Strain localization along the Atlantis Bank oceanic detachment fault system, Southwest Indian Ridge



Microstructural observations and mineral thermometry from in situ samples collected from the Atlantis Bank oceanic core complex (SW Indian Ridge) indicate that detachment faulting was initiated under hypersolidus conditions in the ductile regime and continued through subgreenschist temperatures through the ductile, semibrittle, and brittle regimes as strain localized along the exposed, now subhorizontal fault surface. Ductile, semibrittle, and brittle fabrics are developed within dominantly gabbroic rocks. Footwall rocks exhibit crystal plastic fabrics distributed over a structural thickness up to 400 m below the denuded fault surface exposed at the seafloor, whereas semibrittle and brittle fabrics are concentrated in the 80 and 30 m immediately below the principal slip surface of the detachment fault, respectively. Sample fabrics suggest that strain localization was achieved by dynamic recrystallization of plagioclase at temperatures between 910°C and 650°C, by amphibole-accommodated dissolution-precipitation creep at temperatures ∼750°C–450°C, by chlorite-accommodated reaction softening at temperatures ∼450°C–300°C, and by brittle fracturing and cataclasis at temperatures <300°C.

1. Introduction

Slow and ultraslow spreading mid-ocean ridges accommodate plate separation by both magmatic and tectonic processes [Karson, 1990; Smith and Cann, 1993; Tucholke and Lin, 1994; Blackman et al., 1998], where tectonic spreading is accommodated by high-angle normal faults and low-angle detachment fault systems associated with oceanic core complex formation [Karson and Dick, 1983; Karson, 1990; Dick et al., 1991a; Mutter and Karson, 1992; Tucholke and Lin, 1994; Blackman et al., 1998]. Detachment faults are identified in both on- and off-axis locations based on their distinctive dome-shaped and corrugated fault surfaces that cap the underlying oceanic core complex [Cann et al., 1997; Blackman et al., 1998; Tucholke et al., 1998; Karson, 1999; Ranero and Reston, 1999; Tucholke et al., 2001; MacLeod et al., 2002; Smith et al., 2006], and may make up to 60% of the seafloor locally [Smith et al., 2008]. Oceanic core complexes are thought to form at or near the ridge axis, where footwall rocks are denuded by detachment faults that dip toward the axial valley [Cann et al., 1997; Tucholke et al., 1998; Ranero and Reston, 1999; Blackman et al., 2001; MacLeod et al., 2002; deMartin et al., 2007]. Corrugated fault surfaces are commonly exposed for ∼15–30 km, and extend up to ∼125 km parallel to the spreading direction, suggesting that active detachment faulting can persist for a few to several million years [Tucholke et al., 1998; Ohara et al., 2001; Escartin et al., 2003; Ildefonse et al., 2007; Harigane et al., 2008]. The recovery of variable proportions of oceanic crust and upper mantle rocks from these exposed fault surfaces has led to the development of several models for oceanic core complex formation that invoke a limited to substantial role of magma supply in the process of strain localization, as summarized by Escartin et al. [2003] and Ildefonse et al. [2007].

Few oceanic core complexes have been sampled comprehensively via seafloor sampling and deep drilling efforts, but these unique samples provide the best opportunities to understand the process of strain localization and core complex development [Dick et al., 2000; Blackman et al., 2002; MacLeod et al., 2002; Escartin et al., 2003; Schroeder and John, 2004; Blackman et al., 2006; Boschi et al., 2006; Karson et al., 2006; Kelemen et al., 2007]. In the Atlantic, the eastern Kane (23°N) region was sampled by ODP Holes 669, 670, and 920–924, Atlantis Massif core complex at 30°N was penetrated by Integrated Ocean Drilling Program (IODP) Holes U 1309B and U1309D, and the 15°45′N core complex was sampled by Ocean Drilling Program (ODP) Holes 1275B and 1275D [Karson and Lawrence, 1997a, 1997b; Blackman et al., 2006; Kelemen et al., 2007]. Core recovered from drill holes near the Kane fracture zone were all collected from the same spreading segment of the ridge; the cores from the ridge-transform intersection massif comprise gabbroic rocks with crystal plastic, semibrittle and brittle deformation fabrics, and the cores collected ∼15 km south of the massif exhibit crystal plastically deformed ultramafic rocks [Shipboard Scientific Party, 1988; Cannat et al., 1997; Ceuleneer and Cannat, 1997; Karson and Lawrence, 1997a, 1997b]. Submersible and seafloor studies from the eastern Kane region reveal upper and lower crust rocks in the vicinity of the intersection massif, and upper crust volcanic rocks and ultramafic rocks ∼15 km south of the massif exhibiting primarily brittle deformation at greenschist grade temperatures [Karson and Dick, 1983; Karson et al., 1987; Mevel et al., 1991]. The cores from Atlantis Massif and the 15°45′N core complexes comprise primitive gabbroic and rare ultramafic rocks, and demonstrate that extensive high-strain granulite and amphibolite grade crystal plastic deformation is lacking, and that fracturing and cataclasis are rare and restricted to narrow zones in the upper 10s of meters of each hole [Shipboard Scientific Party, 2004; Blackman et al., 2006; Kelemen et al., 2007]. The rock types and deformation fabrics observed in these cores contrast with samples collected by submersible, dredging and seafloor coring. These seafloor studies indicate that the corrugated surfaces of the massifs and the footwall rocks immediately beneath it (<70 cm) are dominated by both semibrittle fault schists and brittle fault rocks derived from both ultramafic and mafic rocks. Structurally deeper footwall rocks exhibit fabrics indicative of high-strain, high-temperature crystal plastic deformation, and lower-temperature semibrittle and brittle deformation [Cannat et al., 1997; Blackman et al., 2002; MacLeod et al., 2002; Escartin et al., 2003; Schroeder and John, 2004; Boschi et al., 2006; Karson et al., 2006]. Together these submersible and drilling studies of the Kane, Atlantis Massif and 15°45′ core complexes suggest that strain localization processes involving both mafic and ultramafic rocks may vary laterally over short (1–10 km) distances, and underscore the importance of combined drilling and submersible/surface sampling in addressing the evolution of oceanic lithosphere.

The Atlantis Bank oceanic core complex (Southwest Indian Ridge) exhibits extensive hypersolidus and high-temperature crystal plastic fabric development, overprinted by localized brittle fabric development within dominantly gabbroic rocks in both drill core (ODP Holes 735B and 1105A) and submersible samples [Shipboard Scientific Party, 1989; Cannat et al., 1991; Dick et al., 1991b; Shipboard Scientific Party, 1999a, 1999b; Arai et al., 2000; Dick et al., 2000; Kinoshita et al., 2001; Matsumoto et al., 2002; Miranda, 2006; Mehl and Hirth, 2008]. The hypersolidus fabrics and extensive crystal plastic deformation documented in the shear zones of ODP Hole 735B demonstrates that core complex development began during crystallization of gabbroic rocks, and continued long after crystallization was complete. The minor percentage of ultramafic rocks and associated serpentinite recovered from the footwall rocks of Atlantis Bank suggests that strain localization processes leading to the development of a discrete detachment fault may differ from those core complexes exhibiting more heterogeneous footwall rock types. Atlantis Bank is therefore an ideal location to investigate the processes of strain localization that lead to development of a core complex associated with extensive crystal plastic fabric development within a dominantly gabbroic section. In this study, we use in situ submersible samples to investigate fabric development associated with strain localization along the Atlantis Bank detachment fault system, and compare these results with earlier studies from the ODP Hole 735B drill core.

2. Study Location

The Atlantis Bank core complex is located ∼80–110 km south of the present-day Southwest Indian Ridge. It is a ∼300 km2 core complex that formed between 10 and 13 Ma near the center of the ridge segment between the Atlantis II Transform and the nontransform discontinuity at 57°40′E (Figure 1) [Dick et al., 1991b; Stakes et al., 1991; Matsumoto et al., 2002; Natland and Dick, 2002; Hosford et al., 2003; Schwartz et al., 2005; Baines et al., 2007; Baines et al., 2008]. Bathymetric mapping across Atlantis Bank shows preservation of a dome-shaped surface dissected by moderately dipping (30–50°) transform- and ridge-parallel normal faults (Figure 2) [Dick et al., 1991b; Arai et al., 2000; Kinoshita et al., 2001; Matsumoto et al., 2002; Baines et al., 2003]. Due to normal slip along these faults, the morphology of this core complex differs from younger core complexes recognized along the Mid-Atlantic Ridge. These later faults make Atlantis Bank an ideal location for studying strain localization associated with detachment faulting as they expose structurally deep portions of the footwall normally inaccessible on younger core complexes.

Figure 1.

Bathymetric map of the Atlantis Bank region, showing the core complex as a bathymetric high characterized by >5000 m of relief between the transform valley floor and the top of the bank at ∼700 m below the sea surface. Color scale represents depth below sea surface (GMT map constructed by G. Baines using data from Matsumoto et al. [2002]).

Figure 2.

Bathymetric map of Atlantis Bank. The fault system and footwall are exposed over 300 km2. Denudation of hanging wall rocks (shaded brown) exposed a dome-shaped footwall, much of which has been dissected by both transform- and ridge-parallel normal faults. Remnants of the dome-shaped detachment fault surface are shaded light gray. The hanging wall initially separated from the footwall at the inferred breakaway; the leading edge of the hanging wall that slipped is now at the detachment termination. Atlantis Bank has been extensively sampled; Figure 2 shows the location of ODP Hole 735B and ODP 1105A, manned submersible dives (Shinkai dives in blue), and unmanned submersible dives (Kaiko dives in red) (GMT map constructed by G. Baines using data from Matsumoto et al. [2002]).

