tectonic and metasomatic mixing in a high-t, subduction ...mdbarton/mdb_papers_pdf...lina schist,...

28
Tectonic and metasomatic mixing in a high-T , subduction-zone me ´lange — insights into the geochemical evolution of the slab–mantle interface Gray E. Bebout a, * , Mark D. Barton b a The Pheasant Memorial Laboratory, Institute for Study of the Earth’s Interior, Okayama University at Misasa, Misasa, Tottori-ken 682-0193, Japan b Center for Mineral Resources, Department of Geosciences, University of Arizona, Tucson, AZ 85721, USA Accepted 8 February 2002 Abstract The Catalina Schist (California) contains an amphibolite-grade (0.8 – 1.1 GPa; 640 – 750 jC) me ´lange unit consisting of mafic and ultramafic blocks in high-Mg, schistose me ´lange matrix with varying modal proportions of talc, chlorite, anthophyllite, calcic-amphibole, enstatite, and minor phases including zircon, rutile, apatite, spinel, and Fe – Ni sulfides. This me ´lange unit is interpreted as a kilometer-scale zone of tectonic and metasomatic mixing formed within a juvenile subduction zone, the study of which may yield insight into chemical mixing processes at greater depths in subduction zones. Relationships among the major and trace element compositions of the mafic and ultramafic blocks in the me ´lange, the rinds developed at the margins of these blocks, and the surrounding me ´lange matrix are compatible with the evolution of the me ´lange matrix through a complex combination of infiltrative and diffusional metasomatism and a process resembling mechanical mixing. Simple, linear mixing models are compatible with the development of the me ´lange matrix primarily through simple mixture of the ultramafic and mafic rocks, with Cr/Al ratios serving as indicators of the approximate proportions of the two lithologies. This conclusion regarding mafic – ultramafic mixing is consistent with the field observations and chemical trends indicating strong resemblance of large parts of the me ´lange matrix with rinds developed at the margins of mafic and ultramafic blocks. The overall process involved development of metasomatic assemblages through complex fluid-mediated mixing of the blocks and matrix concurrent with deformation of these relatively weak rind materials, which are rich in layer silicates and amphibole. This deformation was sufficiently intense to transpose fabrics, progressively disaggregate more rigid, block-derived materials in weaker chorite- and talc-rich me ´lange, and in some particularly weak lithologies (e.g., chlorite-, talc-, and amphibole-rich materials), intimately juxtapose adjacent lithologies at the (sub-)cm scale (approaching grain scale) sampled by the whole-rock geochemical analyses. Chemical systematics of various elements in the me ´lange matrix can be delineated based on the Cr/Al-based mixing model. Simple mixing relationships exhibited by Al, Cr, Mg, Ni, Fe, and Zr provide a geochemical reference frame for considerations of mass and volume loss and gain within the me ´lange matrix. The compositional patterns of many other elements are explained by either redistribution (local stripping or enrichment) at varying scales within the me ´lange (Ca, Na, K, Ba, and Sr) or massive addition from external sources (Si and H 2 O), the latter probably in infiltrating H 2 O-rich fluids that produced the dramatic O and H isotopic shifts in the me ´lange. Me ´lange formation, resulting in the production of high-variance ultramafic assemblages with 0009-2541/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved. PII:S0009-2541(02)00019-0 * Corresponding author. Permanent address: Department of Earth and Environmental Sciences, Lehigh University, Bethlehem, PA 18015, USA. Tel.: +1-610-758-5831; fax: +1-610-758-3677. E-mail address: [email protected] (G.E. Bebout). www.elsevier.com/locate/chemgeo Chemical Geology 187 (2002) 79 – 106

Upload: others

Post on 15-Feb-2021

3 views

Category:

Documents


0 download

TRANSCRIPT

  • Tectonic and metasomatic mixing in a high-T, subduction-zone

    mélange—insights into the geochemical evolution of the

    slab–mantle interface

    Gray E. Bebout a,*, Mark D. Barton b

    aThe Pheasant Memorial Laboratory, Institute for Study of the Earth’s Interior, Okayama University at Misasa, Misasa,

    Tottori-ken 682-0193, JapanbCenter for Mineral Resources, Department of Geosciences, University of Arizona, Tucson, AZ 85721, USA

    Accepted 8 February 2002

    Abstract

    The Catalina Schist (California) contains an amphibolite-grade (0.8–1.1 GPa; 640–750 jC) mélange unit consisting ofmafic and ultramafic blocks in high-Mg, schistose mélange matrix with varying modal proportions of talc, chlorite,

    anthophyllite, calcic-amphibole, enstatite, and minor phases including zircon, rutile, apatite, spinel, and Fe–Ni sulfides. This

    mélange unit is interpreted as a kilometer-scale zone of tectonic and metasomatic mixing formed within a juvenile subduction

    zone, the study of which may yield insight into chemical mixing processes at greater depths in subduction zones. Relationships

    among the major and trace element compositions of the mafic and ultramafic blocks in the mélange, the rinds developed at the

    margins of these blocks, and the surrounding mélange matrix are compatible with the evolution of the mélange matrix through a

    complex combination of infiltrative and diffusional metasomatism and a process resembling mechanical mixing. Simple, linear

    mixing models are compatible with the development of the mélange matrix primarily through simple mixture of the ultramafic

    and mafic rocks, with Cr/Al ratios serving as indicators of the approximate proportions of the two lithologies. This conclusion

    regarding mafic–ultramafic mixing is consistent with the field observations and chemical trends indicating strong resemblance

    of large parts of the mélange matrix with rinds developed at the margins of mafic and ultramafic blocks. The overall process

    involved development of metasomatic assemblages through complex fluid-mediated mixing of the blocks and matrix concurrent

    with deformation of these relatively weak rind materials, which are rich in layer silicates and amphibole. This deformation was

    sufficiently intense to transpose fabrics, progressively disaggregate more rigid, block-derived materials in weaker chorite- and

    talc-rich mélange, and in some particularly weak lithologies (e.g., chlorite-, talc-, and amphibole-rich materials), intimately

    juxtapose adjacent lithologies at the (sub-)cm scale (approaching grain scale) sampled by the whole-rock geochemical analyses.

    Chemical systematics of various elements in the mélange matrix can be delineated based on the Cr/Al-based mixing model.

    Simple mixing relationships exhibited by Al, Cr, Mg, Ni, Fe, and Zr provide a geochemical reference frame for considerations

    of mass and volume loss and gain within the mélange matrix. The compositional patterns of many other elements are explained

    by either redistribution (local stripping or enrichment) at varying scales within the mélange (Ca, Na, K, Ba, and Sr) or massive

    addition from external sources (Si and H2O), the latter probably in infiltrating H2O-rich fluids that produced the dramatic O and

    H isotopic shifts in the mélange. Mélange formation, resulting in the production of high-variance ultramafic assemblages with

    0009-2541/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved.

    PII: S0009 -2541 (02 )00019 -0

    * Corresponding author. Permanent address: Department of Earth and Environmental Sciences, Lehigh University, Bethlehem, PA 18015,

    USA. Tel.: +1-610-758-5831; fax: +1-610-758-3677.

    E-mail address: [email protected] (G.E. Bebout).

    www.elsevier.com/locate/chemgeo

    Chemical Geology 187 (2002) 79–106

  • high volatile contents, may aid retention of volatiles (in this case, H2O) to greater depths in subduction zones than in original

    subducted mafic and sedimentary materials. The presence of such assemblages (i.e., containing minerals such as talc, chlorite,

    and Mg-rich amphiboles) would impact the rheology of the slab–mantle interface and perhaps contribute to the low-velocity

    seismic structure observed at/near the slab–mantle interface in some subduction zones. If operative along the slab–mantle

    interface, complex mixing processes such as these, involving the interplay between fluid-mediated metasomatism and

    deformation, also could impact slab incompatible trace element and isotopic signatures ultimately observed in arc magmas,

    producing ‘‘fluids’’ with geochemical signatures inherited from interactions with hybridized rock compositions. D 2002

    Elsevier Science B.V. All rights reserved.

    Keywords: Subduction zone; Mélange; Slab–mantle interface

    1. Introduction

    Although geochemical studies of arc magmatism

    commonly call upon metasomatic ‘‘fluids’’ (hydrous

    fluids or silicate melts) as agents for transporting

    ‘‘slab signatures’’ to sites of arc magmatism (e.g.,

    Gill, 1981; Pearce and Peate, 1995), little is known

    about the structural and geochemical processes that

    operate along the slab–mantle interface. At shallow

    levels ( < 10 km), large volumes of dominantly sedi-

    mentary material accumulate in accreted wedges,

    occurring in many complexes as mélange-like mate-

    rial between the slab and the hanging wall (see Shreve

    and Cloos, 1986). The scarcity of appropriate expo-

    sures has made it difficult to infer the nature of the

    slab–mantle interface zone at greater depths. Recent

    geophysical study of the slab–mantle interface has

    identified zones of low seismic velocity at or near the

    top of the subducting oceanic lithosphere (e.g., Fukao

    et al., 1983; Hori et al., 1985; Matsuzawa et al., 1986;

    Helffrich et al., 1989; Helffrich, 1996; Helffrich and

    Abers, 1997). These low-velocity domains have gen-

    erally been interpreted as containing extensively

    hydrated assemblages in the subducting oceanic litho-

    sphere; however, Abers et al. (1999; also see discus-

    sion by Abers, 2000) have speculated that they might

    also represent hydrated hanging-wall materials. The

    generation of zones rich in layered hydrous minerals,

    perhaps in part a mechanical mixing zone, could also

    promote aseismic behavior at great depths in subduc-

    tion zones (Peacock and Hyndman, 1999). The Cata-

    lina Schist, exposed on Santa Catalina Island

    (California), contains a kilometer-scale amphibolite-

    grade, dominantly ultramafic, mélange (Fig. 1). Pet-

    rological considerations indicate that the mélange

    formed at 0.8–1.1 GPa and 640–750 jC (Sorensenand Barton, 1987; Sorensen, 1988). The mélange

    contains mafic and ultramafic blocks showing varying

    metasomatic alteration and, in the case of some mafic

    blocks, migmatization (Sorensen, 1988). The unit is

    thought to represent a zone of tectonic and metaso-

    matic mixing near the slab–mantle interface (Bebout

    and Barton, 1989), and may, thus, yield insights into

    the complex structural and metasomatic processes

    operating at depth in subduction zones.

    The amphibolite mélange unit was first mapped by

    Bailey (1941), who recognized that it consists predom-

    inantly of schists containing assemblages of talcFanthophylliteF chloriteF actinoliteF enstatiteFquartz. Sorensen (1988) and Sorensen and Grossman

    (1989), in their detailed petrologic and geochemical

    studies, documented metasomatic exchange between

    mafic blocks and the surrounding mélange matrix to

    produce ‘‘rinds’’ on these blocks and suggested that

    infiltrating aqueous fluids facilitated this exchange.

