Metasomatism During Subduction: Products and Possible Paths in the Catalina Schist, California

ChemicaIGeology, 108 (1993)61-92 61 Elsevier Science Publishers B.V., Amsterdam

Metasomatism during subduction: products and possible paths in the Catalina Schist, California

Gray E. Bebout a and Mark D. Barton b aDepartment of Earth and Environmental Sciences, 31 14~lliams Drive. Lehigh UniversiO,. Bethlehem, PA 18015. USA bDepartment ofGeosciences, LSm,ersity of Arizona. Tucson, AZ 85287. USA (Received February 15, 1993; revised and accepted April 8, 1993

ABSTRACT

On Santa Catalina Island, southern California, lawsonite-albite to amphibolite facies metasedimentary, metamafic, and metaultramaflc rocks show veining and chemical alteration that reflect fluid flow and mass transfer at 15 to 45 km depths in an Early Cretaceous subduction zone. In many exposures, multiple generations of cross-cutting syn- and post-kinematic veins record fluid transport and metasomatism during various stages of prograde metamorphism and uplift. Mineralogy and whole-rock compositions demonstrate chemical redistribution, especially of Si, AI, and alkali elements (Na, K ), but also of many trace elements, particularly B and LILE (Rb, Cs, St, and Ba). Evidence exists for mass transfer, at both local and larger scales, via mechanical mixing, diffusional, and fluid-mediated transfer processes. Highest-grade, amphibolite facies rocks contain feldspar + quartz _+ mica _+ amphibole leucosomes and pegmatites attributed to migmatization; the leucosomes and pegmatites reflect high-P/7" mass transfer in felsic silicate liquids. Veining and replacement in blueschist grade rocks comprise three contrasting types of assemblages: ( 1 ) silica-saturated (quartz-rich), ( 2 ) potassic (white-mica _+ quartz-rich ), and ( 3 ) sodic and silica-undersaturated (albite/Na-amphibole- rich, quartz-absent). Evidence for silicification and alkali exchange also occurs in greenschist and amphibolite facies units. In all units, the evidence for metasomatism (e.g., veins; stable isotope homogenization; rinds on blocks) is particularly abundant in melange zones, in which melange matrix compositions resulting from mechanical mixtures of mafic, ultramafic, and sedimentary rocks were shifted by metasomatic additions and subtractions during melange formation. Geo- chemical evidence (particularly stable isotope data) indicates that the blueschist, greenschisL, and amphibolite units ex- changed with fluids of similar compositions. The diverse metasomatic features in the Catalina Schist provide evidence regarding fluid sources and paths. Based on the stable isotope data, the H20-rich, low-salinity ( ~ 1 to 2 equivalent wt. % NaCI), C-O-H-S-N fluids are believed to have been derived from low-grade, largely sedimentary parts of the subduction zone (analogs lbr fluid sources are the low- grade units). Metasomatic changes could be driven by flow across boundaries between contrasting lithologies and by variations in pressure and temperature along the fluid flow paths. Simple predictions of mass changes along different P-T paths suggest that both mechanisms could be effective at producing the range of observed features, even though the required equilibrium constants are only poorly estimated at the relevant P-T conditions. Decreasing T and P favors fixing of K, Si, C, and H in rocks, whereas increasing T ( _+ moderately decreasing P) should fix Na but leach most other components. The Si-rich, K _+ Si-rich, and Na-rich/Si-poor assemblages are thus consistent with differing fluid P-T flow paths. Regular differences are expected in silica precipitation/dissolution, alkali exchange, and hydrogen-alkali exchange reactions, among others. Silica_+ carbonate addition, consistent with the majority of veins observed, is likely the consequence of cooling _+ decompression whereas sodic ( _+ silica-undersaturated ) assemblages would be expected for rarer, but geolog- ically plausible up-T fluid flow paths. A composite fluid flow path, first up-grade, then down P and T. is indicated for the silica addition to the largely ultramafic amphibolite-facies melange. Although mass balance and physical constraints appear to preclude pervasive major element metasomatism on large scales, focussing of fluids would likely produce pervasive changes in significant volumes (e.g., up to kin-scale melange zones). Vein mineralogy would record the paths even at small fluxes. Study of the Catalina Schist demonstrates the significance of metasomatism at all scales, but indicates that large-scale changes in vein mineralogy and bulk composition are in some cases attributable to fluid flow over large distances. Comparison with other areas and elementary theoretical considerations suggest that these processes may be widely developed and that their petrographic and[ geochemical effects potentially give insight into the dynamics of subduction zones.

I. Introduction sitional changes during: metamorphism may have resulted from fluid infiltration (e.g., Subduction of oceanic lithosphere results in Ernst, 1965; Essene and Fyfe, 1967; Fyfe and the transfer of volatile-rich, hydrothermally- Zardini, 1967; Liegeois and Duchesne, 1981 ; altered mid-ocean ridge basalts and sediments Moore and Liou, 1979; Moore et al., 1981; to the upper mantle (e.g., Ito et al., 1983). In- Moore, 1984; Maruyama and Liou, 1987, terstitial pore fluids make up a large fraction 1988). Stable isotopic data from subduction of the volatiles initially subducted ( > 50 vol- complexes (Taylor and Coleman, 1968; Ma- ume % in some sediments), but much of this garitz and Taylor, 1976; Rumble and Spear, is probably expelled at shallow levels in sub- 1983; Matthews and Schliestedt, 1984; Agri- duction zones and escapes toward the surface nier et al., 1985; Barton et al., 1987; Bebout ( Langseth and Moore, 1990). At greater depths and Barton, 1989a; Bebout, 1991a,b) demon- in subduction zones ( > 15 km), volatiles are strate extensive isotopic exchange and fluid primarily those bound in minerals. These vol- mobility. Radiogenic isotope studies of high- atiles, when released by metamorphic devola- P/T metamorphic complexes (Barton et al., tilization reactions, may play a key role in the 1987; Nelson, 1991 ), together with trace ele- geochemical evolution of the slab-mantle in- ment analyses of veins and other metasomatic terface and overlying mantle wedge. Fluids may features (Gillet and Goffe, 1988; Sorensen and flux partial melting in arc magma source re- Grossman, 1989; Philippot and Selverstone, gions (70 to 150 km) and contribute slab-de- 1991; Bebout et al., 1993a,b), document mo- rived chemical components to such regions bility of Sr (and other large-ion lithophile ele- (Gill, 1981; Wyllie and Sekine, 1982; David- ments, LILE), rare-earth elements (REE), and son, 1987: Ellam and Hawkesworth, 1988; other trace elements in metamorphic fluids Morris et al., 1990). Fluids may also strongly during subduction-zone metamorphism. influence deformation in subduction zones In this paper, we present field and geochem- ( Dumitru, 1991 ). ical evidence for local and large scale fluid Once released from the subducted rocks, by transport in the Catalina Schist Terrane, an either mechanical processes or by devolatili- early Cretaceous subduction complex in zation reactions, volatiles (supercritical fluids) southern California. These results, and those are potential agents of metasomatism and for other subduction-related metamorphic could potentially redistribute mass over large complexes, are used along with theoretical ar- distances. Abundant evidence exists for mass guments to propose that different fluid P-T transfer associated with fluid mobility in shal- flow paths may produce distinctive metaso- low parts of accretionary complexes (see matic events (Barton and Bebout, 1989; Be- Moore and Vrolijk, 1992 ). Although many ob- bout and Barton, 1989a). We first describe the servations imply the presence of abundant field and petrographic evidence for fluid mo- metamorphic fluid during high-pressure/tem- bility and metasomatism in the Catalina Schist perature (P/T) metamorphism at intermedi- (summarized by Bebout and Barton, 1989a), ate depths of subduction zones ( 15 to 70 km), emphasizing veining systematics and altera- few studies have documented the fluid-related tion within rocks of various bulk compositions mass transfer. Subduction-zone metamorphic as functions of metamorphic grade. We then terranes commonly show complex vein sets and present major and trace element data which local evidence for bulk chemical alteration. A document compositional changes of mafic, ul- number of studies have suggested that compo- tramafic, and sedimentary protoliths during METASOMATISM DURING SUBDUCTION: PRODUCTS AND POSSIBLE PATHS IN CATALINA SCHIST, CA 63 high-P/T metamorphism. Also, we summarize Schist contains lithologies that range in grade O, H, and C isotope data (see Bebout and Bar- from lawsonite-albite- and blueschist-facies to ton, 1989a; Bebout, 1991a,b) which demon- amphibolite-facies (Platt, 1975, 1976; Bebout, strate trends toward isotopic homogenization 1989). Three distinct metamorphic units are in rocks; these trends can be explained only by juxtaposed along low-angle faults, resulting in extensive fluid-rock interaction over large dis- an inverted metamorphic gradient, with struc- tances. The stable isotope systematics, partic- turally-lowest blueschist-facies rocks overlain ularly for O, C, and N, indicate that the low- by greenschist-facies rocks, which are in turn grade metasedimentary rocks in the subduc- overlain by the structurally-highest amphibol- tion zone were fluid sources (Barton et al., ite-facies unit (see inset, Fig. 1 ). Bailey ( 1941 ) 1987; Bebout and Barton, 1989a; Bebout, first mapped the Catalina Schist and suggested 1991 a; Bebout and Fogel, 1992 ) and thus im- high fluid mobility and metasomatism, based plicate the same rocks as plausible sources for on the abundance of veins and siliceous ultra- many of the other chemical components mo- mafic rocks. Sorensen (1988), Sorensen and bilized during high-P/T metamorphism (e.g., Barton (1987), and Sorensen and Grossman trace elements such as B, LILE). (1989) subsequently considered local, and Finally, we consider the relative significance possibly larger scale, fluid- and melt-mediated of diffusional processes, mechanical mixing, mass transfer in studies of trace element distri- and infiltration metasomatism in producing butions in metasomatized and migmatized the disparate metasomatic styles in subduc- metamafic blocks in the amphibolite-facies tion-related metamorphic rocks and the con- melange. sequences of varying fluid P-T flow trajecto- In all units of the Catalina Schist, relatively ries for styles of metasomatic alteration in coherent metasedimentary (+ metamafic) subduction zones. The results of this study, domains are interleaved with melange zones, combined with reports of metasomatic fea- which are, in most cases, dominantly ultra- tures in other high-P/T metamorphic com- mafic in composition. The volumetric propor- plexes and the record of metasomatism from tions of metasedimentary and metamafic rocks petrological and geochemical studies of arc change with increasing metamorphic grade magmatism, suggest some general systematics (metasedimentary rocks comprise ~70% of of fluid and mass transfer in subduction zones, blueschist unit exposures and ~ 10% of am- particularly for intermediate depths ( 15 to 45 phibolite unit exposures); metaultramafic km ). rocks are most abundant in the amphibolite facies unit. Metasedimentary rocks include both 2. Geologic setting metamorphosed, pelagic pelitic and siliceous rocks and metagraywacke and metaconglom- The Catalina Schist, an Early Cretaceous erate. Metabasaltic rocks of up to greenschist subduction complex exposed on Santa Catal- grade show relict pillow structures either in in- ina Island (Fig. 1 ), southern California, contact sequences ofunfoliated pillows or in vari- tains an unusually complete sequence of high- ably foliated pillow breccias. P/T metamorphic rocks. The Catalina Schist In all units, the melange matrix represents contains rocks similar to those in many other mechanical mixtures of ultramafic, mafic, and subduction-related metamorphic complexes sedimentary rocks in varying proportions (see (cf. Bailey et al., 1964; Ernst et al., 1970). Bebout, 1989; Bebout and Barton, 1989a). However, unlike these other dominantly low- These melange domains are unusual among grade subduction complexes, the Catalina high-P/T metamorphic terranes (e.g., the 64 G.E. BEBOUT AND M,D. BARTON

SANTA CATALINA ISLAND Isthmus 2 3 4mi , j , : , J j, __~ 1 2 3 4 5 6 km

LOS Angeles k ~'~..~ Southern

"... . Littl / I P.c tic r~ UltrarnaficMelange Hart .~ ~ ~r,;::'C~ (with tectonicblocks) ~ [~ CoherentAmphibolite NN [ ~i Island

W Airport E Fault North Side ~500 m ~, Feult ] OIIs~ f Thrust

0 1 2 3km after Platt (1976)

Fig. 1. Simplified geological map of Santa Catalina Island (after Plait, 1976); insets show schematic cross-section (also after Platt, 1976 ) and the location of the island.

