Geological Society of America Bulletin-2012-Whitney-B30754.1

27
5/21/2018 GeologicalSocietyofAmericaBulletin-2012-Whitney-B30754.1-slidepdf.com http://slidepdf.com/reader/full/geological-society-of-america-bulletin-2012-whitney-b307541 Geological Society of America Bulletin doi: 10.1130/B30754.1  published online 21 December 2012; Geological Society of America Bulletin  Donna L. Whitney, Christian Teyssier, Patrice Rey and W. Roger Buck  Continental and oceanic core complexes  Email alerting services articles cite this article  to receive free e-mail alerts when new www.gsapubs.org/cgi/alerts click Subscribe America Bulletin  to subscribe to Geological Society of www.gsapubs.org/subscriptions/ click Permission request  to contact GSA http://www.geosociety.org/pubs/copyrt.htm#gsa click official positions of the Society. citizenship, gender, religion, or political viewpoint. Opinions presented in this publication do not reflect presentation of diverse opinions and positions by scientists worldwide, regardless of their race, includes a reference to the article's full citation. GSA provides this and other forums for the the abstracts only of their articles on their own or their organization's Web site providing the posting to further education and science. This file may not be posted to any Web site, but authors may post works and to make unlimited copies of items in GSA's journals for noncommercial use in classrooms requests to GSA, to use a single figure, a single table, and/or a brief paragraph of text in subsequent their employment. Individual scientists are hereby granted permission, without fees or further Copyright not claimed on content prepared wholly by U.S. government employees within scope of Notes articles must include the digital object identifier (DOIs) and date of initial publication. priority; they are indexed by GeoRef from initial publication. Citations to Advance online prior to final publication). Advance online articles are citable and establish publication yet appeared in the paper journal (edited, typeset versions may be posted when available Advance online articles have been peer reviewed and accepted for publication but have not Copyright © 2012 Geological Society of America  as doi:10.1130/B30754.1 Geological Society of America Bulletin, published online on 21 December 2012

Transcript of Geological Society of America Bulletin-2012-Whitney-B30754.1

  • Geological Society of America Bulletin

    doi: 10.1130/B30754.1 published online 21 December 2012;Geological Society of America Bulletin

    Donna L. Whitney, Christian Teyssier, Patrice Rey and W. Roger Buck

    Continental and oceanic core complexes

    Email alerting servicesarticles cite this article

    to receive free e-mail alerts when newwww.gsapubs.org/cgi/alertsclick

    SubscribeAmerica Bulletin

    to subscribe to Geological Society ofwww.gsapubs.org/subscriptions/click

    Permission request to contact GSAhttp://www.geosociety.org/pubs/copyrt.htm#gsaclick

    official positions of the Society.citizenship, gender, religion, or political viewpoint. Opinions presented in this publication do not reflectpresentation of diverse opinions and positions by scientists worldwide, regardless of their race, includes a reference to the article's full citation. GSA provides this and other forums for thethe abstracts only of their articles on their own or their organization's Web site providing the posting to further education and science. This file may not be posted to any Web site, but authors may postworks and to make unlimited copies of items in GSA's journals for noncommercial use in classrooms requests to GSA, to use a single figure, a single table, and/or a brief paragraph of text in subsequenttheir employment. Individual scientists are hereby granted permission, without fees or further Copyright not claimed on content prepared wholly by U.S. government employees within scope of

    Notes

    articles must include the digital object identifier (DOIs) and date of initial publication. priority; they are indexed by GeoRef from initial publication. Citations to Advance online prior to final publication). Advance online articles are citable and establish publicationyet appeared in the paper journal (edited, typeset versions may be posted when available Advance online articles have been peer reviewed and accepted for publication but have not

    Copyright 2012 Geological Society of America

    as doi:10.1130/B30754.1Geological Society of America Bulletin, published online on 21 December 2012

  • Continental and oceanic core complexes

    Donna L. Whitney1,, Christian Teyssier1, Patrice Rey2, and W. Roger Buck31Department of Earth Sciences, University of Minnesota, Minneapolis, Minnesota 55455, USA2School of Geosciences, University of Sydney, Sydney NSW 2006, Australia3Lamont-Doherty Earth Observatory, Columbia University, Palisades, New York 10964, USA

    ABSTRACT

    Core-complex formation driven by litho-spheric extension is a fi rst-order process of heat and mass transfer in the Earth. Core-complex structures have been recognized in the continents, at slow- and ultraslow-spreading mid-ocean ridges, and at continen-tal rifted margins; in each of these settings, extension has driven the exhumation of deep crust and/or upper mantle. The style of ex-tension and the magnitude of core-complex exhumation are determined fundamentally by rheology: (1) Coupling between brittle and ductile layers regulates fault patterns in the brittle layer; and (2) viscosity of the fl owing layer is controlled dominantly by the synextension geotherm and the presence or absence of melt. The pressure-tempera-ture-time-fl uid-deformation history of core complexes, investigated via fi eld- and mod-eling-based studies, reveals the magnitude, rate, and mechanisms of advection of heat and material from deep to shallow levels, as well as the consequences for the chemical and physical evolution of the lithosphere, includ-ing the role of core-complex development in crustal differentiation, global element cycles, and ore formation. In this review, we provide a survey of ~40 yr of core-complex literature, discuss processes and questions relevant to the formation and evolution of core com-plexes in continental and oceanic settings, highlight the signifi cance of core complexes for lithosphere dynamics, and propose a few possible directions for future research.

    INTRODUCTION

    When the lithosphere is under extension, the brittle upper crust breaks and is displaced along normal faults. When extension is concentrated on one or a few faults in a narrow region, ductile material ascends from deeper levels of the litho-sphere, resulting in exhumation of deep crustal

    rocks upper mantle in the footwall of the nor-mal fault(s). The resulting structure is a core complex, which occurs in both continental and oceanic lithosphere (Figs. 1 and 2). Extension is the direct driving force for core-complex devel-opment, but in continental settings, the far-fi eld tectonic regime may be one of convergence, and therefore continental core complexes may occur in orogenic settings under an overall regime of plate convergence.

    As extension proceeds, heat and material are transferred from deep (hot, ductile) to shallow (cool, brittle) levels, driving vigorous fl uid fl ow and strongly infl uencing the location and mag-nitude of subsequent extension. Interactions among minerals, fl uids, and/or magma may pro-duce economically important mineral deposits, and young extensional zones may be sources of hydrothermal activity during and after active faulting.

    Core complexes were fi rst recognized in the continents (e.g., Anderson, 1972; Coney, 1974, 1980; Crittenden et al., 1980; Lister and Davis, 1989), and they have been identifi ed in the geo-logic record from the Precambrian (Holm, 1996) to the present (Hill et al., 1992). Core complexes were later identifi ed at slow- and ultraslow-spreading oceanic divergent zones (e.g., Cannat, 1993; Cann et al., 1997; Blackman et al., 1998; Tucholke et al., 1998; Karson, 1999; Ranero and Reston, 1999; Dick et al., 2000). Continen-tal and oceanic core complexes have similar di-mensions, fault geometry, and kinematics (Figs. 1 and 3), and both involve exhumation of deeper levels of the lithosphere to shallow levels (John and Cheadle, 2010).

    In this review, we discuss the origin and significance of continental core complexes and oceanic core complexes. Although the term metamorphic core complex is a common

    For permission to copy, contact [email protected] 2013 Geological Society of America

    1

    GSA Bulletin; Month/Month 2013; v. 1xx; no. X/X; p. 126; doi: 10.1130/B30754.1; 13 fi gures.

    E-mail: [email protected]

    oceanic detachment system

    gabbro

    sea water

    fluidflow

    brittle-ductiletransition: ~600C(olivine rheology)

    ridgeaxis

    Oceanic core complex

    ~10 km

    fluidflow

    13

    meteoric water

    fluids derived from crystallization of metamorphic/igneous rocks

    brittle layer

    ductilelayer

    mylonitezone

    meteoric water

    brittle-ductiletransition: ~300400C

    (quartz rheology)

    continental detachment system

    ~15 km

    melt

    domi-nantlybrittle layer

    ductilelayer

    mylonite

    sea waterserpentinite

    13

    A

    B

    fluidflow ?

    Option 1Option 2 ?

    ?

    ??

    shearedgabbro

    Figure 1. Continental (A) and oceanic (B) core complexes dif-fer in their primary rock types (granitic and metasedimentary rocks in continental core com-plexes vs. gabbro and serpenti-nite in oceanic core complexes), and therefore in the minerals that infl uence the rheology of the complexes. Nevertheless, many fi rst-order processes of their origin and evolution are similar, and therefore there are many similarities in their archi-tecture. In B, the detachment fault roots in gabbro magma at depth (option 1); option 2 con-siders a dry spreading center in which the brittle detachment transitions to a ductile shear zone at depth (lithosphere boudinage).

    Invited Review

    CELEBRATING ADVANCES IN GEOSCIENCE

    1888 2013

    as doi:10.1130/B30754.1Geological Society of America Bulletin, published online on 21 December 2012

  • Whitney et al.

    2 Geological Society of America Bulletin, Month/Month 2012

    description of core complexes on the conti-nents, for the sake of simplicity (and symme-try), in this review we use the terms continental core complex (equivalent to metamorphic core complex) and oceanic core complex. We do not discuss in any detail the formation of continental margin core complexes, although some locations are highlighted on a world map (Fig. 2A).

    Over the past ~40 yr, interest in core com-plexes has remained high because these struc-tures are common in extending orogens and along slow-spreading oceanic divergent zones, and because they record fundamental thermo-mechanical processes in extending lithosphere. An understanding of the uplift and exhuma-tion of ductile rocks below low-angle normal faults, as well as the dynamics of the faults, is

    relevant to crustal evolution and seismogenesis in extending lithosphere, and core complexes have therefore been intensively studied using a variety of techniques, e.g., fi eld-based stud-ies (e.g., Davis and Coney, 1979; Miller et al., 1983; Bozkurt and Park, 1994; Gessner et al., 2001), numerical modeling (e.g., Buck et al., 1988; Lavier et al., 1999; Tirel et al., 2004, 2008; Rey et al., 2009a, 2009b; Allken et al., 2011), and analog modeling (e.g., Brun et al., 1994; Tirel et al., 2006).