    Sorensen and Barton (1987) and Sorensen (1988)

    suggested that this high-T infiltration event also

    resulted in varying degrees of hydration and metaso-

    matism leading to migmatization of some mafic blocks

    floating in the mélange. Detailed isotopic studies of the

    mélange (Barton et al., 1987; Bebout, 1991) led to an

    infiltration model whereby large amounts of aqueous

    fluid, previously equilibrated with metasedimentary

    rocks, entered themélange, leading to large-scale stable

    isotope homogenization and producing abundant meta-

    somatic features within the mélange unit. In this paper,

    we combine petrological, field (including mapping),

    petrographic, and geochemical (major and trace ele-

    ment; isotopic) evidence relevant for consideration of

    the larger-scale development of this kilometer-scale

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–10680

  • mélange unit. We argue that the rind-forming process

    demonstrated by Sorensen (1988) and Sorensen and

    Grossman (1989) for mafic blocks, and illustrated in

    our study for ultramafic blocks, is representative of the

    initial mixing processes involving combinations of

    diffusive and infiltrative metasomatism and mechan-

    ical (tectonic) mixing, which ultimately resulted in the

    production of the far more voluminous mélange matrix

    in this mélange unit. Finally, we discuss some impli-

    cations of these mixing processes for rheology and

    geochemical hybridization along the forearc and subarc

    slab–mantle interface.

    2. Analytical techniques and normalization of

    mineral compositional data

    X-ray fluorescence (XRF) analyses were per-

    formed at the University of California, Los Angeles

    (for major elements) on a Phillips–Norelco instru-

    ment and at the University of Southern California (for

    trace elements) on a Rigaku instrument. Samples were

    disintegrated in a jaw crusher to a size range appro-

    priate for use of tungsten carbide and steel shatter

    boxes. Loss-on-ignition data were obtained by heating

    of samples in crucibles to 900 jC for 1 to 1.5 h;

    Fig. 1. Geologic map of a part of the amphibolite-facies mélange unit near the airport on Santa Catalina Island, illustrating the distributions and

    proportions of contrasting mélange matrix compositional types and tectonic blocks. Large serpentinite zones are, in general, enveloped by, or

    otherwise spatially related to, more siliceous ultramafic rocks similar in composition to metasomatic zones along shear zones and veins in

    serpentine-rich zones (see Fig. 2b). Dark lines are roads.

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–106 81

  • results were reproducible to F 0.1 wt.%, and the LOIfor nearly monomineralic whole-rock samples (e.g.,

    chlorite schists) agree to within 10% of the stoichio-

    metric H2O contents of the appropriate hydrous min-

    eral. Major element samples were fused with Li2B4O7at 1050 jC for 20 minutes. Samples for trace elementanalyses (for Ba, Sr, Cr, and Ni data) were prepared as

    powdered samples, bound by mixing sample and

    cellulose binder in a 4:1 weight ratio.

    Electron microprobe analyses were performed on a

    Cameca Camebax instrument at UCLA. All analyses

    were obtained at a 15 kV accelerating voltage. Elec-

    tron microprobe analyses were normalized using the

    schemes of Laird and Albee (1981) for amphibole and

    chlorite, and the nomenclature of Leake (1978) is used

    for amphibole.

    3. Geologic setting and field observations

    The Catalina Schist consists of tectonometamor-

    phic units ranging in grade from lawsonite–albite to

    amphibolite facies and is believed to represent pro-

    cesses of underplating at 15–45-km depths in an

    Early Cretaceous accretionary complex (Platt, 1976;

    Sorensen, 1986; Grove and Bebout, 1995). Each unit

    contains metasedimentary, metamafic, metaultramafic

    rocks, and mélange in varying proportions. The mél-

    ange domains contain intact, variably metasomatized

    blocks of mafic, ultramafic, and sedimentary litholo-

    gies within heterogenous, penetratively deformed

    matrices. In each unit, the mélange zones contain

    mineral assemblages consistent with equilibration at

    P–T conditions represented by the peak metamorphic

    mineral assemblages in more coherent (less deformed)

    sections of metasedimentary and metamafic rocks (see

    Grove and Bebout, 1995). The mélange zones have

    been identified as zones of preferential metamorphic

    fluid infiltration based on their highly homogenized

    stable isotope compositions (O, H, and N; see Bebout,

    1991, 1997) and the abundance of metasomatic fea-

    tures such as veins and layers with high-variance

    mineral assemblages (Bebout and Barton, 1993).

    The amphibolite-grade ultramafic mélange unit of

    the Catalina Schist occupies an approximately 8 km2

    area and is separated, along a sheared contact, from

    more coherent exposures of metagabbroic and meta-

    sedimentary schist and gneiss (see map of a part of this

    unit in Fig. 1). The mélange contains variably foliated

    assemblages of talc, chlorite, anthophyllite, clinoam-

    phibole (ranging in composition from actinolite to

    magnesiohornblende), enstatite, and (rarely) quartz.

    Trace phases include Ni–Fe sulfides, chromite–mag-

    netite, ilmenite, rutile, zircon, apatite, ankeritic dolo-

    mite, and clinopyroxene. On a large (tens of meters to

    kilometer) scale, the mélange consists of domains of

    more aluminous, dominantly chlorite-schist containing

    both mafic and ultramafic blocks (shown as ‘‘alumi-

    nous matrix’’ on Fig. 1) and more siliceous domains of

    dominantly amphibole (clinoamphibole and/or

    orthoamphibole)- or talc-schist which contain far fewer

    blocks of aluminous mafic material (shown as ‘‘sili-

    ceous matrix’’ on Fig. 1). Blocks of metasedimentary

    materials are conspicuous by their scarcity in the

    mélange unit.

    Fig. 2a demonstrates schematically the textural

    and mineralogical relationships between more alumi-

    nous parts of the ultramafic mélange matrix and

    blocks floating in the mélange, based in part on the

    work by Sorensen (1988) on the mafic blocks. Ultra-

    mafic blocks, likely to have been serpentinized dur-

    ing retrogradation of the Catalina Schist (see Grove

    and Bebout, 1995), are relatively homogeneous

    bodies (of up to kilometer scales) containing no mafic

    material and with whole-rock compositions suggest-

    ing dunite to harzburgite protoliths (see Sorensen,

    1988; Sorensen and Grossman, 1989; Bebout and

    Barton, 1993). Foliation in the mélange matrix com-

    monly wraps around blocks. The most common block

    types are the retrograded ultramafic blocks, variably

    metasomatized blocks of garnet + omphacitic clino-

    pyroxene (the latter described by Sorensen, 1988),

    and to a far lesser extent, metagabbroic gneiss blocks

    that lithologically resemble the coherent metagab-

    broic section of the amphibolite unit. The mafic

    and ultramafic blocks show a continuum of variable

    metasomatic exchange with the surrounding mélange

    matrix, evident by the presence of metasomatic

    ‘‘rinds’’ at the block margins, and the metasomatized

    lithologies show varying degrees of incorporation

    into the mélange foliation (see field photograph in

    Fig. 3 demonstrating the strong deformation of mate-

    rials interpreted as highly metasomatized block mate-

    rial). Sorensen (1988) described in detail the

    metasomatic assemblages in rinds developed on

    mafic blocks (see Fig. 2a), inferring fluid-mediated

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–10682

  • additions to the blocks of elements such as Rb +

    Th +BaF SrFREE. The less metasomatized maficblocks have spherical to discoidal shapes suggesting

    that they behaved as rigid objects within the mélange

    until they were extensively metasomatized to more

    easily deformed assemblages containing chloriteFamphibole. Discoidal masses dominantly of clinoam-

    phibole-rich material, strongly resembling (in miner-

    alogy and mineral chemistry) the rind material

    developed at the margins of mafic blocks, are com-

    mon in the more aluminous parts of the mélange (see

    Fig. 1). These discoids (e.g., samples 7-2-26b1 and 7-

    2-26b6, for which electron microprobe data are

    presented in Fig. 4 and Table 2) range in size from

    several mm’s to several m’s in diameter (see Fig. 3).

    At the thin section scale, some of the more aluminous

    Fig. 2. Sketches illustrating field relations (at scales of cm to m) in the mélange unit. (a) Cartoon demonstrates schematically the textural and

    mineralogical relationships between the siliceous ultramafic mélange matrix and mafic and ultramafic blocks floating in the mélange. The rinds

    developed on mafic blocks were studied in detail by Sorensen and Barton (1987), Sorensen (1988), and Sorensen and Grossman (1989). (b)

    Sketch illustrating the complex mineralogical zonations at the rims of ultramafic blocks in the mélange, indicating additions of Si and other

    components. A conspicuous feature of the zonations is the common presence of fibrous zones of nearly monomineralic clinoamphibole -

    F orthoamphibole or enstatite (up to 0.25 m thick; e.g., sample 6-4-100FR used on many of the geochemical diagrams in this paper).

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–106 83

  • samples contain interlayered domains (0.1 mm to

    several millimeter thick across the foliation) rich in

    either chlorite or amphibole.

    Fig. 2b illustrates schematically the mineralogical

    zonations that occur at rims of ultramafic blocks and

    the textural and mineralogical relations of veins and

    shear zones which cross-cut the blocks. The general-

    ized progression from contacts with sheared zones

    inward toward cores of blocks is chlorite + clinoam-

    phiboleF talc! orthoamphibole! clinoamphibole +orthoamphiboleFenstatite! orthoamphiboleF ensta-tite! serpentiniteF orthoamphiboleF enstatite. Thezonations are similar to those reported for metaso-

    matic zones in other amphibolite-grade ultramafic

    suites and interpreted as reflecting metasomatic trans-

    fer of silica and other components (e.g., Carswell et al.,

    1974; Sanford, 1982; Pfeifer, 1987).

    Meter- to kilometer-scale compositional domains in

    the mélange closely resemble in mineralogy the lith-

    ologies developed in metasomatic zones (rinds) on

    either mafic or ultramafic blocks (see Fig. 1). Large

    parts of the more aluminous matrix (aluminous matrix

    unit on Fig. 1) containing clinoamphiboleF chloriteassemblages are similar mineralogically to parts of

    metasomatic rinds on mafic blocks showing varying

    degrees of deformation and incorporation into the

    mélange (see Fig. 2a). In many outcrops mapped as

    Fig. 4. Mineral compositions of the mélange matrix. (a) Clinoam-

    phibole compositions, demonstrating varying tschermak substitution

    and more aluminous compositions in more Al-rich whole rocks.

    Note that the compositions of clinoamphiboles in the small discoids

    from chlorite schists (samples 7-2-26b1 and 7-2-26b6) and in

    extremely aluminous mélange matrix sample 6-4-9 overlap those for

    clinoamphibole in rinds developed on mafic blocks (data for rinds

    from Sorensen, 1988). (b) Chlorite compositions, demonstrating

    varying tschermak substitution and relationship with whole-rock Al

    contents (see text). Numbers below some sample numbers are

    whole-rock wt.% Al2O3. Representative mineral chemical compo-

    sitions, obtained by electron microprobe techniques, are provided in

    Table 2. On (b), note data (for sample 6-4-7) for a single chlorite

    grain with zoned chemical composition toward a contact with an

    adjacent amphibole grain (indicated by large arrow).

    Fig. 3. Field photograph of aluminous mélange, demonstrating its

    makeup of highly deformed domains (lenticular in two dimensions)

    containing either chloriteF clinoamphibole schist (darker regions)or talcF orthoamphibole regions (lighter regons), the two litholo-gies similar mineralogically to rinds on mafic and ultramafic rocks,

    respectively. Note handle of geologic rock hammer for scale. The

    horizontal dimension of the photograph represents approximately

    1.25 m.