The footwall to the detachment fault at Atlantis Bank was first sampled in 1987 with the initial drilling of ODP Hole 735B to ∼500 mbsf; the Hole was later deepened to 1508 mbsf in 1997 [Shipboard Scientific Party, 1999b]. The 1508 m of core from Hole 735B is almost entirely composed of gabbroic rocks, with minor diabase and troctolite (Figure 3). The bulk of the core (77%) comprises undeformed rocks with little to no crystal plastic or brittle fabric development [Dick et al., 2000]. Strain is localized in the remaining intervals of the core within ductile shear zones and brittle fault zones, with well-developed crystal plastic and brittle fabrics, the thickest and most significant such zone is located in the upper 100 m of the Hole [Cannat et al., 1991; Stakes et al., 1991; Shipboard Scientific Party, 1999c].

Figure 3.

Lithostratigraphic composition and crystal plastic deformation intensity in ODP Hole 735B. The ∼100 m thick shear zone at the top of Hole 735B is structurally below the detachment fault surface and is defined by crystal plastic fabrics with intensity >1. The intensity of crystal plastic deformation is classified as follows [Cannat et al., 1991]: 1, weakly deformed, no penetrative foliation; 2, penetrative foliation, dynamic recrystallization limited; 3, well foliated, extensive dynamic recrystallization; 4, mylonitic bands (1 mm), extensive dynamic recrystallization; 5, mylonitic foliation, extensive dynamic recrystallization. Modified from Natland and Dick [2002].

Submersible samples used in this study were sampled from broadly distributed exposures of the detachment surface at the seafloor and the footwall directly below, and likely correspond to the shear zone between 0 and 100 mbsf in 735B. We present microstructural observations from these samples representative of the granulite, amphibolite, greenschist and subgreenschist grade fabrics observed in the main shear zone (0–100 mbsf) associated with detachment faulting and discuss lateral variations associated with the detachment fault system to compare with earlier studies [Cannat et al., 1991].

3. Methods

3.1. Sample Acquisition

To confirm the lateral extent and variability of processes associated with strain localization and development of the detachment fault system, manned and unmanned submersibles (Shinkai 6500 and Kaiko, respectively) were used to sample the detachment surface and its footwall and probable hanging wall in a total of 21 dives during three research cruises: the JAMSTEC/WHOI 2001–2002 ABCDE cruise, the JAMSTEC/WHOI Mode '98 cruise, and the JAMSTEC/WHOI MODE 2000 cruise (Figure 2) [Arai et al., 2000; Kinoshita et al., 2001; Matsumoto et al., 2002]. The widely distributed submersible samples provide the basis for a comprehensive study of the mechanisms of strain localization that led to development of the Atlantis Bank oceanic detachment fault system.

Ninety-eight in situ samples from the ABCDE cruise, the Mode '98 cruise, and MODE 2000 cruise were used in this study. Accurate sample locations were determined using the Long Base Line and the Super Short Base Line navigational systems, with vertical sample spacing between 50 m and 200 m. Samples were categorized as “from outcrop” if they were broken off of the outcrop with the robotic arms of the submersible. Samples that were dislocated from the contiguous outcrop by joints, but that were otherwise in place were categorized as “on outcrop”; they were lifted out of their location within the outcrop using the robotic arms of the submersible. Finally, loose samples collected from talus piles or from sediment-covered seafloor surfaces were described as “float.” We designate “from outcrop” and “on outcrop” samples as in situ samples. A complete discussion of sample acquisition techniques and cruise results are discussed by Arai et al. [2000], Kinoshita et al. [2001], and Matsumoto et al. [2002].

3.2. Sample Selection, Preparation, and Analytical Methods

Preliminary hand sample and microstructural observations were made on all in situ samples collected by University of Wyoming shipboard scientists from the 3 cruises; these include 67 samples from the 2001–2002 ABCDE cruise, 29 samples from the Mode '98 cruise, and 2 samples from the MODE 2000 cruise (Table 1). A detailed description of our sample selection criteria is located in the auxiliary material. A subset of 17 gabbroic rocks was chosen for detailed analyses based on well-developed, representative microstructures and mineral assemblages appropriate to document the T-t deformation associated with the development of the detachment system. A summary of the characteristics of these samples is included in Table 2. Their protoliths range in composition from olivine gabbro, oxide gabbro, oxide olivine gabbro, biotite hornblende oxide gabbro, hornblende gabbro and gabbro (following the nomenclature outlined by Shipboard Scientific Party [1999b]), and are characteristic of the rock types recovered in ODP holes 735B and 1105A. The submersible samples allow both evaluation and comparison of strain localization processes between oxide and nonoxide gabbro suites (Figure 3).

Table 1. In Situ Sample Fabric Developmenta
SampleSample DepthStructural Depth Below DetachmentRock TypeSample AcquisitionObserved Textural DevelopmentbDeformation Intensity
GranuliteUpper AmphiboliteLower AmphiboliteGreenschistSubgreenschistDuctilecSemibrittleBrittled
  • a

    Bold text designates samples chosen for electron microprobe analyses and thermometry. Deformation intensities are maximum values observed within a given thin section.

  • b

    Crosses indicate that textural development was observed.

  • c

    1, slight crystal plastic deformation; 2, visible foliation; 3, protomylonite (10%–50% matrix); 4, mylonite (50%–90% matrix); 5, ultramylonite (>90% matrix).

  • d

    1, fractures with visible offset; 2, moderate fracturing, no significant grain size reduction; 3, dense, anastomosing fracturing, incipient breccia (<20% matrix); 4, well-developed fault brecciation, rotation of clasts (20%–70% matrix); 5, cataclasite (>70% matrix).

172-103691400oxide gabbro and gabbroon outcrop xx  local 4–5 in gabbro30
172-24304010oxide gabbroon outcrop xx  440
643-033396 serpentinitefrom outcrop     001
643-152605 serpentinitefrom outcrop     001
644-12197970olivine gabbroon outcropxx   410
645-01273130oxide gabbroon outcropxx   500
645-02273130oxide gabbroon outcropxx   4.520
645-03270030oxide gabbroon outcropxx   510
645-04a270030oxide gabbrofrom outcropxx   500
645-05261130oxide gabbro and olivine gabbroon outcropxxx  320
645-06261130oxide gabbroon outcropxxx  320
645-07249330oxide gabbroon outcropxxx  220
645-08238430olivine oxide gabbroon outcropxx   200
645-09238430olivine oxide gabbroon outcropxxx  2.520
645-10230520oxide olivine gabbroon outcropxx   420
645-1121910chlorite schistfrom outcropxx x 4.550
645-1221910amphibole-chlorite schist and diabasefrom outcropxxxx 4.550
645-16190620gabbronoritefrom outcropxx   400
645-18179210gabbroon outcrop     000
646-064458 brecciaon outcrop     004
647-051949120oxide gabbroon outcropxxx  320
647-08185730oxide gabbroon outcropxxx  local 2–320
647-16151350olivine oxide gabbrofrom outcropxxx  3.520
647-17151350olivine oxide gabbroon outcropxxx  420
648-012698260olivine gabbroon outcropxx   510
648-032691260oxide gabbrofrom outcropxx   1.500
648-042572250olivine gabbroon outcropx    400
648-052571250olivine oxide gabbro with troctoliteon outcropxx   local 3–400
648-062442220olivine gabbrofrom outcropxx   400
648-072296190gabbroon outcropxx   41 
648-102153 diabaseon outcrop     000
648-111997150olivine oxide gabbro and olivine gabbroon outcropxxx  3.510
648-121997110olivine gabbrofrom outcropxxx  340
648-131998110fault brecciafrom outcrop xxxx453
648-14187960fault brecciaon outcrop x xx452.5
648-15187960gabbroon outcropxx   420
648-16177920amphibole schiston outcrop x   451
648-17177920amphibole schiston outcrop x   450
648-191750 diabasefrom outcrop     000
648-2017520amphibole schistfrom outcrop x   450
649-012609 serpentinitefrom outcrop     000
649-022609 serpentinitefrom outcrop     000
649-032593 serpentinitefrom outcrop     000
649-042577 brecciafrom outcrop     004
649-052564 serpentinitefrom outcrop     000
649-062561 brecciafrom outcrop     003
649-092541 brecciafrom outcrop     004
649-112502 gabbrofrom outcrop xx  43.50
650-072244 troctoliteon outcropx    400
650-131895 olivine gabbroon outcropx    3.500
650-141895 gabbroon outcropx    3.500
650-161719 olivine gabbroon outcropx    10local 3
651-093918340oxide gabbroon outcropxxx  420
651-103918340oxide gabbrofrom outcropx    400
651-113848310olivine gabbrofrom outcropxx   40 
653-022974100olivine gabbrofrom outcropxx   2.520
653-04286560olivine gabbro and gabbroon outcropxxx  420
653-06265430gabbroon outcrop xx  32local 2–3
653-072562 diabasefrom outcrop     000
653-08247120oxide gabbroon outcropxx   1.500
653-09246720fault brecciaon outcropxxxxx504
653-1622570fault brecciaon outcrop x xx404
653-1822570fault brecciafrom outcrop x   453
653-2022570fault brecciafrom outcrop x  x4.504
655-103584 diabasefrom outcrop     000
655-113496 fault brecciafrom outcrop x  x003–4
655-143369 epidositefrom outcropxx x 22local 2–3
655-153366 epidositeon outcropxx x 12local 2–3
655-203128 fine-grained amphibolefrom outcropxx   10local 2
458-034609 serpentiniteon outcrop     1  
459-07260180oxide gabbroon outcropxxx  420
459-08247850gabbro and oxide gabbroon outcropxxx  3.520
459-09245550gabbrofrom outcropxxxx 220
459-12229440oxide gabbrofrom outcropx    210
459-13222620oxide gabbrofrom outcropxxx  3.520
459-14222620oxide gabbrofrom outcropxxx  420
460-02281120fault brecciaon outcrop x xx032
460-10215730oxide gabbroon outcropxxx  41.50
460-11215730olivine gabbroon outcropxx   310
460-13193220olivine gabbroon outcropx    200
461-011791 diabasefrom outcrop     000
461-02167735olivine gabbrofrom outcropx    200
461-03167735gabbrofrom outcropxx   320
461-04161310olivine oxide gabbrofrom outcropxx   320
461-0892310oxide gabbrofrom outcropxxx  32.50
461-1177710olivine gabbrofrom outcropxx   1.520
462-03111510fault brecciafrom outcrop xxxx453
462-04109810incipient fault brecciafrom outcrop xxxx452
462-06997 diabasefrom outcrop     000
462-07100215incipient fault brecciafrom outcrop xxxx451.5
462-08100215fault brecciafrom outcrop xxxx453
466-08373880oxide gabbroon outcrop     000
466-10337540oxide gabbroon outcropxxx  420
467-13219750olivine gabbrofrom outcropxx   310
467-15184750gabbroon outcropx    200
467-17184750olivine gabbroon outcropx    3.500
467-18184750olivine oxide gabbroon outcropx    3.500
467-19175520gabbroon outcrop     010
Ultramafic rocks (serpentinites), N = 7
Mafic plutonic rocks (gabbroic rocks), N = 78
Table 2. Samples Chosen for Electron Microprobe Analyses
SampleSample Depth (m Below Surface)Protolith Rock TypePrimary Modal MineralogySecondary Modal MineralogyPlag Grain Size (μm)Cpx Grain Size (μm)Opx Grain Size (μm)Ol Grain Size (μm)Oxide Grain Size (μm)Sample Acquisition
172-103691oxide gabbro7025  5  60   535  2–385020–3850  <10–100on outcrop
172-243040oxide gabbro50–6530–45  5  20   575  15–30040–200  <10on outcrop
460-151640biotite hornblende oxide gabbro55220.5 1840.5     float
467-082526hornblende gabbro61264 36      on outcrop
645-032700oxide gabbro2050  30  2010  3040  10–15050–4000  <10on outcrop
645-102305oxide olivine gabbro6035 41  6015 4120  5–1000100–3000 1000<10on outcrop
645-112191oxide gabbro6035  5      53560 200–3000  10–400from outcrop
645-171792biotite hornblende oxide gabbro5034  1051     on outcrop
647-081857oxide gabbro6525  10  655  1020  20–5000100–2000  10–2000on outcrop
647-171513oxide gabbro6035  5  555  535  10–40010–2000  <10–100on outcrop
648-042572olivine gabbro653055   20–400020–400020–40050–6000 on outcrop
648-062442olivine gabbro5540 5   5530 5 10  20–10020–2000 50 from outcrop
648-072296gabbro5050     5010   40  <10–100<10–500   on outcrop
648-151879gabbro5545     5510   35  5–200020–4000   on outcrop
648-171779oxide gabbro50–6530–45  5  30   565  10–5040–500  <10–100on outcrop
649-112502gabbro6535     55    45  <10–500<10–600   from outcrop
653-162257gabbro (and basalt)50–6550–35     30    655 <10–50<10–100   on outcrop