    There remain important questions about core-complex initiation and evolution. In this review, we integrate knowledge derived from different types of investigations (fi eld, mod-eling) of continental and oceanic core com-plexes and discuss some of these unresolved issues.

    DEFINITIONS

    Herein, we use the following general defi ni-tion of a core complex and the processes that drive core-complex formation:

    A core complex is a domal or arched geologic structure composed of ductilely deformed rocks and associated intrusions underlying a ductile-to-brittle high-strain zone that experienced tens of kilometers of normal-sense displacement in response to lithospheric extension.

    The lithospheric extension that results in core-complex formation is commonly driven by plate divergence, such as at mid-ocean ridges and along rifted continental margins. Extension also occurs in plate convergence settings by slab rollback (e.g., the backarc of an oceanic subduc-tion zone) or by orogenic collapse under fi xed

    HhLi

    HoDI

    Da, Nb

    Pa

    Sh

    Ma

    Lo

    DNCV

    Xi

    HaYOH

    La SC

    PoAA

    SELR Ed

    VeTo

    SB

    GK ChNi KS

    ADGM

    Re

    LfPL

    + Antarctica: Fosdick core complex, Marie Byrd Land

    N. Am

    erican Cordillera

    Aegean

    Mid-AtlanticRidge

    Ka

    At

    AB

    SouthwestIndianRidge

    Ba

    No

    Go

    Australian- Antarctic Discordance

    Nx

    Rh

    AMOR

    M-Ca

    BB

    MC-Pyr

    continental core complexoceanic core complexcontinental margin core complex

    Fig. 2B Fig. 2C

    A

    60N

    0

    60S

    0120W 120E

    60N

    0

    60S

    0 120E

    Figure 2 (on this and following page). (A) Map of the world showing the locations of some Phanerozoic core complexes in the continents and oceans. Key to abbreviations: AAAlpi Apuane (Italy); ABAtlantis Bank (SW Indian Ridge); ADAma Drime (Nepal); AMORArctic segment of Mid-Atlantic Ridge; AtAtlantis Massif (Mid-Atlantic Ridge); BaBaja (Mexico); BBBay of Biscay; ChChapedony (Iran); DaDayman (Papua New Guinea); DIDoi Inthanon (Thailand); DNCVDay Nui Con Voi (Vietnam); EdEdough (Algeria); GKGrand Kabilye (Algeria); GMGurla Mandhata (Pamirs); GoGodzilla; HaHarkin (China/Mongolia); HhHohhot (China); HoHongzhen (China); KaKane (Mid-Atlantic Ridge); KSKongur Shan (Pamirs); LaLaojunshan (China); LfLofoten (Norway); LiLiaodong Peninsula (China); MaMalino (Indonesia); LoLouzidian (China); LRLora del Rio (Spain); M-CaMid-Cayman spreading center; MC-PyrMassif Central (FrancePyrenees, France, Spain; includes Montagne-Noire); NbNormanby Island (Papua New Guinea); NiNide (Turkey); NoNorway rifted continental margin; NxNaxos (Greece); PaPaparoa (New Zealand); PLPayer Land (Greenland); PoPohorje Mountains (Slovenia); ReRechnitz (Austria); RhRhodope (Greece, Bulgaria); SBsouthern Brittany (France); SCSong Chay (China); ShShaerdelan (China); SESierra de las Estancias (Spain); ToTormes (Spain); VeVeporic (Slovenia); XiXiaoqinling (China); YOHYagan-Onch-Hayrhan (China/Mongolia).

    as doi:10.1130/B30754.1Geological Society of America Bulletin, published online on 21 December 2012

  • Continental and oceanic core complexes

    Geological Society of America Bulletin, Month/Month 2012 3

    boundary conditions or even during slow plate convergence (Rey et al., 2001). Core complexes occur in all of these settings.

    In continental core complexes, the normal-sense high-strain zone corresponds to a pro-found metamorphic and/or stratigraphic dis-continuity typically called a detachment fault (Fig. 1), which is so named because rocks above and below the fault zone record different pressure-temperature-time-deformation his to-ries. Although the uppermost part of the fault zone may be a prominent brittle fault surface, below this fault there may be a broad (hundreds of meters) zone in which ductilely deformed rocks have been kinematically and thermally linkedfor some or all of their defor ma tion historyto the structurally higher, brittle fault

    during exhumation of the footwall. Some de-tachment faults record a progression from high-temperature (mylonitic) deformation to much lower-temperature brittle deformation through time.

    The concept of detachment has also been applied to normal faults that are not associated directly with core-complex structures, e.g., the South Tibetan detachment system (e.g., Burg et al., 1984), which is the northern boundary of the Himalayan crystalline wedge, and normal faults in exhumed subduction complexes, such as on the island of Crete (e.g., Ring et al., 2001). There are also detachment-like faults in some conti-nental arcs, such as the Andes (e.g., Mpodozis and Allmendinger, 1993) and the North Cas-cades (Paterson et al., 2004). These detachments

    are normal faults with signifi cant displacement, but they do not bound core complexes.

    Detachment zone, detachment shear zone, and detachment system are terms that have been used to describe the diffuse zone of strain, up to 1.5 km thick, that underlies the upper most fault plane in some core complexes, such as those of the northern U.S. and Canadian Cordillera (e.g., Mulch et al., 2006; Gbelin et al., 2011). Characteristic features of detach-ments and ideas and debates about the dynamics of low-angle normal faults are discussed in later sections.

    Although detachment is widely used to describe core complexbounding faults, other terms are also used: for example, low-angle normal fault (e.g., Axen, 2007). The term

    Figure 2 (continued ). (B) Schematic map of the core complexes in the North American Cordillera. (C) Schematic map of the core com-plexes in the Aegean Sea and surrounding regions.

    zmirAthens

    Istanbul

    Rhodes

    Crete

    NaxosParos

    Samos

    Gulf of Corinth

    Syros

    Tinos

    Thasos

    IosMilos

    Kea

    Andros

    Sifnos

    Kazda Massif

    24 26 28

    40N

    Menderes Massif

    Rhodope Massif

    Aegean Sea

    GREECE

    TURKEY

    C

    ChemehueviWhipple Mts

    Buckskin-RawhideHarcuvarHarquahala Mts

    Catalina

    Sierra Mazatan

    South Mts

    Rincon

    Picacho

    SnakeRange

    Ruby-Humboldt

    Albion-Grouse Creek

    SNAKE RIVER PLAIN

    COLUMBIA PLATEAU

    COLORADO PLATEAU

    SIERRANEVADABATHOLITH

    COAST RANGES-NORTH CASCADES

    IDAHOBATHOLITH

    PENINSULARRANGES

    BATHOLITH

    PacificOcean

    MEXICOUSA

    USACANADA

    Priest River

    Kettle

    OkanoganValhalla

    Thor-Odin

    Frenchmans Cap

    San Andreas Fault

    Pioneer

    AnacondaBitterroot

    Lewis & Clark Clearwater

    Shuswap Complex

    NO

    RTHERN

    BELTCEN

    TRAL BELT

    SOU

    THERN

    BELT

    RaftRiver

    ROCKY M

    OU

    NTA

    I N FO

    LD &

    THR U

    ST BELT

    B

    110 W

    120W

    120W

    110W

    50N

    40N

    30N

    40N

    30N

    as doi:10.1130/B30754.1Geological Society of America Bulletin, published online on 21 December 2012

  • Whitney et al.

    4 Geological Society of America Bulletin, Month/Month 2012

    dcollement has also been used (Coney, 1980; Davis et al., 1980) and can be synonymous with detachment in the context of a core complex, in some cases indicating a low-angle master fault or shear zone. Breakaway fault refers to the fault identifi ed as the detachment that initially intersected Earths surface.

    Some detachment faults are listric; this in-dicates a curving fault with a dip that changes from steep at shallow crustal levels to sub hori-zontal at depth. In contrast, some detachment faults (or parts of detachment faults) in both continental and oceanic settings are downward concave and accommodate large offset yet moderate topography (e.g., van den Driessche and Brun, 1992; Lavier and Manatschal, 2006).

    DEVELOPMENT AND EVOLUTION OF THE CONCEPT

    As a result of debate, primarily in the 1960s1970s, about the tectonic history of basement uplifts (exposures of middle- and lower-crustal rocks) bounded by low-angle faults in the North American Cordillera (Figs. 2A and 2B), a model of core complexes as extensional structures was developed (Anderson, 1972; Coney, 1974; Wright et al., 1974; Proffett, 1977; Davis and Coney, 1979; Coney and Harms, 1984). The concept has been applied to similar structures elsewhere, and some of these are described

    as Cordilleran-style metamorphic core com-plexes (e.g., Aegean [Fig. 2C]Lister et al., 1984; West AntarcticaRichard et al., 1994; Norwegian CaledonidesSteltenpohl et al., 2004; IranVerdel et al., 2007).

    The existence of detachment faults and core complexlike structures was proposed for slow-spreading mid-ocean ridges before such struc-tures were observed in detail (Karson and Dick, 1983; Cannat, 1993; Tucholke and Lin, 1994). Fault-bounded domal structures resembling continental core complexes were later identi-fi ed in seafl oor images of the Mid-Atlantic Ridge (Cann et al., 1997; Blackman et al., 1998; Tucholke et al., 1998; Ranero and Reston, 1999; Tucholke et al., 2001) and confi rmed by ob-servation of samples collected from low-angle normal fault zones bounding these structures (MacLeod et al., 2002; Escartn et al., 2003; Schroeder and John, 2004; Ildefonse et al., 2007). Oceanic core complexes have now been recognized along segments of the Southwest Indian Ridge (Dick et al., 2000; Baines et al., 2003) and at other divergent zones (e.g., the CaribbeanNorth American RidgeHayman et al., 2011; the Australian-Antarctic Discor-danceChristie et al., 1998) (Fig. 2A).

    Many models for core-complex develop-ment have invoked isostatic rebound beneath the detachment fault to explain the arching of the fault and exhumation of the footwall (e.g.,

    in the continents: Spencer, 1984; Wernicke and Axen, 1988; Brun and van den Driessche, 1994; in the oceans: Lavier and Manatschal, 2006; MacLeod et al., 2009). Crustal fl ow beneath the extending upper crust has also been proposed to explain the domal structure of many conti-nental core complexes, as well as the moderate topographic relief of the exhumed footwall de-spite tens of kilometers of offset and the exis-tence of a fl at Moho in many highly extended regions (e.g., Block and Royden, 1990; Buck, 1991; McKenzie et al., 2000). Exploration of the thermal and mechanical consequences of large-magnitude crustal fl ow from deep to shallow crustal levels, including consideration of the re-lationship between continental core complexes and gneiss domes (Teyssier and Whitney, 2002), shows that core complexes can be signifi cant sites for heat and mass transfer and have played a role in differentiation of continental crust through geologic time (e.g., Rey et al., 2009a, 2009b; Thbaud and Rey, 2012).