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–10684

  • Table 1

    Summary of mineral assemblages of samples for which whole– rock geochemical data (and/or electron microprobe data) are presented in this

    paper

    Sample Tca Oam Cam Opx Chl Qtz Op/Rut Other Retrograde

    6-3-7b X X trc X btt

    6-3-9 X X X X X rut mg, tc

    6-3-21 X X X rut ap tc

    6-3-22 X tr btt chl

    6-3-31Cb X+ d X+ X+ X� e X yesf6-3-32 tr + X+ X� X tc6-3-33 tr X X X

    6-3-35a X X X

    6-4-3 X X

    6-4-6B X X yes tc, serp

    6-4-7b X X ilm ap

    6-4-9b X X ilm ap, zirc, btt chl

    6-4-11 X X X tr tc

    6-4-12b X X tr X ilm, mt ap, zirc

    6-4-19 X X X

    6-4-23C tr X tr tc, serp

    6-4-25b X X carb (dolo) tc, serp

    6-4-28b X X tc, serp

    6-4-64 X X yes chl

    6-4-65 X X yes

    6-4-94 X X X

    block X X X

    6-4-103a X X X yes tc

    6-4-103b tr X

    8-3-49Vb X tr X8-4-10ab X ilm zirc, ap

    8-4-10b X ilm ap

    8-4-11b X X X X spinel chl, tc

    8-4-13b X + X X� tr8-4-15ab X X X tr rut, ilm zirc tc, serp

    8-4-15bb X + X� tr8-4-16b X X rut! ilmg chl8-4-19ab X X X tr ilm

    8-4-19bb X X X X ilm

    8-4-19cb X X ilm tc

    Discoids

    7-2-26b1b X X X X rut ap, zirc

    7-2-26b6b X X tr X rut sulfide, btt chl

    Ultramafic rind

    6-4-100FR X X tr tc, serp

    a Mineral abbreviations are as follows: Tc = talc; Oam= orthoamphibole; Cam= clinoamphibole; Opx= orthopyroxene; Chl = chlorite;

    Qtz = quartz; Op/Rut = unidentified opaque phase or rutile; ilm = ilmenite; carb = prograde carbonate (dolomite); zirc = zircon; ap = apatite;

    btt = biotite; mt =magnetite; mag =magnesite.b Probe data exist for these samples.c Present in trace amounts.d Replaces other minerals.e Is replaced by other minerals.f Unidentified opaque mineral present.g Replacement of rutile by ilmenite at rutile rims.

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–106 85

  • Table 2

    Representative mineral compositions for mélange matrix samples, ultramafic rinds, and discoids in chlorite schist

    Sample 6-3-7 6-4-7 6-4-9 Discoids Coexisting amphiboles

    7-2-26b1 7-2-26b6 8-4-19a 8-4-19a 6-4-12

    Mineral Clinoamph Clinoamph Clinoamph Clinoamph Clinoamph Clinoamph Orthoamph Orthoamph

    !More aluminous whole rocksSiO2 50.58 48.51 46.71 49.42 50.61 56.84 57.32 56.17

    MgO 17.61 18.27 17.43 16.51 17.4 22.65 25.15 24.81

    Na2O 1.98 1.89 2.46 2.1 1.65 0.15 0.00 0.12

    Al2O3 9.00 7.90 9.50 10.92 9.48 0.40 0.18 0.89

    FeO* 6.63 7.52 8.90 6.29 5.86 4.65 12.82 13.17

    MnO(NiO) 0.16 0.20 0.19 (0.06) 0.15 0.22 0.39 0.46

    Cr2O3 0.22 0.10 0.00 0.25 0.84 0.00 0.03 0.18

    K2O 0.09 0.39 0.42 0.11 0.11 0.04 0.00 0.00

    CaO 10.84 11.99 11.40 10.92 11.13 11.95 0.86 0.45

    TiO2 0.43 0.30 0.43 0.47 0.30 0.04 0.03 0.01

    Total 97.54 97.07 97.44 97.05 97.53 96.94 96.78 96.26

    Normalizations (Laird and Albee, 1981)

    Si 7.033 6.872 6.613 6.936 7.060 7.875 8.034 7.922

    AlIV 0.967 1.128 1.387 1.064 0.940 0.065 0.000 0.078

    AlVI 0.507 0.192 0.199 0.742 0.619 0.000 0.030 0.069

    Ti 0.045 0.032 0.046 0.049 0.031 0.004 0.003 0.001

    Fe +3 0.583 0.639 0.887 0.339 0.434 0.211 0.000 0.034

    Cr 0.008 0.004 0.000 0.009 0.031 0.000 0.001 0.007

    Fe +2 0.189 0.252 0.167 0.398 0.249 0.328 1.503 1.519

    Mn 0.019 0.024 0.022 0.007 0.018 0.025 0.046 0.055

    Mg 3.650 3.857 3.679 3.454 3.617 4.676 5.254 5.214

    Ca 1.614 1.820 1.729 1.641 1.664 1.774 0.128 0.067

    NaM4 0.386 0.180 0.271 0.359 0.336 0.041 0.000 0.033

    NaA 0.149 0.338 0.404 0.214 0.110 0.000 0.000 0.000

    K 0.017 0.070 0.076 0.019 0.019 0.007 0.000 0.000

    Sum 15.166 15.409 15.480 15.233 15.129 15.007 15.000 15.000

    Sample Rind on ultramafic block Chlorites

    6-4-100FR 6-4-100FR Chlorite-rich Talc-rich Discoids UM rind

    6-4-9 6-4-12 7-2-26b1 7-2-26b6 6-5-56A

    Mineral Clinoamph

    fibrous zone

    Orthoamph

    fibrous zone

    Chlorite Chlorite Chlorite Chlorite

    SiO2 56.97 57.81 30.79 28.89 27.86 28.26 31.85

    MgO 23.86 28.02 28.94 28.07 23.93 24.21 33.03

    Na2O 0.37 0.13 0.00 0.00 0.00 0.00 0.00

    Al2O3 1.03 0.71 18.37 20.24 22.02 21.48 14.54

    FeO* 5.06 9.12 9.64 7.51 12.72 12.83 3.02

    MnO(NiO) (0.06) (0.14) 0.06 0.08 0.00 0.10 0.00

    Cr2O3 0.06 0.09 0.03 1.52 0.16 0.44 3.13

    K2O 0.04 0.00 0.00 0.00 0.00 0.00 0.03

    CaO 9.10 0.54 0.02 0.02 0.01 0.01 0.00

    TiO2 0.06 0.05 0.02 0.05 0.11 0.06 0.04

    Total 96.61 96.61 87.87 86.38 86.81 87.39 85.64

    Normalizations (Laird and Albee, 1981))

    Si 7.872 7.953 SiIV 2.967 2.850 2.776 2.808 3.134

    AlIV 0.128 0.047 AlIV 1.033 1.150 1.224 1.192 0.866

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–10686

  • more aluminous mélange matrix (see Fig. 1), highly

    deformed domains contain either clinoamphibole + ch-

    lorite-rich assemblages resembling parts of mafic rinds

    (darker regions on Fig. 3) or talc + orthoamphibole +

    clinoamphibole assemblages resembling ultramafic

    rinds (lighter regions on Fig. 3). Larger expanses of

    the mélange mapped on Fig. 1 as siliceous matrix and

    containing mineral assemblages rich in talc and

    orthoamphibole resemble assemblages observed in

    metasomatic zones on rims of ultramafic blocks (see

    Fig. 2b). These large domains of siliceous mélange

    matrix contain few or no mafic blocks and, in general,

    envelope large bodies of ultramafic rocks (see Fig. 1),

    consistent with their generation through metasomatism

    of ultramafic blocks.

    4. Mineral assemblages, textures, and compositions

    in the mélange

    Table 1 summarizes mineral assemblage data for the

    samples of mélange matrix for which whole-rock geo-

    chemical data are presented and discussed in this paper,

    and for two samples of amphibole-rich discoids col-

    lected from chlorite schist in one part of the mélange

    unit (samples 7-2-26b1, 7-2-26b6) and one rind devel-

    oped on an ultramafic block (sample 6-4-100FR).

    Sorensen and Grossman (1989) provide the mineral

    assemblages of the rinds developed on mafic blocks.

    Also contained in Table 1 is information regarding

    retrograde assemblages, deduced using petrographic

    observations, and inferred replacement textures among

    high-T phases. Samples selected for whole-rock geo-

    chemistry show relatively little retrograde overprinting.

    Retrograde alteration of the high-T parageneses in-

    cludes minor replacement of high-temperature miner-

    als by chloriteF talcF serpentineFmagnesite at theirrims and cross-cutting of the high-T assemblages by

    serpentineF talc-bearing vein networks to variable andsometimes pervasive replacement of the earlier assem-

    blages by chloriteF talcF serpentineFmagnesite.Table 2 presents representative mineral chemical

    analyses for the clinoamphiboles and chlorite. The

    compositions of these minerals covary with whole

    rock compositions to a greater degree than the com-

    positions of the other major phases of the mélange

    matrix (talc, orthoamphibole, enstatite). With the

    exception of the extremely talc- and enstatite-rich

    schists, all samples contain amphibole, in some cases

    coexisting ortho- and clinoamphiboles. Orthoamphi-

    boles are all Al-poor, magnesian anthophyllite with

    limited Mg–Fe + 2 variations. Clinoamphiboles range

    from actinolite to magnesiohornblende in composition

    (Fig. 4a; cf. Sorensen, 1988) and are generally the

    only Ca-rich mineral present, with the local exception

    of scarce ankeritic dolomite and apatite. Clinoamphi-

    bole compositions correlate with whole-rock compo-

    sitions in more aluminous samples, clinoamphiboles

    are more pargasitic (see data for sample 6-4-9, which

    contains 13.7 wt.% Al2O3, and ‘‘discoids’’ on Fig.

    Table 2 (continued )

    Sample Rind on ultramafic block Chlorites

    6-4-100FR 6-4-100FR Chlorite-rich Talc-rich Discoids UM rind

    6-4-9 6-4-12 7-2-26b1 7-2-26b6 6-5-56A

    Normalizations (Laird and Albee, 1981)

    AlVI 0.040 0.068 AlVI 1.055 1.203 1.363 1.323 0.820

    Ti 0.006 0.005 Ti 0.001 0.003 0.009 0.004 0.003

    Fe +3 0.165 0.000 Fe +3 0.000 0.000 0.000 0.000 0.000

    Cr 0.002 0.003 Cr 0.001 0.040 0.004 0.011 0.081

    Fe +2 0.419 1.049 Fe +2 0.777 0.619 1.060 1.066 0.248

    Mn(Ni) (0.007) (0.007) Mn(Ni) 0.005 0.006 (0.011) 0.009 0.000

    Mg 4.914 5.745 Mg 4.157 4.126 3.554 3.585 4.844

    Ca 1.348 0.079 Ca 0.002 0.002 0.001 0.002 0.000

    NaM4 0.098 0.033 Na 0.002 0.000 0.000 0.000 0.000

    NaA 0.000 0.000 K 0.000 0.000 0.000 0.000 0.004

    K 0.006 0.000 Sum 10.000 9.999 9.980 10.000 10.000

    Sum 15.006 15.000

    *All Fe reported as FeO.

