We have already discussed the possible mode of emplacement of the Atlantis Massif and largely agree with the interpretation of Dick et al. (1991, 2002) that the massif was unroofed as a core complex at the northern RTI between the Atlantis II Transform and the Southwest Indian Ridge. The discussion that follows represents an analysis of the structural features of the core from microstructures to mesoscopic structures. Miller et al. (Chap. 3, this volume) described successfully mapping structural features in the core using FMS logging images and correlating features with the structural and lithologic descriptions in the VCDs. We take two approaches that include examination of the microstructure of each sample and identification of structures within the core.
Microstructural analysis of Hole 735B gabbros during ODP Leg 118 (Robinson, Von Herzen, et al., 1989; Dick et al., 1991) is interpreted to indicate that some gabbros were deformed in the crystal mush stage before they completely crystallized. This magmatic deformation is believed to have produced foliation and a strong plagioclase crystallographic fabric. Later, solid-state deformation occurred at progressively lower temperatures and increased availability of hydrothermal water, probably as the gabbros moved away from the ridge axis (Cannat, 1991). Microstructural studies showed a sharp change of deformation processes in the gabbros as the temperature decreased from the stability conditions of granulite-facies metamorphic assemblages to the stability conditions of amphibolite-facies metamorphic assemblages. Temperature-dependent diffusion processes may have controlled the deformation of all minerals in granulite-facies metamorphic conditions. Disorganized plagioclase and olivine fabric patterns characterize this deformation.
Dick, Natland, Miller, et al. (1999), during Leg 176 on the same gabbroic massif, suggested that deformation was localized from hypersolidus to low-temperature subsolidus conditions and affected the migration of late crystallizing melts. They reported that thick intervals of the core in Hole 735B are either isotropic or contain local domains with weak to moderate magmatic foliations, often overprinted by a weak parallel crystal-plastic fabric. Further, they indicated that magmatic and crystal-plastic foliations have a consistent strike over the entire core but show no systematic distribution or change in dip with depth.
Similar microstructural study in gabbroic shear zones from the MARK area (Agar et al., 1997) revealed that dynamic and static recrystallization in gabbroic rocks occurred at relatively high temperatures, probably >700°C. Deformation was accommodated primarily by dislocation creep. It is possible that some recrystallization of plagioclase occurred at lower temperatures (>450°C), but more evidence of synkinematic cracking of olivine and pyroxene during bulk deformation at these temperatures would be expected. Therefore, it was suggested that most of the deformation in the samples ceased under upper amphibolite facies conditions during which brown amphibole was precipitated around pyroxene recrystallized grain boundaries. During this high-temperature deformation, crosscutting relations indicate that late-stage magmatic fractionates were mobilized into deformed regions (that may have still been deforming). Possible synkinematic shearing at hypersolidus conditions could have generated the crystallographic preferred orientations observed in leucocratic veins. The introduction of fractionated melts apparently changed the bulk composition of the deformed (or deforming) regions.
Dick, Natland, Miller, et al. (1999), in describing the results of Leg 176, suggested that the most fractionated oxide gabbros are generally systematically associated with shear zones, and deformation-driven differentiation may explain the apparent association of crystal-plastic deformation with oxide-rich gabbroic rocks. In this hypothesis, shear zones propagate through a crystal-rich mush and act as low-stress conduits that assist in the migration of fractionated melts into the upper portions of the crust (Dick et al., 1991, 1992). To test this model they tried to evaluate whether a correlation between crystal-plastic fabric intensities and the degree of fractionation as measured by whole-rock Mg# exists. If deformation drives the differentiation, then the more deformed samples should have lower Mg#s, reflecting their more fractionated character. They found only a very weak correlation, at best, between deformation and differentiation.
An alternative hypothesis is that late-crystallizing intercumulus oxide-rich melts behaved as deformation localization zones. The presence of melt lowers the effective normal stress by the amount of pore fluid pressure, and a decrease in the normal stress may cause the rock to cross the failure envelope and crack along shear planes. Melts can then migrate along shear fractures.
Strain localization could also be evidenced under subsolidus conditions by the presence of more abundant oxide minerals, which are generally considered the weakest mineral under crystal-plastic conditions. In addition, oxide minerals are likely to behave in a ductile manner to temperatures of ~450°C (Agar et al., 1997), significantly lower than the silicate phases would remain ductile. Thus, significant oxide abundances could significantly lower the ductile strength and help to localize deformation without the existence of a melt phase.