Dive samples were cut perpendicular to foliation and parallel to lineation when possible to allow characterization of textural development within the motion plane. Detailed interpretation of fabric development for the subset samples was accomplished by petrographic, microstructural, and electron microprobe (EPMA) analyses.

3.3. Fabric Intensity Estimates

Metamorphic grade and intensity of fabric development within the ductile, semibrittle and brittle regimes were documented for each in situ sample collected along individual dive tracks, and are given in Table 1. For each sample, the fabric intensity is classified on a scale from 0 (lowest) to 5 (highest). For ductile fabrics, the intensity of deformation is based on the degree of grain size reduction and mylonitic foliation development, and follows from the shipboard classification used for core from ODP Hole 735B [Shipboard Scientific Party, 1999b; Dick et al., 2000]. Mylonitic fabric development is classified based on the relative proportions of matrix and porphyroclasts [Spry, 1969; Sibson, 1977], and listed from 0 (undeformed) to 5 (ultramylonite). The intensity of semibrittle deformation is based on the degree of fracturing and syntectonic mineral growth; the intensity of brittle deformation is based on the percentage of grain size reduction from brittle fracturing after Schroeder and John [2004].

3.4. Deformation Temperature Estimates

3.4.1. Qualitative and Quantitative Temperature Estimates

Deformation fabrics and mineral assemblages were described in thin section. Temperature estimates for the development of each texture were made on the basis of mineral assemblages and the microstructures developed in the constituent minerals [Brodie and Rutter, 1985]. These qualitative deformation temperature estimates were then compared to quantitative temperature determinations made on the 17 sample subset using mineral thermometry.

3.4.2. Thermometry Methods

Plagioclase, amphibole, pyroxene and chlorite mineral compositions were determined using the JEOL JXA-8900 electron microprobe at the University of Wyoming. Temperatures were calculated for granulite grade mylonites using two-pyroxene QUIlF thermometry [Frost et al., 1988; Frost and Lindsley, 1992; Lindsley and Frost, 1992; Andersen and Lindsley, 1993] on adjacent pairs of recrystallized orthopyroxene and clinopyroxene neoblasts, and for amphibolite grade mylonites using the edenite-richterite amphibole-plagioclase thermometer [Holland and Blundy, 1994] on adjacent grains of amphibole and recrystallized plagioclase. We use this thermometer because it is calibrated for both silica-saturated and silica-undersaturated rocks in the temperature range of 500–900°C, with plagioclase compositions of XAn >0.1 and <0.9. Temperatures were calculated for greenschist grade deformation using chlorite thermometry [Cathelineau, 1988; Jowett, 1991] on chlorite grains in the matrix material of mylonites and microbreccias. Complete descriptions of analytical procedures, thermometry calculations, sources of error, and mineral analyses are included in the auxiliary material and described separately by Miranda [2006].

4. Results

4.1. Microstructural Fabric Description

Our microstructural observations and calculations of average deformation intensity suggest that gabbroic rock samples are more intensely deformed than serpentinized peridotite within the ductile, semibrittle, and brittle regimes (Table 1). Owing to the essentially undeformed ultramafic samples, we then focused our investigation on 17 samples of minimally to intensely deformed olivine gabbro, oxide gabbro, oxide olivine gabbro, biotite-hornblende oxide gabbro, hornblende gabbro and gabbro (Table 2). These samples were collected from as deep as 600 m below the present detachment fault surface, and over a distance ∼22 km along a lithospheric flow line and ∼20 km parallel to the strike of the axial valley. The samples display textures characteristic of fabric development (based on mineral deformation mechanisms) at granulite, amphibolite, greenschist and subgreenschist grade temperatures.

4.2. Microstructures Developed During Granulite Grade Deformation

Granulite grade (800–950°C) metamorphic assemblages in mafic rocks typically include olivine + clinopyroxene + orthopyroxene + plagioclase ± magnetite ± ilmenite ± apatite. In low-strain granulite grade protomylonites, the incipient foliation is manifested by alternating layers of pyroxene and plagioclase (Figure 4a). Some clinopyroxene porphyroclasts are kinked or folded and mantled by coarse-grained recrystallized neoblasts (Figure 4b). Magmatic plagioclase exhibits serrated grain boundaries mantled by recrystallized neoblasts (Figure 4c), similar to microstructures associated with regime 3 high-temperature grain boundary migration deformation in quartz [Hirth and Tullis, 1992; Post and Tullis, 1999].

Figure 4.

Photomicrographs showing textures indicative of granulite grade deformation. The box shows the location of area photographed in Figure 4d. (a) Plane light image of sample 644-12 showing the orientation of foliation. (b) In sample 650-07, a kinked clinopyroxene porphyroclast exhibits undulose extinction and is mantled by pyroxene neoblasts. (c) In sample 460-13, an elongate plagioclase grain exhibits bulging grain boundaries (red arrow). (d) In sample 644-12, brown amphibole-filled fractures cut obliquely across the protomylonitic foliation (double-headed arrow).

Brown amphibole-filled fractures locally cut obliquely across the foliation in some granulite grade samples (Figure 4d). The orientation of the protomylonitic foliation is not disturbed by the amphibole-filled fractures, but the mineral phases that define compositional foliation change with proximity to the fractures. Nearest the fractures, the protomylonitic foliation is manifested by alternating layers of brown amphibole and plagioclase. With increasing distance (>1 cm) away from the fractures, the protomylonitic foliation is defined by alternating pyroxene and plagioclase layers, suggesting that amphibole replaces pyroxene.

A well-developed mylonitic foliation is defined by alternating layers of recrystallized plagioclase and pyroxene porphyroclasts in high-strain granulite grade mylonites (Figure 5a). Both clinopyroxene and orthopyroxene porphyroclasts are bent, resulting in folded cleavage planes and exsolution lamellae. These porphyroclasts typically show sweeping extinction and are mantled with equant recrystallized neoblasts (Figure 5a). Within the foliation, lozenge-shaped olivine porphyroclasts with sweeping extinction and subgrain development are sometimes observed (Figure 5b). Plagioclase is extensively recrystallized, and few porphyroclasts remain. The margins of elongate porphyroclasts contain subgrains similar in size to adjacent recrystallized grains (Figure 5c), indicative of regime 2–style subgrain rotation recrystallization [Hirth and Tullis, 1992]. The recrystallized grains (former subgrains) exhibit bulging grain boundaries, similar to microstructures indicative of regime 1–style bulging recrystallization. The recrystallized plagioclase neoblasts surrounding the elongate porphyroclasts locally form foliation oblique to the compositional banding (Figure 5d).

Figure 5.

Photomicrographs showing granulite grade textures in sample 648-04. (a) Compositionally banded foliation. (b) Asymmetric olivine fish. (c) Elongate plagioclase porphyroclast mantled by recrystallized grains define an oblique foliation. (d) Close-up photo from Figure 5c showing subgrain development (red arrow) along the margins of the porphyroclast.

4.3. Microstructures Developed During Amphibolite Grade Deformation

Upper amphibolite grade deformation is characterized by the assemblage plagioclase + brown amphibole ± pyroxene ± magnetite ± ilmenite ± sphene ± apatite. A well-developed compositionally layered mylonitic foliation is defined by alternating layers of recrystallized plagioclase and amphibole (Figure 6a). Amphibole porphyroclasts are commonly mantled by amphibole neoblasts. We also observe rare amphibole porphyroclasts with cores of relict pyroxene, and infer that the mantled amphibole porphyroclasts were originally pyroxene porphyroclasts (Figure 6b).