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–106 87

  • 4a). Clinoamphiboles in aluminous samples com-

    monly are zoned from aluminous cores to lower Al

    rims. Some samples contain two clinoamphibole

    compositional types; coarse-grained aluminous

    amphibole cross-cut by fibrous low-aluminum (acti-

    nolitic) amphibole (e.g., sample 8-3-49V; see Fig. 4a).Compositions of many clinoamphiboles in relatively

    aluminous samples overlap with those reported by

    Sorensen (1988) for rinds on mafic blocks (see data

    on Fig. 4a for ‘‘discoids’’, samples 7-2-26b1 and 7-2-

    26b6, and chlorite-rich mélange matrix sample 6-4-9;

    representative analyses of clinoamphibole and chlorite

    in these lithologies are provided in Table 2).

    Chlorite is abundant in the mélange, and some

    samples are virtually pure chlorite (e.g., sample 6-4-

    9). Like the clinoamphiboles, the chlorites generally

    have higher Al contents in the more aluminous rocks.

    Chlorite shows strong tschermakite substitution com-

    positional trends, as demonstrated in Fig. 4b. Chro-

    mium content varies widely (0.1–1.6 wt.% Cr2O3),

    with the highest contents found in the less aluminous,

    low-calcium, more siliceous samples (e.g., sample 6-

    4-12 in Table 2).

    Chlorite and amphibole contain less Al and have

    higher Mg/Fe and Cr contents in mineralogical zona-

    tions at rims of ultramafic blocks (ultramafic rinds;

    see Fig. 2b). Clinoamphiboles in these zones tend to

    be more actinolitic or even tremolitic than clinoam-

    phiboles in more aluminous mélange matrix away

    from ultramafic blocks (Fig. 4a). Chlorites are more

    magnesian and more Cr-rich than any measured in

    mélange matrix (for ultramafic rind chlorites, up to

    3.13 wt.% Cr2O3; e.g., 6-5-56A in Table 2), and show

    decrease in tschermakite substitution toward the

    blocks (e.g., C to B to A in sample 6-5-56, Fig. 4b).

    5. Major and trace element evidence for the

    evolution of the mélange matrix

    Major and trace element compositions were used to

    delineate the behavior of various elements during

    mélange matrix formation and to compare the mél-

    ange matrix compositions with the compositions of

    ‘‘rinds’’ developed on mafic and ultramafic blocks.

    The plots of Cr, Al2O3, and SiO2 compositions in Fig.

    5a and b introduce the approach used in the following

    discussions to document and interpret the composi-

    tional variations. On both plots, the mélange matrix

    samples show an extremely wide range in composi-

    tion that extends toward, but does not overlap with the

    compositional ranges for the mafic and ultramafic

    blocks. Data for a clinoamphibole + orthoamphibole

    rind (fringe; sample 6-4-100FR) on an ultramafic

    block are plotted on these and later diagrams, as are

    Fig. 5. Chemical discrimination diagrams, demonstrating the

    conceptual model for mixing in the mélange unit. (a) Cr vs.

    Al2O3, (b) SiO2 vs. Al2O3. On (a), the darker-shaded lines

    emanating from the origin represent the Cr–Al compositions

    resulting from varying ultramafic:mafic proportions (labelled for

    each). The envelopes of simple mixing of ultramafic and mafic

    compositions are schematically indicated on Fig. 5a and b by the

    regions between the two subparallel, hand-drawn, solid lines. The

    two small, unfilled circles on (a) are the mean mafic and ultramafic

    end-member compositions, with the dashed line connected the two

    circles representing linear mixtures of the two end members. Large,

    shaded arrows on (a) and (b) indicate the trends from mafic and

    ultramafic end members toward the compositions of rinds

    developed on each block type.

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–10688

  • data for rinds (including ‘‘inner rinds’’, ‘‘outer rinds’’,

    and ‘‘other rinds’’) developed on mafic blocks (from

    Sorensen, 1988; Sorensen and Grossman, 1989). The

    comparisons of the mélange matrix compositions with

    those of the rinds are key, given that the rinds on

    mafic blocks are clearly the result of mafic:ultramafic

    mixing (Sorensen and Grossman, 1989) and, further-

    more, the rinds on ultramafic blocks appear to have

    resulted from infiltrative–diffusive metasomatic alter-

    ation of the dunite or harzburgite. Whole rock anal-

    yses for 37 samples of the mélange matrix thought to

    be representative of the present exposures are given in

    Table 3; (cf. Table 1). These data were used to derive

    an average mélange matrix composition (Table 3,

    Figs. 5–11). The average composition is useful in

    evaluating the overall open- and closed-system behav-

    ior of various elements within the ultramafic mélange

    unit as a whole. In the following considerations of

    mafic–ultramafic mixing, mafic block compositions

    are taken to be compositions of the ‘‘non-migmatitic

    blocks’’ of Sorensen and Grossman (1989). Field

    observations demonstrate that the metagabbroic

    gneiss, which is another candidate for a mafic end

    member, is far less abundant as block material in the

    mélange.

    Fig. 5a shows the covariation of Al2O3 and Cr

    content of mélange matrix and demonstrates that a

    significant number of the matrix samples fall in an

    envelope that represents possible simple mixtures of

    these components from mafic and ultramafic blocks.

    This mixing envelope is indicated schematically on

    Fig. 5a and b by the regions between the two subpar-

    allel, hand-drawn, solid lines; on Fig. 5a, the dotted

    line within the mixing envelope is a mixing line

    between the mean compositions of the two block types

    (see data for end members in Table 4 and discussion in

    Section 5.1). On Fig. 5a, significant loss or gain of

    elements other than Cr and Al2O3 would be expected

    to shift compositions away from or toward the origin,

    respectively, without changing the Cr/Al2O3 ratios of

    the samples. In fact, the samples that fall outside this

    envelope are either (1) extremely siliceous schists

    (labeled on Fig. 5a as Amphibole and Talc Schists),

    most of which are regarded as having been ultramafic

    rocks with significant amounts of added Si (and

    possibly other components), or (2) nearly monominer-

    alic chlorite schists (labeled on Fig. 5a, b) regarded as

    having experienced significant mass loss.

    The massive addition of SiO2, at least at the local

    scale, is indicated by comparison of Fig. 5a and b. The

    talc- and amphibole-rich, chlorite-poor schists that are

    common in the ultramafic mélange unit have Cr/

    Al2O3 ratios that would indicate high ultramafic:mafic

    mixing ratios greater than f1:1, and mostly > 3:1, onFig. 5a, but their compositions on this plot are shifted

    significantly toward the origin from the envelope of

    simple ultramafic:mafic mixtures. This shift is con-

    sistent with the siliceous character of the samples

    (talc- or amphibole-rich schists), indicating possible

    bulk metasomatic addition of SiO2 to these samples

    and dilution of both Cr and Al2O3 in a constant ratio.

    On Fig. 5b, these high-Cr, low-Al2O3 samples plot to

    the right of, and at slightly higher Al2O3 than most of

    the ultramafic data. Some shift to SiO2 concentrations

    exceeding 60 wt.%. Chlorite-rich schists, which on

    Fig. 5a, fall to the right side of the simple mixing

    zone, plot as the high-Al2O3, low-SiO2 samples far to

    the left of the simple mixing envelope in Fig. 5b.

    Perhaps most important for the overall interpreta-

    tion of the mélange chemical evolution, the compo-

    sitions of many of the mélange matrix samples

    overlap with those of the rinds formed on mafic and

    ultramafic blocks (see data for rinds on Fig. 5a, b).

    This demonstrates that rind-forming processes (e.g.,

    for mafic blocks, described by Sorensen and Gross-

    man, 1989) could have produced many of the mél-

    ange matrix compositions. Data for the inner and

    outer rinds of Sorensen and Grossman (1989) plotted

    on these diagrams show mixing trends at the margins

    of the mafic blocks. These trends, which are indicated

    by the large, shaded arrows on Fig. 5, encompass

    many of the mélange matrix compositions. Inner rinds

    come close to the mafic block compositions. In

    contrast, the outer rinds plot toward the ultramafic

    block compositions, but only partly overlap the enve-

    lope for simple mixtures. Some approach the average

    mélange matrix composition (Fig. 5). The metaso-

    matic rind from the ultramafic block (indicated by the

    filled circle) is shifted from the ultramafic block data

    toward lower Cr and higher SiO2 and is as siliceous

    as the most siliceous mélange matrix sample (see Fig.

    5b).

    Fig. 6 shows the compositional variability of mél-

    ange matrix samples, relative to the compositions of

    tectonic blocks and rinds, for elements that show

    overall compositional variation similar in style to that

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–106 89

  • of Cr and Al2O3; that is, showing variability in con-

    centration largely explained by simple mixing, but with

    superimposed effects of mass loss or, more commonly,

    mass gain. The plot of MgO vs. Al2O3 (Fig. 6a)

    resembles that of Cr vs. Al2O3, whereas the plot of

    Ni vs. Al2O3 (Fig. 6b) differs in that a larger number of

    the mélange matrix samples plot to the low Ni–Al side

    of the simple mixing envelope (see later discussion). In

    the plot of Fe2O3 vs. Al2O3 (Fig. 6c), the mélange

    matrix samples also plot either in the envelope of

    Table 3

    Major and trace element compositions of mélange matrix samples and the average mélange matrix (in wt.%, unless otherwise specified)