    There has long been debate about the dynam-ics of low-angle normal faults, primarily fo-cused on the question of whether faults of low (

  • Continental and oceanic core complexes

    Geological Society of America Bulletin, Month/Month 2012 5

    1981). However, weak zones (faults) could re-sult from fl uid-rock interaction that increases pore pressure or generates phyllosilicates and other weak minerals (Morrison, 1994; Boschi et al., 2006), leaving open the question of which comes fi rst: the faults or the weak min-erals in the fault zones (e.g., Grasemann and Tschegg, 2012).

    As the core-complex concept was devel-oped, the relationships among core-complex formation, magmatic and/or hydrothermal activity, and ore deposition were recognized. Continental and oceanic detachment faults are common sites for metallic ore deposits, owing to the interaction of minerals and hot fl uids (Roddy et al., 1988; Beaudoin et al., 1991; Smith et al., 1991). For example, Cu-Fe sul-fi de and oxide deposits occur in the core com-plexes of SE California and western Arizona ( Whipple-Buckskin-Rawhide; Spencer and Welty, 1986), and other core complexes are associated with Au or Au-Ag deposits (South Mountain, Arizona, USA; Valhalla Complex, Canada; Rhodope, Greece; Massif Central, France) or uranium mineralization (Chapedony complex, Iran; Yassaghi and Masoodi, 2011). In some oceanic core complexes, hydrothermal Cu-Zn-Co-Aurich massive sulfi de deposits are associated with ultramafi c rocks in detach-ment fault zones (Mid-Atlantic Ridge; Fouquet et al., 2010).

    In the following sections, we survey the pri-mary structural and petrologic features of conti-nental and oceanic core complexes, followed by discussion of the dynamics of core complexes in different tectonic settings.

    CONTINENTAL CORE COMPLEXES

    Some of the most studied continental core-complex belts are in the North American Cor-dillera (where core complexes formed during diachronous collapse of thickened crust), the Aegean Sea/western Turkey (where core com-plexes formed in the backarc setting associated with rollback of the Hellenic subduction zones), and Mongolia-China-Korea (where core com-plexes extend far into Eurasia and their tectonic setting is not clear) (Fig. 2). Core complexes have also been described along the entire length of the Alpine-Himalayan orogenfrom the Pyrenees to SE Asiaas well as in the older orogens of Europe (Caledonian, Variscan) and Asia (Fig. 2A). In addition, core complexes are reported in Papua New Guinea, New Zealand, Antarctica, and various Precambrian terranes; it remains controversial whether the Appalachian orogen contains core complexes.

    The belt of core complexes in the North American Cordillera, from Mexico to Canada

    (Fig. 2B), has been used as a type locality for understanding continental core-complex dynamics of the crust and lithosphere, although there are important differences in core-complex development along the length of the belt. The Cordilleran core complexes record differences in the nature of the interaction between the shal-low and the deep crust, as shown by variations in tectonic evolution in three regions (Fig. 2B): (1) a southern core-complex belt (Mexico; Ari-zona, southern California, USA); (2) a central belt, from the Snake Range (Nevada, USA) to the Raft River complex (Utah, USA); and (3) a northern belt, from the Pioneer Mountains (Idaho, USA) to the Shuswap Complex (British Columbia, Canada). These three regions of core complexes vary in age of extension (as young as Miocene in south, Eocene in north), but, more importantly, in magnitude of exhuma-tion (with some exceptions: least in south and most [tens of kilometers] in north), and in the presence/involvement of partially molten crust (least in south, most in north) (Vanderhaeghe and Teyssier , 2001; Rey et al., 2009a).

    Continental core complexes are typically ellip-tical, with a long axis of ~1040 km (Fig. 3A); the footwall of core complexes is typically elevated above the surrounding rocks, in some cases by 12 km of relief. There are a few core complexes that are substantially larger (e.g., the Shuswap complex, Canada/United States; the Menderes complex, Turkey; and some dome complexes in the Pamirs, central Asia), but these typically con-tain several subsidiary core- complex/dome struc-tures within them.

    In the following sections, we survey structural and petrologic features relevant to understand-ing the origin and evolution of continental core complexes from the upper crust, through detach-ment faults and shear zones, to the lower crust (Fig. 4). The hanging wall and footwall in core complexes are mechanically coupled in various ways, such that the geometry of the hanging wall (e.g., multiple or single normal faults, gra-ben, half graben, tilted blocks) is inherently tied to the ability of the deep crust to fl ow and gener-ate a core complex (Block and Royden, 1990; Brun and van den Driessche, 1994; Lavier et al., 2000; Rey et al., 2009a, 2009b).

    An important parameter controlling lower-crustal viscosityand therefore coupling of deep and shallow crustis the geotherm. An elevated geotherm appears to be necessary to the development of continental core complexes, but models of the infl uence of geotherm on core-complex generation can be divided into three categories (Fig. 4). (1) In a warm crust, the deep crust is able to fl ow, but the strong coupling between deep and upper crust results in multiple upper-crust faults (Fig. 4A). (2) In a

    hot crust, extension localizes in a single, large-offset detachment fault system that arches as the low-viscosity deep crust develops a core complex in the footwall (Fig. 4B). (3) In the hottest crust case, the combination of localized upper-crust extension and reduction of lower-crust viscosity by partial melting results in exhumation of the deep crust; partially molten material is exhumed nearly isothermally and undergoes complex deformation during ascent, with contractional structures overprinted by ex-tension (Fig. 4C).

    Continental Core Complexes: Hanging-Wall Characteristics and Processes

    In most continental core complexes, hanging-wall rocks are present, although these typically have been at least partially removed by tectonic and/or erosional processes. In the North Ameri-can Cordillera, hanging-wall rocks of some core complexes are composed of unmetamorphosed to low-grade (meta)sedimentary and volcanic rocks; in other core complexes in the Cordillera, hanging-wall rocks are medium- to high-grade metamorphic rocks and intrusions in which metamorphism and intrusion predated core-complex development. In some of the Aegean core complexes, the hanging wall consists of ophiolitic rocks and unmetamorphosed sedi-mentary rocks, in places fi lling structural basins (Gautier et al., 1993).

    In some continental core complexes, supra-detach ment basins make up a signifi cant frac-tion of the hanging wall, indicating that the detachment came close to the surface during ex-tension. For example, in the southern Basin and Range of the North American Cordillera, some detachment faults intersect the surface, and ad-jacent syntectonic basins contain sediments de-rived from the footwall of the detachment (e.g., Miller and John, 1988; Miller and John, 1999). In addition, the western detachment that bounds the Shuswap core complex (Okanagan detach-ment in British Columbia) is near the base of some basins and is itself cut by normal faults and tilted to the east, so that the detachment here has the geometry of a thrust (Vanderhaeghe et al., 1999, 2003). The normal faults that cut the mylonitic detachment may root at depth into another detachment system that formed as the core complex cooled during exhumation. In this case, the basins, the detachment, and a part of the footwall became the hanging wall for this new, hypothetical detachment.

    The basal units of supradetachment basins commonly record a high paleogeothermal gradient during and after their deposition. For example, study of coal units in a half graben above the detachment on the north fl ank of the

    as doi:10.1130/B30754.1Geological Society of America Bulletin, published online on 21 December 2012

  • Whitney et al.

    6 Geological Society of America Bulletin, Month/Month 2012

    Montagne Noire core complex (France) record paleogeothermal heat fl ow between 150 and 180 mW/m2 (Copard et al., 2000). Similarly, thermo-chronologic studies of some continental core complexes show that heat may be conducted (or advected via fl uids) from footwall to hanging wall (e.g., Zeffren et al., 2005), resulting in re-setting (or partial resetting) of isotopic systems in hanging-wall rocks and minerals. Such a high

    heat fl ow is consistent with the strong gradient that develops during fast extension at the con-tact between hanging wall and footwall (Rey et al., 2009b) and shows that, although detach-ment faults represent a profound discontinuity in pressure-temperature-time history of footwall relative to hanging wall, rocks above and be-low the detachment fault may share a late-stage thermal history.

    Continental Core Complexes: Detachment Fault Characteristics and Processes

    In some continental core complexes, the detachment is not a single fault but is made up of multiple, closely spaced and anastomos-ing faults (Wernicke and Burchfi el, 1982). The upper most detachment fault may have a particu-larly well-defi ned fault plane, typically dipping

    ~ 50 km

    extensionstrain

    Moho

    flowing partially

    molten crust

    solidus

    solidus

    contraction strain

    "domino" rotation of upper crust blocks

    flow of lower crust

    ROLLING-HINGE MODEL FOR CORE COMPLEX FORMATIONBLOCK-ROTATION MODEL FOR CORE COMPLEX FORMATION

    flow of lower crust

    1 2 3 4 5Order of faulting:

    1

    CONVERGING CHANNEL FLOW OF LOW-VISCOSITY(PARTIALLY MOLTEN) LOWER CRUST

    C1. C2.