A representative suite of deformation textures in gabbroic shear zones can be observed in a wide variety of modal compositions. Shear zones range from isolated structures at the centimeter scale to >1 m thick. The main scope of this study included
Also, it is emphasized that correlation between qualitative estimates of finite strain and modal composition are of limited value unless the spatial relationships of strain localization to compositional and textural variations are carefully documented. FMS data were used to constrain these spatial relationships. Also, the scale of observation can change the modal composition estimates of some samples (e.g., thin section vs. core); therefore, an integrated observation of a wide range of scales (microscopic, mesoscopic, and macroscopic) is required to understand the influence of modal composition and primary texture on strain localization.
Textures of the plutonic rocks were characterized on the basis of grain shape, mutual contacts, and preferred mineral orientation. Rock textures such as equigranular, inequigranular, intergranular, and granular were used to describe the overall texture of each lithologic interval. Poikilitic, ophitic, subophitic, and interstitial textures were distinguished according to the predominant grain shapes in each interval. The textural type attributed to each thin section is not absolutely representative of the whole thin section but represents only the predominant texture of that particular thin section (i.e., >60% of the thin section demonstrates a particular texture). There are examples where two or three different textural types can be identified in each thin section, but only the predominant one is selected as the representative texture. For this study, parameters such as the development of microstructures associated with crystal plasticity (e.g., subgrains, neoblasts, porphyroclasts, sutured grain boundaries, deformation twins, kink bands, percentage of recrystallization subgrains formed, recrystallized grain sizes, degree of shape-preferred orientation, and extent of crystallographic preferred orientation) were recorded for each thin section and used to classify textural types. The textural types outlined here represent only those samples used in this study and do not represent the entire spectrum of deformation textures within Hole 1105A. Similar textures in different samples do not necessarily imply that the samples followed the same deformation paths because there are variable starting protolith textures and compositions and there may be different extents of hypersolidus to solid-state deformation that lead to the same textural type. Igneous lamination can be taken as an indicator of possible hypersolidus processes such as laminar flow or in situ crystallization. The presence of inclusions, overgrowths, and zonation was noted, and the apparent order of crystallization was suggested in the comments section for samples with appropriate textural relationships. The presence and relative abundance of accessory minerals such as Fe-Ti oxides were noted, and modal analysis allowed clear differentiation of the rocks based on oxide modal abundances. The percentage of alteration was also recorded. Deformation textures or categories were assigned to each thin section. The eight textural categories assigned generally reflect increasing strain from 1 to 8, 1 indicating no strain and 8 indicating maximum strain, typically within a mylonitic shear zone. The increase in strain from one category to the next is not by any means linear, and we emphasize that the measure of deformation extent is simply tied to textural categories that we outline below. We used the textural type as a qualitative measure of the extent of deformation downhole and to examine the influence of modal mineralogy on the extent of deformation in the sample. We emphasize that the textural categories are designed to provide qualitative information on the extent of strain and recrystallization downhole.
No foliation or lineation is apparent in thin sections or core samples. Crystallographic preferred orientation is not evident with the gypsum plate inserted. Evidence of recrystallized grains of plagioclase or olivine, if observable, is <5% and <2%, respectively. No evidence of clinopyroxene recrystallization can be observed. Intact igneous texture with subhedral to euhedral morphologies and straight exsolution lamellae in clinopyroxene grains indicate that they are relatively undeformed. Some plagioclase grains show local tapering deformation twins but many have blunt ends typical of igneous twins. No kink bands in plagioclase are detected. Although the thin section dominantly appears unaltered and undeformed, some small percentage of alteration, especially within the olivine crystals, can be observed in most thin sections. Most of the textures within this category have granular texture with different grain sizes ranging from fine to coarse grains with euhedral to subhedral mineral shapes.
Figure F39A represents a medium- to coarse-grained olivine gabbro with equigranular texture. Plagioclase grains are elongated to equant, clinopyroxene grains are elongated, and olivine grains have an amoeboidal appearance. Figure F39B is a magnified part of Figure F39A. Note the subhedral shape of olivine grains and the straight plagioclase polysynthetic twins. Also, no marginal alteration or bulging is detected around the grain boundaries. Embayment of long plagioclase laths in large clinopyroxene and olivine grains delineates a local subophitic texture in the sample. Little alteration exists in this section, and most of the minerals appear unaltered. Some plagioclase shows strong zoning, but a lack of zoning is typical. Also, undulose extinction can be observed in plagioclase laths, but no significant tapering is observed in plagioclase twins. Very straight exsolution lamellae in clinopyroxene grains at the bottom of the thin section can be clearly identified.
Figure F39C shows a coarse-grained equigranular olivine gabbro with elongated plagioclase laths, equant clinopyroxene grains, and equant olivine grains. Figure F39D is a magnified portion of Figure F39C, where fresh clinopyroxene minerals with no signs of deformation are clearly present. This sample is one of the least deformed thin sections of the whole collection. Even the hexagonal euhedral shapes of the clinopyroxene crystals are preserved.