    Sample 6-3-7 6-3-9 6-3-21 6-3-22 6-3-31c 6-3-32 6-3-33 6-3-35a 6-3-60 6-4-3

    SiO2 55.17 40.24 53.04 50.30 55.95 56.01 47.97 60.18 48.37 60.31

    TiO2 0.20 0.07 0.56 0.37 0.03 0.05 0.09 0.05 0.04 0.07

    Al2O3 5.59 7.57 7.97 8.89 2.58 2.89 6.55 1.15 3.37 1.18

    Fe2O3 6.37 5.62 7.64 6.07 5.74 7.33 8.36 6.07 8.81 2.33

    MnO 0.10 0.07 0.13 0.13 0.16 0.23 0.19 0.14 0.15 0.08

    MgO 22.30 35.82 19.74 20.72 32.22 29.28 27.95 27.89 32.52 25.39

    CaO 7.17 0.53 3.97 7.40 1.10 0.32 3.82 0.06 0.09 9.38

    Na2O 0.89 0.03 0.33 0.96 0.02 0.01 0.00 0.01 0.07 0.03

    K2O 0.15 0.01 0.44 0.52 0.02 0.03 0.04 0.01 0.02 0.07

    Cr (ppm) 1324 5672 1259 817 2356 2559 1603 939 2626 584

    Ni (ppm) 893 1498 800 513 1769 1621 1213 1116 1525 872

    Sr (ppm) – 10 – 0 16 10 – – 12 93

    Zr (ppm) – 5 – 0 5 5 – 90 5 34

    Ba (ppm) – 4 – 0 131 58 – – 180 17

    LOI 1.41 9.5 5.9 2.9 2.1 4.1 4.4 3.9 5.9 0.6

    Total 99.56 100.17 99.89 98.41 100.34 100.66 99.68 99.69 99.75 99.56

    UM/M (Cr/Al) 1.06 3.90 0.67 0.35 4.95 4.78 1.10 4.32 4.10 2.41

    Mass change 32 � 55 13 24 13 3 11 177 � 2 295Volume Change � 42Density dataa

    6-4-65 6-4-94 6-4-103a 6-4-103b block 6-5-53 8-3-49’ 8-4-10a 8-4-10b 8-4-11

    SiO2 61.38 52.68 51.24 41.31 58.87 59.02 57.07 27.28 28.08 53.23

    TiO2 0.01 0.44 0.54 0.36 0.37 0.42 0.22 1.30 0.58 0.04

    Al2O3 0.69 6.46 4.35 8.05 6.34 6.34 4.93 18.14 18.67 2.64

    Fe2O3 5.10 7.07 10.67 11.63 7.33 11.43 7.28 9.28 11.88 5.10

    MnO 0.09 0.15 0.22 0.13 0.17 0.22 0.20 0.07 0.12 0.11

    MgO 28.31 18.57 26.37 29.68 16.48 14.31 16.96 33.11 29.72 28.48

    CaO 0.09 9.24 1.00 0.28 7.31 5.34 10.37 0.07 0.26 7.17

    Na2O 0.01 1.44 0.05 0.06 1.09 1.12 1.30 0.00 0.00 0.38

    K2O 0.03 0.20 0.01 0.01 0.11 0.09 0.39 0.01 0.03 0.05

    Cr (ppm) 1493 1437 1783 1742 1426 920 1079 187 167 1701

    Ni (ppm) 1389 654 999 660 659 501 810 209 173 1060

    Sr (ppm) 12 68 12 15 31 – 33 7 7 33

    Zr (ppm) 6 22 58 78 41 – 18 375 65 7

    Ba (ppm) 45 16 14 18 32 – 152 19 8 23

    LOI 3.62 3.02 5.06 8.51 1.29 1.11 0.83 11.48 11.37 2.39

    Total 99.61 99.47 99.79 100.26 99.55 99.54 99.75 100.81 100.74 99.86

    UM/M (Cr/Al) 16.24 0.99 1.96 0.96 1.00 0.61 0.97 0.02 0.01 3.29

    Mass Change 100 18 22 –4 20 47 56 –21 –22 45

    Volume change 147 24 54

    Density dataa talc chlorite clamp

    a Density for this mineral used in calculation of volume change in this nearly monomineralic sample.b Average mélange matrix composition.

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–10690

  • simple mixing, or to the left of it; this relationship is

    also apparent on the plot of Zr vs. Cr (Fig. 6d).

    Other alkalis and alkaline earths (CaO, Na2O, K2O,

    Ba, and Sr) show erratic concentration variations

    within the mélange matrix and rinds, seemingly

    requiring considerable mobility of these elements

    (Fig. 7). Na2O, K2O, and Ba are locally higher in

    some rinds of mafic blocks than in the blocks them-

    selves (Sorensen and Grossman, 1989); however,

    along with CaO and Sr, they are typically depleted

    6-4-6b 6-4-7 6-4-9 6-4-11 6-4-12 6-4-19 6-4-23c 6-4-25 6-4-28 6-4-64

    39.78 41.07 34.52 60.75 58.89 63.26 55.94 56.16 56.19 54.33

    0.04 1.11 0.29 0.05 0.04 0.55 0.08 0.01 0.06 0.30

    7.86 9.88 12.40 1.24 1.08 10.39 3.25 0.79 1.03 5.72

    5.04 9.13 13.20 4.86 8.68 7.78 5.30 7.63 6.24 12.99

    0.07 0.10 0.09 0.08 0.21 0.08 0.08 0.07 0.09 0.20

    32.97 25.27 27.20 26.50 27.38 7.88 22.08 28.06 29.82 18.57

    6.15 6.18 1.47 2.53 0.17 4.27 7.69 0.60 0.04 4.86

    0.03 0.42 0.20 0.04 0.01 0.65 0.60 0.00 0.01 0.62

    0.02 0.12 0.03 0.02 0.01 2.73 0.12 0.01 0.01 0.12

    2178 878 3054 1937 1828 334 1585 1433 1504 1689

    1804 781 2428 1837 1491 187 1402 1033 1348 811

    50 67 33 14 – 59 83 21 9 36

    8 120 57 6 – 98 8 4 5 22

    11 – 16 15 – 948 7 7 11 497

    7.7 6.4 10.4 4.3 2.9 2.5 4.3 7.5 7.1 1.7

    100.03 99.85 100.33 100.75 99.66 100.24 99.74 101.06 100.85 99.66

    1.27 0.33 1.11 9.87 11.05 0.06 2.39 12.16 9.00 1.36

    � 14 13 � 42 49 59 33 45 105 91 1410 � 25 85 97 46 59 153 136chloritea chlorite talc talc clamp clamp talc talc

    8-4-13 8-4-15a 8-4-15b 8-4-16 8-4-19a 8-4-19b 8-4-19c Melmataveb 6-4-100FR

    53.37 57.47 53.69 55.55 52.81 46.16 29.13 50.75 61.17

    0.05 0.12 0.07 0.51 0.12 0.27 0.53 0.27 0.03

    2.18 1.59 2.14 0.69 2.66 4.84 12.99 5.63 1.21

    7.11 4.40 11.74 12.59 7.71 8.94 14.88 8.12 6.07

    0.16 0.15 0.19 0.22 0.22 0.22 0.11 0.14 0.10

    30.86 23.62 28.43 28.53 25.07 27.94 32.08 25.99 21.36

    2.58 11.57 0.28 0.42 9.05 7.44 0.10 3.73 7.28

    0.31 0.00 0.00 0.00 0.00 0.17 0.19 0.29 0.01

    0.03 0.06 0.01 0.04 0.07 0.07 0.02 0.15 0.10

    1479 216 1944 359 490 1358 332 1474 825

    1195 571 1535 842 669 1096 404 1038 956

    19 66 11 8 46 37 8 31 49

    7 12 5 89 13 36 158 48 6

    32 68 145 44 18 21 11 84 7

    2.94 0.51 3.06 1.06 1.85 4.24 11.08 4.56 2.55

    99.86 99.58 99.98 99.74 99.68 100.55 101.21 99.91 100.05

    3.48 0.56 4.88 2.56 0.80 1.28 0.03 1.18 3.50

    70 503 37 551 216 39 9 24 203

    289

    clamp

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–106 91

  • compared to the simple mixing lines in the mélange

    matrix.

    5.1. A mixing model

    Simple mixing calculations allow partial quantifi-

    cation of the contrasting compositional systematics in

    the mélange matrix. In these calculations, the mafic

    and ultramafic blocks were taken as the mixing com-

    ponents, using the mean concentrations provided in

    Table 4. The distinctive characteristics of these two

    components are: (1) the serpentinites, derived from

    dunitic/harzburgite protoliths, have low concentrations

    of Al, Zr, Ca, Na, and the LILE (K, Sr, Ba), and high

    concentrations of Mg, Cr, and Ni; (2) mafic blocks

    (‘‘non-migmatitic clinopyroxene-bearing blocks’’ and

    ‘‘non-migmatitic blocks with rinds’’ of Sorensen and

    Grossman, 1989) have higher concentrations of Al, Zr,

    Ca, Na, and the LILE, and lower concentrations of Mg,

    Cr, Ni. Excluded are metasedimentary rocks which

    have low Mg and Ca and high Al, Si, Na, and the LILE

    (documented in Bebout et al., 1999) because they are

    virtually absent as tectonic blocks in the mélange unit,

    and not needed to explain the observed variations.

    Given that the metasedimentary rocks are easily dis-

    aggregated and, thus, might have been fully incorpo-

    rated into the mélange matrix, we later explore this

    possibility from a geochemical perspective.

    Aluminum and Cr were selected as the initial test

    ‘‘geochemical reference frame’’ because of their per-

    ceived relative immobility in fluids in previous studies

    of metasomatic alteration (see discussions by Ague,

    1994; Baumgartner and Olsen, 1995; Ague and van

    Haren, 1996; Yang et al., 1998) and because of their

    starkly contrasting concentrations in the two mixing

    end members (Figs. 5 and 6). Sorensen and Grossman

    (1989) similarly suggested that the concentrations of Cr

    and Al in rinds onmafic blocks reflect simple mixing of

    mafic and ultramafic components, and contrasted their

    behavior with that of elements such as Ba, Rb, and Th

    that appear to have been added to rinds by metasomatic

    fluids. Mean concentrations of Cr and Al2O3 in the

    mafic and ultramafic end members (see Table 4) were

    used to calculate the Cr/Al2O3 that would result from

    mixing the end members in varying proportions.

    Employing this calculated relationship, a model ultra-

    mafic:mafic mixture for each of the samples was

    obtained using its observed Cr/Al2O3 (see the mixing

    line and labeled lines of constant ultramafic:mafic

    mixing on Fig. 5a). On Fig. 5a, the several lines repre-

    senting constant ultramafic:mafic proportions (using

    themean compositions of the two block types; Table 4),

    demonstrate that the ultramafic/mafic ratios vary

    between >10:1 to < 1:1 for domains within the mél-

    ange. The average matrix composition corresponds to a

    ultramafic:mafic ratio of f 1.2:1, suggesting this ratioas a approximation of the time-integrated ultramafic–

    mafic mixing proportions for the mélange matrix.

    Given the mean Cr/Al2O3 for the two end-member

    block types (Table 4), the nominal concentrations of

    the other major and trace elements were calculated

    based on simple mixing. Comparisons of the observed

    compositions with those calculated from simple mix-

    ing (hereafter referred to as ‘‘observed/calculated’’)

    affords an evaluation of the degree to which simple

    mixing explains the contents of individual compo-

    nents and, conversely, the need for metasomatic trans-

    fer. This analysis, although limited by the uncertainty

    in the mafic and ultramafic end members, allows some

    broad delineations to be made regarding the composi-

    tional behavior of the different major and trace ele-

    ments during mélange matrix formation.

    5.2. Relative mobilities of elements within the mélange

    Using the method described above, and even with

    the scatter shown, it is possible to place the composi-

    tional variations of the elements into one of several

    Table 4

    Compositions of the ultramafic and mafix mixing components

    (wt.% unless otherwise specified)

    Ultramafic Mafic

    Mean 1r Mean 1r

    SiO2 48.58 2.88 44.64 2.05

    TiO2 0.02 0.03 2.00 0.71

    Al2O3 0.55 0.21 14.64 1.24

    Fe2O3 9.02 1.27 15.00 2.73

    MgO 40.00 2.75 9.30 2.01

    CaO 0.10 0.10 11.61 3.75

    Na2O 0.02 0.01 1.26 0.21

    K2O 0.04 0.07 0.27 0.1

    Cr (ppm) 3146 688 271 158

    Ni (ppm) 2765 422 102 17.8

    Sr (ppm) 8 3 228 259

    Zr (ppm) 5 0.4 151 41

    Ba (ppm) 11 2 65.5 33.8

    LOI 0 0 0.63 0.37

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–10692

  • groups (see Fig. 8). The first group, consisting of Ni,

    Fe2O3, MgO, and perhaps also Zr, behaves coherently

    with Cr and Al2O3 and, thus, is consistent with simple

    mixing and relative fluid immobility. On Fig. 8a, these

    elements show similarity in both the overall range of

    observed/calculated for mélange matrix samples (ver-

    tical lines) and the average mélange matrix observed/

    calculated (filled circles), the latter falling at or just

    below the 1:1 line. The observed/calculated values for

    Mg, shown in order of increasing calculated ultra-

    mafic block proportions [expressed as ultramafic/

    (ultramafic +mafic)] to the right-hand side on Fig.