    incipient normal faulting, block rotation,and lower crustal flow

    extensional basins

    incipient normal faulting, fault rotation, lower crustal flow; fault 1 stops slipping, fault 2 takes over, resulting in passive rotation of fault 1

    channel flow

    Moho

    Moho

    Moho

    Moho

    moho

    pre-extension Moho

    pre-extension Moho

    pre-extension Moho

    2

    kinematic hinge

    PLATEAU: THICK, HOT CRUST FORELAND: COLD CRUST

    channel detachment

    CORE COMPLEX DEVELOPED AT EDGE OF OROGENIC PLATEAU

    pre-extension Moho

    upper crust

    lower crust

    incipient kinematic hinge

    transfer of thick crust toward foreland

    flowing partially molten crust

    rolling-hinge detachment

    solidu

    s

    fixed

    bo

    un

    dar

    y

    T

    P

    solid

    us

    shearing in channel

    contraction

    extension

    A. WARM CRUST B. HOT CRUST

    C. HOTTEST CRUST

    strai

    n hi

    stor

    y

    P-T

    pat

    h

    Figure 4. Modes of development of continental core complexes in warm, hot, and hottest crust. (A) Warm crust exhumes continental core complexes in footwall of normal faults that are distributed in upper crust; for example, exhumation by domino-style rotation of upper-crust blocks. (B) Hot crust focuses faulting in upper crust, lead-ing to large-offset fault and exhumation of lower crust by development of rolling-hinge detachment (after Brun and van den Driessche, 1994). (C1) Hottest crust has signifi cant partial melt; the low-viscosity lower crust fl ows in channels attracted by focused zone of upper-crust extension; channels collide and move upward to fi ll the gap created by upper-crust extension; deep crustal rocks record signifi cant decompression and deformation from con-traction to extension as they are exhumed (after Rey et al., 2011). (C2) At the edge of an orogenic plateau, partially molten crust fl ows owing to lateral gradients in gravitational potential energy; note expected reversal of sense of shear (kinematic hinge) between channel and rolling-hinge detachments (after Teyssier et al., 2005).

    as doi:10.1130/B30754.1Geological Society of America Bulletin, published online on 21 December 2012

  • Continental and oceanic core complexes

    Geological Society of America Bulletin, Month/Month 2012 7

    30 (Figs. 1 and 3). A region of brecciation and greenschist-facies alteration (recorded by secondary growth of chlorite epidote) may characterize the structurally highest (brittle) re-gions of detachment zones if suitable lithologies are present (e.g., granitic gneiss). In addition, many detachment faults record an early history as a ductile shear zone that evolved into a brittle fault during tectonic and erosional denudation, as well as during cooling driven in part by circu-lating fl uids (Malavieille et al., 1990). The upper brittle fault zone is typically underlain by a re-gion of high strain (and temperature) gradient, up to ~1.5 km thick in the footwall (Wernicke, 1981; Miller et al., 1983; Mueller and Snoke, 1993; Wells et al., 2000; Foster and Raza, 2002; Mulch et al., 2006).

    Although some of the deformation in this re-gion of high strain may have developed during pre-extensional tectonism, the deformation his-tory related to core-complex development can be discerned through integrated structural and isotopic studies, including geochronology. For example, although some shear zones bounding Cordilleran core complexes record Mesozoic deformation, the majority of Cordilleran core complexes show widespread exposures of my-lonitized Tertiary rocks that formed during ex-tension and core-complex development (e.g., Foster and Fanning, 1997; Vanderhaeghe et al., 1999; Wells et al., 2000; Gbelin et al., 2011).

    In the northern Cordilleran core complexes, the mylonite zone is several hundred meters thick (Hyndman, 1980; Mulch et al., 2006; Brown et al., 2012). It affects all lithologies but has an affi nity for quartzite or marble if these units are present in the extended crust. Sense of shear is typically unambiguous and indicates footwall up, unless the detachment zone has been tilted by late normal faulting or arched as a result of a rolling-hinge or doming effects.

    Detachment zones are self-exhuming struc-tures, and therefore detachment-related my-lonite zones display a range of fabrics, from ductile to brittle. A prevailing view is that high-temperature mylonitic fabrics are progressively overprinted by lower-temperature ductile fabrics followed by brittle processes such as cataclastic fl ow and brecciation (e.g., Davis et al., 1980). Based on the degree of refrigeration of footwall during extension, mylonitic rocks may become incorporated into the hanging wall and preserve a history of deformation fabrics formed at vari-ous temperatures (Mulch et al., 2006).

    Some core complexes have planar detach-ment faults (e.g., Bitterroot continental core complexes), but many continental and oceanic core complexes are characterized by a corrugated (undulating) detachment fault surface in which the corrugation axis is oriented parallel to the

    displacement (extension) direction and the undu-lations have amplitudes of approximately tens to hundreds of meters and wavelengths of hun-dreds of meters to tens of kilometers (John, 1987; Richard et al., 1990; Spencer and Reynolds, 1991; Cann et al., 1997; Yin, 2004; Cannat et al., 2009). In oceanic core complexes, these corruga-tions have been referred to as megamullions (Christie et al., 1998; Tucholke et al., 1998).

    Although some continental core complexes are bivergent, with symmetric, oppositely dip-ping detachments on either side of a footwall core (e.g., Hetzel et al., 1995a), many show structural asymmetry in their footwall, particu-larly in those exhumed at the edge of a continen-tal plateau (Teyssier et al., 2005). For example, in the Shuswap core complex, British Columbia (Fig. 2B), the continental core complex is asym-metric, with a series of gneiss domes located in the immediate footwall of the eastern (Columbia River) detachment system. The gneiss domes contain the highest-grade metamorphic rocks exposed in the core complex, including high-melt-fraction migmatite (Vanderhaeghe et al., 1999) and gneiss containing sillimanite and cor-dierite pseudomorphs after kyanite. These rocks recorded rapid near-isothermal decompression at ~750 C to 800 C from 1 GPa to

  • Whitney et al.

    8 Geological Society of America Bulletin, Month/Month 2012

    (e.g., 40Ar/39Ar in hornblende, biotite, musco-vite, K-feldspar; apatite zircon fi ssion-track; and/or apatite zircon U-Th/He) are used to capture temperature-time information as a function of exhumation history on the detach-ment. Some detachment zones record a slow cooling stage (515 C/m.y.) followed by more rapid cooling (70100 C/m.y.) (Scott et al., 1998; Wells et al., 2000). This trend has been interpreted to indicate a steepening of the fault through time or other changes in detachment zone geometry. To interpret thermochronol-ogy data in terms of detachment evolution, care must be taken to understand the thermal struc-

    ture of the detachment zone (e.g., position of isotherms, calculation of geothermal gradients) and contribution of erosion to denudation his-tory (Ketcham, 1996; Fayon et al., 2000).

    Thermochronology-based estimates of aver-age slip rate for detachments bounding Cor di-lleran core complexes range from ~1 mm/yr to 12 mm/yr (Foster et al., 1993; Scott et al., 1998; Foster and John, 1999; Wells et al., 2000; Carter et al., 2004, 2006; Foster et al., 2010), and simi-lar rates have been determined for Aegean core complexes (John and Howard, 1995; Brichau et al., 2006; Thomson et al., 2009). Studies involving multiple thermochronometers are

    able to detect possible changes in slip rate with time, such as a signifi cant increase in slip rate proposed for some detachments in the southern Basin and Range (Carter et al., 2004, 2006). The increase has been ascribed to a change in the regional tectonic regime, e.g., presence of a slab window beneath part of the Basin and Range at ca. 15 Ma.

    Detachment fault zones may be sites of sig-nifi cant fl uid circulation and hydrothermal al-teration (Bartley and Glazner, 1985; Kerrich and Hyndman, 1986; Kerrich and Rehrig, 1987; Kerrich, 1988; Fricke et al., 1992; Famin et al., 2004; Person et al., 2007). Fluid-mineral inter-action in detachment zones is relevant to under-standing the chemical, thermal, and physical evolution of detachment systems, the origin and location of ore deposits, and the interpretation of low-temperature thermochronometry results.

    Locally, and likely transiently, water-rock ratios are high during deformation at tempera-tures suffi cient for (re)crystallization of miner-als in the fault zone, and detachment zones are therefore characterized by greenschist-facies and lower-grade minerals: typically, hydrous miner-als such as chlorite, white mica, and epidote in continental core complexes, and talc, chlorite, tremolite, and serpentine in oceanic core com-plexes, as well as metalliferous ore deposits in both settings (Smith et al., 1991; McCaig et al., 2010). Hydrous minerals and other alteration products (e.g., from K-metasomatism) that form in the fault zone may be important in the initia-tion and subsequent structural evolution of the core complex, such as by promoting strain lo-calization and/or affecting the thermal state and therefore mode and pattern of faulting of the brit-tle crust (Lavier and Buck, 2002). In addition, a vigorous hydrothermal system in which mete-oric water circulates through faults in the upper 1015 km of the extending crust (i.e., to the level of the detachment fault) drives effi cient advec-tive removal of heat (Morrison and Anderson, 1998; Famin et al., 2004; Person et al., 2007).

    The effects of fl uid circulation in extended upper crust can be seen in fi eld and isotopic records. Vein systems and mineralized fault zones are common in the hanging wall of de-tachments and indicate that minerals precipi-tated from hot fl uids during ascent and cooling. Stable isotope (particularly hydrogen) values of hydrous minerals such as white mica, biotite, chlorite, and epidote show that the water-rich fl uids that interact with minerals in detach-ment zones are derived from various sources, including meteoric sources at structurally high levels and metamorphic/magmatic sources at lower levels (Kerrich and Hyndman, 1986; Spencer and Welty, 1986; Kerrich and Rehrig, 1987; Wickham and Taylor, 1987; Baker et al.,

    A

    B

    C D

    Figure 5. (A) West-dipping, Eocene Okanogan detachment zone, eastern Washington; the detachment footwall grades downward from mylonite to migmatite across a 23 km sec-tion. (B) East-dipping, Miocene Raft River detachment with remnants of hanging wall on mylonitic quartzite; view is looking south from the Idaho-Utah state line. (C) Photomicro-graph of mylonitic quartzite from the Snake Range detachment, Nevada; mica fi sh, S-C fabrics, and quartz crystallographic preferred orientation (c-axis, electron backscatter diffraction measurements of >1000 grains) are well developed and indicate top-to-the-east shear; stretched quartz grains are partially to entirely recrystallized by subgrain rotation in the dislocation creep regime. (D) Photomicrograph of mylonitic quartzite from the Colum-bia River detachment that bounds the Shuswap core complex to the east; mica fi sh and S-C fabrics indicate top-to-the-east shear; quartz grains are recrystallized by combination of subgrain rotation and grain boundary migration in the dislocation creep regime.

    as doi:10.1130/B30754.1Geological Society of America Bulletin, published online on 21 December 2012

  • Continental and oceanic core complexes

    Geological Society of America Bulletin, Month/Month 2012 9

    1989; Nesbitt and Muehlenbachs, 1995; Losh, 1997; Holk and Taylor, 2007; Mulch et al., 2004; Gbelin et al., 2011). Studies of detach-ments that are localized in simple lithologies (quartzite, marble) have shown that, in general, meteoric water equilibrates during detachment activity with neo/recrystallized hydrous phases such as white mica at moderate temperatures (~350450 C) (Famin et al., 2004).