Franciscan Complex; see Cloos, 1986) in that (1987), Sorensen (1988), and Sorensen and the mineral assemblages in the matrix indicate Grossman ( 1989 ). The melange matrices typ- peak metamorphic recrystallization at the same ically have isotopic values that are more conditions as those that affected the more co- strongly shifted than more coherent domains, herent parts of the complex and most blocks in indicating that melange zones localized fluid the melange (of metamafic, metaultramafic, flow in all units (Barton et al., 1987; Bebout, and to a lesser extent, metasedimentary rocks ). 1991a,b). The blueschist unit and amphibolite unit me- Petrologic studies by Platt( 1975, 1976 ) and lange matrices contain veins and high-vari- Sorensen ( 1984 ) indicated peak metamorphic ance mineral assemblages indicative of mass conditions of 7-12 kbar and 350-750°C. Ad- transfer (see discussions in later sections). The ditional petrography shows that rocks on the amphibolite facies melange contains variably west end of the island represent sub-bluesch- metasomatized and mechanically incorpo- ist-facies, lawsonite-albite-type metamor- rated blocks ( 1 cm to 1 km in length) of me- phism at somewhat lower pressures (Bebout, tamafic and metaultramafic rocks and is inter- 1989). Platt (1976) suggested that the rocks preted as a zone of mechanical and were metamorphosed in an inverted thermal metasomatic mixture near the slab-mantle in- gradient developed during early stages of sub- terface (see Fig. 3 in Bebout and Barton, duction; this model would explain the anoma- 1989a). Metasomatic alteration and partial lously high geothermal gradients of up to melting of the metamafic blocks have been de- 30°C/km inferred for the Catalina Schist scribed in detail by Sorensen and Barton compared with other high-P/T metamorphic METASOMAT1SM DURING SUBDUCTION: PRODUCTS AND POSSIBLE PATHS IN CATALINA SCHIST, CA 65 terranes (see Ernst, ! 988 ) and thermal models 4. Evidence for fluid flow and mass transfer of subduction (e.g., Anderson et al., 1978; Peacock, 1990). This interpretation is consistent with reconstructions of plate motions that All units of the Catalina Schist contain fea- indicate renewed convergence approximately tures that reflect fluid flow and mass transfer 125 Ma between the North American and Far- during various stages o f metamorphism. These allon plates (Engebretson et al., 1985) about features include veins, reaction zones between 10 to 15 Ma before the apparent peak of Catal- disparate lithologies (which do not necessarily ina metamorphism. Recent 4°Ar/39Ar study of require the involvement of fluid), changes in the Catalina Schist indicates that peak meta- bulk composition, and large-scale changes in stable isotopic composition. Individual out- morphic recrystallization in the three major crops commonly show multiple veining or al- metamorphic/tectonic units may not have teration events. Using textural and petrologic been synchronous (Grove and Harrison, criteria, the metasomatic features can be placed 1992 ). However, the wealth of geochronologi- into the context of the: metamorphic and de- cal data shows that all units of the Catalina formational histories of the rocks. The critical Schist were metamorphosed during a pro- observations are the relationships between the tracted Early Cretaceous subduction episode fabrics and mineral assemblages of the veins (Suppe and Armstrong, 1972; K-Ar, 98-112 (and overprinting mineral parageneses) to Ma; Mattinson, 1986; U-Pb, 112-114 Ma). those of the host rocks. Units mapped in Fig. 1 as the blueschist and greenschist units are subdivided to distinguish variable metamorphic 3. Analytical techniques grade within these areas (glaucophanic- greenschist and epidote-amphibolite parts of Whole-rock and mineral separate major and greenschist unit described by Sorensen, 1986; trace element compositions were analyzed by lawsonite-albite parts of blueschist unit de- XRF, INAA, and ICP techniques. Except scribed by Bebout, 1989 ). where otherwise noted, INAA were acquired Table 1 lists the common individual vein as- through the Oregon State University Reactor semblages and other characteristic alteration features for each bulk composition at the var- Sharing Program; ICP analyses were per- ious metamorphic grades. Also listed are the formed either at the Carnegie Institution of textural and petrological criteria used for Washington (Geophysical Laboratory and placement of these metasomatic features into Department of Terrestrial Magnetism), or by the deformational and metamorphic histories Activation Laboratories, Ltd. (Ontario, Can- of the host rocks. The photographs in Fig. 2 il- ada); some XRF analyses were performed at lustrate several examples of metasomatism in the University of California, Los Angeles. Data the Catalina Schist. For' the geochemical study, obtained for individual samples by multiple emphasis was placed on features regarded as techniques are satisfactorily comparable (gen- "prograde" based on mineralogy and texture; erally to ~<5%). Electron microprobe mea- these features were presumably produced prior surements were performed at UCLA on a Ca- to or during peak metamorphism of the rocks. meca Camebax instrument. Techniques for the Some veins and other rnetasomatic features are stable isotope analyses, performed at UCLA more likely products of either pre-subduction and Carnegie Institution of Washington (Geo- alteration (e.g., diagenesis, seafloor hydro- physical Laboratory), are described elsewhere thermal alteration ) or post-peak metamorphic (Bebout, 1989; Bebout and Fogel, 1992). alteration (retrograde fluid flow). 66 G.1E. BEBOUT AND M.D. BARTON

TABLEI

Textural and mineralogical characterization of metasomatic features (units in order of metamorphic grade )

Bulk comp. Feature a Mineralogy b Prograde/ Texture relative Syn/post Cross-cuts/ (V, etc.) retrograde c to host rock fabric kinematic replaces

La wsonite-albite Sedimentary Ct-c carbonate (now Cc ) PS-i relict carbonate cement PS V-l-t WM P-c concordant S V-2-a Qz + Cc + Lw P transposed/concordant S V- 1 V-3-a Qz+Cc_+ Lw P boundin fracture filling S V-4-a Qz+Cc+ Pp(f) P-i discordant P V-l, 2, 3 V-5-c Qz ? discordant P V- 1 to 4 V-6-c Cc ? discordant P V-1 to 4 Mafic V-l-t Cc + Ab_+ Ac (pyrite) PS pods/discontinuous veins PS V-2-r Cc_+Gr+_Ab (black) ? unfoliated host ? V-3-c Cc+Ab (white) ? unfoliated host ,7 V-2 V-4-c Ch + Lw + Ab + Ac _+ P unfoliated host ? allanite +_ Cc V-5-c Cc + Ab + WM ? unfoliated host 9 (pyrite) Melange matrix MA-a Ac +_ Sph _+ Tc P matrix assemblage S - (Ch, Ab, Cc ) V-l-u Qz ? transposed/segregation S V-2-u Cc ? discordant P MA, V- 1 Blueschist Sedimentary Ct-c carbonate (now Cc) PS relict carbonate cement PS (pelitic. V- I a-a Ab +_ Gr P-i transposed/concordant S - psammitic ) V- I b-a Qz _+ Cc P-i transposed/concordant S - V-lc-a Qz_+Cc (NA, Lw, St) P boudinage fracture filling S - V-2a-a Qz+Cc+_Ab P-i discordant vein sets P V-la, b, c V-2b-u Qz _+ Pp _+ Cc R discordant vein sets P V- I a, b, c Sedimentary V-l-a Qz+Ab (NA) P-i unfoliated host ? - (psammitic V-2-c NA_+ Qz_+Ab (f) P unfoliated host ? - only) Qz or Ab envelopes Gr-absent envelopes Mafic V- l-c Cc _+ NA _+ St (pyrite) PS pods/discontinuous veins PS - breccia fillings V-2-c Cc+Lw_+Ab P discordant/transposed S/P - V-3-u St _+ Cc _+ Ab (pyrite) P unfoliated host/discordant ? - V-4-c NA+Cc+Lw+Ab (f) P unfoliatedhost ? - Ultramafic (blocks in Rp-B-c Antigorite P-i replacement of ultramafic blk ? (ultramafic blk ) melange matrix ) V- l-u Cc ? vein network/ultramafic blk ? Rp-B Melange matrix MA-c NA + Qz _+ Sph P high-variance layers/domains S - R-B-c Ac + Sph _+ Ab P rind developed on mafic blk S (mafic blk ) V-2-c NA P discordant/transposed S/P MA V-3-u Ab + Ch + Cc? P discordant -undeformed P MA V-4-r Qz 9 discordant/transposed S/P MA/blocks

Glaucophanic greenschist Sedimentary V-l-a Qz ? discordant/transposed S/P - Mafic V- l-a Cc+Ep+NA (WM, PS pods/discontinuous veins PS - Ab) V-2-c Ep_+ Gt _+ Ab_+ Ch P discordant/transposed S/P - (Sph, Ap, WM ) V-3-u Cc + NA + Ab P discordant P V- 1.2 V-4-u NA+Chl (f) P-PS? veins near pillow rims ? - V-5-t Qz + NA + Ch + P boudin fracture fillings in S V-2 WM _+ Ep boudinaged V-2 V-6-r Pp + Ab R discordant P V- 1 to 3 META~SOMATISM DURING SUBDUCTION: PRODUCTS AND POSSIBLE PATHS IN CATALINA SCHIST, CA 67

Bulkcomp. Feature a Mineralogy b Prograde/ Texture relative Syn/post Cross-cuts/ (V, etc.) retrograde c to host rock fabric kinematic replaces

Greenschist Sedimentary V- 1-a Qz + Ab_+ Ch +_ Ac P? discordant/concordant/ S/P transposed Mafic V- l-r Ac + Ep + Ab P unfoliated host ? V-2-c Pp + Ab R unfoliated host ? Melange matrix MA-c Ac + Ch +_ Bt + Tc P matrix assemblage S Epidote amphibolite Sedimentary V-l-a Qz ? discordant/concordant S/P V-2-u Pp + Qz + Ch R discordant P V-I Mafic V- l-c Cc _+ Ep _+ WM PS pods/breccia fillings PS V-2-u Ep + Qz P discordant? ( unfoliated host) ? V-3-c Ep + Ab _+ Hb P discordant? ( unfoliated host) ? V-4-c Ep + Ab + WM P unfoliated host-networks ? V-5-c Qz (marg. hyd. ) d R unfoliated host ? V-4 V-6-c Ep ? unfoliated host 9 V-7-u Cc "~ unfoliated host ? V-8-c Pp + Ab _+ Ac R unfoliated host ? Amphibolite Coherent V- l -a Qz P segregations S sedimentary V-2-c Qz+PI+WM (Gr, P segregations/pegmatites S/P Czo, Gt, Tr, Ky, Hb, (Bebout and Barton, 1988) Ap, Rut) V-3-u Qz ? discordant P V-l,2 V-4-t Qz + Pp + Ch R discordant P V-I.2 Rp-l-c WM+Zo R replacement of plagioclase P V-2 Rp-2-c WM R replacement of kyanite P V-2 Rp-3-c Ch R replacement of garnet P V-2 Coherent mafic V-l-c Pl+Czo+Cx (Hb) P segregations/transposed veins S/P V-2-c Pl+Qz+Czo P segregations/pegmatites S/P (Gt, Hb, Ap, Rut) (Bebout and Barton, 1988) V-3-c Pp+Ab+Ac R discordant P V-l,2 V-4-u Ab and Qz veins ? discordant P V-I,2 Rp-l-c? Lw R replacement of zoisite P V-2 Ultramafic melange MA-a At, Ac, Hb, Tc, En, Ch P melange matrix assemblages S (Cb, Ilm, Zirc, Sf, Ap) Rp-2-c Tc + serpentine R replacement of siliceous P MA (Cb-magnesite) assemblages V- l-r Qz P? transposed S MA V-2-c Qz+PI+WM (CA, P discordant, concordant S/P MA Czo, Tr, Ap, Gr, pegmatites Gt) Ultramafic blocks V-l-c Ac+Tc+ En (f) P veins cutting unfoliated host S (zoned) V-2-c Ac, At, En (zoned) P mineralogical zonations S/P at rims of blocks Rp-l-a lizardite + chrysotile R pervasive replacement S/P? V-I,2 + magnetite _+ chl of dunite/harburgite Rp-2-c Tc + serpentine R replacement of siliceous P MA, V-l, 2 assemblages V-3-a chrysotile_+ Tc (f) R veins in serpentinized zones P V-I,2 V-4-c SiO2 polymorphs+ R veins in serpentinized zones P V-I,2 serpentine Sedimentary blocks V- l-r Ch d ? hydration seams P Rp-r Ch ? replacement of garnet, biotite P