Figure 6.

Photomicrographs showing microstructures associated with upper amphibolite grade deformation. (a) Sample 465-03 contains a foliation defined by alternating bands of brown amphibole and plagioclase. (b) An amphibole porphyroclast with a core of relict pyroxene from sample 465-03. (c) In sample 645-10, an elongate plagioclase porphyroclast is microfaulted. (d) In sample 645-10, an elongate plagioclase porphyroclast exhibits core and mantle structure.

In upper amphibolite grade mylonites, the coarsest (>2 mm) plagioclase porphyroclasts are elongate with extensive undulose extinction and deformation twinning. Microfaults across these large porphyroclasts truncate deformation twins and locally exhibit neoblast development along their traces (Figure 6c). Large (∼1 mm), angular, elongate plagioclase porphyroclasts (Figure 6d) exhibit subgrain development in narrow (∼50 μm) zones along grain boundaries, and are surrounded by a matrix of very fine grained (∼10 μm) neoblasts. Matrix neoblasts are similar in size to the subgrains, and show bulging grain boundaries indicative of regime 1–style bulging grain boundary migration recrystallization [Hirth and Tullis, 1992].

Amphibole schists are also characteristic of upper amphibolite grade deformation. In these schists, abundant light brown amphibole exhibits a strong grain shape preferred orientation that defines the schistose foliation (Figure 7a). Fine-grained (∼50 μm) tabular amphibole does not display undulose extinction, deformation twins, bulging grain boundaries or other microstructural evidence of crystal plastic deformation (Figure 7b). Plagioclase, when present, is confined to foliation-parallel lensoid-shaped regions (Figure 7b). The plagioclase grains rarely form porphyroclasts; rather the grains are extremely fine grained (∼10 μm) and exhibit bulging grain boundaries indicative of regime 1–style bulging grain boundary migration recrystallization [Hirth and Tullis, 1992].

Figure 7.

Photomicrographs showing textures observed in amphibole schists. (a) Cross-polarized image of thin section from sample 648-17. The brownish-gold colored regions are amphibole-rich areas; the gray regions are dominated by plagioclase. The locations of Figures 7b, 7c, and 7e are designated by the black boxes. (b) A lens-shaped region of recrystallized plagioclase is surrounded by strongly oriented, tabular amphibole in sample 648-17. (c) A sphene-filled fracture cuts at high angle across the foliation. (d) Oxide-filled fractures (enclosed by black box) in amphibole schist sample 648-16 also cut at high angle to the foliation. (e) A partially healed fracture in sample 648-17 (enclosed by black box) cuts obliquely across the foliation. Sphene grains (red arrows) remain oriented in the trend of the former fracture.

Fractures oblique to the schistose foliation are partially filled by Fe-Ti oxide minerals and sphene (Figure 7c), and show little offset of the foliation across the fracture (Figure 7d). Locally, amphibole has grown parallel to foliation across the fractures that do not contain oxide minerals or sphene, and effectively conceal portions of the fracture trace. We infer the existence and orientation of former fractures by the presence of linear trails of fracture-filling minerals that are not incorporated into the schistose foliation (Figure 7e).

Middle amphibolite grade deformation is characterized by the assemblage plagioclase + brown amphibole ± olive green amphibole ± magnetite ± ilmenite ± sphene ± apatite. Mylonitic foliation is manifested by compositional layering of dynamically recrystallized plagioclase and brown amphibole, and the absence of relict pyroxene in the cores of amphibole porphyroclasts. Evidence for crystal plastic deformation of plagioclase and amphibole is largely similar to that observed in upper amphibolite grade mylonites, barring evidence for semibrittle deformation in the mylonites. Large (>2 mm) fragmented plagioclase porphyroclasts with parallel sets of amphibole-filled microfaults are common (Figure 8). These amphibole-filled fractures commonly cut across plagioclase porphyroclasts and their surrounding matrix of fine-grained recrystallized plagioclase, but not adjacent layers of amphibole neoblasts (Figure 8). We infer that the sinuous grain boundaries of amphibole neoblasts indicate crystal plasticity and grain boundary mobility leading to local fracture healing within amphibole layers. The crosscutting relationships imply multiple fracture events during dynamic recrystallization.

Figure 8.

Photomicrograph showing fabric observed in middle amphibolite grade mylonitic shear zone. In sample 460-09, amphibole-filled microfaults are cut by the matrix of recrystallized plagioclase grains surrounding the porphyroclast (red arrow at right). A different amphibole-filled fracture (red arrow at left) cuts across the plagioclase porphyroclast and the surrounding recrystallized matrix; it does not, however, cut across the finely recrystallized amphibole zone (red box).

Lower amphibolite grade mylonitic fabrics are defined by alternating layers of recrystallized olive green amphibole and plagioclase (Figure 9a). Synkinematic alteration of brown amphibole porphyroclasts to olive green amphibole is indicated by porphyroclast cores of brown amphibole being increasingly altered to olive green amphibole toward the rim, and are mantled by green amphibole neoblasts. Large (>2mm) amphibole grains form fragmented porphyroclasts offset along parallel microfaults (Figure 9b). Mantled amphibole porphyroclasts exhibit asymmetric wings of amphibole neoblasts that extend from porphyroclast wings, and help define the amphibole layers of the compositional foliation. Amphibole neoblasts exhibit bulging grain boundaries (Figure 9c) that resemble regime 1–style dynamic recrystallization [Hirth and Tullis, 1992], which we interpret to result from extensive amphibole recrystallization.

Figure 9.

Photomicrographs showing microstructures observed in lower amphibolite grade samples. (a) Plane light image of sample 460-09 showing well-developed, compositionally layered foliation, manifested by alternating bands of plagioclase and olive green amphibole. The larger amphibole porphyroclasts have brown cores and olive green rims, indicative of down-temperature synkinematic growth. (b) A large fragmented amphibole (after pyroxene) porphyroclast in sample 466-10 exhibits microfaults. (c) Recrystallized amphibole grains exhibit bulging grain boundary migration recrystallization structures in sample 460-09.

Amphibole-filled fractures that cut obliquely across lower amphibolite grade mylonitic foliation (Figure 10a) are lined with brown amphibole, whereas the fracture centers are filled with green amphibole (Figure 10b). Extensive foliation-parallel amphibole lies in the mylonitic foliation nearest the fractures, but decreases away from these planes. Lens-shaped regions of recrystallized plagioclase are enveloped by green amphibole adjacent to the fractures and by brown amphibole further away (>0.5 cm). At lowest amphibolite grade temperatures, the foliation development is defined by lens-shaped domains of recrystallized plagioclase enveloped by anastomosing bands of fine-grained amphibole (Figure 10c). The fine-grained domains of plagioclase resemble the matrices of recrystallized neoblasts observed around porphyroclasts in higher-grade mylonites. The abundant amphibole isolates the lens-shaped regions of recrystallized plagioclase, creating an interconnected network of amphibole.

Figure 10.

Microstructures observed in lower amphibolite grade samples. (a) Plane light image of mylonitic fabric in sample 647-17. Large amphibole-filled veins cut the mylonitic foliation. Amphibole growth in the foliation is more extensive within 0.5 cm of the fracture. The location of Figure 10b is designated by the black box. (b) The fracture walls are lined with brown amphibole; green amphibole fills the center of the fracture. Large amphibole porphyroclasts within the foliation are zoned with light brown amphibole in the core and green amphibole on the rim. (c) Cross-polarized image of spaced foliation in lowest amphibolite grade shear zone within sample 460-09. Lens-shaped regions of recrystallized plagioclase are enveloped by fine-grained green amphibole.

4.4. Microstructures Developed During Greenschist and Subgreenschist Grade Deformation

Greenschist grade metamorphic assemblages include green amphibole + epidote + chlorite + plagioclase ± sphene ± apatite. Brittle greenschist grade deformation is manifested by prehnite-, epidote-, and serpentine-filled fractures across granulite and amphibolite grade mylonitic fabric. Semibrittle greenschist grade deformation is manifested in the growth of amphibole, chlorite and epidote at the expense of plagioclase.

In samples 467-19 and 650-16, plagioclase cores are rimmed with fine-grained chlorite, epidote and amphibole (Figures 11a and 11b). These replacement textures are also observed in greenschist grade gabbro mylonites, though the alteration was complete or nearly complete, and the chlorite and epidote often show semibrittle or brittle deformation. Mylonitic foliations comprising alternating layers of amphibole porphyroclasts and fine-grained chlorite or epidote (Figure 11c), are inferred as relict high-temperature mylonites, based on the well-rounded, coarse-grained porphyroclastic texture of the amphibole grains. Comparison with compositional foliation in upper amphibolite grade mylonites suggests that the layers of chlorite and epidote in sample 650-16 were once layers of dynamically recrystallized plagioclase. Amphibole-chlorite schists (Figure 11d), contain fine-grained amphibole and chlorite that exhibit a strong grain shape preferred orientation, but little evidence for crystal plastic deformation of either mineral.

Figure 11.

Photomicrographs showing greenschist grade textures. (a) In undeformed gabbro sample 467-19, chlorite replaces plagioclase on the margins of a plagioclase grain. (b) In undeformed gabbro sample 650-16, epidote replaces plagioclase in grain centers. (c) A relict high-temperature mylonitic fabric is preserved in sample 645-11. Chlorite forms the matrix around an asymmetric amphibole porphyroclast. (d) In sample 645-12, fine-grained chlorite and amphibole define the schistosity.