    Hydrogen isotopic values in synkinematic white mica (e.g., Figs. 5C and 5D) and other hydrous minerals in detachment zones have been interpreted to indicate interaction of mica with a fl uid derived from a high-elevation catch-ment (e.g., Columbia River fault at the latitude of Thor-Odin, British Columbia, and at the lati-tude of the Kettle dome, Washington; detach-ments in the Ruby Range and Snake Range, Nevada; Fricke and ONeil, 1999; Mulch et al., 2007; Gbelin et al., 2011). Hydrous minerals in detachment mylonites may therefore contain the paleoelevation record over the time scale (~15 m.y.) of mylonite formation (Mulch et al.,

    2004). Studies that combine data from detach-ment-basin pairs have verifi ed that the isotopic composition of mylonite minerals matches that measured in basin strata deposited at high eleva-tion (Mulch and Chamberlain, 2007).

    Detachment zones typically accommodate large lateral displacement and considerable thin-ning. Tens of kilometers of displacement are ac-commodated by deformation in the detachment zone over the time scale of 15 m.y., implying high average strain rates. Quartz microstruc-ture in detachment zones provides information on fl ow paleostress through the analysis of re-crystallized grain size using paleopiezometers; temperature of deformation can be derived from quartz-mica oxygen isotope thermometry or from titanium-in-quartz thermometry. In the quartzite mylonites of some core complexes in the North American Cordillera, quartz dynamic recrystallization is dominated by subgrain ro-tation, which is consistent with low recovery under high-fl ow stress conditions in the disloca-tion creep regime (Figs. 5C and 5D). Whatever

    the temperature of deformation (350500 C) inferred from quartz-mica stable isotope pairs (Mulch et al., 2006; Gottardi et al., 2011; Gbelin et al., 2011), calculated fl ow stress is typically high and quite constant over the en-tire mylonitic section. This behavior suggests that localization or delocalization of strain, cor-responding to increasing or decreasing strain rates, respectively, is a response of the system to maintain fl ow stress near the critical crustal strength, a property that is best exemplifi ed in extensional systems (Mulch et al., 2006).

    There has been much discussion of the dy-namics of low-angle normal faults during core-complex development. Debate stems from predictions from rock mechanics theory that extension produces high-angle faults (~60; Ander son, 1942, 1951) and that low-angle faults cannot slip, consistent with the proposal that moderate to large earthquakes (M 5.5; Jackson, 1987; Jackson and White, 1989) do not occur along these faults. This conclusion apparently confl icts with fi eld observations that low-angle, and in some cases subhorizon-tal, normal faults have been active in the brittle crust (Wernicke , 1981; Reynolds and Spencer, 1985; Davis et al., 1986; John, 1987; Wernicke and Axen, 1988; Scott and Lister, 1992; John and Foster, 1993; Lecomte et al., 2010); for example, the recognition that horizontal fi eld markers are close to the detachment surface (e.g., Scott and Lister, 1992; John and Foster, 1993), and therefore that displacement on the detachment fault occurred at low dips. In addi-tion, some seismically active, low-angle (

  • Whitney et al.

    10 Geological Society of America Bulletin, Month/Month 2012

    Continental Core Complexes: Footwall Characteristics and Processes

    Footwall rocks may record a wide range of agesfrom pre- to synextensionof meta-morphic, magmatic, and deformation events. Metamorphic grade may also vary within the footwall, not only as a function of structural level exposed, but also owing to the complex pressure-temperature (P-T ) paths that rocks fol-low before and during extension. It is important to determine the thermal state of the lithosphere during extension because the ambient synexten-sion geothermal gradient is an important factor in controlling the evolution of core complexes, as well as the role of crustal fl ow in creating and maintaining a fl at Moho, as is observed in many highly extended regions (Block and Royden, 1990; Buck, 1991; Rey, 1993) (Fig. 4).

    The footwall of most core complexes has a domal structure (Fig. 3). The domal geom-etry has been explained for some continen-tal core complexes as resulting from uplift of the detachment during isostatic rebound as the hanging-wall rocks are extended, thinned, and denuded by tectonic and erosional processes (Spencer, 1984; Buck, 1988). These models as-sumed that the lower-crust fl owed at fast rates, whereas later studies considered how weak the lower crust must be to allow dome formation (Block and Royden, 1990; Kruse et al., 1991; Wdowinski and Axen, 1992; McKenzie et al., 2000; Rey et al., 2009a, 2009b, 2011).

    Evidence for fl ow of weak crust can be seen in the complex internal structure of gneiss (or mig-matite) domes, which occur within some conti-nental core complexes. Some core complexes contain one or more gneiss domes, typically beneath a carapace of high- to medium-grade metamorphic rocks (Brun and van den Driessche, 1994; Vanderhaeghe and Teyssier, 2001; Whitney et al., 2004b); these have been called migmatite-cored metamorphic core complexes (Rey et al., 2009a, 2009b), and their origin relates to regional extension and fl ow of deep crust beneath detach-ment faults. Studies have shown that the crystal-lization of the magmatic portions of migmatite domes in core complexes was followed by rapid cooling to T < 300 C (Fig. 6); cooling ages co-incide with ages of synkinematic minerals (e.g., mica) in detachment fault zones (Malavieille et al., 1990; Maluski et al., 1991; Kruckenberg et al., 2008). These results show that high-tem-perature crustal fl ow occurred during core-com-plex formation but ended when hot rocks reached shallow crustal levels and cooled rapidly.

    Gneiss (migmatite) domes commonly dis-play a relatively simple external surface but a complex internal structure (e.g., Kruckenberg et al., 2011) that varies with level of exposure.

    Some domes have a double-dome pattern con-sisting of two main compartments defi ned by foliation divided by a steep, median high-strain zone (Fig. 4C1); examples include the Naxos (Greece) and Montagne Noire (France) core complexes (Rey et al., 2011). Other domes show nappe-like recumbent folds that overprint earlier, steeper structures (e.g., McFadden et al., 2010). These double-dome and more complex three-dimensional structures are signifi cant for understanding crustal fl ow under extension, and they provide a framework for interpreting defor-mation features as a function of time and space.

    Numerical modeling provides insights to help understand the relative infl uence of geothermal gradient, crustal thickness, and crustal fl ow in core-complex initiation and evolution (e.g., Tirel et al., 2008; Rey et al., 2009a, 2009b). For example, Tirel et al. (2008) showed that for the model assumptions and parameters used, a Moho T > 800 C is required to produce a core complex in 60-km-thick crust. At T < 800 C, a strong upper mantle contributes to crustal-scale boudinage (necking). Insights from numerical modeling relevant to the fl ow of deep crust, as well as the origin of double-domes, are dis-cussed in a later section (Dynamic Models of Continental Core-Complex Development).

    Although some core complexes are associ-ated with regions of low heat fl ow (e.g., south-ern California and Arizona; Lachenbruch et al., 1994), others are associated with regions of high heat fl ow (e.g., the islands of Naxos and Paros, in the central Aegean; Gautier et al., 1993; Keay et al., 2001; Brichau et al., 2006; Seward et al., 2009). These Aegean islands contain cores of syntectonic migmatite and granite and have been proposed as the site of a thermal anomaly. However, these core complexes do not neces-sarily represent an anomalously hot region of the crust. Instead, they may have formed in a zone of large-scale extension or transtension that triggered the ascent of hot, ductile crust that fl owed from deep to shallow levels and became involved in the high-strain zone beneath the de-tachment, where isotherms collapsed (Krucken-berg et al., 2011; Rey et al., 2011). According to this idea, hot ductile crust was present region-ally, but only locally exhumed.

    Interpretation of the P-T conditions and paths of metamorphism relevant to core-complex de-velopment (e.g., Buick and Holland, 1989) re-quires knowledge of the age of metamorphic events that affected footwall rocks. As noted, footwall rocks may have experienced multiple metamorphic events prior to metamorphism related to extension and core-complex develop-ment, so care must be taken (particularly with zircon) to determine the age of the last metamor-phic event. In some Cenozoic core complexes,

    for example, zircons in gneiss yield pre-Cenozoic ages, recording protolith crystallization and/or later metamorphism associated with pre-exten-sion crustal thickening (e.g., Kruckenberg et al., 2008). The age of extension and accompanying exhumation of the footwall rocks is indicated by the youngest U-Pb zircon and monazite ages, and is further bracketed by cooling ages deter-mined by thermochronometers such as 40Ar/39Ar for hornblende, micas, and K-feldspar (Fig. 6). In migmatite-cored core complexes, melt that collected in boudin necks or extensional shear zones may provide an indication of the timing of onset and/or cessation of major extension (Gordon et al., 2008; McFadden et al., 2010).

    Even in cases in which the extension-related temperature-time path of footwall rocks is known, it can nevertheless be challenging to understand the pressure (depth) evolution of footwall rocks, and in particular it is diffi cult to determine the maximum pressure experienced by rocks exhumed in a core complex; however, this is important information for reconstructing pre-extension crustal thickness and unravel-ing particle paths in footwall rocks. Some core complexes exhume rocks from great depths, e.g., high-P rocks such as eclogite or kyanite-bearing gneiss, although much of the core-complex footwall may consist of metamorphic rocks recording only a low-P, high-T history (Rey et al., 2009a, 2011). In some cases, high-P rocks exhumed in a core complex record a much older (pre-extension) metamorphic history (e.g., Precambrian eclogite in the Cenozoic Menderes core complex, western Turkey; Candan et al., 2001), although it is likely that exhumation of the high-P rocks occurred during extension, and therefore the presence of these rocks is relevant to understanding core-complex dynamics. It is therefore particularly important to know the P-T-time history of these rocks so as to be able to track the timing, magnitude, and paths of their exhumation.

    Some conceptual and numerical models for core-complex development assume that the major motion of footwall rocks is horizontal, except for some vertical motion related to arch-ing of the footwall beneath the detachment (Brun and van den Driessche, 1994). Based on this as-sumption, predictions are made about expected changes (or lack of changes) in metamorphic grade in footwall rocks in the direction of tec-tonic transport for core complexes with different thermal histories (Gessner et al., 2007). This as-sumption is valid for some core complexes but not for those in which rocks registered a major component of vertical motion, as shown by a record of isothermal decompression, e.g., from >2030 km to

  • Continental and oceanic core complexes

    Geological Society of America Bulletin, Month/Month 2012 11

    nental core complexes, this decompression has been interpreted as driven by extension (Soto and Platt, 1999; Viruete et al., 2000; Norlander et al., 2002). Additional insights about pressure history come from observations of reaction tex-tures in footwall metamorphic rocks (Krucken-berg and Whitney, 2011; Goergen and Whitney, 2012) and from numerical modeling (Ruppel et al., 1988; Rey et al., 2009a, 2009b).