(continued on p. 68 ) 68 G.E. BEBOUT AND M.D. BARTON

TABLE I (continued)

Bulk cornp. Feature" Mineralogy b Prograde/ Texture relative Syn/post Cross-cuts/ IV, etc. l retrograde c to host rock fabric kinematic replaces

Marie blocks V- 1-c Gt _+ Qz p segregations/transposed veins S - V-2-t Hb + Zo P replacement along fractures S? V- 1 Rp- l-a Hb + Zo P pervasive replacement S? - R- l-a Hb, Ac, Tc, Ch, Qz, P rinds on marie blocks I see S - Musc, Bt (Ap. Rut) Sorensen, 1984, 1988) V-3-a Qz + P1 + WM + P migmatitic segregates ? - Czo (Hb, Ap, Gt ) ( Sorensen and Barton, 1987 )

Note: Units are listed in order of increasing metamorphic grade. Qz -- quartz; Ab -- albitic plagioclase; Ch --- chlorite; WM -- white mica; Pp -- pumpellyite; Cc -- calcite: NA -- sodic amphibole; Lw -- lawsonite; St -- stilpnomelane; Gr -- graphite: Ep -- epidote; Ac -- actinolitic amphibole; Hb -- hornblende, usually magnesio-hornblende; PI -- plagioclase, not necessarily albitic; Gt -- garnet; Bt -- biotite: Ky -- kyanite: Zo -- zoisite or clinozoisite; Cx -- clinopyroxene; At -- anthophyllite; Tc -- talc: En -- enstatite; CA -- calcic clinoamphibole; Sf-- sulfide. Common vein accesso~' minerals include apatite, sphene, graphite, white mica and, in the lawsonite-albite and blueschist units, stilpnomelane. "Features include: veins (V); replacement assemblages (Rp); metasomatic rinds (R); cements formed during earlier stages of subduction or on the seafloor (Ct): metasomatic assemblages in melange matrix (MA). Indicators of abundance are: trace (t); rare (r) ; uncommon (u); common (c); abundant (a). bMinerals in parentheses are those that occur locally in such features (generally trace phases). (f) indicates that veins may show fibrous mineral growth, generally perpendicular to vein walls. Tr=tourmaline. CPrograde (P) refers to features developed during subduction but before attainment of peak metamorphic P-T conditions. Retrograde (R) refers to features developed after attainment of peak conditions. For the ultramafic melange in the amphibolite unit, prograde refers to amphibolite-grade assemblages. PS denotes features formed during early stages of subduction or on the seafloor. P-i refers to designations based on stable isotope fractionations between coexisting minerals (In some cases temperatures near petrologically inferred peak metamorphic temperatures are indicated. ). P-c indicates that the feature is regarded as prograde due to similarity in mineral composition of minerals in veins with those in host rocks. aRetrograde hydration of prograde minerals occurs along the vein.

4.1. Veining systematics veins, corresponds closely with host-rock composition (see Table 1). Within less siliceous Veins and their metasomatic envelopes are mafic and ultramafic lithologies, prograde common features of all units, typically com- veins rarely contain quartz and generally show prising 0.1 to 10% and, locally, greater than the clearest metasomatic envelopes. For ex- 50% of the outcrop. Coalescence of vein envel- ample, in the amphibolite-grade ultramafic opes in some areas results in pervasive bulk melange, envelopes of prograde veins and shear compositional change. Veins may be either zones in serpentinite zones show zonations prograde or retrograde and syn- or post-kine- from siliceous assemblages of talc _+ antho- matic (Table 1). Prograde veins commonly phyllite _+ actinolite nearest veins or block rims contain minerals diagnostic of the highest grade to anthophyllite_+ enstatite-bearing assem- of metamorphism preserved in the host rocks. blages and finally to dunitic compositions Retrograde veins commonly contain minerals (now serpentinized) farthest from vein cen- requiring lower-T ( _+ P) conditions than their ters (Bebout and Barton, 1989a). Figure 3a host. Approximately 35% of the veins lack di- shows schematic textural relationships be- agnostic mineral assemblages (e.g., veins con- tween mafic and ultramafic blocks and the sur- tain only calcite, quartz, albite ); in many cases, rounding amphibolite melange matrix; Fig. 3b these veins may be placed in sequence by cross- show the mineral zones that occur in veins and cutting relationships or stable isotopic temper- shear zones that cross-cut ultramafic blocks (cf. ature estimates (see textural observations in similar veining relations, in other areas, de- Table 1 ). scribed by Pfeifer, 1987; Carswell et al., 1974). Vein mineralogy, particularly of prograde In metasedimentary rocks, quartz-bearing METASOMATISM DURING SUBDUCTION: PRODUCTS AND POSSIBLE PATHS IN CATALINA SCHIST, CA 69

Fig. 2. Photographs illustrating significant metasomatic systematics in the Catalina Schis:L. (a) Photograph of typical exposure of blueschist-facies metasedimentary rocks, showing various vein generations discussed in the text (hammer for scale). (b) Photomicrograph of a blueschist-facies metagraywacke, showing cross-cutting of" quartz veins (light colored) by sodic amphibole + stilpnomelane vein (darker, through-going vein). The long dimension of the photomicrograph is approximately 5 cm. (c) Photograph of typical epidote-amphibolite-facies metamafic exposure, showing dense network of epidote + muscovite + albite veins and abundant albite porphyroblasts in host rocks (camera lens cap for scale ). veins rarely have obvious envelopes, although omineralic quartz, calcite, or albite veins or some Na-amphibole-bearing veins have quartz- contain quartz + calcite + albite, regardless of or albite-bearing and/or graphite-absent en- the host-rock bulk composition (e.g., Fig. 2a). velopes (see examples of veining relations in Textural relationships of veins are complex; blueschist-grade metasedimentary exposures in individual outcrops commonly show several Fig. 2a,b). Graphite-beating veins and peg- generations of veins (see Table 1 ). Many of the matites mainly occur in graphite-bearing me- mineralogically prograde and some of the min- tasedimentary rocks. Vein minerals commonly eralogically indeterminant veins are synkine- are similar in composition to the same min- matic, showing variable transposition into the erals in the host rock (see electron microprobe fabric (cf. Yardley, 1!)86; Fyson, 1987). In data in Bebout, 1989 ). Retrograde veins show some outcrops, veins with indeterminant min- less correspondence to host-rock composition eral assemblages are cross-cut by veins that than prograde veins and may or may not con- contain prograde mineral assemblages (e.g., tain diagnostic metamorphic minerals. Veins quartz veins in blueschist unit cross-cut by classified as "late-stage" based on textures sodic-amphibole+stilpnomelane veins; see (cross-cutting relations) are invariably mon- Fig. 2b), linking the formation of the indeter- 70 G E. BEBOUT AND M.D. BARTON

Serpentinite En vein

En + Anth Rosettes in (serp) Massive Anth (Tc, serp) IIIIl[llll[llllllActinolite-+AnthFringe (Yc, ser~)IIllll]~~]~

ChloriteL-_An~o&Ac tin~_+Tc Schist

En + Anth Rosettes in (serp) "~ ...... ~.~

Albite + ~ite + Graphite

I ransposed

......

~ Boudinaged Black Albite More Continuous ~ Veins with Lineated Black Albite Vein Quartz-Rich Veins --~ Extensional Quartz-rich Veins

~piidote up to l0 cm length

tigh-Na, K Metabasaltic Rock ]

Muscovite Porphyroblasts near Veins

Carbonate Breccia-Filling

Fig. 3. Sketches illustrating veining and other metasomatic systematics in the Catalina Schist. (a) Block-matrix relations in amphibolite-facies melange (scale variable; blocks range from lcm to several km in diameter). (b) Mineral zones ,in veins and shear zones cross-cutting ultramafic blocks in amphibolite melange (horizontal scale approximately 3 meters). Minerals and mineral assemblages in parentheses occur as overprinting retrograde parageneses. (c) Veining in blueschist- facies metasedimentary exposures (horizontal scale approximately 1 meter; see Fig. 2a) ). (d) Veining and other recta- somatic features in dominantly sedimentary melange zones in the blueschist unit (horizonlal scale approximately 0.5 meters). (e) Veining and other metasomatic features in dominantly ultramafic melange in the blueschist unit (scale variable: blocks range from 1 cm to 0.5 km in diameter). (f) Veining relations in exposures ofepidote-amphibolite-facies metamafic rocks (horizontal scale approximately 1 meter). minant veins to earlier parts of the subduction evidence of inward growth from the walls; some history. In some exposures, multiple genera- vein minerals topotaxially overgrow prograde tions of veins with the same mineralogy show host rock minerals. Crack-seal fabrics are com- cross-cutting relationships (e.g., multiple mon, as are other textures indicative of high pumpellyite + quartz + calcite vein sets in law- pore fluid pressures (e.g, veins which show fi- sonite-albite metasedimentary rocks). brous quartz microstructures perpendicular to Many prograde and retrograde veins show vein walls; see Etheridge et al., 1984; Fisher and METASOMATISM DURING SUBDUCTION: PRODUCTS AND POSSIBLE PATHS IN CATA.LINA SCHIST, CA 71