Subgreenschist grade samples are variably altered and intensely deformed within the brittle regime. Clast- and/or matrix-supported fault breccias contain clasts of amphibolite grade mylonites (Figure 12a), amphibolite grade fault schists (Figures 12b and 12c), greenschist grade fault schists, and diabase. The amphibole and chlorite schist clasts retain their internal schistose texture defined by the grain shape preferred orientation of amphibole. Though plagioclase is extremely rare in the relict schistose foliation of these highly altered clasts, a few clasts preserve the compositional banding of amphibole and plagioclase characteristic of amphibole schists (Figures 12b and 12c). Pumpellyite and chlorite commonly make up to 50% of the matrix between clasts and give the fault breccias a green color in thin section and hand sample. Clast size varies from ∼2 mm to clay-sized particles, indicating substantial mechanical grain size reduction. In clast-supported breccias, visible offset can be observed across fractures. In matrix-supported breccias, clasts show rotation relative to one another.

Figure 12.

Photomicrographs showing fault breccia fabric development. (a) Sample 653-20 contains clasts of oxide gabbro mylonite; they are identified by oxide-rich relict foliation. (b) Sample 653-16 contains clasts of amphibole schist. The amphibole in the clasts is similar in appearance to that in amphibole schist samples 648-17 and 648-16. (c) Clasts of amphibole schist in sample 653-16 contain oxide-filled fractures at high angle to the relict foliation, also similar to textures in samples 648-17 and 648-16.

5. Mineral Chemistry and Thermometry

Mineral compositions (pyroxene, plagioclase, amphibole and chlorite) from 17 samples were analyzed to estimate temperatures associated with all grades of fabric development, and are listed in the auxiliary material.

5.1. Mineral Chemistry

Clinopyroxene is augitic in composition, with orthopyroxene more Mg-rich relative to Fe. Amphibole data define a linear trend from actinolite to pargasite compositions on a plot of AlIV cations versus alkali (Na + K) cations (Figure 13). Plagioclase composition varies over the entire data set from An88 to An7, whereas the average An composition for each sample varies from An61 to An17. The average, median, maximum, and minimum An composition for plagioclase compositions are shown in Table 3.

Figure 13.

Amphibole chemistry based on total alkalis (Na + K) versus Al(IV). The graph symbols are color coded by dive track: 172 (red), 645 (blue), 647 (pink), 648 (green), and 649 (black). The shape of the symbol is unique for each sample. Amphibole in samples 172-10, 172-24, 645-03, 645-10, 647-08, 647-17, 648-06, 648-07, 648-15, and 648-17 is actinolitic to pargasitic in composition. The amphibole in sample 649-11 varies in composition from tremolite to magnesio-hornblende.

Table 3. Summary of Plagioclase Analyses Statistics
Average An #171728462938453756474461
Median An #171735463038453639484461
Standard deviation4494792424446
Maximum An #253038713554524488535680
Minimum An #971639712412714303048
Number of analyses79802514033451623040808160

5.2. Thermometry

In situ samples exhibiting distinctive textures described above for granulite, amphibolite and greenschist grade deformation were selected for temperature estimates using two-pyroxene QUIlF, amphibole-plagioclase, and chlorite thermometry (auxiliary material), and help quantify the down-temperature evolution of detachment fault evolution.

Temperatures associated with granulite grade fabric development were estimated using two-pyroxene QUIlF thermometry on sample 648-04 (Figure 14). Two-pyroxene thermometry from orthopyroxene and clinopyroxene neoblast pairs indicates dynamic recrystallization of pyroxene occurred at temperatures ranging from 862 to 910 (±20)°C (Table 4).

Figure 14.

Regions of two-pyroxene QUILF thermometry analyses in sample 648-04. (a) Cross-polarized image of a portion of the thin section. (b) Cross-polarized image of area 4 (out of 7) pyroxene analyses area within the thin section (n = 53). (c) Cross-polarized image of area 5 region of pyroxene analyses (n = 40). (d) Cross-polarized image of area 6 region of pyroxene analyses (n = 29).

Table 4. Representative Pyroxene Analyses Used in QUIlF Thermometry Calculations
Study Area in Sample 648-04Number of AnalysesSiO2Al2O3TiO2Cr2O3MgOFeOMnOCaONa2OK2OTemperature (°C)Error
Clinopyroxene analyses average area 3n = 552.0531.6940.3980.00814.5298.8380.25220.8730.4350.002862±20
Orthopyroxene neoblast analyses average area 3n = 653.1030.7300.5940.01123.45520.8200.5280.8830.0180.012  
Clinopyroxene analyses average area 4n = 2151.7532.2080.5170.01514.4348.7810.24420.6700.4390.009874±14
Orthopyroxene neoblast analyses average area 4n = 1553.4540.9480.1850.01124.02820.0960.4740.8630.0170.005  
Clinopyroxene analyses average area 6n = 1651.5872.3180.5030.01514.3758.8110.26820.5380.4500.009910±20
Orthopyroxene neoblast analyses average area 6n = 953.0321.1830.1850.01523.63420.2280.4951.0320.0200.004  
Clinopyroxene analyses average area 7n = 152.3181.6300.3410.00014.6809.0940.24121.0840.3900.017885±20
Orthopyroxene neoblast analyses average area 7n = 1853.1811.0630.2280.01023.78120.1970.4900.9390.0170.010  

Amphibole-plagioclase thermometry was used to estimate temperatures of deformation preserved in amphibolite grade compositionally banded mylonites, mylonites with spaced foliation, and amphibole schists (see Tables 5 and 6). Though all of these fabric types contain dynamically recrystallized plagioclase, not all contain dynamically recrystallized amphibole. Consequently, dynamically recrystallized amphibole and plagioclase pairs and undeformed amphibole and dynamically recrystallized plagioclase pairs were analyzed. In the case of the undeformed amphibole and dynamically recrystallized plagioclase pairs, calculated temperatures likely reflect the temperature of amphibole growth. Average temperatures from the entire data set range between 592 to 818 (±40)°C.

Table 5. Representative Amphibole and Plagioclase Analyses
Amphibole analyses (wt %)
   Total cations15.6515.6215.4015.5215.5015.1615.8515.7315.6215.7815.7315.4315.9815.9415.7315.8115.7515.9215.5015.5915.3315.23
Plagioclase analyses (wt %)
   Total cations4.984.974.995.005.014.995.
   X An0.
Table 6. Amphibole-Plagioclase Thermometry Calculations Where Each Study Area Is a Fabric Element
SampleAmphibole TexturePlagioclase Texture, Foliation TypeT (°C)
AverageMedianStandard DeviationMaximumMinimumN
Entire Data Set
172-10no crystal plastic deformationcompositionally banded mylonite684.7692.255.6772.6495.666
172-24no crystal plastic deformationspaced mylonitic foliation662.8676.342.0725.9530.775
645-03dynamically recrystallizedspaced mylonitic foliation592.1609.080.8707.4445.516
645-10areas 1–6, dynamically recrystallized; areas 6a–8, no crystal plastic deformation; area 8, porphyroclastcompositionally banded mylonite729.8757.3114.5887.2208.2106
647-08no crystal plastic deformationspaced mylonitic foliation737.4745.946.1789.1616.820
647-17no crystal plastic deformationspaced mylonitic foliation725.7754.5123.2938.1341.537
648-06no crystal plastic deformationcompositionally banded mylonite817.8806.629.1886.4770.927
648-07no crystal plastic deformationspaced mylonitic foliation806.6805.857.4908.4618.240
648-15no crystal plastic deformationcompositionally banded mylonite767.2765.949.6858.9556.461
648-17no crystal plastic deformationamphibole schist721.2720.123.4771.9674.558
649-11no crystal plastic deformationspaced mylonitic foliation700.5698.532.9792.7631.159
Study Area 1A
172-10  678.8692.770.3772.6495.617
Study Area 1
172-24  636.2643.842.2710.9530.738
645-03  629.8614.053.9707.4552.69
645-10  622.1671.8145.0757.2374.310
647-08  763.0766.417.8789.1736.19
647-17  777.7778.720.1820.5748.48
648-06  827.7823.631.7886.4790.615
648-07  818.2827.570.1908.4618.220
648-15  776.5771.330.0835.4736.47
648-17  728.2722.723.0763.7700.118
649-11  690.0694.215.7719.7664.719
Study Area 1B
172-10  650.6657.044.9720.2547.318
648-15  759.6762.568.9858.9556.415
Study Area 1C
648-15  776.4788.754.4846.0585.519
Study Area 1D
648-15  760.8753.031.4824.3718.320
Study Area 2
172-24  690.1690.317.2725.9654.337
645-03  543.5526.586.9651.2445.57
645-10  751.4747.831.5794.2696.17
647-08  716.4727.852.0773.2616.811
647-17  823.6817.855.2938.1744.311
648-06  805.3803.720.6839.1770.912
648-07  795.0804.039.8844.1654.420
648-17  718.1718.223.2771.9674.540
649-11  664.3657.023.5718.4631.110
Study Area 2, Zoom 1
649-11  716.5713.723.6769.1678.110
Study Area 2, Zoom 1 Sub
649-11  687.7695.329.6712.1637.35
Study Area 2, Zoom 2
649-11  731.5728.029.0792.7682.115
Study Area 2A
172-10  719.4718.541.9767.8614.517
Study Area 2B
172-10  693.7696.035.8769.5645.814
Study Area 3
645-10  662.6665.662.2742.3580.37
647-17  580.2616.1134.4739.1341.510
Study Area 4
645-10  744.8758.134.4788.5686.316
647-17  713.1706.834.6754.5670.07
Study Area 5
645-10  677.2739.4146.4766.2327.810
Study Area 6
645-10  742.8752.427.1777.0682.710
Study Area 6a
645-10  658.3724.8200.0783.7208.27
Study Area 7a
645-10  797.9803.620.8821.4761.010
Study Area 7b
645-10  813.1814.528.0863.9773.99
Study Area 8
645-10  760.6790.4134.4887.2220.820

Estimated temperatures of mylonitization for gabbro sample 645-10 vary with fabric development associated with the early history of the detachment fault system; the compositionally layered protomylonitic fabric is overprinted by high-strain mylonitic shear zones (Figure 15). Amphibole and plagioclase porphyroclasts in the early protomylonitic fabric have an average temperature of 880 ± 40°C (Figure 15d). In contrast, recrystallized amphibole rimming amphibole porphyroclasts and recrystallized plagioclase defining the mylonitic fabric yields a lower average temperature of 795 ± 40°C. Temperature estimates from very fine grained recrystallized amphibole and plagioclase in folded mylonitic fabric developed under an average temperature of 665 ± 40°C. These crosscutting relationships and thermometry therefore demonstrate down-temperature deformation consistent with denudation by normal faulting.