    8b, support its use as one of the geochemical reference

    frame elements, as all but three samples straddle the

    1:1 observed/calculated line, with a slightly greater

    number of samples falling below this line. The three

    anomalous samples exert a significant effect on the

    average observed/calculated shown on Fig. 8a; a more

    realistic value may be represented by the unfilled

    circle which excludes these three samples from the

    average. These relationships stand out in isocon plots

    (after Grant, 1986) in Fig. 9a–f for each of the

    elements proposed for the geochemical reference

    frame. In these plots, the observed compositions of

    the mélange matrix and rind samples are compared

    with the compositions calculated as reflecting simple

    Fig. 6. Chemical discrimination diagrams, comparing the compositions of mélange matrix samples with those of mafic and ultramafic blocks

    and metasomatic rinds developed on each of the two block lithologies. (a) MgO–Al2O3, (b) Ni–Al2O3, (c) Fe2O3–Al2O3, and (d) Zr–Cr. On

    (d), one mélange matrix samples plots at higher Zr concentrations at Zr = 375 ppm, 18.14 =Al2O3 wt.%, and one inner rind sample plots at

    Zr = 531 ppm, 13.40 =Al2O3 wt.%. On each diagram, the envelope of simple mixing of ultramafic and mafic compositions is crudely indicated

    by the regions between the two hand-drawn, solid lines.

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–106 93

  • mixing between the mafic and ultramafic end mem-

    bers in proportions estimated by the Cr/Al2O3 ratios

    (see Fig. 5a and discussion above; see similar

    approach by Yang et al., 1998). From these plots, it

    is evident that Fe and Zr (Fig. 9e and f) scatter more

    than Al, Cr, and Mg (see Fig. 9a, b, and c), whereas

    the Ni data (see Fig. 9d) fall systematically below the

    1:1 line when compared with the Al, Cr, Mg, and Zr

    compositions. In each plot, nearly all of the rind data

    plot within the range of mélange matrix compositions.

    Also in each plot, the samples that fall above the 1:1

    line tend to be the more chlorite-rich schists which are

    depleted in Si, Ca, Na, and the LILE presumably by

    significant mass loss with the result that the less

    mobile elements are concentrated (see Figs. 5a and

    7a–d). Although TiO2 is probably relatively immobile

    (Ague, 1994; Baumgartner and Olsen, 1995), it does

    not exhibit simple mixing relationships. This likely

    reflects the large original spread in values in the mafic

    end member ( < 1.0 to 3.5 wt.%; Sorensen and Gross-

    man, 1989; this study), perhaps with a contribution

    from sampling issues caused by the rather coarse size

    (up to 5 mm) and spotty distribution of rutile aggre-

    gates in the mélange matrix.

    On Fig. 8a, Si shows a relatively tight range in

    observed/calculated and an average mélange matrix

    Fig. 7. Chemical discrimination diagrams, comparing the compositions of mélange matrix samples with those of mafic and ultramafic blocks and

    metasomatic rinds developed on the two block lithologies for oxides and trace elements showing more erratic variation relative to apparent simple

    mafic–ultramafic mixing envelopes. (a) CaO–Al2O3, (b) Na2O–Al2O3, (c) K2O–Al2O3, and (d) Ba–Sr. On (a), (b), and (c), the envelopes of

    simple mixing of ultramafic and mafic compositions are indicated schematically by the regions between the two hand-drawn, solid lines.

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–10694

  • observed/calculated somewhat higher than 1:1. The

    significant degrees of enrichment of mélange matrix

    samples in SiO2 relative to the calculated composi-

    tions for simple mafic–ultramafic mixing are empha-

    sized in Fig. 10a (in an isocon diagram) and in Fig.

    10b, the latter a plot of the observed/calculated SiO2vs. the observed/calculated Al2O3. One of the princi-

    pal conclusions derived from the mixing models is

    that mechanical mixtures of mafic blocks with the

    ultramafic rocks cannot account for the present, sili-

    ceous state of the mélange (see Fig. 5b and the

    discussions above). The evidence for SiO2 additions

    in mineralogical zonations at rims of mafic blocks

    (Sorensen, 1988) and ultramafic blocks (this study;

    Fig. 2b) is compatible with contributions of SiO2mobilized in fluids. Overall addition to the mélange

    unit is also evident for H2O (here represented by the

    LOI data), as nearly every mélange matrix samples is

    significantly enriched in H2O relative to the calculated

    ultramafic–mafic mixtures (see Figs. 8a and 10c).

    Other elements, such as Ca, Na, and Sr, show

    scatter in observed/calculated values (see the vertical

    lines on Fig. 8a, indicating overall ranges in observed/

    calculated for mélange matrix samples), and Na, and

    particularly Sr, have average mélange matrix ob-

    served/calculated significantly lower than that of the

    geochemical reference frame elements and the 1:1 line

    (the latter line indicating the results of simple mixing

    of the mafic and ultramafic end members; see Fig.

    10d–f). Potassium and Ba also show much larger

    ranges in observed/calculated concentrations, but

    (particularly for Ba) have average mélange matrix

    observed/calculated higher than that of the geochem-

    ical reference frame elements and the 1:1 line indicat-

    ing addition. For Ca, Na, and K, a large number of the

    mélange matrix samples have observed/calculated

    significantly lower than the observed/calculated for

    the mafic rind samples. For Ca, K, and Sr, the

    mélange matrix observed/calculated values are lower

    than that of the ultramafic rind.

    6. Discussion

    6.1. The nature of the tectonic mixing

    A key inference from the data is that some elements

    (Al, Cr, Fe, Ni, Mg, and Zr) are best explained by

    Fig. 8. Diagram illustrating the differing compositional systematics

    of selected major and trace elements. (a) Diagram showing the

    overall range in observed/calculated for individual mélange matrix

    samples (broad vertical lines) and the observed/calculated value for

    the averaged mélange matrix composition (filled circle for each

    element or LOI). (b) Plot of ultramafic/(ultramafic +mafic) vs. Mg

    observed/calculated for mélange matrix samples and rinds on mafic

    and ultramafic blocks, demonstrating that Mg observed/calculated

    shows similar variation to that of the other geochemical reference

    frame elements. Three samples with extremely high observed/

    calculated (>3) strongly influence the average mélange matrix

    composition observed/calculated for Mg, as shown on (a), but do

    not as strongly affect the average mélange matrix composition

    observed/calculated for other elements. On (b), the symbols are the

    same as those used in Figs. 5–7. Note that, for CaO, Na2O, K2O, Sr,

    and Ba, some of the extremely chlorite- and talc-rich schists have

    concentrations below the analytical detection limits (concentrations

    of 0 wt.% or ppm on Table 3), or (for the trace elements) were not

    analyzed (denoted by ‘‘– ’’ on Table 3).

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–106 95

  • simple mixing (cf. similar conclusion reached by

    Sorensen and Grossman, 1989, for the mafic rind

    compositions). Field observations in the Catalina

    Schist amphibolite-grade mélange unit clearly show

    that the mélange matrix developed in mafic-block-

    bearing shear zones between domains of low-Al ultra-

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–10696

  • mafic materials. Progressive hydration and metaso-

    matic alteration of both mafic and ultramafic blocks

    produced weak rinds rich in sheet silicates and amphib-

    oles. The similarity of the rinds and the mélange matrix

    in mineralogy and chemical compositions is consistent

    with production of the matrix through progressive

    incorporation of block rinds. Rind-like materials can

    be seen to be progressively incorporated into the

    mélange matrix foliation (see Fig. 3), resulting in

    adjacent, mineralogically/chemically disparate litholo-

    gies at scales of meters down to millimeters. At the

    lower size limit, this process is represented by small

    discoids of clinoamphibole-rich material within the

    more aluminous schists, and by the existence of layers

    with contrasting mineralogy at thin-section scale (e.g.,

    layers of amphibole and talc in chlorite-dominated

    schist). Deformation was sufficiently intense to trans-

    pose fabrics, disaggregate rigid block-derived materi-

    als in weaker chorite- and talc-rich mélange, and in

    some particularly weak lithologies (e.g., chlorite-, talc-,

    and amphibole-rich materials), intimately juxtapose

    disparate lithologies at the (sub-)cm scales sampled

    by the whole-rock geochemical analyses.

    Yang et al. (1998) similarly argued for the com-

    bined effects of tectonic mixing and metasomatism to

    produce an upper-amphibolite facies, ductile shear

    zone, concluding that certain elements (in that case,

    HREEs, Ti, V, and Sc) fit a simple mixing model

    (representing ‘‘mechanical mixing due to shearing’’),

    whereas other elements exhibited mobile behavior.

    Deformation in the Catalina Schist amphibolite-facies

    mélange unit should have enhanced chemical

    exchange, particularly by increasing chemical poten-

    tial gradients through juxtaposing contrasting litholo-

    gies (cf. Wintsch, 1985), and also enhancing the

    permeability of the mélange (cf. Rutter and Brodie,

    1985) facilitating infiltration-driven metasomatism.

    Deformation-enhanced metasomatism could explain

    the apparently greater alteration of some mélange

    matrix samples relative to rind samples, notably the

    chlorite-rich schists with that have exceptionally low

    concentrations of Si, Ca, Na, K, Sr, and Ba.

    6.2. Estimating mass gains and losses in the mélange

    matrix

    Use of the observed/calculated for the geochemical

    reference frame elements affords estimates of mass

    gains and losses during mélange formation. We

    emphasize that these calculations serve to qualita-

    tively delineate the contrasting evolution of the con-

    trasting compositional domains in the mélange matrix,

    although they do not fully treat the continuum of

    coupled tectonic and metasomatic mixing proposed

    based on the field and petrographic observations (as

    discussed above). Total mass changes (Table 3; after

    Gresens, 1967; Grant, 1986) were calculated using the

    ratio of the concentrations of Cr or Al2O3 predicted

    from simple mixing (Ci; normalized using Cr/Al2O3)

    to the observed concentrations (Cf):

    % change in mass ¼ ½Ci=Cf � 1� � 100The results indicate a wide range in mass loss and

    gain, ranging from f 55% mass loss, in the mostchlorite-rich schists, to >100% mass gain in the most

    siliceous schists. For the rinds on mafic blocks,

    calculated mass gains range from small in the inner

    rinds (f 8–10% gain) to gains approaching those ofmany of the mélange matrix samples in the outer rinds

    (2–44% mass gain; see Table 5). An approximately

    200% mass gain is calculated for the rind on the

    ultramafic block (sample 6-4-100FR). For the average

    mélange matrix composition, a mass gain of f 24%is calculated, and based on an analogous calculation

    for changes in individual components, we suggest that

    the majority of the mass changes are in SiO2 and H2O

    (see Figs. 5b and 10a–c).