    Temperature-time paths for footwall rocks during extension-related exhumation can be deter mined by application of an array of thermo-chronometers with different closure tempera-tures in different minerals (cf. similar methods applied to detachment zone rocks; Fig. 6). In-terpretation of these data in the context of the exhumation path (i.e., changes in depth accom-panying cooling) requires estimates of the geo-thermal gradient during extension. Similar to the T-time history of some detachment zones, footwall rocks (below the detachment zone) may record rapid cooling (>50 C/m.y.) fol-lowed by a more protracted cooling history (

  • Whitney et al.

    12 Geological Society of America Bulletin, Month/Month 2012

    Another possibility is that the detachment con-tinues along-axis, but it is covered by extrusive rocks and rider blocks (Escartn et al., 2008; Reston and Ranero, 2011).

    In some regions of the Mid-Atlantic Ridge, the hanging wall is affected by intense hydro-thermal circulation (Fig. 1B); this has led to the suggestion that hydrothermal activity at slow-spreading systems relates to the dynamics of the underlying detachment fault and does not directly relate to magmatism (Petersen et al., 2009). However, it has also been proposed that the detachment serves as a conduit for fl uids and links the fractured hanging wall and associated hydrothermal fi elds at the ocean fl oor to magma chambers at depths of ~510 km (e.g., Escartn et al., 2003; McCaig et al., 2007, 2010; McCaig and Harris, 2012).

    Intrusion and extrusion of basalt clearly add to the hanging wall. Extrusion is important in the formation of hanging-wall rider blocks (Reston and Ranero, 2011; Choi and Buck, 2012), and the formation of oceanic detach-ments may depend on the amount of dike intru-sion into the hanging wall (Buck et al., 2005). This topic is discussed more in the section Dynamic Models of Oceanic Core-Complex Development.

    Oceanic Core Complexes: Detachment Fault Characteristics and Processes

    Geological sampling of oceanic detachments by coring and dredging indicates that the zone of brittle deformation in some core complexes may be

  • Continental and oceanic core complexes

    Geological Society of America Bulletin, Month/Month 2012 13

    Oceanic core-complex detachments gener-ally display a domal structure as well as shorter-wavelength corrugations with axes parallel to the extension direction; these corrugations have amplitudes up to several hundred meters (Cann et al., 1997; Tucholke et al., 1998, 2008; Ranero and Reston, 1999; Cannat et al., 2006). The longer wavelengths, which generate domal oceanic core complexes, may be related to mag-matic versus tectonic extension, and the shorter wavelength (corrugations) may be related to variations in both space and time of melt bodies that feed the spreading axis (Lin et al., 1990) and that ultimately control the rheology of detach-ment root zones.

    Fault rocks associated with detachment faults in some oceanic core complexes record similar temperatures to those in continental core complexes: ~550300 C (amphibolite to greenschist facies; Karson, 1999); in others (Atlantis Bank, Southwest Indian Ridge), defor ma tion conditions ranged from granulite to greenschist facies (from >900 C to

  • Whitney et al.

    14 Geological Society of America Bulletin, Month/Month 2012

    spreading center is between ~3500 and 4800 m depth. Shallower seafl oor is likely to have too large a magma supply, and deeper seafl oor is likely to have too little magma. Fast-spreading ridges have very thin lithosphere (~1 km thick), so even a modest supply of magma could ac-commodate all lithospheric spreading. For oce-anic core complexes to form, the magma supply apparently cannot be too high or too low (i.e., the Goldilocks condition of Tucholke et al., 2008). If faults on opposite sides of the axis are active simultaneously and slip at the same rate, then a core complex can form on both sides of the ridge (even in cases in which the plate sepa-

    ration rate is not accommodated by magma-tism in the form of dike opening, as suggested by Schouten et al., 2010). However, any rate-dependent weakening (or healing, as in Buck et al., 2005) leads to only one fault being active at a given time. The inference that low values of magma supply result in a complex pattern of moderate-offset, crosscutting faults argues for there being one dominantly active fault at a time in the axial region (Cannat et al., 2006).

    However, as little of the seafl oor is mapped with suffi cient resolution to detect oceanic core complexes, it is likely that they are more com-mon than the Tucholke et al. (2008) study in-dicates. For example, the correlation of oceanic core complexes with higher than normal rates of seismicity and hydrothermal activity may indi-cate that ~50% of a long (75100 km) section of the Mid-Atlantic Ridge could involve detach-ments (Smith et al., 2006; Escartn et al., 2008).

    CORE COMPLEXES AND LITHOSPHERE DYNAMICS

    Core complexes form in regions of extension driven by surface forces at plate boundaries, or volume forces in relation to lateral variation of gravitational potential energy, or both. World-wide, most core complexes form in regions of extension in collapsed orogens and at relatively slow-spreading mid-ocean-ridge systems. Oro-genic collapse may occur under free boundary conditions, such as driven by slab rollback at convergent margins, or under a fi xed bound-ary, owing to spreading of thick, weak crust (Rey et al., 2001). In these settings, extension is lithosphere scale, but the major expression of extension may be in the crust because normal faulting of the brittle upper crust is coupled with ascent of the ductile crust, resulting in regions of extreme thermal and strain gradients across detachment zones (meters to ~1.5 km thick).

    During lithospheric extension, strain is natu-rally partitioned in the crust into weak fault zones in the brittle layers, and shear zones and homogeneous strain in the lower crust. Strain localization may result from physical processes involving thermodynamic energy fl uxes even in the absence of any particular rheological anomaly. Whatever its origin, strain localiza-tion in the upper crust is essential to initiate core complexes, and it is easily achieved around rheological and/or density anomalies such as a preexisting fault in the brittle upper crust (Buck, 1993; Lavier et al., 1999; Koyi and Skelton, 2001; Gessner et al., 2007), a rheological and/or density anomaly in the lower crust (Brun et al., 1994; Tirel et al., 2004, 2008), or a strong den-sity discontinuity along the brittle-ductile transi-tion (Wijns et al., 2005).

    In the rest of this section, we fi rst use model results that focus on the brittle layer to explore the mechanics that lead to the development of detachment faults. We then examine more com-plete lithospheric models in which the coupling of rheologically realistic layers is investigated (brittle, temperature-dependent viscous, par-tially molten) for the case of continental core complexes, followed by discussion of the dy-namics of oceanic core complexes. Regard-ing the broader geodynamic settings in which continental core complexes form, we then ad-dress two cases: (1) the role of mantle wedge dynamics in Cordilleran-type orogens, where continental core complexes develop in the con-tinental overlying plate; and (2) the transition between an orogenic plateau and its foreland.

    Mechanics of Core-Complex Faults

    Mohr-Coulomb fracture mechanics imply that faulting occurs most easily at an angle of ~30 to the maximum principal stress (e.g., Anderson, 1942). Assuming an Andersonian extensional stress fi eld in which the minimum principal stress is horizontal, normal faults in the brittle upper crust should initiate at dips ~60 (Anderson, 1942) and should be active at dips of no less than 30 (e.g., Sibson, 1985). Several authors have suggested that normal faults could initiate with low dips if particular loads reorient the tectonic stress fi eld (Spencer and Chase, 1989; Yin, 1989; Parsons and Thompson, 1993). However, Wills and Buck (1997) showed that even with these purpose-built stress fi elds, low-angle faults would not form before high-angle faults.

    An alternative to slip on low-angle normal faults is that the upper parts of some actively slipping high-angle normal faults rotate to shal-lower dips. Spencer (1984) suggested that the isostatic response to offset of a low-angle nor-mal fault would tend to decrease the dip of the fault. Although there is ample fi eld evidence to suggest that detachment faulting occurs on low- angle faults (e.g., Scott and Lister, 1992; John and Foster, 1993), large rotation of high-angle faults is consistent with structures seen in many continental core complexes (Hamilton, 1988; Wernicke and Axen, 1988; Buck, 1988). More recent paleomagnetic studies of oceanic core complexes have demonstrated a minimum of 4050 rotation of footwall rocks below the detachment (Morris et al., 2009; MacLeod et al., 2011), validating the rolling-hinge model for low-angle normal and fault formation asso-ciated with plate spreading at slow and ultraslow mid-ocean ridges.

    In their conceptual rolling-hinge models, Wernicke and Axen (1988) assumed local

    crus

    t

    moho

    mantlelithosphere

    melt

    shear zones

    crus

    t

    axial valley

    moho

    corrug

    ated

    seafloo

    rvolcanic

    layer

    mantlelithosphere

    axial valley

    crus

    t

    moho

    0

    5

    ~20 km

    volcanic seafloor

    dike

    mantlelithosphere

    0

    5

    ~20 km

    0

    5

    ~20km

    A VOLCANIC-VOLCANIC SEAFLOOR PAIR

    B CORRUGATED-VOLCANIC SEAFLOOR PAIR

    C SMOOTH-SMOOTH SEAFLOOR PAIR

    (after Cannat et al., 2006)

    Figure 7. Sketches of axial regions for three proposed modes of slow to ultraslow spread-ing, shown in order of inferred decreas-ing melt supply: (A) volcanic seafl oor and (B) corrugated seafl oor occur at both slow- and ultraslow-spreading ridges, whereas (C) smooth seafl oor occurs only at ultraslow-spreading ridges. Horizontal dimensions are ~80 km across axis and ~40 km along axis (modifi ed from Cannat et al., 2006).

    as doi:10.1130/B30754.1Geological Society of America Bulletin, published online on 21 December 2012

  • Continental and oceanic core complexes

    Geological Society of America Bulletin, Month/Month 2012 15

    isostasy , whereas Buck (1988) calculated the fl exural response of lithosphere to loads caused by the offset of a high-angle normal fault. Nor-mal fault offset is supported regionally by fl ex-ure, and since the Moho remains fl at in many continental core complexes owing to fl ow of low-viscosity lower crust, the elastic lithosphere in which bending occurs is restricted to the upper crust (Masek et al., 1994). By analogy with bending at subduction zones (e.g., McAdoo et al., 1978; Bodine et al., 1981), applying a reasonable rock yield strength (the stress needed to break and slip on a fault) based on labora-tory measurements for an ~10 km brittle layer gives the observed range of domal wavelengths and topographic relief (Buck, 1988). For oce-anic core complexes, thin crust combined with a small density contrast between crust and mantle reduce the importance of crustal thickness varia-tions compared to topographic variations.