Figure 15.

Regions of amphibole-plagioclase thermometry analyses in sample 645-10. (a) Cross-polarized image of the sample thin section. Regions of thermometry analyses are outlined with white boxes. (b) Fine-grained region of amphibole and plagioclase analyses (two yellow dots). In this region, n = 20. (c) Additional region of amphibole and plagioclase analyses, n = 20. (d) Coarse-grained region of amphibole and plagioclase analyses, n = 20.

Estimated temperatures associated with greenschist grade fabrics are based on chlorite thermometry. In sample 645-11, a static greenschist grade overprint of the high-temperature mylonitic foliation is manifested by near-complete replacement of plagioclase by chlorite. Chlorite within the foliation yields an average temperature of 328 ± 50°C (Figure 16). In sample 653-16, chlorite-altered clasts in fault breccia yield a slightly lower average temperature of 303 ± 50°C, implying continued down-temperature deformation/fracturing and fluid flow. Representative chlorite analyses and accompanying temperature calculations are shown in Table 7.

Figure 16.

Regions of chlorite thermometry analyses in sample 645-11, collected directly from the detachment fault surface on the seafloor. (a) Plane light image of sample 645-11. The box encloses the region of both sets of chlorite microprobe analyses. (b) Cross-polarized image of one region of chlorite analyses (n = 50). (c) Cross-polarized image of the other region of chlorite analyses (n = 50).

Table 7. Representative Chlorite Analyses
Weight percent
   Al (total)2.412.522.542.382.002.101.681.56
Temperature (°C)330338358308287296242365

6. Discussion: Strain Localization Processes

In this study, we use in situ submersible samples to investigate fabric development associated with the localization of strain along the Atlantis Bank oceanic detachment fault system in concert with microstructural analyses and thermometry to describe a protracted history of down-temperature deformation and strain localization. Microstructures initially developed in granulite grade mylonites exposed at Atlantis Bank suggest that plagioclase is the rheologically weak mineral phase during deformation. Extensive dynamic recrystallization of plagioclase contributed to the development of a fine-grained matrix with competent pyroxene and olivine porphyroclasts that have undergone limited recrystallization at temperatures between 862 and 910 (±20)°C. In rare instances, plagioclase porphyroclasts show textures suggestive of regime 2–style subgrain microstructures overprinting regime 1–style bulging grain boundary migration microstructures, implying continued recrystallization during decreasing temperature or increasing strain rate conditions, consistent with progressive strain localization [e.g., Kohn and Northrup, 2009]. An interconnected network of one mineral phase has been found an effective means of controlling the rheology of a sample [Jordan, 1987; Handy, 1989]; similarly, we suggest that the plagioclase matrix largely controls the rheology of the analyzed granulite grade mylonites. This result is consistent with studies of naturally deformed, high-grade mafic rocks where plagioclase is often the most easily deformed phase, leading to significant grain size reduction during dynamic recrystallization, implying a relatively weak rheology [Brodie and Rutter, 1985; Rutter, 1999]. Further, the presence of recrystallized anhydrous phases (pyroxene, olivine, and plagioclase) suggests a dry rheology for our samples, and that “dry” experimental flow laws for either plagioclase aggregates (<0.005% H2O) or two-phase anorthite-diopside aggregates (∼0.004% H2O) [e.g., Rybacki and Dresen, 2004; Dimanov and Dresen, 2005; Dimanov et al., 2007] may be representative of lower crustal deformation at Atlantis Bank for granulite grade temperatures, consistent with recent studies of Atlantis Bank granulite grade mylonites [Mehl and Hirth, 2008].

Based on observations of amphibole-filled fractures in granulite grade mylonites, fracture formation and hydrous fluid infiltration began at the transition from granulite to amphibolite grade temperatures (∼800°C) and continued through amphibolite grade temperatures (∼820–600°C), leading to a shift from plagioclase- to amphibole-accommodated strain localization with decreasing temperature. This was accomplished by the growth of hydrous amphibole at the expense of pyroxene followed by plagioclase, resulting in the modal reduction of the granulite grade rheology-controlling phase, and the increased modal abundance of amphibole. As water has a known weakening effect in shear zones [Tullis et al., 1996], we infer that the infiltration of hydrous fluids at temperatures near the granulite to amphibolite grade transition promoted strain localization at temperatures <800°C by first enhancing dynamic recrystallization of plagioclase and then its subsequent reaction to hydrous amphibole. This led to a shift from largely crystal plastic to semibrittle deformation mechanisms in plagioclase and amphibole.

The addition of hydrous fluids during upper amphibolite grade deformation first promotes replacement of pyroxene by secondary amphibole. Pyroxene porphyroclasts mantled by brown amphibole indicate increasingly hydrous conditions during upper amphibolite mylonitic fabric development. We find that the proportion of amphibole to plagioclase in upper amphibolite mylonites is largely similar to the proportion of pyroxene to plagioclase in granulite grade mylonites; these ratios suggest a direct replacement of pyroxene by amphibole. In lower amphibolite grade rocks, the ratio of amphibole to plagioclase exceeds the ratio of pyroxene to plagioclase in granulite or upper amphibolite grade rocks. With decreasing temperature at amphibolite grade conditions, the modal proportion of plagioclase decreases while the modal abundance of amphibole increases, suggesting that amphibole grows at the expense of both pyroxene and plagioclase (Tables 1 and 2).

The growth of abundant amphibole in the mylonites requires an influx of hydrous fluids channeled into the shear zones, as well as a mechanism to keep fluid supplied to the shear zone for the duration of amphibolite grade fabric development. The presence of fractures that are either partially healed by bulging dynamic recrystallization of plagioclase at upper amphibolite grade temperatures, and those that are partially or fully healed by growth of secondary amphibole grains at lower amphibolite grade temperatures suggests that the fractures are synkinematic with respect to mylonitic fabric development, and were continuously generated and healed during progressive, down-temperature amphibolite grade deformation [e.g., McCaig and Knipe, 1990; Knipe and McCaig, 1994]. The dynamic process of synkinematic fracture generation, fluid infiltration, and subsequent healing is therefore a viable mechanism for supplying fluids to the detachment fault shear zone.

The change in foliation geometry from compositional to spaced to schistose is largely the result of increased, down-temperature growth of amphibole. In samples with foliation, growth of amphibole around domains of dynamically recrystallized plagioclase suggests that hydrous fluids wetted grain boundaries between recrystallized plagioclase grains, resulting in disruption of the interconnected plagioclase grain networks that effectively localize strain at higher temperatures. Taken with the decrease in modal plagioclase and progressively lower temperatures, the anastomosing amphibole network reduces the influence of plagioclase rheology on the localization of strain by isolating plagioclase, and pinning individual plagioclase grain boundaries, preventing regime 1 grain boundary migration recrystallization, and further crystal plastic deformation of plagioclase. In schistose samples, the near-absence of plagioclase, the absence of evidence for crystal plasticity in amphibole, and the strongly oriented, foliation-defining amphibole grains together are consistent with microstructural evidence for dissolution-precipitation creep processes in amphibole. Similar microstructures have been observed in amphibole in both experimentally and naturally deformed samples, and they have been used to demonstrate that an amphibole lattice preferred orientation can develop without the operation of crystal plastic processes during reaction-controlled dissolution-precipitation creep [Bons and den Brok, 2000; Imon et al., 2004; Díaz Aspiroz et al., 2007; Miranda, 2006]. The evolution in foliation geometry during amphibolite grade deformation suggests a transition from dynamic recrystallization-accommodated strain localization in plagioclase to dissolution-precipitation creep processes in amphibole. Further, these results imply a wet rheology for our samples at upper amphibolite grade temperatures where plagioclase still deforms under dislocation creep, and that “wet” (≥0.07% H2O) experimental flow laws for plagioclase aggregates [e.g., Dimanov et al., 1999; Rybacki and Dresen, 2000; Offerhaus et al., 2001; Rybacki and Dresen, 2004; Rybacki et al., 2006] may be more representative of lower crustal deformation at Atlantis Bank at these temperatures, consistent with other studies of amphibolite grade mylonites from Atlantis Bank [Miranda, 2006].

At greenschist grade conditions, the proportion of plagioclase in recovered fault rocks is further diminished; with the addition of Mg-rich fluid, reaction of plagioclase to chlorite contributes to the formation of well-foliated, fine-grained chlorite-rich fault schists. There is little evidence of crystal plastic deformation of amphibole or chlorite in the schists, despite well-developed foliation. Our thermometry shows that chlorite growth occurs at lower temperatures than amphibole growth, and we interpret this as evidence that chlorite dissolution-precipitation creep occurs after amphibole-accommodated dissolution-precipitation creep. Both brown and green amphibole-filled fractures in lower amphibolite grade rocks, and chlorite- and prehnite-filled fractures in greenschist grade rocks demonstrate that synkinematic fracture development and fluid infiltration continued with decreasing temperature. Strain localization along the detachment fault at lower amphibolite and greenschist temperatures is achieved by dissolution-precipitation creep growth of amphibole and chlorite parallel to foliation in the presence of hydrous fluids.

Below greenschist grade conditions, brittle deformation dominates strain localization. Breccias from the detachment surface are dominated by clasts of amphibole and chlorite schist; they rarely contain clasts of gabbro mylonite, oxide gabbro mylonite, and undeformed diabase. Most of the breccia clasts are therefore derived from rocks deformed within the ductile and semibrittle regimes. Mylonitic and schistose textures in the clasts indicate semibrittle and ductile deformation occurred at temperatures from ∼600 to ∼850°C. Subgreenschist grade brecciation overprints the higher-temperature ductile and semibrittle deformation, with matrix chlorite growth in fault breccias at temperatures between ∼250–350°C. Either brecciation initiated at temperatures above or similar to this temperature range in order for chlorite to grow in the matrix, or chlorite growth resulted from circulation of fluids after the brecciation was complete.