Byrne, 1990). Veins with open space provide matic exchange between blocks and enclosing the most compelling evidence for sustained melange matrix. Veins are variably transposed high pore fluid pressures. Such veins, which into the melange matrix fabric (see Fig. 3e). primarily occur in low-grade exposures, com- Higher-grade exposures also commonly con- monly contain 1 to 10 cm vugs with 1 to 10 tain evidence for complex, multiple metaso- mm euhedual crystals of albite or other min- matic events. Figure 3f illustrates vein rela- erals (e.g., Na-amphibole, lawsonite, pumpel- tions in a typical .exposure of epidote- lyite) grown inward into the veins; in some amphibolite-grade metamafic rocks (see pho- cases, the remaining vein space is filled with tograph of similar exposure in Fig. 2c). In ep- other minerals (e.g., late-crystallized calcite or idote-amphibolite-facies exposures, meta- quartz). mafic rocks are weakly foliated to unfoliated, Figures 3c, d, and e illustrate the veining re- and contain large epidote +_ quartz_+ albite lations in blueschist unit metasedimentary and veins and finer networks of epidote + al- melange matrix exposures. In many blueschist bite+white-mica+quartz veins (Fig. 3f). unit metasedimentary exposures (Figs. 2a,b, Breccia fillings of calcite _+ epidote + white- and 3c), abundant quartz_+ calcite + Na-am- mica are common in ~Lhese exposures; C iso- phibole veins are preserved in psammitic bou- tope compositions suggest that these features dins; the surrounding pelitic material contains resulted from low-T seafloor alteration and not only younger, cross-cutting quartz + calcite + high-P fluid flow (see discussions below). pumpellyite veins and transposed al- Highest-grade, amphibolite-facies metasedi- bite _+ graphite (black albite) veins. In some mentary and metamafic exposures contain areas, metagraywacke layers contain abundant abundant concordant leucosomes and discor- discordant Na-amphibole _+ albite _+ quartz dant pegmatites of quartz + feldspar + mus- veins. Figures 3d and e illustrate veining rela- covite + biotite + hornblende (see Sorensen tions in blueschist-grade melange, Fig. 3d in and Barton, 1987; Bebout and Barton, 1988). relatively sediment-dominated parts of the In such exposures, other high-variance layers melange (those interleaved with metasedi- (e.g., quartz-rich layers in metasedimentary mentary material and containing metasedi- exposures; clinozoisite- or hornblende-rich mentary blocks), and Fig. 3e in more ultra- layers and pods in metamafic exposures) may mafic melange domains. Exposures of represent transposed, earlier vein sets. sediment-dominated melange are relatively Although prograde features are typically micaceous or albitic; they contain variably penetratively deformed, the prograde veins in transposed black albite and quartz-rich veins; some exposures are post-kinematic. Variations black albite veins commonly show develop- in the extent of deformation of prograde veins ment of mineralized extensional fractures may in some cases be explained by the hetero- (containing quartz _+ calcite). Some quartz + geneous nature of the cleformation (i.e., vary- calcite veins cross-cut the transposed and ing between discrete domains). In blueschist- veined black albite seams and continue into the facies metasedimentary exposures (see Figs. surrounding matrix. The more ultramafic ex- 2a,b, and 3c ), less penetratively deformed me- posures ofblueschist melange matrix (Mg- and tagraywacke layers typically contain unde- Cr-rich; Fig. 3e) contain high-variance assem- formed veins linked to prograde metamor- blages rich in talc and chlorite, but also fuch- phism (e.g., quartz + calcite + lawsonite or Na- site, Na-amphibole, and Na-Ca-amphibole. amphibole _+ albite + quartz + lawsonite veins). Less altered metamafic and metaultramafic All but the texturally latest veins are deformed rocks occur as blocks in the melange with var- to varying extents in surrounding metapelitic iably developed rinds representing metaso- layers (e.g., transposed albite + graphite veins 72 G.E. BEBOUT AND M.D. BARTON cross-cut by late-stage quartz _+ calcite veins; see Barton, 1987). All units contain late-stage, Figs. 2a and 3c). Prograde veins may show re- pumpellyite-bearing veins (see Table 1 ). trograde features such as replacement of prograde vein minerals by retrograde assem- 4.2. Evidence for whole-rock major and trace blages; however, retrograde veins are usually element metasomatism texturally distinct from prograde features. Re- trograde veins are rarely deformed, tending to be planar structures, cross-cutting maximum- Many rocks in the Catalina Schist have ma- grade fabrics and prograde veins. Retrograde jor- and trace-element bulk compositions that veins include serpentine + talc + magnesite are different from likely protolith composi- veins which cross-cut prograde parageneses in tions. Textures, diagnostic mineral assem- the ultramafic melange and pumpellyite-bear- blages, and isotopic compositions require that ing veins ( +_ quartz + albite _+ calcite ) in meta- some of these changes took place at high P~ T. mafic and metasedimentary rocks in all units. Locally, evidence points to earlier alteration, Many exposures contain veins along which probably on the seafloor or at shallower levels prograde host-rock minerals show retrograde of the accretionary complex. Volatiles and hydration/replacement (e.g., hornblende --, some relatively "fluid-rnobile" trace elements chlorite+actinolite along veins in epidote- (e.g., N, B, Cs) decrease in concentration with amphibolite metamafic rocks). increasing metamorphic grade, particularly in As a further indication of the heterogeneous the metasedimentary rocks (see Bebout and nature of the prograde deformation, prograde Fogel, 1992; Bebout et al., 1993a,b); these de- veins in mafic lithologies of up to epidote-am- creases reflect loss of these elements during phibolite grade (see Figs. 2c and 3f) tend to progressive devolatilization of the rocks. be developed in fractures which cross-cut Metabasaltic rocks in all units show evi- weakly foliated to unfoliated hosts, suggesting dence for metasomatism, although much of the brittle behavior of the rocks during prograde metasomatic alteration may be related to seaf- metamorphism. Nearby exposures of metase- loor processes (cf. studies of Shuksan Schist dimentary rocks and melange matrix of the high-PIT metamafic rocks by Dungan et al., same grades are strongly foliated, indicating 1983; Owen, 1989). Many metabasaltic rocks their ductile deformation. In these exposures, in each unit are enriched in Na, LREE and all but the latest-stage veins are variably trans- LILE, such as K, Ba, Rb and Cs, relative to posed into the strongly developed foliation of MORB protoliths (Figs. 4a and b; e.g., epi- the rocks. dote-amphibolite rocks with > 2 wt. % K20 Widespread retrograde hydration _ carbon- and glaucophanic greenschist rocks with > 6 ation overprints the high-P~ T mineral assem- wt. % Na20). In epidote-amphibolite meta- blages of all units (see Table 1 ). In the amphi- mafic exposures, pervasive addition of Na and bolite-grade ultramafic melange, prograde K relative to A1 and Ti is petrographically in- assemblages are in places pervasively replaced dicated by abundant albite, muscovite and by serpentine + talc _ magnesite assemblages. biotite porphyroblasts which occur in areas In metamafic and metasedimentary rocks, with abundant discordant epidote+albite prograde assemblages are locally replaced by + quartz + muscovite + biotite veins (see Figs. chlorite, pumpellyite, lawsonite, or white-mica; 2c and 3f). Metamafic blocks in the domi- in the amphibolite unit, oligoclase in nearly all nantly ultramafic melange in the amphibolite plagioclase-bearing lithologies is replaced by unit are variably metasomatized as a result of varying proportions of albite, phengite, and varying degrees of fluid-rock interaction and zoisite + quartz symplectite (see Sorensen and block-matrix exchange (see Sorensen and METASOMATISMDURING SUBDUCTION: PRODUCTS AND POSSIBLEPATHS IN CATALINASCHIST, CA 73

13 1000

AheIe,l Basallq / horn v ,'u.iou s sludies ] . , • , . , • , . , . , . , • j • .1 4 5 6 7 8 9 10 11 12 3 B Cs RbBa K Ta LaCe Sr NdZrSmTi Y Yb Weight % MgO

3000 ' , • , ' , ' I - 6 , Plagioclase -> Phengite • ~._. 5 - Hornblende -> Glaucophane

4 2ooo

2 ~[~ lOOO'

1 i i i L o o 1 2 3 4 5 0 0 1 2 3 4 (wt. %) (wt. %)

Fig. 4. Evidence for major and trace element metasomatism of metamafic rocks of the Catalina Schist. (a) Plot of CaO vs. MgO, demonstrating variability in the compositions of metabasaltic rocks of the Catalina Schist and the similarity of these compositions to those of many seafloor-altered metabasaltic rocks (data from Sorensen, 1984; this study). The ranges for unaltered MORB are from Humphris and Thompson ( 1978; labelled "'Fresh Basalts" ) and Basaltic Volcanism Study Project ( 1981 ; in box). Ranges of altered MORB are from Humphris and Thompson ( 1978; area with pattern) and other studies referenced in Humphris and Thompson (1978; outlined range without pattern). (b) Element abundances in metabasaltic rocks of the Catalina Schist normalized to MORB (values for MORB from Hofmann, 1988), illustrating enrichments of LILE in metamafic rocks relative to MORB. (c) Co-enrichments of K20, Cs, and Ba in igneous cobbles in blueschist metaconglomerate. These enrichments accompany replacement of feldspars by phengitic white mica and replacement of hornblende by glaucophane.

Barton, 1987; Sorensen, 1988; Sorensen and change with seawater, perhaps at the mid-ocean Grossman, 1989; see Fig. 3a). ridge (Barton et al., 1987; see also Hart et al., Identification of high-P alteration in meta- 1974; Nelson, 1991 ). Some metamafic rocks mafic rocks, particularly in the absence of ob- of the Catalina Schist show enrichments in vious high-PIT vein envelopes and metaso- MgO and depletions in CaO consistent with matic rinds, is complicated by uncertainties relatively high-T alteration on the seafloor regarding the nature and extent of presubduc- (Fig. 4a; cf. Humphris and Thompson, 1978; ,ion alteration. Enrichments in trace elements Seyfried et al., 1978; Alt et al., 1986; Erzinger, similar to those in Catalina mafic rocks (par- 1989); however, many of the compositional ticularly in LILE, relative to MORB protolith systematics are consistent with lower-T sea- compositions; Fig. 4b) have been reported as floor alteration (e.g., coupled decrease in CaO the result of hydrothermal alteration on the and MgO contents, enrichments in LILE and seafloor (Hart et al., 1974; Frey et al., 1974; B; see Figs. 4a and b; cf: Donnelly et al., 1980; Hart, 1976; Ludden and Thompson, 1979). Alt et al., 1986; Erzinger, 1989 ). Stable isotope Shifts to more radiogenic 87Sr/86Srinitial in me- data for some highly metasomatized meta- tamafic rocks of all units could represent ex- mafic exposures (see Figs. 2c and 3f) indicate 74 G.E. BEBOUT ,~ND M.D. BARTON local pervasive reequilibrati0n of such expo- transposed into the foliation of the rocks and sures with the isotopically distinct high-P/T do not have obvious envelopes. Metasedimen- fluids (Bebout and Barton, 1989a; see discus- tary whole-rock analyses do show significant sion below). That evidence for exchange with decreases in the concentrations of some trace the high-P/T metamorphic fluids is consistent elements (e.g., B, Cs, N) with increasing met- with some related major and trace element amorphic grade. These variations accompany modification during infiltration. Metasoma- decreases in H20 content and have been at- tism related to the formation of rinds on me- tributed to removal by H20-rich fluids during tamafic blocks in melange of all grades (see progressive devolatilization (Bebout and Fo- Sorensen and Barton, 1987 for blocks in am- gel, 1992; Bebout et al., ][993a,b). phibolite melange; Figs. 3a and e) is more Many of the whole-rock compositional clearly related to high-P/T metasomatism. changes linked to high-/Y/T metasomatism in Blueschist-facies, dioritic and gabbroic me- metamafic and metasedimentary rocks do not taconglomerate clasts (igneous affinities require mass transfer from beyond the bound- judged using textures and the proportions of aries of their lithologic unit and thus do not re- felsic and mafic phases) show evidence of Na- quire voluminous fluid l]ow. Melange matrix, K metasomatism (e.g., intermediate An pla- particularly in the amphibolite-facies unit, is gioclase--,phengite + glaucophane; inagnesio- an exception. A chemical mass balance can be hornblende--,glaucophane; Tenore-Nortrup done for the melange matrix in the amphibol- and Bebout, 1993 ). Many of these cobbles have ite unit. Using XRF and INAA whole-rock high concentrations of LILE compared to likely major/trace element data and the abundances igneous protoliths (up to 1415 ppm Ba, 101 of the various rock types (see map of litholo- ppm Rb, 5.5 ppm Cs, and 3.9 wt.% K20), and gies in Fig. 5a), it appears that the ultramafic show co-enrichments of K20, Cs, Ba, Rb, and melange originated by mixing of ultramafic and NH4+ which correlate with degrees of metaso- mafic rocks+small amounts of sedimentary matic alteration deduced petrographically (see material, accompanied by massive addition of Fig. 4c). Textural and mineralogical observa- Si by infiltrating fluids (Bebout, 1989; Bebout tions regarding metasomatism of metacon- and Barton, 1989a). Some compositional glomerate cobbles were also reported by Fyfe variations in the melange (e.g., A1, Cr) can be and Zardini (1967) and Moore et al. (1981). explained by mechanical mixing (i.e., linear Moore et al. ( 1981 ) reported enrichments in combinations of mafic and ultramafic rocks alkali elements in some cobbles in metacon- can explain these variations; see Fig. 5b). As- glomerates of the Franciscan Complex, suming that A1 and Cr were immobile, a mix- California. ture of metamafic blocks (wt. % SIO2=50 % Pervasive compositional alteration in meta- on anhydrous basis) with dunitic protolith sedimentary rocks is yet more difficult to doc- ( ~ 40 wt. % SiO2) cannot explain the Si con- ument because of highly variable protolith tent of the melange matrix (present average wt. compositions (see discussion by Bebout and % SIO2~54%), necessiitating an external Si Fogel, 1992 ) and extensive mechanical mixing source. Figure 5c shows this trend toward in- (+ local chemical redistribution) accompa- creasing Si content inferred as the result of fluid nying ductile deformation. The metasedimen- infiltration and metasomatism. This conclu- tary rocks do not show regular trends in con- sion is consistent with Si-addition to perido- stituents normalized to "relatively immobile" rite along veins and shear zones which cross- constituents (e.g., ratios of Si, K, and Na to AI cut ultramafic blocks and at margins of blocks or Ti. Bebout et al., 1993b). Particularly in with melange matrix (see Figs. 3a and b). higher-grade exposures, most early veins are Large parts of the melange mineralogically and METASOMATISM DURING SUBDUCTION: PRODUCTS AND POSSIBLE PATHS IN CATALIN4 SCHIST. ('A 75

4000 25 % eo i.i,es Ib Eclogite 1(2 Mafic 2O *(T'N.x ~ f~ 30OO ge Mela2ge Matrix 5 .-~ 15 ~-~ 20O0 < k~ "•~*~.j AverageofPresent "x Melange ~."~ MatrixCompositions ds on ~'~ 10 1~ "\ ~ Eclogites Matrix ~~ ] I ,,...-/ ~ ~"TSedimen tary ~'Ma.... I000 ] " ~'~ e;~. • Rocks ~" lvlaI~lCNOCKS 5 /hi " Ultramafic ~ ~ .1 0 .ocks • 0 10 20 30 40 50 60 70 Wt. % A1203 Wt. % SiO 2