In the weakly deformed samples, relict igneous textures are evident. Yet again, no foliation or lineation is apparent in thin sections. Crystallographic preferred orientation is generally not evident with the gypsum plate inserted, except for a few small (2–3 mm) domains of weak crystallographic preferred orientation in plagioclase and olivine (but no shaped-preferred orientation). Maximum recrystallization of plagioclase is ~15%, and as much as 5% olivine is recrystallized. Clinopyroxene lacks recrystallization. Relict plagioclase grains commonly display weak undulose extinction and contain thin, tapering deformation twins, primarily on albite but also on pericline. The twins are deformed by gentle bands, kink bands (as wide as 1 mm), and microfaults. A high angle can be measured between most of the kink bands or microfaults and albite twins aligned close to (001), as commonly observed in other studies of plagioclase deformation (e.g., Oleson, 1987; Oleson and Kohlsted, 1985; Agar et al., 1997). Plagioclase neoblasts are generally equant with gently curved boundaries that commonly intersect at 120° triple junctions. There are some lobate protrusions on the boundaries of some neoblasts and relict plagioclase grains, but they are not strongly serrated. Plagioclase neoblast grain size is relatively large (generally >0.1 mm). More than 50% of the neoblasts display twinning, but very few have undulose extinction. Subgrains are rare, but where identified they rotate deformation twins by 2°–15° within the outer margin of relict plagioclase grains. The subgrain size is slightly smaller than those of neoblasts.
A gradual grain size layer contact is observed from one side of the thin section to the other side on Figure F40A. Recrystallization of plagioclase is restricted to the lower left quadrant of the section, where small neoblasts of plagioclase and olivine can be recognized along a narrow shear zone. Strong deformation bands in all olivine crystals are clearly obvious. Figure F40B is a magnified view of deformation bands within an olivine grain. Although the overall texture is granular to subophitic, consertal texture between clinopyroxene-clinopyroxene and clinopyroxene-olivine can be detected. Examples of magmatic twinning of clinopyroxene are also observed.
A coarse-grained olivine gabbro with equigranular texture is presented in Figure F40C. Coarse grains of plagioclase, olivine, and clinopyroxene are undeformed, but at the contacts of these coarse grains small neoblasts of pyroxene and plagioclase of different sizes can be observed. In some portions, a gradation in neoblast grain size from coarse to fine grains can be observed. Plagioclase is predominantly fresh and shows clear kink bands, subgrain boundaries, and deformation twins. Olivine is highly fractured, and most oxides are concentrated along olivine fractures. Olivine grains exhibit strong deformation bands. In general, plagioclase neoblasts are relatively large, as are olivine neoblasts.
Another example of weakly recrystallized texture is shown in Figure F40D, where, in fact, the majority of the thin section is covered with a large pegmatitic plagioclase grain that includes olivine and clinopyroxene chadacrysts. At the margins of plagioclase contacts with other grains, very small neoblasts are observed. Tapered deformation twins are present in large plagioclase, along with very strong undulatory extinction in olivine and plagioclase.
Oikocrystic relict igneous textures are preserved in clinopyroxene, but some grain boundaries are serrated and surrounded by narrow recrystallized phases. Subplanar zones of the plagioclase neoblasts define weak foliation. Weak alignment of plagioclase neoblasts, albite twins, and homogeneous interference colors with a gypsum plate indicate domains of moderate crystallographic preferred orientation in plagioclase, but no significant crystal shape-preferred orientation is evident. The percentage of recrystallized plagioclase varies between 15% and 30%, and this extent is distinctive for the category. Olivine can be as much as 20% recrystallized, whereas clinopyroxene is <5% recrystallized and typically is not recrystallized. Undulatory extinction is common in relict plagioclase grains that also contain curved, tapering albite and pericline twins. Microfaults and kink bands at high angles to albite twins are present. Kink bands tend to be narrower and more numerous than those of the weakly recrystallized texture. Plagioclase neoblast grain sizes are still relatively large. Neoblast grain boundaries tend to be more irregular than those in the weakly recrystallized textures. More than 50% of the neoblasts are twinned, but these twins, in contrast to those in weakly recrystallized textures, tend to be more tapered and discontinuous. Subgrains are rare within neoblasts but can be seen in the margins of relict plagioclase grains. Aggregates of recrystallized olivine are generally located toward the margins of olivine porphyroclasts. An igneous morphology is still at least partly evident in most olivine grains. A few domains of recrystallized olivine show crystallographic preferred orientation defined with a gypsum plate. Wherever kink bands are present, subgrains are also identifiable. Clinopyroxene remains largely undeformed except for narrow recrystallized margins in some grains. Clinopyroxene neoblasts, if present, generally have rounded to weakly serrated margins. The recrystallized grain size is noticeably smaller than that of either olivine or plagioclase. Clinopyroxene oikocrysts are generally not recrystallized, but some of them are individually separated into two or more large grains with slightly different extinction angles.