    A similar calculation for volume changes illumi-

    nates the differing evolution of the contrasting com-

    positional domains in the mélange matrix. Selected

    Fig. 9. Isocon diagrams (after Grant, 1986) presenting results of the Cr–Al-based mixing calculations for elements which show overall

    systematics similar to those of Cr and Al. This plot compares the observed concentrations with those calculated as the result of simple mixing of

    the two end-member lithologies, using the mixing proportions based on Cr/Al2O3. These elements, together with Cr and Al, constitute the best

    ‘‘geochemical reference frame’’ with which assessments of mass loss and gain can be made. (a) Al2O3 isocon plot, (b) Cr isocon plot, (c) MgO

    isocon plot, (d) Ni isocon plot, (e) Fe2O3 isocon plot), and (f) Zr isocon plot. Note that, for each of these elements or oxides, the data straddle the

    1:1 line on these plots, and that the point representing the average mélange matrix falls on or slightly below the 1:1 line (shaded diagonal line).

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–106 97

  • G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–10698

  • values of volume change were calculated for samples

    with simple mineralogy by using mineral densities as

    a proxy for whole rock values (Table 3), specifically,

    the nearly monomineralic chlorite, talc, or amphibole

    schists. Densities used were 3.4 g/cm3 for the anhy-

    drous ultramafic and eclogitic end members and 2.65,

    2.75, and 3.1 g/cm3 for clinochlore, talc, and amphib-

    ole, respectively. Volume changes come from the

    following relationship where qi and qf represent theinitial (calculated) and final (observed) densities,

    respectively.

    % change in volume ¼ ½ðCi=Cf Þðqi=qf Þ � 1� � 100

    Chlorite schists show volume changes of � 42%to + 10%, talc schists show volume changes of + 85%

    to + 153%, and the three clino- and orthoamphibole-

    rich schists show volume changes of + 46% to + 59%.

    The negative to low positive volume changes for the

    chlorite schists reflect their loss of mobile elements as

    indicated by their enrichment in the geochemical

    reference frame elements. Volumes still increase

    slightly in some chlorite schists because of the much

    lower density of chlorite relative to the mixing end

    members. Calculated volume change for the rinds on

    mafic blocks, using the specific gravity data of Sor-

    ensen (1988) for these samples, demonstrate that, on

    the average, the outer rinds gained more mass and

    volume than the inner rinds (see Table 5), and that all

    rinds have volume changes that fall on the low end of

    the positive volume gains calculated for mélange

    matrix samples.

    The talc- and amphibole-rich mélange matrix sam-

    ples that show the largest gains in mass are those that

    contain the smallest concentrations of Al2O3 ( < 3

    wt.%, cf. Table 1, Table 3). These resemble ultramafic

    rind sample 6-4-100FR (with 1.21 wt.% Al2O3), for

    which a f 200% mass gain was calculated (see Table5). These samples with the largest gain also show the

    greatest variation among the observed/calculated val-

    ues for the geochemical reference frame elements (Cr,

    Al, Mg, Ni, Fe, and Zr). The most likely interpretation

    is that these samples represent significant metasomatic

    additions of Si, Mg, FCa to the other reference frameelements. The ultramafic rind sample (6-4-100FR) is

    fibrous and undeformed (see sketch in Fig. 2b) and

    appears to have formed via metasomatic processes at

    the rim of an ultramafic body (with little or no

    deformation). Alternative interpretations, for example,

    a protolith depleted in the reference frame elements,

    are possible but seem less likely given the field

    relationships and restricted Cr and Al concentration

    ranges for the serpentinites.

    6.3. Implications of mixing for stabilities of accessory

    mineral trace element hosts

    The work by Sorensen and Grossman (1989, 1993)

    on variably migmatitized mafic blocks and their

    metasomatic rinds, together with the mineral assem-

    Fig. 10. Diagrams presenting results of the Cr–Al-based mixing calculations for elements which show overall systematics dissimilar to those of

    the elements belonging to the ‘‘geochemical reference frame’’ (see Figs. 8 and 9). (a) SiO2 isocon plot, (b) Si (observed/calculated) vs. Al

    (observed/calculated), (c) LOI (observed/calculated) vs. Si (observed/calculated), (d) CaO (observed/calculated) vs. Na2O (observed/calculated),

    and (e) K (observed/calculated) vs. Ba (observed/calculated), and (f) K (observed/calculated) vs. Sr (observed/calculated). On these plots, the

    1:1 lines represent compatibility with simple mixing, without depletion or enrichment relative to the simple mixtures of mafic and ultramafic

    rocks. Note the compatibility (and even resemblance) of (b) with Fig. 5b, the two emphasizing the significance of Si losses and gains in the

    mélange matrix.

    Table 5

    Calculated mass and volume loss for rinds (data for mafic rinds

    from Sorensen and Grossman, 1989)

    Sample UM/M DMa DVb

    Inner rinds

    CA866A 0.01 8 16

    CA866B 0.12 7 19

    L11786 0.05 10

    Outer rinds

    CA865 0.40 2 15

    CA864 0.92 16 31

    L11781 0.99 44

    Other rinds

    714822 0.63 13 30

    1113842 0.15 23 39

    712847 0.35 2 15

    Ultramafic rindc

    6-4-100FR 3.50 203 289

    a Change in mass relative to calculated compositions.b Change in volume, using densities from Sorensen (1988).c Ultramafic rind from thi s study (composition in Table 3).

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–106 99

  • blage data presented in this paper (Table 1), allows

    consideration of the role of accessory minerals on

    trace element behavior during mélange formation.

    One consequence of tectonic (mechanical) mixing is

    that it shifts bulk compositions such that accessory

    trace element mineral hosts can be destabilized. For

    example, mechanical mixing alone or combinations of

    mechanical mixing and fluid addition can destabilize

    titanite, a significant trace element host in metamafic

    rocks (cf. Green, 1981; Sorensen and Grossman,

    1989). In the Catalina Schist amphibolite-grade mél-

    ange, titanite occurs in mafic blocks and in rinds

    developed at the margins of mafic blocks, but is

    replaced by rutile and ilmenite as the Ti-rich phases

    in the mélange matrix samples. Sorensen and Gross-

    man (1989) demonstrated the importance of phases

    such as titanite in the trace element budget of the

    metasomatized blocks (particularly for elements such

    as the LREE, U, and Th). Release of these trace

    elements due to breakdown of titanite to rutile or

    ilmenite could result in the removal/redistribution of

    the elements by fluids, depending on the evolving

    mélange mineral assemblage during mechanical and

    metasomatic mixing. Apatite, common in trace

    amounts in the mélange (Table 1), concentrates LREE

    in the mafic blocks (Sorensen and Grossman, 1989)

    and should also play a role in governing the behavior

    of the REE and U.

    Rutile may play a similar role, incorporating

    instead HFSE (e.g., Nb, Ta; see recent discussion by

    Johnson and Plank, 1999). Rutile, commonly in

    various stages of replacement by ilmenite (based on

    petrographic evidence), occurs in numerous samples

    of the mélange matrix (see Table 1). Sorensen and

    Grossman (1989) reported that rutile from migmatitic

    mafic blocks in the mélange contains 20 to 160 ppm

    Ta. The common replacement of rutile by ilmenite is

    suggestive that increases in FeO activity (e.g., varia-

    tions in FeO/MgO) could result in its destabilization.

    This is consistent with addition of mafic rocks which

    have higher FeO/MgO through mechanical mixing.

    6.4. Degree of involvement of metasedimentary rocks

    Field relations suggest that sedimentary rocks were

    not important in formation of the mélange matrix.

    Metasedimentary blocks, mostly chert, are rare within

    the mélange, and semipelitic to pelitic schist occurs as

    blocks in only one or two exposures near the contact of

    the mélange unit with the coherent metasedimentary

    and metagabbroic section. In contrast, field relations

    indicating mechanical and metasomatic interactions

    involving mafic and ultramafic rocks are ubiquitous

    within the complex.

    Unequivocal geochemical evidence to preclude

    involvement of sediment is difficult to produce. The

    mélange matrix compositions could accommodate a

    moderate, though unrecognized amount of sediment.

    One of the largest differences between mafic and

    sedimentary compositions is the higher content of

    LILE and Na in the latter (Bebout et al., 1999; Fig.

    11a). Stripping of LILE (e.g., K2O) and Na would be

    required if sediments were involved in mélange for-

    mation (see data for the Catalina Schist metasedimen-

    tary rocks relative to the mélange matrix data and the

    mafic–ultramafic simple mixing region in Fig. 11a).

    Also, because of the low CaO content of the meta-

    sedimentary rocks compared to the CaO content of the

    similarly aluminous, somewhat more Mg-rich mafic

    end member (Fig. 11b), the production of mélange

    matrix by large amounts of sedimentary-ultramafic

    mixing would likely require substantial addition of

    CaO from external sources. The relatively good geo-

    chemical fit obtained from mafic–ultramafic mixtures

    alone (see Fig. 11b) and the clear field evidence for

    their mutual interaction also argues against a major

    sedimentary component (on Fig. 11b, see the large

    arrows indicating rind-formation trends). Nonetheless,

    incorporation of small amounts of sediment, partic-

    ularly the more SiO2-rich material, cannot be pre-

    cluded.

    6.5. Integrated model for the evolution of the mélange

    Field and petrographic evidence suggest that the

    mélange matrix was developed through metasomatic

    interactions and mechanical mixing of mostly ultra-

    mafic material and mafic lithologies. In the field,

    intermediate stages of this process are clear. Large

    parts of the mélange matrix compositionally overlap

    with rinds on the mafic (hornblende-rich) or ultramafic

    (actinolitic clinoamphiboleF orthoamphiboleF ensta-tite) blocks. Nowhere in the mélange can similar

    relationships be found related to minor metasedimen-

    tary (metagreywacke and metachert) blocks. Intense

    veining by high-T assemblages indicates fracturing of

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–106100

  • blocks as they were entrained in the mélange. Within

    larger expanses of serpentinite, schistose zones con-

    taining Al-rich, high-T mélange matrix and showing

    development of metasomatic mineralogical zonations

    in ultramafic materials at their contacts, represent

    localized shear zones developed during block incorpo-

    ration (Fig. 2b). The mineralogical zonations evidence

    substantial fluid movement along these structures with

    concomitant metasomatic transfer of soluble constitu-

    ents (e.g., Ca, Si, in addition to H2O). Although the

    contents of most elements in the mélange matrix fit a

    mixing model based on reaction of mafic and ultra-

    mafic blocks and their rinds, some elements (e.g., Na,

    K, Ba, Sr) appear to require fluid-mediated metaso-

    matic redistribution or stripping after mixing (Fig. 10).