Fig. 5. Evidence for compositional evolution of the amphibolite-facies melange. (a) Map of amphibolite-facies melange unit. illustrating the textures and proportions of contrasting compositional types. 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. 3b); material mapped as "Melange Matrix" is relatively aluminous schist (chlorite-rich) with varying metamafic component. (b) Plot of A1203 vs. Cr. illustrating regular mixing relations for melange in amphibolite unit. Samples labelled "Siliceous Ultramafic Rocks" are believed to have had large amounts of Si added by fluids. (c) Plot of A1203 vs. SiO2, illustrating the need for massive addition of Si to the melange in the amphibolite unit (see arrow between inferred composition due purely to rnechanical mixing, labelled "'Initial") and present average composition of the melange matrix, based on whole-rock data and mapping. Within the field for melange, high-Al203, low-SiO2 samples are chlorite-rich schists [mapped as "Melange matrix" on (a)]; low- AI~O3, high-SiO2 samples are talc- or actinolite-rich schists and hornfelses [mapped as "Siliceous ultramafic rocks ~' on (a)]. 76 G.E. BEBOUT AND M.D. BARTON compositionally resemble metasomatic assem- side the melange is not required; see Bebout, blages developed at the rims of mafic and ul- 1989). Pervasive retrograde metamorphism tramafic blocks in the melange (e.g., nearly (hydration to produce serpentinite; see areas monomineralic actinolite and hornblende do- mapped as serpentinite on Fig. 5a) apparently mains), suggesting complex evolution of the had little effect on the amphibolite unit ultra- melange through combined mechanical and mafic bulk compositions other than the addi- metasomatic processes. Significant mechani- tion of H20, although somewhat lowered Mg/ cal addition of high-Si materials such as sedi- Si ratios of serpentinites suggest some metaso- ment (including chert) is discounted based on matic Si addition, possibly during serpentini- field evidence. Using an estimated average zation (data in Sorensen, 1984; Bebout, 1989 ). composition for the melange matrix, mechan- Although major and trace element data do ical mixing models suggest an approximately not exist for melange matrix in the blueschist 2:1 molecular proportion ofultramafic to mafic and greenschist units, all melange shows abun- rocks for the melange matrix as a whole. Other dant evidence for pervasive chemical altera- compositional variations suggest either addition during fluid-rock interactions. The tion from external reservoirs (e.g., Si, O- and blueschist melange (see Figs. 3d and 3e) con- H-isotope systematics) or redistribution within tains abundant layers (up to several meters in the melange (e.g., Ca, Na, K, rriany trace thickness) with high-variance assemblages elements). (e.g., nearly monomineralic Na-amphibole and In addition to Si, concentrations of Ca, Na, talc layers ). Greenschist facies melange locally and K (and many trace elements; e.g., Ba) in contains meter-scale domains of nearly pure the melange matrix cannot be explained by actinolite; some individual actinolite crystals mechanical mixing and may reflect metaso- are ,-- 20 cm in length. matic redistribution. Enrichments of Ca in Veins are typically enriched in many of the metasomatic zones at rims of ultramafic blocks elements enriched in metasomatized meta- (i.e., in actinolite-rich rind assemblages; see mafic and metaultramafic rocks and melange Figs. 3a and b) and Na+K enrichments in matrix, particularly the., LILE, B, and N, but rinds on mafic blocks (cf., Sorensen, 1988; see also in rarer cases, the LREE. Calcite veins in Fig. 3a) also suggest that these elements were low-grade exposures contain up to 3000 ppm mobile in the metamorphic fluids. Trace ele- Sr, and other veins containing silicate assem- ment and Nd isotope data for variably meta- blages are commonly enriched in other LILE. somatized mafic blocks in the melange suggest One chlorite+actinolite vein in a lawsonite- addition of a sedimentary component to the albite metamafic exposure contains an "allan- blocks (Sorensen and Grossman, 1989; Bar- ire-like" phase with approximately 15 weight ton et al., 1987). The proposed addition of % (LREE)203 (Bebout, 1989; based on elec- trace elements to the blocks is compatible with tron-microprobe analyses). The occurrence of the stable isotope evidence for massive influx trace-element enriched allanite in rinds on me- of fluid equilibrated with lower-T, metasedi- tasomatized metamafic blocks in amphibolite mentary rocks (see Barton et al., 1987; Bebout melange has been documented by Sorensen and and Barton, 1989a; Bebout, 1991a; discussion Grossman ( 1989 ). below). However, the trace element systematics in the amphibolite melange matrix indicate 4.3. Summary of stable isotope evidence for that combined mechanical and metasomatic fluid-rock interactions processes may have resulted in mobilization of many of these trace elements from within the Stable isotope systematics indicate exten- melange unit (i.e., addition by fluids from out- sive fluid-rock exchange and large-scale fluid METASOMATISM DURING SUBDUCTION: PRODUCTS AND POSSIBLE PATHS IN CATALINA SCHIST, CA 77 flow from lower-T, sediment-rich parts of the mentary rocks like those in the Catalina Schist accretionary complex into higher-T, domi- over a temperature range of ~ 350 to 600°C nantly mafic/ultramafic parts of the complex. (see Fig. 2 in Bebout, 1991a). Hydrogen-iso- In the following summary, ~180 and dD are re- tope compositions are consistent with the ox- ported relative to SMOW, ~13C relative to ygen isotope systematics, indicating that pro- PDB, and ~15N relative to atmospheric N. grade veins and other metasomatic features in Calculations of fluid ~180 using petrologi- all units equilibrated with H20-rich fluid with cally and, in some cases, isotopically inferred similar dD of - 15 + 15°o0 (Barton et al., 1987; temperatures indicate that prograde veins and Bebout and Barton, 1989a; see Fig. lb in Be- melange matrix equilibrated with H20-rich bout, 1991a). This range in dD suggests an fluids with ~180 of + 13+ 1%o (Fig. 6). Host- evolved seawater source either from isotopi- rocks in more coherent exposures show vary- cally modified pore waters or through more ing shifts in ~80 toward equilibrium with vein complex hydration/dehydration processes (see minerals and melange matrix (Fig. 6). Calcu- Barton et al., 1987; Bebout and Barton, 1989a; lated fluid ~ t 8O approximates O-isotopic equi- Bebout, 1991 b; cf. Magaritz and Taylor, 1976 ). librium with metasedimentary rocks in the Ankeritic dolomite in the amphibolite-facies lower-grade units, consistent with the deriva- ultramafic melange has ~13C of -- 10 to --9%o tion of these fluids in the sediment-rich, lower- consistent with equilibrium at 650-750°C with T parts of the complex (Barton et al., 1987; CO2 with ~ ~3C of - 8 to - 7%0 similar to cal- Bebout and Barton, 1989a). The calculated culated CO2 values for calcite-bearing veins in fluid fi~80 range of + 13 + 1%o is consistent the blueschist unit. This similarity in calcu- with the equilibration of fluids with metasedi- lated fluid C isotope composition is consistent

I I I:::::::::::::::::::::::::::::::::::::::::::::::::::::::::: I I _I n = 60, mean = 13.1%o Arnphibolite