As an example, Figure F41A displays plagioclase laths similar those in subophitic textures that crosscut larger clinopyroxene grains with smaller enclosed euhedral plagioclase chadacrysts. In general, plagioclase chadacrysts remain euhedral, apparently protected by the low internal strain in the clinopyroxene. The presence of the plagioclase neoblasts along the margins creates an incipient porphyroclastic appearance for this thin section. Locally, clinopyroxene is also recrystallized and marginally altered to brown amphibole. In zones where amphibole is not present, clinopyroxene fails to recrystallize, suggesting importance of high-temperature hydration in the deformation response of clinopyroxene. Plagioclase grains show mostly serrated boundaries. A weak crystal-plastic foliation can be detected subvertically parallel to the plagioclase neoblasts.
Another example for this category is shown in Figure F41B. Recrystallization is mostly restricted to the grain contacts, and most of the clinopyroxene is altered to brown and green amphiboles. However, no specific orientation of the foliation defined by shape fabrics can be detected. Most of the oxides and green amphiboles are located along the margins of larger porphyroclasts. Olivine and clinopyroxene are both recrystallized, mostly along the grain contacts. Undulatory extinction is common in relict plagioclase grains. Most of the plagioclase grains are curved and tapered.
Subgrain rotation, serrated boundaries, and narrow zones of kink bands are clearly visible in Figure F41C. Again, most of the neoblasts occur along the grain contacts in narrow zones. Narrow zones of kink bands cut through the relict plagioclase grains. Very coarse plagioclase and clinopyroxene grains dominate the section. On Figure F41D, elongated coarse plagioclase and clinopyroxene grains dominate most of the section, and clinopyroxene is highly altered to pale green amphibole and oxide minerals. Most of the deformation twins in plagioclase laths are curved, and plagioclase grains have sutured boundaries and show strong subgrain rotation. Olivine is recrystallized into aggregates. Crude crystal-plastic foliation can be detected parallel to the curved and elongated plagioclase laths.
Moderately aligned porphyroclasts of plagioclase (with aspect ratios up to 1:5) surrounded by mantles of dynamically recrystallized plagioclase define this category. Moderate preferred dimensional orientation of porphyroclasts and weak to moderate crystallographic and shape-preferred orientations of recrystallized plagioclase and olivine resemble a more pronounced porphyroclastic texture. A total of 30%–50% of plagioclase is recrystallized; 30%–50% recrystallization occurs in olivine but <20% in clinopyroxene. Plagioclase porphyroclasts contain numerous deformation twins that are strongly curved or kinked. Margins of the porphyroclasts are commonly serrated and irregular and are enveloped by recrystallized plagioclase.
Figure F42A shows an oxide-bearing olivine gabbro with brittle deformation within a plagioclase porphyroclast. Left lateral sense of displacement along the brittle fractures is very common. Brittle deformation was apparently stronger in plagioclase than clinopyroxene. Highly deformed and curved plagioclase porphyroclasts in this section are quite obvious. Deformation twins are common. Brittle deformation appears to be parallel to the crude crystal-plastic foliation, which is defined by the preferred orientation of elongated clinopyroxene and olivine grains. Most of the neoblasts are concentrated along margins of larger grains. Olivine and clinopyroxene show less recrystallization than plagioclase.
Figure F42B represents an oxide-bearing olivine gabbro with very coarse plagioclase and clinopyroxene grains. Neoblasts occur along most of the contact boundaries, producing a porphyroclastic appearance of the rock. Olivine is highly altered in the upper left side of the section but is recrystallized to smaller neoblasts in the lower right portion. It is difficult to define any foliation because of the large size of the porphyroclasts; however, a crude diagonal crystal-plastic foliation may be present. One very coarse clinopyroxene grain includes large plagioclase laths. Clinopyroxene is altered to brown amphibole along margins, and recrystallized portions of the grain edges are also highly altered. Exsolution lamellae and consertal texture are observed in clinopyroxene. Olivine, although altered, still displays strong deformation bands. Plagioclase has curved and tapered deformation twins on the right side of the section, whereas on the opposite side plagioclase displays strong subgrain rotation with serrated boundaries, strong undulose extinction, and a higher percentage of recrystallization due to the higher strain.
