or plagioclase grain sizes 20–100 µm cut weakly deformed peridotites and gabbroic rocks at Sites 1268, 1270, 1271, and 1274. Illustrations can be found in the Leg 209 Initial Reports volume (Shipboard Scientific Party, 2004a: figs. F20, F22; 2004b: figs. F48, F49, F50; 2004c: figs. F12, F20, F21, F39, F51, F60, F65, F77, F80; 2004d: figs. F39, F41, F42 2004f: figs. F6, F40). These mylonites formed by localized high-strain ductile deformation. Many of the Leg 209 peridotite mylonitic shear zones are substantially coarser than peridotite mylonites previously dredged from the Mid-Atlantic Ridge that formed at ~600°C and high stress (e.g., Jaroslow et al., 1996). We infer that the coarser mylonites recovered during Leg 209 formed under lower stress conditions at temperatures >900°C because they have olivine grain sizes in the same range as in well-documented high-temperature shear zones in ophiolites and mantle massifs (Dijkstra et al., 2002; Kelemen and Dick, 1995; Newman et al., 1999; Vissers et al., 1991). Most, though not all, of the Leg 209 mylonites formed along contacts between residual peridotite and gabbroic rocks.

In addition to ductile shear zones, we recovered fault gouge and cataclasites at Sites 1268, 1270, 1271, 1272, 1274, and 1275. These formed in numerous brittle fault zones (Shipboard Scientific Party, 2004a: figs. F41, F48; 2004b: figs. F57, F58, F64; 2004d: fig. F53; 2004e: figs. F47, F48; 2004f: figs. F7, F32, F47, F48; 2004g: figs. F56–F60).

It is striking that we recovered samples from more than one ductile shear zone at Sites 1268, 1270, and 1271 and more than one brittle fault zone at Sites 1268, 1270, 1271, 1272, and 1274. In addition, based on bathymetry and dive observations, at Sites 1270, 1274,and 1275 the seafloor is interpreted as a fault surface. Thus, several zones of localized deformation were recovered at all gabbro and peridotite sites except Site 1275, where only one significant fault—at the seafloor—was observed. This indicates that the "typical" spacing between adjacent shear zones and faults in the 14°–16°N region—at least close to the seafloor—is <100–200 m.

Shipboard and Shore-Based Analysis of Paleomagnetic Data

Garcés and Gee (2007) synthesized paleomagnetic data from Leg 209 core using shipboard observations (Kelemen, Kikawa, Miller, et al., 2004) and subsequent shore-based laboratory analyses. Expanding on a hypothesis formed at sea (Shipboard Scientific Party, 2004a: figs. F16, F17, F18, F28, F29, F30, and accompanying text), Garcés and Gee (2007) demonstrate that large tectonic rotations (>90°) have occurred in the footwall(s) to currently low-angle faults recovered in core and thus that the paleomagnetic data are consistent with nucleation and slip along high-angle normal faults, followed by passive rotation of the high-angle faults to their current, nearly horizontal orientations. Original fault orientations dipped steeply toward the spreading axis.

Lattice-Preferred Orientation of Minerals in Site 1274 Peridotite

In their 2005 American Geophysical Union abstract, Achenbach et al. report on preliminary measurements of olivine and pyroxene lattice-preferred orientation in a few samples from Hole 1274A. They find a weak shape foliation in olivine and orthopyroxene and weakly developed olivine lattice fabrics consistent with dislocation creep at ~1200°C with flow in or near the plane of the foliation. Using paleomagnetic data and assumptions about ridge-parallel tectonic rotation (e.g., Garcés and Gee, 2007; Kelemen, Kikawa, Miller, et al., 2004) they attempt to restore the microstructural observations to their geographical orientation at the time that the remnant magnetization was acquired (probably, during serpentinization at ~200°–300°C). They conclude that there was subhorizontal foliation with horizontal ridge-parallel olivine a-axis lineation at the time of magnetization. With the additional assumption that there were no significant tectonic rotations of the rocks during the time for cooling from ~1200°C to ~300°C, these results are consistent with high-temperature ridge-parallel ductile flow in the mantle beneath the Mid-Atlantic Ridge near Site 1274.

Discussion of Structure and Tectonics Results

The geometry of plate spreading, together with the observation of residual mantle peridotites and high-pressure igneous cumulates on the seafloor, demands that some rocks underwent tectonic uplift and rotation during corner flow within the upper 15–20 km below the seafloor. The paucity of ductile deformation fabrics in most peridotites, coupled with the abundance of localized mylonitic shear zones and faults, suggests that blocks of peridotite were passively uplifted and rotated along localized shear zones extending deeper than 15 km. Whereas some faults observed in the region, particularly at Site 1275, could have formed at shallow depth—for example, at the "dike–gabbro transition" as inferred by Escartín et al. (2003) for this same area—denudation of nearly undeformed residual peridotites and high-pressure cumulates requires uplift along localized shear zones and faults that extend to depths of more than 15 km.