unit (650 °C) less altered altered & unrragrnatizedmeta- matrix & mafic migmatized sedimentaryrocks rocks rocks Greenschist unit (525 °C) meta,;edimentary rocks away from veins:& melange Blueschist @ _1 unit (400 °C) n - 70, mean= 12.9%~ i'i'i'i'i'i?i'i'i'i'?i'i?i'i'i'i'i? i'ii i', i':i i I I I 6 8 10 12 14 16 18 6'80s,ow Mineral and Mafic ~ Pegmatitic and migmatitic O whole rock: (vein & host-rock) Ultramafic O Metasedimentary Fig. 6. Calculated H20 d~80, using 8t80 data for various metasomatic features and host-rocks and petrologically and/or isotopically inferred peak metamorphic temperatures. Mineral-fluid fractionation data used in these calculations are from references given in O'Neil (1986) and Clayton and Kieffer ( 1992 ). In all units, 3~80 ,~alues are the most consistent with exchange involving H20-rich fluids with 3~8OsMow of + 13 + 1%0 (shaded region ) in melange zones (both melange matrix altered mafic and ultramafic blocks in melange) and in/near prograde veins (see data for blueschist unit in Be- bout, 1991b). In the blueschist and greenschist units, some siliceous schists (metacherts) have c~80 of >20%0 (see Bebout, 199 l a). For the amphibolite unit, data are subdivided as representing "less altered matrix and mafic rocks", "'altered and migmatized rocks" (see statistical data for these samples), or "unmigmatized metasedimentary rocks". 7~ (i.E. BEB()UT AND M.D, BARTON with the large-scale fluid communication in- low-grade metasedimentary exposures, cal- ferred from the O and H data. cite-bearing veins petrographically related to Evidence exists locally for the retention of high-P fluid flow have ,413C of - 13 to -6%0 stable isotope signatures reflecting either seaf- consistent with equilibrium at peak metamor- loor processes or diagenetic processes at shal- phic temperatures with the carbonaceous mat- lower depths in the subduction zone. The ~ 180 ter, and ~ 1s O values of -t- 14.5 to + 17%0, in O- values of vein calcite in metabasaltic rocks of isotopic equilibrium wilLh the + 13 + 10/00 fluid up to epidote-amphibolite grade (+ 13 to reservoir (see Bebout, 1991b). The ~3C val- +17%o) indicates equilibrium with the ues of most of the more continuous calcite- + 13 _+ 10/00aqueous fluids; however, their c~~ 3C bearing veins in mafic lawsonite-albite and values (-7 to +2.5%0) show a similar range blueschist exposures indicate that they were to those of veins in unsubducted oceanic crust produced by fluids in C-isotope equilibrium ( Stakes and O'Neil, 1982; Alt et al., 1986; data with organic matter in the metasedimentary in Bebout and Barton, 1989b) and could, in rocks. However, as discussed above, less con- part, be explained by retention of seafloor val- tinuous veins and breccia fillings in these me- ues. Similarly, isotopic compositions of finely tamafic exposures have higher c~13C values disseminated carbonate in some low-grade which are similar to those of veins in seafloor metasedimentary rocks (~ 13C = - 12 to - 8%o; altered basaltic rocks. Nitrogen-isotope sys- ~O= + 13.5 to +28%0; Bebout and Barton, tematics likewise indicate a sediment-sourced 1989b; Bebout, 1991 b) resemble those of car- fluid signature (Bebout, 1992). In each unit, bonate cements in lower-grade, shallower parts metasomatized metamafic and metaultra- of accretionary complexes (Vrolijk, 1987), mafic rocks and melange matrices have ~15N suggesting earlier deposition of the finely dis- similar to that of the melLasedimentary rocks of seminated carbonate as diagenetic cements that unit (including all units, range is ~ + 1 to ( Bebout, 199 lb). + 6%0; Bebout and Fogel, 1992 ). Carbonaceous matter in metasedimentary In all units, stable isotopic compositions (for rocks (up to 1.7 wt. % in all units) ranges from O, H, and N) are most uniform in veins and poorly crystalline ( < 1/tin size) in the lowest- melange zones (i.e., relative to more coherent grade rocks to well-ordered graphite ( ~ 1 mm) metamafic and metasedimentary exposures; in the amphibolite unit (determined by XRD). Bebout and Barton, 1989a; Bebout, 1991b: Be- The d~3C (-27 to -24O/o0; Bebout and Bar- bout, 1992; see Fig. 6). In the amphibolite-fa- ton, 1989b), grain size, and poor crystallinity cies melange (see Figs. 3a,b and 5a), ~O of of the carbonaceous matter in the lowest-grade individual minerals is uniform over distances rocks require organic origin. The ~13C values of several kilometers (see map in Fig. 5a; Fig. of the carbonaceous matter increase with in- 6; Barton et al., 198"7). Hydrogen-isotope creasing metamorphic grade to values of -21 compositions in the melange are likewise rela- to -19%o in the amphibolite unit. Leuco- tively uniform over large distances; fifteen cli- somes and pegmatites in the amphibolite unit noamphiboles from throughout the ultramafic derived through vapor-saturated (high aH2o) melange matrix and from metasomatic rinds partial melting of mafic and sedimentary rocks on mafic blocks in the amphibolite melange (Sorensen and Barton, 1987; Bebout and Bar- (Barton et al., 1987; Bebout, 1991a) have ton, 1988 ) contain graphite with ~3C of - 25 mean oq) of -410/00 with a standard deviation to - 190/0o interleaved with muscovite. ~ ~3C of ( l a) of 6%. In lawsonile-albite- and bluesch- albite + graphite veins in the blueschist unit is ist-facies exposures, melange domains and similar to that of graphite in enclosing meta- veins are more uniform in calcite ~80 and in sedimentary rocks ( ~ -250/o0; see Table 1 ). In ~SN than nearby coherent metasedimentary METASOMATISM DURING SUBDUCTION: PRODUCTS AND POSSIBLE PATHS IN CATALINA SCHIST, CA 79 rocks (Bebout, 1991 b, 1992 ). Tenore-Nortrup ceous veins cut silica-undersaturated host rocks and Bebout (1993) and Bebout (1992) re- (mafic and ultramafic). In the amphibolite ported relatively uniform 61sO, 013C, and 0 ~5N unit, abundant evidence exists for the mobility for metasomatized igneous cobbles in tabular of elements in felsic silicate liquids released by metaconglomerate bodies in blueschist me- partial melting. lange zones. The calcite O isotopic composi- Melange zones are relatively homogeneous tions and whole-rock N isotopic compositions isotopically over distances of up to several kil- of the cobbles (data summarized in Tenore- ometers (as in the case of the amphibolite me- Nortrup and Bebout, 1993) are indistinguish- lange; see Figs. 5a and 6 ). In contrast, the more able from those of veins and melange matrix coherent metasedimentary and metamafic ex- collected from throughout the blueschist unit posures and some tectonic blocks in melange in metasedimentary, metaultramafic, and me- zones (see Figs. 3a and e) may retain hetero- lange exposures. This uniformity in isotopic geneous isotopic signatures reflecting either compositions indicates possible equilibration seafloor processes or earlier stages of subduc- of rocks in veins and melange zones with fluids tion-zone metamorphism (i.e., at shallower of similar isotopic compositions over dis- depths ). Within melange zones, abundant evi- tances of up to kilometers. dence exists for element mobility, particularly of Si and the alkali elements. The mass balance 5. Synthesis of metasomatic alteration in the relations in the amphibolite melange demon- catalina schist strate the extreme modification of rock compositions through coeval mechanical mixing The metasomatic alteration in the Catalina and metasomatic addition/subtraction; the Schist may be characterized as falling into sev- lower-grade melange units are believed to have eral major types. For example, blueschist-fa- formed similarly based on the prominence of cies rocks contain (1) siliceous (typically spatially extensive high-variance assemblages. quartz-rich) assemblages, (2)potassic (white- Alkalic mineral assemblages are, in general, mica_+quartz-rich) assemblages, and (3) more prominent in metasomatized low-grade sodic, silica-undersaturated (albite/Na-am- rocks (e.g., albite- and Na-amphibole-rich phibole-rich, quartz-poor) assemblages. The veins and melange layers ); in high-grade units, higher-grade units contain analogous evidence mass balance relation,; demonstrate the re- for contrasting metasomatism. Individual ex- moval of alkalis, but silica is in some cases dra- posures at all grades may show grossly con- matically enriched (e.g., amphibolite melange trasting metasomatic assemblages with ob- matrix). vious cross-cutting relationships (see examples Stable isotope data fbr the Catalina Schist for blueschist unit in Table 1; Fig. 2b). In gen- indicate that fluid flow occurred at the scale of eral, the diversity of vein types diminishes at kilometers and likely facilitated large-scale higher grades, as do the critical textural rela- chemical redistribution, particularly along me- tionships that allow the veins to be placed into lange zones and fractures. Stable isotope data a relative temporal sequence. In the high-grade (particularly for O) indicate a relatively low- units, veins are typically transposed into the T, metasedimentary source for the fluids which Jbliation of the host rocks; many veins which metasomatized higher-T parts of the Catalina precipitated at earlier stages of metamorphism Schist. Some fluid-mobile elements were ap- have presumably undergone further metaso- parently removed and mobilized during this matic exchange with the host rocks during con- devolatilization. For example, B, Cs, and N are tinued prograde metamorphism. Metasomatic enriched in many metamafic rocks and veins; vein envelopes are best-developed where sili- in melange zones, there is evidence for the re- 80 G.E. BEBOUT AND M.D. BARTON distribution of many major and trace ele- nate (see Bailey et al., 1964; Ernst et al., 1970), ments, including alkali elements (Na, K), Ca, as they do in the Catalina Schist (Table 1 ), but and trace elements such as the LILE (Sr, Ba, other mineral associations can be significant. Cs). Petrologic, stable isotopic, and fluid in- In particular, these include sodic assemblages clusion evidence (heating/freezing and quad- (commonly quartz-free) consisting of jad- rupole mass spectrometry; see Sorensen and eite_+ blue amphibole + albite found in the Barton, 1987; Bebout, 1991a) indicates that Franciscan Complex (,e.g., Moore, 1984; Ma- the high-PIT metamorphic fluids were H20- ruyama and Liou, 1987, 1988; this study) and rich, low-salinity (~ 1 to 2 equivalent wt. % in Turkish subduction complexes (Okay, NaCI; Sorensen and Barton, 1987; Bebout, 1982), and also calcic assemblages containing 1989 ), C-O-H-S-N fluids. pumpellyite and lawsonite found in the Fran- ciscan Complex (this study; Ernst et al., 1970; 6. Comparison with evidence for metasomatism Nelson, 1991 ). In eclogitic rocks (from the in other subduction-related metamorphic Austrian Alps) that represent metamorphism complexes at 60 to 70 km depths, Philippot and Selver- stone ( 1991 ) and Selw:rstone et al. ( 1992 ) re- Many of the metasomatic features described ported both discontinuous segregations (of for the Catalina Schist have been reported for quartz, dolomite or magnesite, omphacite, other occurrences of subduction-related rocks. kyanite, zoisite/clinozoisite, phengite, rutile, These features include abundant veining, bulk- and apatite) and more continuous vein sets rock metasomatic changes, and large-scale iso- (vein minerals include albite, chlorite, carbon- topic homogenization. Their fluid-related ori- ate, phengite, amphibole, and quartz). They gin is corroborated by expulsion of 'metamor- interpreted the contrasting textures and min- phic' fluids and heat anomalies in modern eralogy of the two vein types as representing a accretionary prisms and by theoretical consid- record of both eclogite-facies fluid processes erations of volatile mass balance (see Lang- (represented by segregations) and infiltrative seth and Moore, 1990; Ito et al., 1983; Pea- fluid transport during uplift of the rocks (rep- cock, 1990; Moran et al., 1992). Meta- resented by more continuous veins). somatism of this magnitude is also implicit in Shifts in the bulk compositions of high-P/T current models for arc magmatism; these metamorphic rocks are well-documented for a models call upon the metasomatic removal of few areas. The best known changes are those components from subducted rocks and the ad- associated with the formation of rinds on mafic dition of these components to arc source re- blocks in melange (see Moore, 1984; Cloos, gions in the mantle wedge (e.g., Gill, 1981; 1986; Sorensen, 1988; this study). Most inves- Tatsumi et al., 1986; Morris et al., 1990). tigators interpret these rind assemblages to re- Mineralogically diverse veins are wide- sult from relatively local chemical exchange. spread in accretionary prisms and have been More pervasive systematic chemical changes described in many paleosubduction com- have been described for only a few areas. In the plexes. At the shallow levels of accretionary Franciscan Complex, sodic-potassic altera- prisms, veins consist primarily of calcite, tion of feldspathic lithologies (medium to quartz, and lesser amounts of materials such as coarse-grained sedimentary rocks and clasts) reduced carbon and sulfur-bearing substances has been proposed by several groups (Essene (e.g., Vrolijk, 1987; Vrolijk et al., 1988; Fisher and Fyfe, 1967; Fyfe and Zardini, 1967; Moore and Byrne, 1990). In rocks metamorphosed at et al., 1981; Maruyama and Liou, 1987, 1988; greater depths ( > 15 km), quartz and carbon- for the Catalina Schist, Tenore-Nortrup and ate (calcite/aragonite) veins typically domi- Bebout, 1993). Moore et al. ( 1981 ) docu- METASOMATISM DURING SUBDUCTION: PRODUCTS AND POSSIBLE PATHS IN CATALINA SCHIST, CA 8 1 mented that the compositions of granodioritic tances. Sparse evidence for radiogenic isotopic and granitic clasts shift from plausible original homogenization is consistent with extensive igneous compositions to highly sodic compo- fluid flow (Barton et al., 1987; Nelson, 1991 ); sitions along compositional trajectories con- however, the scale of such exchange has yet to sistent with molecular exchange of Na for K be convincingly demonstrated. (Fig. 7). Sorensen (1988) suggested that Na In summary, field geochemical evidence and Si deficiencies in mafic blocks from sev- from a number of circum-Pacific and Tethyan eral western-US subduction complexes re- subduction complexes demonstrates that fluid- sulted from high-pressure fluid-rock mediated mass transfe, r can be significant. interactions. Transfer of carbonate, silica, and stable iso- Considerable evidence for the scale and ex- topes appears to be widespread in many com- tent of fluid-mediated isotopic exchange dur- plexes (this study; Taylor and Coleman, 1968; ing subduction has been presented for north- Magaritz and Taylor, 1976; Rumble and Spear, ern Pacific accretionary prisms (Vrolijk, 1983; Moore, 1984; Matthews and Schliestedt, 1987), the Franciscan Complex (Taylor and 1984). Finally, sodic vein and alteration as- Coleman, 1968; Magaritz and Taylor, 1976; semblages are common in many high-P/T Rumble and Spear, 1983; this study), the metamorphic complexe, s (this study; Essene Mediterranean (Matthews and Schliestedt, and Fyfe, 1967; Fyfe and Zardini, 1967; Okay, 1984), and Norway (Agrinier et al., 1985). 1982; Moore, 1984; Maruyama and Liou, Each of these areas shows evidence of isotopic 1987, 1988; Nelson, 1991; Philippot and Sel- exchange at scales too large to be caused purely verstone, 1991 ). by diffusional processes. Based on the stable isotope systematics and other evidence for in- 7. Interpretation of the evidence for contrasting filtrational mass transport (veins, etc.), each metasomatism of these areas is interpreted to have experienced substantial fluid flow over large dis- In this section, we consider the potential im- pact of large-scale fluid flow on chemical evo- 7 I I I I I I lution in subduction environments. Although [] observed '~K~+%z the evidence locally for metasomatism from 6- composmons ~ Y~ field observations and chemical and isotopic data is compelling, it is difficult to quantify the ~5~4[]~~e contributions from various mechanisms of mass transfer (e.g., local-scale diffusional exchange vs. larger-scale,, fluid-mediated ex- (~3- change). It is appropriate to consider possible Z [] original~te 2 - compositions - systematics of infiltration metasomatism because of the abundant evidence for large-scale Compositional shifts in [ 1 - Franciscan clast compositions - fluid transport in the Catalina Schist (e.g., the (Moore et al., 1981, ChemicalGeology,) stable isotope systematics). 0 I I I I i I 0 1 2 3 4 5 6 7 Three processes may contribute to changes K20 (wt %) in bulk composition during subduction: diffusional processes, mechanical mixing, and infil- Fig. 7. Compositional shifts in granitoid clast composi- tration metasomatism. Diffusion of chemical tions from the Franciscan complex, illustrating sodic al- species can be driven by chemical gradients, as teration due to Na exchange for K in feldspars (data from Moore et al., 1981 ). The double arrow indicates the slope between contrasting lithologies. Mass transfer of the 1 : 1 molar exchange vector. by such processes in high-pressure rocks is well- 82 G.E. BEBOUTAND M.D. BARTON documented in rinds around blocks in me- transfer can be estimated by considering the lange (see Moore, 1984; Sorensen, 1988). Me- stability of probable mineral assemblages and chanical mixing during melange formation, ac- aqueous species. For example, Fig. 8, a dia- companied by local diffusion and larger-scale gram showing the log of the activity ratio of infiltrative metasomatism, can significantly CaO to MgO versus the log of the activity of alter compositions as demonstrated for the Ca- silica, and calculated for the peak metamor- talina Schist (Bebout and Barton, 1989a; see phic P-Tconditions of the amphibolite unit of discussion above; Figs. 3a,b and 5a,b,c). In the the Catalina Schist, illustrates that there are following discussion, we concentrate on infil- substantial chemical potential differences be- tration metasomatism, where chemical changes tween the different lithologies. Consequently, are driven by transfer of materials as solutes in fluids flowing between such units at constant P an H20-rich metamorphic fluid. Infiltration and T will react with the rocks, causing meta- metasomatism, though extensively docu- somatic changes in proportion to changes in mented in lower-pressure environments (see solubility. For silica, where the activity, asio2, compilation by Barton et al., 1991), has re- is directly proportional to the concentration, ceived relatively little consideration in studies ms,o2, (e.g., Walther and Helgeson, 1977), the of high P~ T environments. A thorough analy- change in concentration can be as much as sis of infiltration metasomatism is beyond the ~ 0.5 log units, as for the reaction: scope of this paper; in this section, we consider some elementary consequences of the process 3Mg2 Si04 + 5SiO2,aq ( ~qz saturalion ) for idealized situations. + 2H20 = 2Mg3 I[Si4Oio ] (OH)2 (1)