Higher aspect ratios of porphyroclasts can be observed on Figure F42C, which shows an olivine porphyroclast with an aspect ratio of 1:4. Elongated pegmatitic plagioclase laths are also observed. Compared to the previous samples, the sample contains a higher percentage of neoblasts, stronger plagioclase subgrain rotation, and more sutured boundaries. A crude subvertical crystal-plastic fabric can be defined parallel to the elongated olivine and plagioclase porphyroclasts.
An igneous layer contact is presented in Figure F42D, where on the right side of the thin section a pegmatitic gabbro shows minor recrystallization and tapered plagioclase laths. This changes gradually to a finer-grained gabbro in which a moderately recrystallized to porphyroclastic texture occurs toward the left across the grain size contact. Strong undulose extinction, subgrain rotation, and recrystallization of plagioclase are observed. In addition to plagioclase, olivine and clinopyroxene are also recrystallized. Aspect ratios of the porphyroclasts are mostly between 1:2 and 1:3.
Strongly aligned porphyroclasts of plagioclase surrounded by mantles of dynamically recrystallized plagioclase define the porphyroclastic texture. Preferred dimensional orientation of porphyroclasts and the crystallographic and shape-preferred orientations of recrystallized plagioclase and olivine characterize the foliation. As much as 50%–70% of plagioclase can be recrystallized, and as much as 50% recrystallization may also occur in olivine but <30% in clinopyroxene. Plagioclase porphyroclasts contain numerous strongly curved or kinked deformation twins. Margins of the porphyroclasts are commonly serrated and irregular and are enveloped by recrystallized plagioclase (core and mantle structure; White, 1975). Plagioclase neoblasts may have moderate shape-preferred orientation and moderate to strong crystallographic preferred orientation. Also weakly to highly serrated boundaries of plagioclase neoblasts form mantles surrounding porphyroclasts and have undulatory extinction. Plagioclase neoblast grain size is significantly smaller than the weakly or moderately recrystallized textures described above. More than 50% of plagioclase neoblasts are twinned. Subgrain concentrations around the margins of the plagioclase porphyroclasts are common. Olivine porphyroclasts have a weak to moderate shape-preferred orientation. The margins of clinopyroxenes are recrystallized with weakly serrated boundaries. Clinopyroxene porphyroclasts do not appear to have any preferred orientation except as aggregates in pressure shadow regions.
Figure F43A represents an oxide gabbro with porphyroclastic texture. Strong alteration of clinopyroxene to green and brown amphibole and oxide minerals is observed. Corona structure is common. Plagioclase porphyroclasts show strong undulose extinction and strong subgrain rotation. Curved deformation twins are observed within medium-sized plagioclase laths. Some subhedral clinopyroxene porphyroclasts are elongated parallel to the oxide and alteration veins and create a foliation pattern. Different generations of plagioclase neoblasts show size grading.
Figure F43B contains strong kink bands, subgrain rotation, and undulatory extinction in olivine porphyroclasts. Plagioclase porphyroclasts also contain deformation twins, which are bent and have strong undulatory extinction. Olivine is highly altered to secondary minerals and oxides. Plagioclase neoblasts are in a range of grain size belonging to a different generation of recrystallization. Clinopyroxene is marginally altered to brown amphibole. Plagioclase and clinopyroxene porphyroclasts have strong shape-preferred orientations and high aspect ratios, as high as 1:4. Some plagioclase neoblasts even show shape-preferred orientation.
Plagioclase neoblasts in Figure F43C have more equant shapes with 120° triple junctions and strain-free appearance. No undulose extinction is observed in plagioclase neoblasts; however, very strong subgrain rotation and undulose extinction is observed in plagioclase porphyroclasts with serrated boundaries. Olivine and clinopyroxene are both highly recrystallized. Most of the oxide minerals are distributed within the matrix of neoblasts. Some oxide minerals are marginal to the olivine and clinopyroxene porphyroclasts.
The porphyroclasts in Figure F43D show varying grain size from very coarse to fine grains from right to left diagonally, probably an inherited grain size layer boundary. Different generations of neoblasts can also be identified. Serrated boundaries are very common in most of the plagioclase porphyroclasts. Oxide minerals occur as lenses of recrystallized neoblasts, mostly parallel to the foliation orientation, and are also marginal to the olivine and clinopyroxene porphyroclasts and neoblasts. This unique texture may be a reaction texture because it appears as if oxide lenses were percolating through the rock and replacing olivine and clinopyroxene. Also, plagioclase tends to be highly zoned in places. These oxide lenses could also result from mechanical segregations. A crude modal layering parallel to the crystal-plastic fabric is observed, as clinopyroxene and olivine porphyroclasts tend to segregate between zones of plagioclase and oxide neoblasts. Some zones verge on a protomylonitic texture with highly bimodal grain sizes.