Down-temperature strain localization within the ductile, semibrittle, and brittle regimes associated with initiation and slip on the detachment fault system at Atlantis Bank is summarized as follows (Figure 17): (1) At granulite grade conditions (∼910–862°C), dynamic recrystallization of plagioclase dominates the initial localization of strain. (2) At amphibolite grade conditions, strain is accommodated by both dynamic recrystallization of plagioclase and dissolution-precipitation creep of amphibole at temperatures ∼820–600°C. (3) At greenschist grade conditions (∼450–300°C), strain is localized by reaction softening due to growth of chlorite. (4) At subgreenschist grade conditions (<300°C), strain is localized by brittle fracturing and cataclasis.

Figure 17.

Schematic diagram highlighting the temperature ranges for the deformation mechanisms that promoted strain localization along the Atlantis Bank detachment fault.

7. Distribution of Deformation Fabrics

7.1. Variations of Fabric Intensity With Rock Type

Though we document the onset of strain localization at near-solidus granulite grade temperatures with development of mylonitic fabrics through amphibolite grade temperatures, the prevalence of crystal plastic fabrics in a range of gabbroic compositions suggests composition-influenced strain localization. We classify the in situ samples as either nonoxide gabbro or oxide gabbro (>0.5% oxide) in order to compare fabric intensity with rock type. We designate gabbro, olivine gabbro, and gabbronorite as “nonoxide gabbro,” and classify oxide gabbro, oxide olivine gabbro, and olivine oxide gabbro as “oxide gabbro.” The highest intensity of crystal plastic fabric development is observed in both nonoxide gabbro and oxide gabbro rock types (Figure 18). Crystal plastic fabric intensity does not appear to correlate with primary rock composition, or structural depth beneath the most recent slip surface. Based on the samples investigated, strain is not preferentially localized in either nonoxide gabbro or oxide gabbro rock compositions.

Figure 18.

Plot of mylonitic fabric intensity versus rock composition. The nonoxide gabbro and oxide gabbro suites show similar intensities of fabric development, suggesting that strain is not localized preferentially into a particular gabbro composition.

7.2. Variations of Fabric Intensity With Structural Depth

We estimated the structural depth of each sample beneath the detachment fault surface to compare the thickness of ductile, semibrittle and brittle deformation zones across the dissected dome of Atlantis Bank. Due to truncation and displacement of the detachment surface by several transform- and ridge-parallel normal faults (Figure 2), sample depth (collected by submersible) cannot be equated directly with true structural depth beneath the detachment fault surface. We used bathymetric maps, dive sample location projections into “dive profiles,” or vertical slices through the bathymetry, and submersible observations to identify exposures of the relict detachment fault surface. The estimated structural depth of each sample beneath the detachment fault was made by measuring the orthogonal distance between this contoured surface and the sample [e.g., Schroeder and John, 2004]. The dive profiles and accompanying structural depth calculations are shown in the auxiliary material.

Footwall rocks exhibit crystal plastic fabrics distributed over a structural thickness up to 400 m below the denuded fault surface exposed at the seafloor, and concentrated within the upper 120 m of this zone. Crystal plastic fabrics vary in intensity with depth, demonstrating localization of ductile deformation into anastomosing shear zones up to tens of meters thick (Figure 19). The presence of high-strain (fabric intensity > 4) fabrics in granulite and upper amphibolite grade rocks shows that strain localization did occur, and was accommodated by dynamic recrystallization of plagioclase within the ductile regime. However, not every granulite grade and upper amphibolite grade shear zone is overprinted by successively down-temperature deformation. Some granulite grade mylonites exhibit only high-temperature assemblages, preserving microstructures indicative of dynamic recrystallization in pyroxene, olivine, and plagioclase. Similarly, some amphibolite grade mylonites exhibit only granulite and amphibolite grade assemblages and microstructural evidence of dynamic recrystallization of pyroxene, brown amphibole, and plagioclase. These observations suggest that strain was localized on a few of the high-temperature shear zones with decreasing temperature, such that many shear zones were left with microstructures developed during high-temperature deformation, and not subsequently deformed.

Figure 19.

Plot of average strain intensity with structural depth where n = number of samples within a given 10 m bin of structural depth. Brittle, semibrittle, and crystal plastic strain intensities are shown as red, blue, and green bars, respectively. Average strain intensity is calculated by the sum of the strain intensities in a given 10 m bin divided by the number of samples in the bin. Ductile deformation is distributed over the 400 m thickness of the shear zone at higher strain intensities than either semibrittle or brittle deformation and is concentrated in the upper 120 m of the detachment fault shear zone. Semibrittle deformation is distributed over the 400 m thick zone beneath the detachment fault but concentrated in the upper 80 m. Brittle deformation is distributed over the upper 110 m of the 400 m thick zone beneath the detachment fault but concentrated in the upper ∼30 m.

Semibrittle fabrics are also distributed over a structural thickness up to 400 m beneath the fault, but concentrated in the upper 80 m of this zone (Figure 19). The greatest intensity of semibrittle fabric development occurs in amphibole schists and amphibole-chlorite schists; these rocks crop out in a zone up to ∼20 m thick directly beneath the detachment fault slip surface (Table 1). Based on textures observed in this study, growth of amphibole in the presence of hydrous fluids lead to the development of amphibole schists. As the schists must preferentially develop in a region of high fluid flow, the incipient detachment fault zone likely was the conduit for the fluids, as fluid flow can be enhanced through fracturing and increased porosity associated with faulting. This is corroborated by the restricted occurrence of amphibole and chlorite schists immediately below (within a few tens of meters) the detachment surface. Similarly, amphibole, chlorite, and talc schists have been identified along and immediately below detachment fault surfaces along the Mid-Atlantic Ridge [MacLeod et al., 2002; Escartin et al., 2003; Schroeder and John, 2004; Blackman et al., 2006; Boschi et al., 2006; McCaig et al., 2007].

The greatest magnitudes of brittle fabric intensity occur in fault breccias, locally exposed at the seafloor associated with the principal slip surface. These exposures are characterized by fault breccias that contain clasts of amphibole and chlorite schists, gabbro mylonites, oxide gabbro mylonites and diabase. The restricted occurrence of fault breccias along the detachment surface indicates that the zone of brittle deformation is not appreciable (<30 m), or may have been eroded during probable subaerial exposure of the fault surface [Shipboard Scientific Party, 1999b]. The clasts contain fabrics indicative of high-temperature deformation in the ductile and semibrittle regimes. The overprinting of high-temperature fabrics by progressively lower-temperature brittle fabrics along the detachment fault, and the structural juxtaposition of brittle rocks above ductile rocks are both traits that are consistent with normal sense motion. These relationships are shown in a schematic diagram for the top 100 m of the fault system (Figure 20).

Figure 20.

Schematic section through the upper 100 m of the detachment fault shear zone. The upper portion of the shear zone comprises high-temperature mylonitic shear zones overprinted by amphibole and chlorite schists; both of these rock types are incorporated into fault breccias immediately below the detachment fault slip surface.

8. Comparison of Strain Localization Processes With Results From ODP Hole 735B

In ODP Hole 735B, a ∼100 m thick mylonitic shear zone is developed in gabbroic rocks at the top of the core, though other mylonitic shear zones are also found downhole (Figure 3) [Cannat, 1991; Cannat et al., 1991; Shipboard Scientific Party, 1999a; Dick et al., 2000; Natland and Dick, 2002]. Granulite grade fabrics in the upper 70 m of the shear zone are overprinted by amphibolite grade fabrics in the interval 0–30 mbsf, and are further overprinted by greenschist grade fabrics in the intervals ∼5–10, 15–20, 45–50, and 60–70 mbsf [Stakes et al., 1991]. The recovery of fault breccia clasts within the upper 25 m of the core suggests the presence of a brittle fault zone at the top of Hole 735B. Overprinting fabric relations demonstrate that the shear zone at the top of Hole 735B was progressively overprinted by thinner, lower-grade shear zones, indicating that strain in the main shear zone associated with the detachment fault was localized with decreasing temperature [Dick et al., 1991b; Stakes et al., 1991].

The detachment fault system at Atlantis Bank is suggested to have rooted at a melt-rich zone based on the presence of hypersolidus magmatic fabrics overprinted by subparallel granulite grade crystal plastic fabrics [Cannat, 1991; Cannat et al., 1991; Dick et al., 1991b; Shipboard Scientific Party, 1999b; Dick et al., 2000]. Many zones of crystal plastic deformation coincide with zones of magmatic foliation development, suggesting a continuum between the onset of deformation in the presence of melt and progressive subsolidus deformation [Shipboard Scientific Party, 1999b; Dick et al., 2000]. Of the 4 core complexes sampled by both drill core and submersible, the Atlantis Bank detachment fault system is unique because (1) both ductile and brittle fault rocks are derived from gabbroic rocks and developed during down-temperature deformation from hypersolidus to subgreenschist grade temperatures and (2) there is little evidence for either high- or low-temperature deformation associated with detachment faulting in serpentinized peridotites exposed across ∼39 km parallel to the spreading direction [Arai et al., 2000; Kinoshita et al., 2001; Matsumoto et al., 2002]. The down-temperature rheology of gabbroic rocks is therefore an important influence on the strain localization processes associated with detachment faulting at Atlantis Bank.