7.2. Consequences of flow across contacts and Given that the solubility of quartz in water along different P-T paths at the Catalina amphibolite P-T conditions is

In the case of infiltration metasomatism, the -3 0 I I chemical potential for reaction comes from " I CaO-MgO-A1203-SJiO,-H20 1 650°C [ system - [ 10 kbar either lithological contrasts or changes in P-T 0 I "" " ."" i: conditions along the fluid flow paths. Specific, 3 51 D)opside / / ~i, --" -~ "-. Garnet / / well-constrained geological examples and good "~ [ ""-<~: ;( Zoisite experimental data for solubilities are needed to I ~ tAllorllte \ \ evaluate these effects quantitatively. The re- -4.0 -] ~-/ Tremolite ".. quired solubility data are not available at appropriate P-T conditions for species other than - I Forsterite ,;~.~ I alc aqueous silica; however, data for lower pres- -4.5 i i il i sures (Johnson et al., 1991) can be extrapo- - 1.0 -0.5 0.0 lated to crudely evaluate the magnitudes and log asio2 signs of possible changes under higher-P Fig. 8. Calculated activity-activity diagram showing min- conditions. eral assemblages approximating the associations found in Infiltration of fluids across contacts between the amphibolite unit of the Catalina Schist. The dashed lines separate the Al-bearing :minerals, the solid lines sep- contrasting lithologies leads to substantial mass arate the Al-free minerals. Rock types plotted are: transfer in many geological environments. In UM = ultramafic, Maf= mafic, Sed = sedimentary com- subduction zones, it is likely to be particularly positions. This figure illustrates the differences in fluid important at boundaries between sedimen- activities/compositions thal can exist between juxtaposed lithologies. Calculated from data in Helgeson et al. tary, mafic, and ultramafic lithologies. The ( 1978; see Johnson et al., 1991 ) and assuming ideal solid signs and approximate magnitudes of mass solutions. METASOMATISM DURING SUBDUCTION: PRODUCTS AND POSSIBLE PATHS IN CATALINA SCHIST, CA 83

~ 5 wt% (cf. Fig. 9), approximately 30 grams perimposed metasomatic or mechanical events of silica could be added to forsterite-bearing would tend to obscure boundaries. In general, rock per kilogram of water infiltrating from a such changes would not be predicted in large quartz-bearing rock. Likewise, smaller amounts areas of reasonably homogeneous materials, for could be added to silica-undersatured mafic example, as in the km-scale exposures of some rocks, or silica could be removed from quartz- lithologies in the Catalina Schist (see Fig. 5a). saturated rocks by fluids moving in from un- Mass transfer along a flow path with varying dersaturated rocks. Transfer of this magnitude P and/or T could lead to changes comparable would require water to rock mass ratios of ~ 2 to the effects predicted for flow between lith- to account for the change in SiO2 contents in- ologies, but with some distinct and perhaps ferred for the amphibolite melange (Figs. 3b powerful implications. In such cases, metaso- and 5a,b,c). matism takes place in response to the change Flow across lithologic contacts should have in the fluid-mineral equilibrium along the flow predictable and, ideally, observable conse- path. The simplest illustration of this relation- quences for mass transfer. Metasomatic zon- ship involves transport of silica. Figure 9 shows ing will develop at the contacts and will reflect the solubility of quartz as a function of pres- the mineral assemblages of the juxtaposed sure and temperature. Mass transport for any units. Asymmetry may be present, although reaction balanced in terms of aqueous silica can diffusional (bimetasomatic) processes and su- be estimated from this diagram assuming local equilibrium. In quartz-saturated rocks, for ex- 16 t~, 3',~ ~o~ ~o% ample, the equilibrium: ] Quartz I .... [ Solubility[ SiO2,quartz = SiO2,aqueous (2) ..... [ in H20 [ governs the amount of silica that is dissolved 12 :' ...... or precipitated per mass of fluid along a fluid P-Tpath. The same principle applies in quartz- ¢ ~'o • ~ UCEP undersaturated rocks. For example, in a hy- drous peridotite, the equilibrium defined by d~ ){e' "~,> ~4eta~orphit' 'ond~io,., | ~,~ 8 7atali Amp, r~olite nit V reaction (1) might buffer aqueous silica at about 1/3 the value of quartz saturation (see Fig. 8). Higher water-to-rock ratios are re- \ [Solubility [f(P,T)] changes [ quired to move the same amount of silica in 4 \ I commensurate with maximum I the peridotite than in the quartz-saturated ex- '1 AIa between lithologies I ample. In both cases the silica concentrations I AT - -2 x 102 degrees, or I vary regularly with changes in pressure and ~--~ AP~-5 x 103 bars ] temperature. The gray arrows in Fig. 9 illus- 0 -.....-----0, I I I 200 LC~EP 600 1000 trate that such changes could be important. T°C Each of the arrows in Fig. 9 represents a change in P-T conditions corresponding to a change Fig. 9. Quartz solubility as a function of pressure and temperature (replotted from data summarized in Holland and in the equilibrium constant comparable to that Malinin, 1979 ), illustrating the path dependence of silica which occurs upon infiltration between con- transport. Each of the gray arrows represents a change in trasting lithologies (Fig. 8). These solubility silica solubility commensurate with likely maximum changes could occur over temperature and changes between lithologies. Silica addition to the ultramafic melange of the Catalina Schist requires up-dip rather pressure intervals of ~ 100 °C or a few kbar-- than up-T flow if the transfer is to be done by fluid well within the P-T intervals anticipated for infiltration. fluid flow in dewatering subduction zones. 84 G.E. BEBOUTAND M.D. BARTON

An important implication of these consid- depends on ZCI and the value of the equilib- erations is that the nature of metasomatic rium constant. changes is sensitive to the fluid flow path (cf. At present, insufficient experimental data are Barton et al., 1991 for discussion of signifi- available to evaluate these changes quantita- cance in lower-P environments). Fluids going tively; however, it is possible to make some away from the arrowheads in Fig. 9, would re- general predictions from known behavior at move corresponding amounts of silica from the lower pressures. Four major types of reactions rocks along their paths. For the Catalina Schist are considered here: net silica transfer, hydrol- amphibolite-facies ultramafic melange (see ysis, alkali exchange, and isotope exchange. Fig. 5a), this implies that hydrothermal silica Examples include: addition would have required a particular path SiO2,mineral • SiO2,aq (net transfer of silica) (Fig. 9 ): up-dip (down-P, down-T) fluid flow; the observed silicification of the melange (Fig. (3) 5c) precludes simple vertical flow (down-P, 4HCI°q + 5Na2 Mg3 AI2 Sis 022 (OH) 2,glaucophane but up-T). In principle, the thermodynamic expression + 5H 2 0 = for the dependence of the equilibrium con- 4NaCl°q + 3Mg5 A12 Si3 Pip (OH) 8,clinochlore stant on P and T may be used to calculate the amount of reaction that would take place for a + 4NaAISi3 O8,albite -~ 19SiO2.qoartz particular flow direction. The appropriate (hydrolysis) (4a) expressions are: 3KAISi3 O8.K-feldspar+ 2HCI°q = GT, p GTo ,p m KAI2 [ AISi301o ] (OH) 2,mascovite+ 2KCI]q RT RT ° + 6SiO2.quartz (hydrolysis) (4b) K~H~,.eo ( l/T- 1/T ° ) + V~,,(P-P ° ) Aln R RT NaAISi3 O8.albite + KCI°q = KAISi3 O8,albite + NaCl~q (alkali exchange ) (5) or 18OmmeraI -]- 16Ofluid = 16OmineraI "~- 18Ofluid OGrxn 0 (Prpath) -- (isotope exchange ) ( 6 ) Although such reactions are plausible, assess- --Srxn[0(~D0-~ath )]dT- ]- Vrxn[.0 (]°0~--ath)ld.P ments of their significance depend on the dem- onstration of appropriate metasomatism in For the case of net transfer reactions (e.g., subduction zone rocks. It must be established dissolution or precipitation of quartz or graph- in each case that isochemical (excepting vola- ite), the equilibrium constant can be simply tile addition/loss) reactions did not produce transformed to the change in mass along the the observed assemblage. For example, it path. For exchange reactions, however, the should be shown that blue amphibole veins mass transfer can be dependent on other fac- represent addition of Na (e.g., by reaction 4a), tors. For example, the amount of reaction in rather than a conventional isochemical reac- the exchange reaction: tion. Evidence for the operation of these types of reactions (reactions 1 to 6) is found in sub- NaA1Si3 O8,albite + KCl~queous= duction zone rocks (see discussions above), yet KmlSi3 O8,K-feldspar "~ NaCl°q ..... the reactions (forward or reverse ) show differing P-T dependencies. Figure 10 qualitatively METASOMAT1SM DURING SUBDUCTION: PRODUCTS AND POSSIBLE PATHS IN CATALINA SCHIST, CA 85