Higher percentages of plagioclase neoblasts (up to 80%) and stronger shape and crystallographic preferred orientation displaying a tendency locally toward mylonitic texture and bands typify this category. Mylonitic textures are commonly spatially gradational with the porphyroclastic textures, and this transition can be observed clearly. The core and mantle structure that characterizes the porphyroclastic texture is only locally present. Exsolution lamellae in clinopyroxene porphyroclasts are commonly bent at subgrain boundaries, and kink bands are oriented at high angles to the foliation. This is the case for the clinopyroxene porphyroclast in Figure F44A, where the exsolution lamellae and magmatic twins of clinopyroxene at high angles to the foliation orientation are bent. Figure F44B is a magnified portion of Figure F44A representing the clinopyroxene porphyroclast. Porphyroclastic to mylonitic texture of the rock is prominent, and strong foliation is observed diagonally across the sample. The long dimensions of clinopyroxene porphyroclasts are oriented parallel to the foliation plane. Aggregates of olivine neoblasts have obviously replaced some former olivine porphyroclasts, perhaps as a consequence of subgrain rotation. In addition to the crystal-plastic foliation, deformation-induced grain size layering is parallel to the foliation orientation. Nearly all the neoblasts show high aspect ratios and strong shape-preferred orientation. More than 80% of plagioclase and most of the olivine grains are recrystallized. Clinopyroxene appears to be the strongest phase in this thin section. Although clinopyroxene porphyroclasts are internally deformed, they are not recrystallized. Magmatic twins in clinopyroxenes, although bent, are still preserved and easily identifiable.
Spatial alternation of porphyroclastic and mylonitic texture can be better observed in Figure F44C, where high–aspect ratio plagioclase, olivine, and clinopyroxene porphyroclasts alternate with very fine mylonitic bands of recrystallized plagioclase neoblasts. Recrystallized oxide neoblasts are parallel to these neoblastic plagioclase zones, and they are typically marginal to olivine and clinopyroxene grains. The highest aspect ratio porphyroclasts of the whole collection may be observed in Figure F44D, where olivine and ribbon plagioclase in an oxide olivine gabbro have aspect ratios as high as 1:8. Strong subgrain rotation and undulose extinction in elongated plagioclase porphyroclasts with serrated boundaries are common. Some olivine porphyroclasts are totally altered to brown and green amphibole. Oxide minerals are interstitial parallel to the crystal-plastic foliation and are potential glide surfaces in the matrix. Clinopyroxene porphyroclasts are marginally recrystallized and altered to green and brown amphibole and have strong shape preferred-dimensional orientation.
More extensive recrystallization and more extensive clinopyroxene recrystallization is the most significant difference between mylonites and porphyroclastic mylonites with respect to the preceding porphyroclastic textures (textural Category 6). In addition, the recrystallized matrix is generally >50% of the rock and ranges as high as 90%. Generally, there are few remaining porphyroclasts of clinopyroxene and the rocks appear more equigranular. Plagioclase and olivine relict porphyroclasts are present, but they are scarce. Strong foliation and in some places lineation characterize the mylonitic texture. Fine grain sizes, strong crystallographic preferred orientation, and serrated grain boundary morphology defines this mylonitic to porphyroclastic mylonitic texture. Plagioclase relict porphyroclast aspect ratios range from 1:1 to 1:5. The core and mantle structure that characterizes the porphyroclastic texture is only locally present. Some mylonitic samples preserve neoblasts with strongly serrated margins, whereas others preserve more equant, polygonal neoblasts. Fewer plagioclase neoblasts are twinned when compared to less deformed samples. Subgrains are evident in both plagioclase porphyroclasts and neoblasts, but they do not show distinct rims around the porphyroclasts as in the porphyroclastic texture (above); they are still concentrated in porphyroclast margins. Olivine and clinopyroxene have finer grain sizes than other textures.
Serrated grain boundaries and subgrain rotation on the margins of plagioclase porphyroclasts are clearly visible, which is a unique texture of crystal-plastic foliation in almost equigranular porphyroclastic textures. Most of the plagioclase neoblasts show pentagonal or hexagonal shape, arguing for a grain boundary migration mechanism of deformation. Also, the porphyroclasts are small in relative size, and no strain is observed in most of the grains. Foliation is subvertical in the section. Aggregates of olivine and clinopyroxene are parallel to the foliation direction. Several generations (at least four) of neoblasts in plagioclase, clinopyroxene, and olivine can be identified. Strain patterns and kinking are only observed in porphyroclasts. Neoblasts are mostly strain free except for the sample presented in Figure F45B, where plagioclase neoblasts exhibit undulatory extinction. Undulose extinction is observed in plagioclase neoblasts when different generations of plagioclase neoblasts are present. Horizontal foliation (parallel to the long edge of the thin section) is defined by the orientation of plagioclase neoblasts and elongated clinopyroxene grains. Also, the assemblage of clinopyroxene and olivine grains is common, sometimes overlapping. Yellowish (in normal light) oxides are present in some olivine fractures, most probably representing hematite or limonite. Sometimes it is difficult to distinguish between neoblasts and porphyroclasts because of the more uniform grain size distribution (pseudoequigranular texture).