    Our earlier stable isotope studies of the Catalina

    Schist, including the amphibolite unit mélange, con-

    cluded that the mélange unit equilibrated with an

    aqueous fluid that had an O-isotopic signature derived

    from a lower-temperature sedimentary source, and

    that this fluid homogenized O and H isotopic compo-

    sitions in the mélange matrix over kilometer-scale

    distances (Barton et al., 1987; Bebout, 1991; Bebout

    and Barton, 1989, 1993). Furthermore, we inferred

    that these sediment-sourced fluids would have been

    capable of transporting large amounts of SiO2 into the

    mélange unit to the point of creating talc-bearing

    assemblages (Bebout and Barton, 1989, 1993; cf.

    Manning, 1995).

    Felsic pegmatites in the mélange formed during the

    high-T evolution of the mélange, as they are observed

    to both cross-cut the mélange fabric and to be

    deformed into the mélange fabric (Fig. 2a; Sorensen

    and Barton, 1987). Sorensen and Barton (1987) pro-

    posed that the pegmatites in the mélange represent

    melts derived from migmatization of mafic blocks in

    the mélange unit, a hypothesis developed in detail by

    Sorensen and Grossman (1989). Our further work has

    indicated that the pegmatites show a wide range of

    trace element compositions which fall between those

    observed for pegmatites obviously derived from the

    coherent metasedimentary sections and those in the

    coherent metamafic sections in the Catalina Schist

    amphibolite-grade unit (Bebout and Barton, 1993,

    unpublished data). Thus, the mélange unit appears to

    preserve effects of the transfer of both H2O-rich C–

    O–H–S–N fluids, which homogenized the O and H

    isotopes, and water-rich felsic melt, the latter sourced

    from diverse mixtures of mafic and sedimentary

    components, some also possibly outside the mélange

    zone.

    Fig. 12 illustrates some possible mineralogical paths

    for mechanical mixing and metasomatic silica. On

    these activity–activity diagrams, a trajectory between

    Fig. 11. Chemical discrimination diagrams comparing the compo-

    sitions of mélange matrix samples with those of mafic, ultramafic,

    and sedimentary rocks in the Catalina Schist (data for metasedi-

    mentary rocks from Bebout et al., 1999). (a) Plot of K2O vs. Na2O,

    demonstrating the far higher concentrations in the metasedimentary

    rocks relative to the mafic mixing tectonic blocks, and (b) plot of

    MgO vs. Cao, demonstrating the low Mg, low Ca character of the

    metasedimentary rocks in the Catalina Schist. Data for metasedi-

    mentary rocks are for samples from all units of the Catalina Schist,

    as detailed geochemical study has indicated no detectable variations

    in the Si–Al–Mg–Ca compositions as functions of differing

    metamorphic grade. The two large, shaded arrows on (b) are the

    trends in composition from the mafic and ultramafic blocks toward/

    through the data for rinds developed on each of the block types. On

    (a) and (b), the extent of the envelope of simple mixing of

    ultramafic and mafic compositions is indicated by the hand-drawn,

    dark, thin lines.

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–106 101

  • mafic rocks (garnet–pyroxene–amphibole) and oli-

    vine-rich rocks broadly reproduces the patterns

    observed. Mechanical mixing can stabilize chlorite +

    Al-rich clinoamphibole assemblages, but cannot pro-

    duce the highly siliceous, nearly monomineralic talc

    schists that are abundant in the mélange (Figs. 5b and

    10a and b). The latter require silica introduction, and on

    the diagram, this is shown as fluid-mediated addition.

    These fluids, which the evidence noted above shows

    must be external to the unit, would also redistribute or

    leach alkalis and other soluble components to yield the

    diverse mélange matrix compositions. Also illustrated

    is the destabilization of titanite along the same path.

    Silica solubility relationships with olivine and talc-

    bearing assemblages require that a minimum of several

    rock masses of water must have passed through the

    ultramafic rocks to produce the talc-rich assemblages

    (cf. Bebout and Barton, 1989, 1993; Manning, 1995;

    Newton and Manning, 2000). Multiple paths are pos-

    sible; thus, there is a large uncertainty in any fluid–

    rock estimate. The most favorable circumstances for

    the silicification would be for waters equilibrated with

    sedimentary rocks in the amphibolite unit of the

    Catalina Schist to move into the ultramafic mélange.

    Least favorable would be significant up-temperature

    flow from lower-grade units, which would require

    Fig. 12. Calculated activity–activity diagrams (in a–c for systems labeled on each diagram; all for 650 jC, 1.0 GPa) illustrating possiblemineralogical consequences of mechanical mixing and fluid-mediated SiO2 additions. Activities are those of the oxides (lime, periclase and

    quartz) at 650 jC and 1.0 GPa. In (d), labeled ‘‘Mass Exchange Processes’’, the various lithologies regarded as mixing components (‘‘Eclogite’’,‘‘Garnet Amphibolite’’, and ‘‘Peridotite’’) and a generalized compositional range for mélange matrix are indicated. Thermodynamic data for

    most species are from Johnson et al. (1992) and adapted from Robie et al. (1978) for sphene (titanite) and rutile.

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–106102

  • considerable layer-parallel flow before silicification

    could take place under amphibolite-facies conditions

    (Barton et al., 1987; M.D. Barton, G.E. Bebout, in

    preparation). These volumes are consistent with the O

    isotopic shifts described above, which also require an

    external source of sediment-equilibrated fluid.

    Fig. 13 illustrates schematically the inferred tec-

    tonic setting of formation of the amphibolite-facies

    mélange unit (also see Fig. 3 in Bebout and Barton,

    1989). The mélange is possibly a fragment of a zone of

    complex mechanical and metasomatic mixing of mafic

    (possibly slab-derived; see discussion by Sorensen and

    Grossman, 1989) components with ultramafic materi-

    als from the hanging wall. Mafic and ultramafic blocks

    in the mélange represent variably but incompletely

    incorporated mafic and hanging-wall materials that

    were entrained in the evolving shear zone. Some of

    the blocks were eventually extensively metasomatized

    and became tectonically (mechanically) incorporated

    into the mélange matrix (at various scales). During this

    development, the mélange was infiltrated by hydrous

    C–O–H fluid enriched in SiO2 and other dissolved

    components. The H2O-rich fluids stabilized the present

    hydrous, siliceous assemblages in the mélange and

    locally stabilized ankeritic dolomite and magnesite.

    The model involving infiltration of the mélange by

    fluids from subducted, dominantly sedimentary sour-

    ces is supported by evidence for extensive SiO2mobility in the lower-grade units of the Catalina Schist

    and isotopic systematics (Barton et al, 1987; Bebout,

    1991; Bebout and Barton, 1989, 1993). A model of

    SiO2 addition to the mélange via infiltrating aqueous

    fluids from the slab and sediments is also supported by

    experimentally and theoretically inferred high solubil-

    Fig. 13. Cartoon showing inferred tectonic setting of mélange formation along the slab–mantle interface. The mixing zone develops between

    the downgoing slab and sediments and the hanging wall, and probably grades into highly deformed, low-Al, hydrated ultramafic rocks (in the

    Catalina Schist, initially dunites and harzburgites, based on major element chemistry) upward into the hanging wall. Fluids (hydrous fluids and

    siliceous silicate melts) derived in the slab and sediment enter the mixing zone and may either be transported along the mélange or, after

    experiencing fluid– rock interactions in the mélange, rise into the hydrated and deforming hanging wall. Beneath arcs, some of the chemically

    evolved fluid could ultimately ascend into and chemically interact with arc source regions in the mantle wedge.

    G.E. Bebout, M.D. Barton / Chemical Geology 187 (2002) 79–106 103

  • ities of quartz in aqueous fluids under subduction-zone

    P–T conditions (e.g., Manning, 1996). Calculated

    mineral–fluid solubility relations (Manning, 1998)

    indicate that other components (Al, Na, in particular)

    may well be mobile in metamorphic fluids at blues-

    chist- to eclogite-facies conditions.

    7. Implications of mélange formation deep in

    subduction zones

    The inferred depths of peak metamorphism for the

    Catalina Schist range from 25 to 45 km, obviously

    shallower than those inferred for the slab–mantle inter-

    face beneath arcs. Inferred metasomatic and structural

    processes for the Santa Catalina rocks must, therefore,

    be considered as analogues to those which operate at

    greater depths, or at least as an indication of the

    complexity of such processes which may affect such

    deeper zones. The Santa Catalina rocks probably rep-

    resent early stages of subduction (Platt, 1976; Grove

    and Bebout, 1995), but the rocks of the highest-grade

    amphibolite unit have experienced 650–750 jC tem-peratures not unlike those that may affect deeper parts

    of subduction zones at later stages (Peacock, 1993).

    The lower-grade mélange matrices in the Catalina

    Schist may be considered analogues to the ultramafic

    mélange in the amphibolite unit but formed through

    intense mechanical mixing and with coeval metaso-

    matic alteration at significantly lower temperatures

    more similar to those predicted for these depths by

    thermal models of mature subduction zones (350–450

    jC; Grove and Bebout, 1995; cf. Peacock, 1993). Thelower-grade mélange exposures show many similar-

    ities with the amphibolite-grade mélange, including

    metasomatic rinds on tectonic blocks, field evidence

    for progressive tectonic incorporation, and domains of

    high-variance mineral assemblages, indicating meta-

    somatic formation (see Bebout and Barton, 1993;

    Bebout, 1997).

    Geochemical models invoking additions from the

    subducting slab and sediments to explain complex

    isotopic and trace element systematics generally call

    upon discrete hydrous fluid or melt phases derived

    from one or the other of the subducting lithologies.

    Interestingly, the Catalina Schist amphibolite-facies

    mélange contains evidence for the migration of both a

    hydrous phase and felsic silicate melts during its

    deformation, perhaps providing a field expression of

    the generation and mobility of multiple slab/sediment-

    derived fluid phases. Presumably, these fluid phases

    must traverse any complex mechanical mixing zone at

    the slab–mantle interface in order to enter the mantle

    wedge, and the potential exists for extensive fluid–

    melt–rock interactions resulting in modification of the

    slab fluid phases prior to their migration into the

    overlying mantle wedge. It is difficult to explicitly

    evaluate such fluid–rock interactions in studies of arc

    geochemistry; however, we suggest that further stud-

    ies could consider the possibility that single, mixed

    (perhaps even homogenized at some large scale; cf.

    Morris et al., 1990; Bebout, 1991) fluid phases

    possessing chemical signatures of mechanically hybri-

    dized lithologies could produce many of the arc geo-

    chemical systematics. It is notable that, in the Catalina

    Schist ultramafic mélange, LILE are highly mobile as

    has been inferred in studies of arc magma sources

    (e.g., Gill, 1981; Pearce and Peate, 1995). In the

    voluminous chlorite-dominated parts of the mélange,

    LILE were stripped but only locally enriched else-

    where as in the rinds on the mafic blocks.

    Mélange formation, resulting in the production of

    high-variance assemblages with high water content

    may also be an important process in facilitating reten-

    tion and storage of H2O at depth in subduction zones

    (Bebout, 1991). Whereas chlorite is unstable in mafic

    and sedimentary bulk compositions at amphibolite and

    greater metamorphic grades, chlorite (containing f 13wt.% H2O) in rocks of bulk compositions near that of

    its own composition may be stable to far greater

    temperatures and pressures (cf. Jenkins and Chernosky,

    1986). Talc (containing f 5 wt.% H2