The 14°–16°N region along the Mid-Atlantic Ridge has commonly been interpreted as unusual. Numerous studies have documented the presence of extensive outcrops of mantle peridotite on both sides of the rift valley, extending for at least 50 km from the 15°20´ Fracture Zone (summarized in Kelemen, Kikawa, Miller, et al., 2004), whereas along other geologically well known parts of the Mid-Atlantic Ridge, exposed mantle peridotite is typically limited to one side of the rift valley, generally on inside corner highs within 10–30 km of a large-offset fracture zone (e.g., Tucholke and Lin, 1994). Thus, although peridotite exposures along the Mid-Atlantic Ridge are commonly interpreted as the result of asymmetric deformation along detachment faults, it has generally been thought that the 14°–16°N region is tectonically more complex (with alternating dips of normal faults) and "magma starved."

For this reason, the processes outlined in the previous section of this paper—igneous crystallization and localized deformation throughout a thermal boundary layer >15 km thick beneath the Mid-Atlantic Ridge—could be unique to this unusual area. However, as noted above, theoretical calculations suggest that the thermal boundary layer beneath most or all slow-spreading ridges could extend deeper than 15 km, and petrological studies of lavas suggest that mantle-derived melts begin to crystallize at depths of 15 km or more in many places along the Mid-Atlantic Ridge. In addition, shipboard observations during Leg 209 suggest that localized shear zones form along contacts between residual peridotite and gabbroic intrusions, perhaps because the rheological contrast between peridotite and gabbro enhances localization of deformation between ~1200° and 600°C. Thus, Kelemen et al. (submitted [N1]) suggest that the 14°–16°N area may simply be an end-member example that reveals typical slow-spreading processes in their best known, clearest expression.

In this context, Kelemen et al. (submitted [N1]) question some of the common interpretations of this region. For example, is this area really "magma starved"? Leg 209 observations, coupled with previous work, suggest that the entire 14°–16°N area may be underlain by mantle peridotite hosting 20%–40% gabbroic intrusions and impregnations. Gabbro of 30% in the upper 21 km of an oceanic plate would correspond to 7 km of "normal" oceanic crust. Gabbro of 30% (7.2 km/s) + 75% peridotite (8.2 km/s) yields a "mantle" compressional wave velocity (VP) (7.9 km/s) greater than or equal to sub-Mohorovicic Discontinuity (Moho) VP observed in about half of the seismic refraction studies of oceanic crust that have been conducted to date (Shipboard Scientific Party, 2004a: fig. F8 and associated references).

Recently, Lizarralde et al. (2004) performed a refraction experiment along a flow line in the Atlantic south of Bermuda. They found that an episode of relatively slow spreading formed seismic crust ~5 km thick, compared to 7-km-thick crust formed at faster spreading rates. Shallow mantle VP is slower beneath the 5-km-thick crust, compared to that beneath the 7-km-thick crust. The difference in VP is consistent with the presence of ~7.5% gabbroic material distributed within the uppermost mantle beneath the 5-km-thick crust, so that the total proportion of gabbroic rocks formed during the slower and faster spreading episodes could be the same. Furthermore, Lizarralde et al. (2004) observed that areas with 5-km-thick crust had Bouger gravity anomalies 20–30 mGal higher than areas with 7-km-thick crust, even though the proportion of gabbroic rocks could be the same in both areas. They modeled the gravity data for the area with 5-km-thick crust as the result of 7.5% gabbroic material distributed over 30–60 km of the uppermost mantle. Generalizing from this result, Kelemen et al. (submitted [N1]) hypothesized that some of the observed variation of seismic crustal thickness and Bouger gravity anomalies along the Mid-Atlantic Ridge may be due to variable depth and distribution of gabbroic intrusions rather than to variable overall proportions of gabbroic rocks.

Finally, the strength of seismic anisotropy in the shallow mantle—formed by alignment of olivine a-axes during viscous deformation—will be smaller where corner flow in the uppermost mantle is accommodated by block rotation along localized shear zones, rather than by penetrative ductile deformation of high-temperature peridotites. At fast-spreading ridges, where the adiabatic geotherm probably extends to the base of the crust, corner flow in the shallow mantle is probably accommodated entirely by ductile deformation of all peridotites (e.g., Nicolas et al., 2000). Beneath slow-spreading ridges, localized deformation and passive rotation of undeformed blocks is likely in the uppermost mantle (Fig. F6). Thus, during Leg 209, we hypothesized that seismic anisotropy in the upper 15–30 km of the mantle would be greater beneath crust formed at the fast-spreading East Pacific Rise compared to the slow-spreading Mid-Atlantic Ridge. We were delighted to discover that this hypothesis is consistent with the recent results of Gaherty et al. (2004).

Amplifying the paleomagnetic results and shipboard observations of shear zone and fault structure, Schroeder et al. (submitted [N3]) present a tectonic synthesis of data from Leg 209. They endorse a "rolling hinge" model for large normal faults that denude mantle peridotite, intruded by gabbroic plutons, to the seafloor, coupled with later development of other normal faults with smaller displacement.

To summarize, recent seismic and gravity observations of old Atlantic seafloor southwest of Bermuda independently suggest that gabbroic material is distributed throughout the uppermost mantle below the Moho during periods of relatively slow spreading and that the role of penetrative ductile deformation is much smaller in the shallow mantle beneath the Atlantic compared to the Pacific. These interpretations are consistent with the central results from Leg 209. Thus, many or most slow-spreading ridges may be characterized by igneous crystallization and localized deformation throughout a thermal boundary layer thicker than 15 km. Continued geological and geophysical studies can test this hypothesis and help to define the regional extent of oceanic plates that form in this way.

NEXT