Ultramylonitic texture is spatially gradational with mylonitic and porphyroclastic mylonite texture because the mylonitic shear zones can be quite thin. The strain gradient can be remarkably sharp, even on a thin section scale (e.g., see thin section Sample 179-1105A-25R-2, 3–8 cm, in "Supplementary Material"). Ultramylonitic bands are typically >90% recrystallized matrix; the major difference is the finer grain size of the plagioclase neoblasts (ribbon shape) and the more complete recrystallization of clinopyroxene porphyroclasts. Relict porphyroclasts are dominated by clinopyroxene. The even grain size distribution of olivine and finer grain sizes of neoblasts define the ultramylonitic textures. Oxide bands parallel to foliation are common, giving the mylonite a banded structure. Pressure shadows are common in this category, and the sense of shear can sometimes be determined.
Figure F45 represents an ultramylonitic texture. Bands within the thin section have extremely fine grain sizes and are dominantly neoblasts, often with ribbon-shaped neoblasts. They are described as ultramylonites if the porphyroclast count is low. These fine bands grade rapidly to porphyroclastic mylonite. Figure F45D represents a magnified portion of Figure F45A, where a small clinopyroxene porphyroclast can be observed within the ribbons of plagioclase neoblasts. Pressure shadow structure defines dextral sense of shear for this porphyroclast. Most of the porphyroclasts are clinopyroxene, which may argue for the stronger rheology of clinopyroxene compared to olivine.
Figure F45E represents similar texture with a relatively finer grained matrix and progressively larger porphyroclasts left to right across the photo. Most of the oxide mineral bands (dark thin bands) are parallel to the shear zone and shearing direction and appear to help localize shear strain. Some core mantle structures can be observed around plagioclase porphyroclasts (upper left). High alteration of some olivine and plagioclase can be also observed as well as a postkinematic alteration veins cutting normal to the foliation direction (middle of photo). Figure F45F represents deformation-induced grain-size layering from coarse porphyroclastic grain sizes in the lower right corner of the thin section to very fine grained ultramylonitic texture toward the upper left corner. Oxide minerals also alternate parallel to the ultramylonitic zones.
As described in the previous section, based on the diagnostic microstructures observed in all the available thin sections (137), eight broad microstructural and textural categories were defined:
These textural categories represent progressively increasing extents of deformation from 1 to 8, which helps to quantify the downhole deformation. Quantification of deformation extents for thin sections from different depth intervals helps to document the variations in deformation within the borehole and also facilitates several correlations between deformation extent and lithologic and, therefore, rheologic intervals, shipboard structural logging, and FMS image logs. For thin sections that exhibit more than one textural category, a deformation extent representing the predominant textural category is assigned. The downhole deformation extent data described above are presented in Figure F46 for the 137 sections described, along with a static downhole FMS image for the entire core, the lithologic log (see "Appendix" in "Supplementary Data" in Pettigrew, Casey, Miller, et al., 1999), the shipboard downhole foliation log, and an FMS interpretation of ductile shear zones in the core. Most ductile shear zones defined directly correlated with the core descriptions where recovery allowed direct correlation; others were recognized by correlation with shear zone characteristics on the FMS log and definition of crystal-plastic foliation via FMS. Several points are obvious from the diagram. Mylonite zones (deformation categories 7 and 8) are concentrated in oxide-rich units (IIA, IIB, and IV). The FMS static images are generally most conductive in oxide-rich zones (darker colors) and most resistive in nonoxide-bearing zones (lighter colors). There is an overall correlation between conductive zones on the FMS image and deformation. There is a similar correlation between the VCD data (foliated gabbro) and deformation extent. Lastly, the FMS-determined inclinations of the shear zones are represented by apparent dip marks in a north-south plane. Based on FMS data, the shear zones and crystal-plastic foliations dip both toward the ridge axes to the north and away from the ridge axes to the south. Dips away from the ridge axis dominate. The strike of the shear zones forms a strong maximum and is essentially ridge parallel. True dips range from 10° to 60° (average = 45°) (J.F. Casey and P. Zarian, unpubl. data). These apparently conjugate shear zones may result from normal inward-dipping ridge axis faults as well as conjugate faults resulting from bending stresses during footwall rollover and development of the core complex (e.g., Tucholke et al., 1998).