    The energetics of fault offset and layer bend-ing (Forsyth, 1992) predict that the initial orien-tation of a fault corresponds to the least friction on the fault per unit of horizontal displace-ment. This energy or work approach yields the same initial orientation for a fault as the clas-sic Ander sonian stress analysis, but when the fault records signifi cant slip, work is done in the bending of the displaced layer. An analytical es-timation of this extra work using an approxima-tion of the lithosphere as a thin, perfectly elastic layer fl oating on an inviscid substrate (Forsyth, 1992) showed that an initially low-angle normal fault could accumulate much more offset than a high-angle fault. If this elastic plate model is correct, the initial dip of a normal fault has to be extremely low to accommodate tens of kilome-

    ters of offset. Such a fault could only initiate on a preexisting very weak zone.

    The inclusion of realistic yield stresses de-creases the wavelength of fault and footwall bending and radically lowers the work that results from fault-related topography (Buck, 1993). A reasonably weak fault may accommo-date an offset suffi ciently large that the inactive part of the fault rolls over and becomes fl at. A fault that is not suffi ciently weak is replaced by another fault before large offset develops (Buck, 1988, 1993). Faults may weaken as they record slip, and the amount of fault weakening needed to allow large fault offset increases lin-early with brittle layer thickness (Buck, 1993). For example, a fault affecting a 10-km-thick layer would have to weaken by ~10 MPa to de-velop a large offset.

    Analog models are useful for simulating the early, small-offset stage of fault development (e.g., Brun et al., 1994; Tirel et al., 2006), but they cannot easily simulate the thermally con-trolled strength evolution that is likely to affect large-offset faults. Early two-dimensional nu-merical studies (e.g., King and Ellis, 1990) assumed a preexisting weak normal fault em-bedded in a purely elastic layer and solved for the topographic relief and stress changes when the fault recorded slip. Given the potential importance of the fi nite brittle yield strength (plastic deformation), early models treated the lithosphere as a viscous-plastic layer and were not concerned with fault localization (Braun and Beaumont, 1989; Bassi, 1991). Subsequent extension models of a viscous-elastic-plastic layer incorporated strain weakening as a func-tion of strain to promote normal fault devel-

    opment (Poliakov and Buck, 1998; Buck and Poliakov, 1998); these models showed that a sequence of high-angle faults might form and accommodate extension at a simple mid-ocean-ridge structure. When more strain weakening is allowed, large- offset faults develop (Lavier et al., 1999, 2000) (Fig. 8). If the amount of fault strain weakening is dependent on the thickness of the brittle layer being extended, then large-offset faults only develop when a layer is thinner than a critical thickness.

    The behavior and characteristics of core com-plexbounding faults may inform not only how much faults weaken, but also how they weaken with offset (Lavier et al., 2000). If faults weaken too fast, the entire layer shatters, and bending around a fi rst fault gives rise to secondary faults that delocalize extension. If faults weaken too slowly with offset, then the initial fault does not become suffi ciently weak before the extra resis-tance to slip owing to topography makes it easier for another fault to break elsewhere. A thin layer can generate a large-offset fault, while multiple faults develop in a thicker layer (Fig. 9; Lavier and Buck, 2002). This is consistent with the ob-servation that large-offset normal faults are only seen in areas of higher-than-average heat fl ow, where one expects thin brittle crust. In areas of high heat fl ow and thick crust, the lower con-tinental crust may fl ow easily (e.g., Block and Royden, 1990; Buck, 1991), allowing a single large-offset fault (detachment) to accommodate signifi cant extension.

    Recent work suggests that relatively small features of core complexes may help bound the amount of fault weakening with offset. Kilome-ter-scale allochthonous rider blocks that are cut

    1050

    plastic strain

    vertical exaggeration 1.5 : 1

    -50 0 +50distance (km)

    1500

    -1500

    0

    1500

    -1500

    0

    1500

    -1500

    0

    vertical exaggeration 3.0 : 1

    dept

    h (k

    m) 0

    5

    10

    dept

    h (k

    m) 0

    5

    10

    dept

    h (k

    m) 0

    5

    10

    Incr

    easi

    ng e

    xten

    sion

    STRAIN TOPOGRAPHY (m)

    -50 0 +50distance (km)

    -50 0 +50distance (km)

    Figure 8. Left panels: Results of numerical model of extension of a fl oating brittle Mohr-Coulomb layer with a single seeded normal fault (top panel). Right panels: Corresponding topographic profi les with vertical exaggeration. Progressive extension is suffi cient to allow foot-wall to rotate; abandoned parts of fault rotate to, and even past, horizontal (bottom-left panel). Cohesion loss of fault is function of strain (decrease of 1/3 of initial brittle yield strength of layer) and occurs linearly with fault offset up to 1.5 km (from Lavier et al., 2000).

    as doi:10.1130/B30754.1Geological Society of America Bulletin, published online on 21 December 2012

  • Whitney et al.

    16 Geological Society of America Bulletin, Month/Month 2012

    off from the hanging wall and related basin in-fi ll, and are transported on top of the footwall, are common in both continental and oceanic core complexes (e.g., Reston and Ranero, 2011). Rider blocks superimposed on large-offset nor-mal faults may not form if a fault loses too much strength (Choi and Buck, 2012); a narrow range of fault weakening relative to intact surrounding rock allows the formation of large-offset faults with rider blocks. These blocks form when the dominant form of weakening is by reduction of fault cohesion, whereas faults that weaken primarily by friction reduction do not generate distinct rider blocks. Either a lack of infi ll or an extreme reduction of friction by serpentinization of exhumed mantle rocks may explain the lack of rider blocks on some oceanic core complexes.

    Dynamic Models of Oceanic Core-Complex Development

    Magmatism and associated hydrothermal effects are likely to be signifi cant factors in oceanic core-complex development, but the in-clusion of magmatism in numerical models is at an early stage. The input of limited amounts of gabbro may be needed to allow alteration of peridotite to weak minerals, as silica-rich magmatic fl uids interact with peridotite (e.g., Ildefonse et al., 2007). Such weakening would allow oceanic core-complex detachments to slip at lower than normal levels of stress, but so far such weakening has not been included in spreading center models, and most modeling effort has focused on dikes.

    One simple way of treating the effect of dike intrusion in numerical models of long-term lithospheric extension was developed by Buck et al. (2005). The rate of dike opening was

    specifi ed by the fraction M of the plate separa-tion rate that is accommodated by dike opening. For M = 0, dikes account for none of the plate spreading; for M = 1, they accommodate all of it. Dikes may supply much of the heat that keeps the axis hotter and the axial lithosphere thinner than the lithosphere farther from the axis.

    A simple geometric argument shows how faults and dikes interact at a ridge with a fi xed position of diking and fi xed thermally defi ned strength structure. If one fault forms due to lithospheric stretching, it should initially cut the thinnest axial lithosphere on one side of the axis. If the fault moves away from the axis into thicker lithosphere, it becomes more diffi cult for this fault to accommodate deformation, even though it is weaker than the surrounding litho-sphere. Eventually it will be easier to form a new fault cutting the axis, and the fi rst fault will be replaced by a new fault.

    Models of large-offset normal faulting have evaluated the infl uence of M on fault behavior (Lavier et al., 1999). For M = 0.5, the hanging wall of the fault does not move away from the axis, so the fault could build up potentially un-limited offset. For M close to 1, the maximum fault offset can be small, whereas for M close to 0.5, the offset should be very large. A numeri-cal model of diking and stretching (Buck et al., 2005) tested this conceptual model using the nu-merical approach of Poliakov and Buck (1998). Results show that normal fault offset varies greatly as a function of M; large fault offset occurs when M = 0.5; for M = 0.95, the model generates a fairly symmetric pattern of mainly inward-dipping, small-offset faults, and a sym-metric axial valley. For values of M < 0.5, the active fault gradually moves across the spread-ing axis and may be cut by later faults. How-

    ever, the fi xed temperature structure assumed by Buck et al. (2005) is questionable when there is little magma input. If the temperature structure is strongly affected by the advection associated with fault offset, then the weakest part of the lithosphere migrates with the active fault. Since the sensible and latent heat of intruding dikes should dominate over this effect, the simple symmetric strength structure assumed in Buck et al. (2005) should be valid for M >~0.4.

    Tucholke et al. (2008) included a more realis-tic, evolving temperature structure that consid-ered advection and diffusion of heat and latent heat of crystallization. This study showed that large-offset faults could form for a range of M values between ~0.3 and 0.6 (M = 0.5 shown in Fig. 10). As noted earlier, several sets of obser-vations indicate that oceanic core complexes de-velop when the supply of magma to a spreading center is not too high and not too low. Olive et al. (2010) found that the formation of large-offset faults and oceanic core complexlike structures depends more on the rate of dike opening than on the rate of gabbro intrusion. There appears to be a reasonable correspondence between the M-based models and the inference of similar modes of spreading based on geological and geophysical data from slow-spreading ridge systems (Cannat et al., 2006, 2009).