Downhole core description from ODP Legs 118, 176 (Hole 735B), and 179 (Hole 1105A) suggest that the main shear zone associated with detachment faulting is ∼100 m thick and that within this zone, the deformation fabric intensity (based primarily on crystal plastic fabric development) increases upward toward the top of Hole 735B, and toward the detachment fault surface (Figure 3) [Cannat et al., 1991; Shipboard Scientific Party, 1999a; Dick et al., 2000; Casey et al., 2007]. Numerous granulite and amphibolite grade crystal plastic shear zones distributed within and below the main shear zone indicate that a thick zone of distributed crystal plastic deformation developed during early, high-temperature detachment faulting [Cannat et al., 1991; Shipboard Scientific Party, 1999b; Dick et al., 2000; Mehl and Hirth, 2008]. An upward increase in the volume percent of amphibole veins has been documented within the upper 200 mbsf of Hole 735B [Dick et al., 1991a]; as intense fracturing and subsequent growth of amphibole are indicative of semibrittle deformation, the upward increase in amphibole content is a manifestation of an upward increase in semibrittle deformation. The low (36%) core recovery within the upper 25 m of Hole 735B [Cannat et al., 1991] likely reflects the difficulty associated with sampling brittle fault rocks, including incohesive fault breccias from the detachment fault surface exposed at the seafloor. Fragments of brecciated rock were recovered in the core from this interval, indicating that brecciation associated with the main shear zone of the detachment fault (0–100 mbsf) was likely limited to the upper 25 m of ODP Hole 735B [Shipboard Scientific Party, 1989].

In the submersible samples discussed, a thick zone of ductile deformation associated with detachment faulting is inferred from the depths at which the samples were acquired. While the intensity of ductile fabric development varies irregularly with depth, the intensity of semibrittle and brittle fabric development increases toward the detachment surface. Therefore, the upward increase in deformation intensity toward the detachment fault surface is a pattern common to both drill core and submersible sample sets.

Initiation of detachment faulting at hypersolidus conditions and the tectonic evolution of the detachment fault system through granulite and amphibolite grade conditions as documented in ODP Hole 735B [Cannat et al., 1991; Mehl and Hirth, 2008] and 1105A [Shipboard Scientific Party, 1999a] is largely corroborated by the textures observed in the submersible samples. Mehl and Hirth [2008] show that granulite grade deformation in ODP Hole 735B occurred at temperatures ranging from 808 ± 14°C to 946 ± 69°C; our granulite grade thermometry calculations for the submersible samples (862 ± 20°C to 910 ± 20°C) are within this range. Though there is evidence for early, high-temperature dynamic recrystallization of plagioclase leading to strain localization at granulite and amphibolite grade temperatures, strain does not appear to be preferentially localized into oxide-rich gabbros at granulite grade and lower temperatures in the submersible samples. In Figure 19, high (>4) ductile fabric intensity values are associated with olivine gabbro mylonites, gabbro mylonites and oxide gabbro mylonites. Further, not all oxide gabbros have fabric intensity values >4. In addition, oxide gabbro mylonites with high-intensity ductile fabric development do not exhibit the complete down-temperature spectrum (from granulite grade temperatures through subgreenschist grade temperatures) of overprinting relationships. Oxide gabbros do not develop deformation textures below amphibolite grade temperatures, suggesting that strain is localized into nonoxide gabbros below temperatures of ∼450°C (Table 1). Amphibole schists rarely contain abundant oxide, and fault breccias contain very few clasts of oxide gabbro mylonite, both suggesting that oxide gabbros do not strongly influence strain localization processes at lower amphibolite, greenschist, and subgreenschist grade conditions.

Finally, results of this work, taken with recent thermochronologic, rheologic, and geophysical studies from the Atlantis Bank detachment, permit estimation of the duration of detachment faulting, strain rates associated with high-temperature deformation, and detachment fault slip magnitude. Cooling rates over Atlantis Bank are interpreted from both ODP 735B core samples and in situ submersible samples using the Ti-in-zircon crystallization temperature (∼850 ± 50°C [Grimes et al., 2007]), the Curie temperature associated with magnetic anomalies (∼540°C [Baines et al., 2008]), and 40Ar/39Ar biotite closure temperature (∼365°C [John et al., 2004]). The range in closure temperatures (∼850–365°C) is similar to the range of deformation temperatures interpreted from mineral thermometry outlined in our study. Geochronologic and thermochronologic ages associated with this temperature range (850–365°C) indicate rapid cooling (>800°C/myr) between ∼11.99 ± 0.05 Ma [Baines et al., 2009] and 11.46 ± 0.35 Ma [John et al., 2004]. Based on these results we suggest that strain localized rapidly at granulite to greenschist grade temperatures within ∼500 kyr of igneous crystallization, consistent with conductive cooling models at depths ranging from 0 to 1500 mbsf [Schwartz et al., 2009].

Rheologic studies of mylonites from Atlantis Bank also suggest rapid strain localization. Mehl and Hirth [2008] examine granulite grade gabbro mylonites from ODP Hole 735B and document dislocation creep and grain boundary sliding accommodated diffusion creep in plagioclase and pyroxene at ∼900–850°C under strain rates of 10−12 to 10−11 s−1. Miranda [2006] examined amphibolite grade gabbro mylonites from in situ submersible samples, and documents a transition from dislocation creep to diffusion creep in plagioclase at ∼750–670°C under strain rates of 10−12 to 10−11 s−1. Given that rates of detachment faulting at Atlantis Bank are equivalent to the half-spreading rate between the Antarctic and African plates (∼14 +1.8/−1.5 km/Myr), and accounted for up to ∼80% of the full plate spreading [Baines et al., 2008], the estimated strain rates at granulite and amphibolite grade temperatures are equivalent to the half-spreading rate accommodated on a <100 m wide shear zone, consistent with the observed thickness of the main detachment fault shear zone recovered from ODP Hole 735B [Miranda, 2006]. Using the half-spreading rate outlined above [Baines et al., 2008], the ∼500 kyr duration of fabric development corresponds to ∼7 km of slip along the detachment fault at strain rates of at least 10−12 to 10−11 s−1. Assuming an initial fault dip of 50–70° equivalent to that outlined by deMartin et al. [2007] based on microseismicity for the active TAG detachment fault along the Mid-Atlantic Ridge, we estimate that granulite to greenschist grade deformation and associated strain localization occurred within ∼2.5 to 4.5 km of the ridge axis. This estimate is within the boundaries of the modern axial valley, interpreted to be ∼5–10 km wide based on bathymetry and magnetic anomalies [Hosford et al., 2003]. The rapid localization of strain implied by combined thermochronology and mineral thermometry suggests that the detachment fault system at Atlantis Bank was initiated and strain was localized over hypersolidus, granulite and greenschist grade temperatures within the confines of the axial valley along the Southwest Indian Ridge in <500 kyr.

9. Conclusions

Microstructural observations and mineral thermometry from in situ samples collected from the Atlantis Bank oceanic core complex indicate that detachment faulting was initiated under hypersolidus conditions in the ductile regime, and continued to subgreenschist temperatures through the semibrittle and brittle regimes with progressive strain localization. The down-temperature sequence of fabric development is manifested within dominantly gabbroic rocks, suggesting that the development of the Atlantis Bank detachment fault system cannot be explained by models for detachment faulting that require strain partitioning between crust and mantle rocks [e.g., Escartin et al., 2003; Ildefonse et al., 2007]. Initiation of detachment faulting at hypersolidus conditions and the extensive high-temperature crystal plastic fabric development in gabbroic rocks at near-solidus temperatures demonstrates that the onset of core complex development was coeval with magma crystallization, inconsistent with models that require strain localization dominantly within the brittle regime along a low-temperature alteration front [e.g., Escartin et al., 2003]. Our model for detachment faulting at Atlantis Bank involves reaction-driven changes in gabbro rheology that promote down-temperature changes in deformation mechanisms. A key feature is development of synkinematic fractures as a mechanism for enhancing reaction of both pyroxene and plagioclase. The onset of synkinematic fracturing at ∼800°C promotes the following sequence of deformation mechanisms that enable strain localization: plagioclase-accommodated dislocation creep at temperatures between 910 and 650°C, amphibole-accommodated dissolution-precipitation creep at temperatures ∼750–450°C, chlorite-accommodated reaction softening at temperatures ∼450–300°C, and brittle fracturing and cataclasis at temperatures <300°C.

The initiation of detachment faulting at near-solidus conditions and the tectonic evolution of the detachment fault system through granulite and amphibolite grade conditions as documented in ODP Hole 735B [Cannat et al., 1991] and 1105A [Shipboard Scientific Party, 1999a] is largely corroborated by the textures observed in the submersible samples. In addition, the progressive overprinting of the 100 m thick main shear zone at the top of ODP Hole 735B by successively thinner and lower-grade shear zones is also similar to the pattern of overprinting relationships observed in the ∼120 m thick main shear zone interpreted from the submersible samples. However, in contrast to the ODP 735B samples, crystal plastic deformation does not appear to be preferentially localized into oxide-rich gabbros at granulite grade and lower temperatures in our sample set. Together, the in situ submersible samples from this study and core from ODP Holes 735B and 1105A show that there is little evidence for fabric development and strain localization in serpentinized peridotites, demonstrating the lateral continuity of strain localization mechanisms developed within gabbroic rocks exposed in the footwall over a distance ∼39 km parallel to the spreading direction.


We thank the captain and crew of the Yokosuka, Shinkai 6500, Kairei, and Kaiko and their respective shipboard scientific parties. We thank B. Ron Frost for assistance with petrographic analysis, advice about the QUILF thermometer, and discussions about the relationship between textures and metamorphic reactions; Susan Swapp for assistance with electron microprobe analyses; Graham Baines for the GMT maps of Atlantis Bank (Figures 1 and 2); Henry Dick for the use of samples 172-10 and 172-24; Graham Baines, Mike Cheadle, Greg Hirth, and Joshua J. Schwartz for discussions during the project; and Graham Baines and Joshua J. Schwartz for initial reviews of the manuscript. Funding for this work was provided by a Wyoming NASA space grant and a California State University Northridge probationary faculty grant to E. A. Miranda and NSF OCE grant 0352054 to B. E. John and M.J. Cheadle. We thank Donna Blackman and Andrew McCaig for constructive reviews that helped improve our manuscript.