Figure F47 represents a series of three plots indicating the deformation extent vs. frequency percentage of the three main groups of lithologies. We modified the grouping because we noted that some rocks with <2% oxides generally contained only poorly disseminated and interstitial varieties and were typically significantly <2%, approaching zero. Although these oxides may represent important petrogenetic criteria in classifying rocks, the small amounts of oxides are unlikely to influence the rheology of the sample. Thus, we grouped rocks with <2% oxides together with "oxide-free" gabbroic rocks for these plots only. Of note is the group that shows by far the highest extent of deformation and strain in the section, the oxide gabbros and oxide olivine gabbro (>5% oxides) grouping, with the majority of samples plotting in textural category 5 or above. The oxide-bearing group (2%–5% oxides) showed a range of deformation states, the majority of which were category 5 or below. The oxide-free grouping (0%–2% oxides) by far showed the lowest degree of deformation, with a maximum of category 5 and nearly 65% of the samples in textural category 1 (undeformed magmatic texture). The positive correlation between deformation extent and oxide abundance is therefore striking, suggesting that oxides impart a weak rheology to the rock that lowers its overall strength and tends to localize deformation (e.g., Agar and Lloyd, 1997; Agar et al., 1997). Whether the oxide is symptomatic of initial deformation being hypersolidus as suggested by Dick, Natland, Miller, et al. (1999) is difficult to determine because of the subsolidus deformation overprint that the rocks have now experienced, making it difficult to determine the entire deformation path of each sample.
In our estimation of trapped melt within the gabbroic samples utilizing modal estimates, many very highly incompatible whole-rock trace element abundances, and estimates of equilibrium melts for mineral chemistry, we examined a ductile shear zone and the transition between undeformed gabbro and porphyroclastic to porphyroclastic mylonitic gabbro. If shear zones were channelways for late fractionating melts, as suggested for Hole 735B (Dick et al., 2002), we expected that the products of the melt crystallization would be identifiable and that the bulk samples would be very enriched in highly incompatible elements. Estimates of trapped melts, which are made using modal analysis, calculation of equilibrium melt compositions based on BLF liquid lines of descent of primitive Atlantis II basalts (Johnson and Dick, 1992), gabbroic mineral compositions, and bulk rock incompatible element abundances from this study are not compatible with the supposition of melt enrichment in the shear zones (Banerji, 2005). Our results are displayed in Figure F48, which shows a transition from little-deformed oxide-free gabbroic rocks to oxide gabbros that have been partly mylonitized in the core interval from 129 to 132 mbsf. Oxide gabbros are characterized by low bulk Mg#s (~40) and oxide-free gabbroic rocks by higher Mg#s (~73–75). Ironically, estimates of trapped melt indicate the highest abundances are within the lowest undeformed gabbroic rocks (classified geochemically as meso- and orthocumulates with >7% and >25% trapped melt, respectively). The oxide gabbroic rocks had the lowest estimates (classified geochemically as adcumulates; <7% trapped melt). In addition to our estimates of trapped melt, we also display a profile of chondrite-normalized Ce, which is one of the highly incompatible elements used in our estimates. The profile clearly shows that the mylonitized oxide gabbros are depleted in trapped melt constituents, especially considering their evolved mineral chemistry, in comparison to the more primitive oxide-free gabbros outside the shear zone. Based on this evidence we assume that the oxide minerals accumulated in the protolith of the shear zones during crystallization of the original cumulates are perhaps similar, in part, to pyroxene and plagioclase as adcumulus phases. Oxides would not represent the dominant products of an evolved trapped melt, which would be highly enriched in incompatible elements like Ce. Alternatively, perhaps the most intense shearing regions eliminated all late-stage melts as shearing progressed. The major question becomes if higher melt percent induces deformation, why did the oxide-poor gabbroic rocks with estimated higher trapped melt not deform initially or at least along with the adjacent oxide-rich gabbroic rock that underwent shear? The question remains, do high modal percentages of oxide minerals simply present a weak rheology after they have crystallized (Agar and Lloyd, 1997; Agar et al., 1997) that, in turn, tends to localize strain in a subsolidus state, or do they represent a symptom (crystallization product) of a highly allochthonous and late-stage melt phase that localized the deformation in a hypersolidus state as envisioned by Dick, Natland, Miller, et al. (1999) and Dick et al. (2002)? In our example, the melt phase that would be evidenced by an abundance of highly incompatible elements has not remained in the samples that are most highly deformed, and the oxide-rich zone thought to be evidence of hypersolidus deformation appears to lack significant crystallization products of a trapped melt.