    A potential problem with these simple mod-els concerns the rate of oceanic detachment faulting that has been estimated using magmatic and thermochronologic data. For example, in the Atlantis Massif, much of the full plate spreading rate was taken up on the oceanic de-tachment (Grimes et al., 2008). Estimates of the rate of oceanic detachment slip range from ~14 km/m.y. at Atlantis Bank (Baines et al., 2008), to ~24 km/m.y. at Atlantis Massif (Grimes

    0 50 100 150

    10

    30

    20

    0

    dep

    th (k

    m)

    Distance (km)

    250C500C

    700C

    0 50 100 150

    10

    30

    20

    0

    Distance (km)

    250C

    500C

    700C

    0

    5

    4

    0

    5

    4

    top

    og

    rap

    hy

    (km

    )

    Total Strain0.7

    1.4

    2.1

    2.8Total Strain

    0.7

    1.4

    2.1

    2.8

    Thin lithosphere Thick lithosphere

    Figure 9. Results of numerical model calculations for extension of thin and thick brittle lithosphere. Rheology is viscous-elastic-plastic; viscous strength depends strongly on temperature. Total strain is the square root of the second invariant of the strain tensor. Hydrothermal circulation cools the shallow lithosphere and infl uences the temperature fi eld. Extension of hot, thin lithosphere leads to single large-offset fault; extension of cooler and thicker layer results in multiple normal faults (from Lavier and Buck, 2002).

    as doi:10.1130/B30754.1Geological Society of America Bulletin, published online on 21 December 2012

  • Continental and oceanic core complexes

    Geological Society of America Bulletin, Month/Month 2012 17

    et al., 2008), and up to ~38 km/m.y. at Godzilla Megamullion (Ohara et al., 2001). John and Cheadle (2010) concluded that, during the time period when the oceanic detachments were ac-tive, they accounted for 60%100% of the total plate spreading. This would give M values be-tween 0.0 and 0.4; these values are lower than predicted to produce large-offset faults.

    Dynamic Models of Continental Core-Complex Development

    Localized thinning of the continental upper crust by extension is isostatically compensated by the fl ow of deep crust into core complexes (Block and Royden, 1990; Wdowinski and Axen, 1992). Partially molten crust is particu-larly likely to facilitate core-complex develop-ment during extension owing to the dramatic decrease in viscosity associated with the pres-ence of melt. In the case of fl uid-absent melt-ing reactions, which can be encountered during heating and/or decompression, a positive feed-back between extension-induced decompres-sion and partial melting may generate migmatite

    domes within continental core complexes (Teys-sier and Whitney, 2002; Whitney et al., 2004b).

    Physical experiments show that a strongly coupled upper and lower continental crust leads to distributed surface extension, whereas me-chanical decoupling between weak lower crust and much stronger upper crust leads to localized surface extension that favors continental core-complex development (Brun et al., 1994). The magnitude of coupling is strongly dependent on the geothermal gradient. In a cold to nor-mal geothermal gradient, the lower crust is me-chanically coupled to the upper crust and upper mantle. In this case, strain in the lower crust is a response to plate boundary forces. In contrast, under hotter conditions, the lower crust is weak and fl ows in response to both tectonic stresses and gravitational stresses.

    Numerical modeling that integrates tempera-ture-dependent viscosity (Fig. 11) addresses the competition between plate boundary extension rate and the rate at which ductile crust can fl ow. The fl ow of ductile crust is driven by gravita-tional stresses and is controlled by the viscosity (temperature) of lower crust and the gradient in

    gravitational potential energy. Under high ex-tension rate and low surface heat fl ow, the duc-tile crust is too strong to respond to gravitational stresses over short time scales. In this case, two-dimensional models predict that the lower crust extends and thins rather homogeneously (Fig. 11, a1, with TMoho = 600 C), redistributing stresses uniformly in the upper crust and upper mantle, where numerous normal faults develop. That is, extension under relatively cool geother-mal conditions results in more normal faults in the brittle upper crust (graben, half graben, and rotated blocks) and relatively homogeneous extension in the ductile crust (Fig. 11A). The ductile crust thins and fl ows to accommodate the topography of the brittle-ductile transition and that of the Moho. Major normal faults in the upper crust evolve into strong shear strain gradients in the lower crust (Figs. 11AB, a1a2, b1b2). In the ductile crust, fl at foliations domi-nate, although some thin regions of vertical foliation occur underneath major normal faults (cf. Fig. 11, a2, b2).

    When the geotherm is warmer, the lower crust is able to fl ow over short time scales in

    M = 0

    M = 0.5

    M = 0.7

    0102030 10 20 30

    0

    10

    5

    dept

    h (k

    m)

    0

    10

    5

    dept

    h (k

    m)

    0

    10

    5

    dept

    h (k

    m)

    distance (km)

    magma injection zone

    strain rate (s1)14 1216

    magmatic accretion seafloor faulted surface seafloormodel original seafloor

    600C

    600C

    600C

    0.9 m.y.

    0.9 m.y.

    0.8 m.y.

    Figure 10. Snapshots of modeled fault behavior and seafl oor morphology for values M = 0, 0.5, and 0.7; model allows thermal evolution throughout run (after Tucholke et al., 2008). Structural interpretation is superimposed on modeled distribution of strain rate (1016 to 1012 s1); model time is indicated in panels at lower right; dashed white line at bottom is 600 C isotherm and approximates the brittle-ductile transition; dashed seafl oor is original model seafl oor, red seafl oor is that formed dominantly by magmatic accretion, and solid bold seafl oor is fault surface.

    as doi:10.1130/B30754.1Geological Society of America Bulletin, published online on 21 December 2012

  • Whitney et al.

    18 Geological Society of America Bulletin, Month/Month 2012

    response to gravitational stresses (Fig. 11, a2, with TMoho = 700 C). Higher geothermal gra-dients result in more strongly localized ex-tension in the upper crust, driving fl ow in the deep crust. This illustrates that the amount of coupling between the upper and lower crust is controlled by the crustal-scale temperature distribution (Block and Royden, 1990; Lavier and Buck, 2002). As the lower-crust viscosity diminishes (for example, when melt fraction increases), the lower crust fl ows upward be-neath the region of upper-crustal necking; the ductile crust is hot and suffi ciently weak to be decoupled from the upper crust. This upward fl ow generates horizontal fl ow in two channels that converge toward the zone of extension and move material in a direction opposite to the mo-tion of the brittle upper crust and upper mantle. This strongly partitioned fl ow results in the formation of contractional structures (upright

    folds, nappes) beneath the zone of upper-crust extension (Rey et al., 2011). Therefore, exten-sion in the upper crust is coeval with contrac-tion in the deep crust. The two-dimensional structure beneath the dome-shaped detachment fault consists of a subvertical high-strain zone that separates two subdomes, creating a double dome (Figs. 4C and 11).

    Double domes (or more complex dome geom-etries in three dimensions) generated in this way may explain the presence of relict high-P metamorphic rocks (typically dismembered layers and pods) in domes that otherwise record low-P, high-T emplacement. Two-dimensional model results predict that deep crustal rocks (eclogite, granulite) and possibly lithospheric mantle will be entrained and carried upward in steep high-strain zones. These simple dy-namic models are run under steady extension but produce a complex deformation sequence

    for material that is exhumed from the lower crust. These deformation stages occur at vari-ous P-T conditions and include shearing in the lower-crust channel, horizontal contraction in the domain of collided channels, and horizontal extension when material moves up beneath the detachment zone.

    In high-geotherm settings, upward fl ow of low-viscosity material is localized, and there-fore the lower crust infl uences upper-crustal stress and strain only in the extended region (necking of upper crust). Upward advection of heat promotes surface fl uid circulation (Mulch et al., 2004; Person et al., 2007; Gbelin et al., 2011), which contributes to further strain local-ization in the upper crust. In the brittle upper crust, extensional strain is strongly focused, and the number of normal faults is limited; typically a single large-offset detachment fault accom-modates upper-crustal extension (Lavier et al.,

    TMoho = 600C

    TMoho = 700C

    TMoho = 800C

    TMoho = 900C

    45 km

    50 km

    55 km

    60 km

    TMoho = 617C

    TMoho = 702C

    TMoho = 767C

    TMoho = 845C

    Temperature

    Viscosity

    Stressa1 b1

    b3

    air

    b2a2

    a3

    a4

    air

    brittle crust

    ductile crust

    mantle

    faults

    A B

    solidus

    VARIABLE TMoho (600900C) CRUSTAL THICKNESS 60 KM VARIABLE CRUSTAL THICKNESS (4560 KM)

    b4

    Figure 11. Infl uence of the geotherm/rheological profi le on strain distribution in continental core complexes, investigated using two-dimensional modeling (Ellipsis) to explore the effect of different geothermal gradients (expressed here as different Moho temperatures) on core-complex development. (A) A 60-km-thick continental crust, in variable thermal state, is submitted to 1.5 m.y. of symmetric exten-sion (1.13 cm/yr at both sides, i.e., 2 1015 s1). (a1, a2) Extension under a cool geothermal regime (i.e., without partial melting) leads to homo geneously distributed extension: There are more normal faults in the upper crust, and deformation in the lower crust is more homo-geneously distributed. (a3, a4) In contrast, warmer geotherms lead to more strongly localized extension in the upper crust and a more local-ized fl ow in the lower crust. The temperature regime controls the amount of coupling between the upper and lower crust. (B) Infl uence of crustal thickness on strain distribution. In this series, a continental crust of increasing thickness is submitted to extension (same velocities as A). The geotherm (steady state when crustal thickness = 40 km) is allowed to evolve for 30 m.y. before extension begins; extension lasts for 2 m.y. This experiment confi rms the dominant role of rheological layering on the strain fi eld.

    as doi:10.1130/B30754.1Geological Society of America Bulletin, published online on 21 December 2012

  • Continental and oceanic core complexes

    Geological Society of America Bulletin, Month/Month 2012 19

    2000). In this tectonic setting, buoyancy of the weak lower crust plays a second-order role, and core complexes form even when the lower crust is denser than the upper crust.

    Subduction Dynamics and Continental Core Complexes

    In active ocean-continent plate margins, the state of stress in the upper plate can rapidly switch from contractional to extensional de-pending on the interplay among (1) trench-nor-mal velocity, (2) friction along the subduction interface, (3) gravitational forces in the thick crust of the overriding plate, and (4) traction imposed at the base of the overriding plate by the buoyant mantle wedge (e.g., Billen, 2008). The effect of trench rollback on the stability of the overriding plate has been studied in detail (e.g., Faccenna et al., 2007; Becker and Fac-cenna, 2011), particularly for the Aegean region (slab rollback and breakoff; Jolivet and Brun, 2010). Here, we focus on the role of the mantle wedge because the transformation of strong lithospheric mantle into weaker and more buoy-ant mantle, driven in part by the release of fl uids from the subducting plate and partial melting of the wedge, must have a signifi cant effect.

    Under dynamic equilibrium (no deforma-tion in the overriding plate), there is a balance between driving forces (ridge push, slab pull, and gravitational forces) and resisting forces (generated by friction along the subduction in-terface as well as other viscous forces). Gravi-tational forces stored in the thick overriding plate