(see Bodnar and Costain, 1991 for discussion Possible met~zvornati(" I Some exchanges assemhlagc,~ along I of variability in low-P environments). The for different dP/dT dillS'rein flow pathis ] likelihood of fluid immiscibility at many subduction-zone conditions (Selverstone et al., 1992) points to further complexity. ~D " ~,~,:~ ~ \o~ '.oo v-..., q, isotope ~ /_/- Qlu:r:z-:~.c. 7. 2. Possible applications cD exchange ~~ ~,~,,~ <,H,i:izar,.... Given sufficient geological and physical ,/~-::' i,:).> i \ , I~> tl)o chemical constraints for particular cases, it should be possible to evaluate the processes and :',~:-aml~h:p.L • paths involved in generating mass transfer ¢l##i=c~/i':;/t during subduction. Observations from the Ca- Temperature talina Schist and other subduction complexes indicate great complexity; in the case of the Fig. 10. Schematic dependence of metasomatic exchanges Catalina Schist, contrasting metasomatic as- on fluid P-T flow paths. Black lines indicate exchanges, gray lines indicate possible product assemblages for flow semblages, in some cases affecting the same in those directions. See text for discussion. rocks (e.g., multiple, cross-cutting vein sets; see Fig. 2b; Table 1 ), require that distinct meta- indicates the types of metasomatic changes that somatic events took place. In this section, ele- are expected for fluids flowing along different mentary considerations regarding mass trans- P-T trajectories. These changes are based on fer are used to predict the contrasting experimental and thermodynamic estimates of metasomatic consequences of some general- such reactions (e.g., as done by Barton et al., ized subduction zone fluid flow scenarios. 1991 for contact metamorphic environments The paths that connate and metamorphic using SUPCRT, Helgeson et al., 1978; John- fluids follow in subduction environments are son et al., 1991 ) and the observation that iso- poorly known. The best constraint is that only topic exchange reactions are sensibly P-inde- a small fraction of the down-going flux is re- pendent (see Clayton and Kieffer, 1992, and turned via arc magmatism (see Ito et al., references therein ). 1983 ); most must reernerge in the fore-arc en- A wide variety of assemblages could form vironment, particularly in the vicinity of the depending on the fluid P-T paths. In the next accretionary prism (see Moore and Vrolijk, section, we consider the consequences of sev- 1992). Theoretical and observational data eral physically plausible fluid paths and in- demonstrate that volatHes are progressively lost sights regarding the distributions and magni- during subduction, perhaps largely in the shal- tudes of the resulting metasomatism. Many lower parts of subduction zones ( < 50 km ), but other reactions could be considered. For ex- that non-trivial quantities continue to be re- ample, carbonates and graphite occur in many leased to greater depths (Delany and Helge- vein assemblages in the Catalina Schist (see son, 1978; Peacock, 1990; Bebout, 1991 a). Table 1); however, their solubility relation- Fluid release and subsequent movement ships are complex (e.g., Holland and Malinin, back towards the surface will be a function of 1979, for a discussion of calcite solubility) and many factors including permeability, the ve- are likely to be coupled to other reactions tak- locity of subduction and thermal structure of ing place in the rock. Differences in the major the subduction zone, the many chemical fac- fluid components (e.g., in the C-O-H-S-N- tors that can impede or enhance fluid move- C1 system ) will dramatically affect solubilities ment, as well as the nature of the fluids them- 86 G.E. BEBOUT AND M.D. BARTON selves (i.e. transport properties; see Brenan, Three schematic fluid paths are illustrated in 1991 ). Intergranular and fracture permeabil- Fig. 11 a for an idealized subduction zone. Each ity is most likely substantial in subduction path begins at a common point near the zones (based on observations and expected blueschist/eclogite boundary, but they diverge hydrofracting where Pnuid>/Plithost,~ic); how- depending on the balance between subduction ever, long-range interconnected permeability velocity and the Darcy velocity of the fluid. seems unlikely except at very shallow levels Cases 1 and 2 are situations where the Darcy (based on field observations and knowledge of velocities are always greater than the subduc- rock ductilities). Fluid movement will be dri- tion velocity (thus the fluids always flow up), ven by the hydraulic gradient, thus the overall whereas Case 3 illustrates a possible path where tendency will be for fluids to move upward (see initial escape dewatering is slow compared to Peacock, 1990; Bebout, 1991 a), although two subduction velocity. The P-T characteristics of factors complicate simple upward flow. First, these three paths are illustrated in Fig. 1 lb. the permeability is likely to be strongly aniso- These paths merely illustrate some logical pos- tropic and highest parallel to the subducting sibilities; actual paths will depend on the inte- plate (e.g., along melange zones; see Bebout, grated devolatilization history and the evolv- 1991b). Secondly, to some extent, fluids will ing thermal and mechanical structure, be entrained in the down-going slab such that including underplating. Cooling and de- their net velocity for parts of some paths may compression are common to all three cases, be downward in spite of their movement up- thus metasomatic effects such as precipitation ward relative to their local (rock) frame of ref- of carbonate and silica would be common (cf. erence (see discussions of fluid entrainment in Fig. 10). In paths 2 and 3, significant heating Magaritz and Taylor, 1976; Bebout, 1991b; plus possible compression are experienced Nadeau et al., 1993 ). This is analogous to bur- along the early stages of the paths. Such paths ial and progressive dewatering in sedimentary could readily cause the sodic alteration (e.g., basins where the net transport of fluid is al- by reaction 4a), silica dissolution, and allied most always up relative to the rocks, but some effects (Fig. 10) documented in subduction- fraction of the fluids are transported to greater related metamorphic complexes. Further- depths and higher pressures with increasing more, the complex cross-cutting relationships burial. observed in areas such as the Catalina Schist

T°C 200 400 600

0-_...... 300°C100°C...... , ~.,,..,...... 0 ...... 500oc ...... :.:....., .,' ,,.,-' E - : ?/fIZZ- ...... 7••°C ...... "i;)~ /" "•..,..--

60 ,""-/ ....."" "" 16 D. .-" .. ;i......

.... • ,,"""" not to scale A, 120 " -32 1.

Fig. 11. Contrasting schematic fluid P-T paths during subduction: (a) Cross-section with isolEherms, (b) P-T diagram• All fluid flow trajectories originate at a hypothetical point near the blueschist-eclogite transition and move "upward" (relative to the subducting slab) toward more nearly stationary materials near the slab-wedge interface• The paths differ in the ratio fluid (Darcy) velocity to the subduction velocity (highest for path 1, lowest for path 3). METASOMATISM DURING SUBDUCTION: PRODUCTS AND POSSIBLE PATHS IN CATALINA SCHIST, CA 87

(see Table 1 and examples in Figs. 2 and 3) TABLE 2 likely record the progressive transport of the Relative efficiency of fluids for mass and energy transfer rocks through differing parts of the evolving hydrologic regime. Component Content in Content in F/R rock fluid (equal masses)

7. 3. Limitations Heat ~1J/g-K ~4 J/g-K -4 Hydrogen <0.1-2% J[ 1% 10->> 100 The potential exists for significant chemical Oxygen 45-50% 89% ~ 2 Silica 40-70% 0.1 -10% < 0.04- ~ 0.25 redistribution and metasomatic alteration in Alkalies < 1-10% <0.1-5% <0.1-> 1 subduction zones. Large volumes of volatiles (SC1=0.1-2 M) are subducted and released during subduction Carbon 0.1-1% and experiments and theoretical predictions indicate high solubilities of many elements in high-P/T metamorphic fluids. Although we gives some approximate factors governing the consider metasomatism along some some pos- mass balance efficacy of thermal, isotopic, and sible generalized flow paths, we neglect the ac- chemical transfer. Conceptually, the mass of tual physics of fluid flow: the nature of perme- rock that can be affected is proportional to the ability, the plausible fluxes, and the heat or mass content of the fluid that causes requirements for convection. Focussing of fluid transfer compared to the content of the same flow, at varying scales, will be of obvious im- component in the rock. Only for H, O, and heat portance in localizing the metasomatic effects are the specific contents of water higher than of the fluids. If fluid flow is sufficiently local- those in rocks (from 2 to > 100 times). The ized along melange zones, effects of fluid-rock integrated effect of other constituents moved interactions involving contrasting lithologies by 1-10 % of fluid by mass will be much will be minimized. Convection of fluids in the smaller, due to the high rock contents com- subduction environment (discussed as a pos- pared to fluid contents. sibility by Criss and Hofmeister, 1991 ) is con- Consequently, while metasomatic effects sidered unlikely due to the lack of intercon- may be profound on a "local" scale (up to km- nected permeability near the slab-mantle scale), the larger volume of subduction zone interface; forced convection due to entrain- rocks will likely exhibit relatively minor chem- ment of fluid in convecting melange matrix ical changes, particularly in the chemical com- (see Cloos and Shreve, 1988 ) might be signif- ponents which are less soluble in the H20-rich icant if Darcy flow is inefficient. fluids. Nevertheless, changes in the constitu- The amount of infiltrative mass transfer that can take place is ultimately limited by the ents which are less soluble in fluids may be amount of volatiles available. Simple mass prominent in the form of veins or in other areas balance considerations indicate that pervasive of localized flow. This distinction is well-dem- effects are to be expected only if fluid flow is onstrated in the Catalina Schist, where iso- localized, and only for constituents such as H topic alteration is widespread but most dra- isotopes and, possibly, heat and O isotopes. matic in melange units which were the primary Crude limits can be placed on the gross chem- loci of fluid transport (see Bebout, 1991a,b); ical and thermal transfer that fluids can cause melange units also show dramatic major and given the likely volatile content of subducted trace element alteration relative to more co- materials (for the upper 10 km of sedi- herent metamafic and metasedimentary expo- ments + oceanic crust + pore waters, H:O con- sures (see above discussions). The observa- tent is on the order of 1-10 weight %). Table 2 tion, in the Catalina Schist, of minimal isotopic 88 G.E. BEBOUT AND M.D. BARTON or other chemical change in most non-melange metasomatism (veins, stable isotope homoge- rocks away from veins and their envelopes is nization, block rinds) is particularly abundant wholly consistent with observations indicating in melange zones. Shifts in the composition of limited pervasive mass transfer in parts of melange matrix from plausible compositions other subduction complexes (e.g., Selverstone derived through mechanical mixing reflect et al., 1992; Nadeau et al., 1993). metasomatic additions and subtractions. Geo- Finally, we have argued above for the likely chemical evidence (stable isotope; whole-rock) importance of contrasting P-T fluid flow paths indicates exchange of all of the major units with in generating some of the metasomatic assem- compositionally similar fluids. For the Catal- blages observed in the Catalina Schist and other ina Schist, the most likely fluid source is in low- subduction complexes. However, as noted ear- grade, sedimentary rocks (analogs are low- lier, compositional changes can easily be gen- grade units) based on the stable isotope data. erated by mechanical mixing of and flow/dif- Given mineral-fluid equilibrium constants, fusion between contrasting lithologies. These it is possible to predict metasomatic change as mechanisms, which will have different geo- a function of fluid flow paths. Decreasing Tand chemical effects (e.g., linear compositional P favors fixing of K, Si, (7, and H in rocks. In- trends in the case of two-component mechani- creasing T (+ moderately decreasing P) cal mixing), must be considered in any inter- should fix Na but may leach most other com- pretation of the processes of subduction zone ponents. The Si-rich, K_+ Si-rich, and Na-rich/ metasomatism. Si-poor assemblages are consistent with these P-T paths. These effects can be comparable to 8. Conclusions other mechanisms for mass exchange, but are likely to be important only if flow is localized Study of deep fluid processes in subduction or in the unlikely case of fluid recirculation. zones is limited by the fragmentary nature of The nature of the mass transfer depends on the high-P~ T metamorphic terranes. The Catalina P-T paths and lithologies, and will thus inter- Schist, California, contains atypically coher- act with physical evolution in subduction ent units of sedimentary, mafic, and ultra- zones. Given sufficiently detailed field obser- mafic rocks metamorphosed over a wide range vations, metasomatic features should ulti- of P- T conditions (lawsonite-albite to amphi- mately be useful in interpreting the paleodyn- bolite facies ) during Early Cretaceous subduc- amics of subduction complexes. Much tion. The range in grade permits assessment of additional field-based study of subduction- fluid communication among spatially dispar- zone metasomatism is needed, particularly ate parts of the accretionary complex and vari- with emphasis on the interpretation of the ations in metasomatism as functions of P-T mineralogy and textures of multiple vein sets conditions. and the interpretation of geochemical evi- In the Catalina Schist, there is evidence for dence for element mobility. This study has fur- compositional changes via mechanical mixing ther emphasized the critical need for reliable (i.e., to produce melange matrix), diffusional solubility data under P-T conditions appro- processes (e.g., development of rinds on priate for deep subduction environments. blocks), and larger-scale, fluid-mediated The scale and complexity of metasomatic transfer processes. Veins are ubiquitous in all evolution of the Catalina Schist may be di- units; multiple vein generations (identified rectly representative of processes leading to texturally and mineralogically) record meta- magmatism in forearc regions (i.e., at depths somatism during both prograde metamor- ~< 50 km) in relatively hot, sediment-rich sub- phism and uplift/cooling history. Evidence for duction zones, and may be analogous to that MET~SOMATISM DURING SUBDUCTION: PRODUCTS AND POSSIBLE PATHS IN CATALINA SCHIST, CA 89 of metasomatic processes deeper in more ther- Thermal model for subduction with dehydration in the mally mature subduction zones. Comparison downgoing slab. J. Geol., 86: 731-739. Bailey, E.H., 1941. Mineralogy, petrology and geology of of the metasomatism in the Catalina Schist Santa Catalina Island, California. Ph.D. Thesis, Stan- with that in other high-P/T metamorphic ford University, Stanford, CA, 193 pp. complexes representing varying depths in sub- Bailey, E.H., Irwin, W.P. and Jones, D.L., 1964. Francis- duction zones (e.g., Vrolijk et al., 1988; Fisher can and related rocks. Calif. Dept. Mines Geol. Bull., 183. and Byrne, 1990; Philippot and Selverstone, Barton, M.D. and Bebout, G.E., 1989. Chemical conse- 1991; Selverstone et al., 1992; Nadeau et al., quences of fluid flow paths in subduction zones 1993) will contribute a better understanding (abstr.). Geol. Soc. Am. Abstr. Progr., 216: 85. of metasomatic evolution as a function of Barton, M.D., Bebout, G.E. and Sorensen, S.S., 1987. Iso- depth, thermal structure, and lithology. topic constraints on the geochemical evolution of an ultramafic subduction zone melange: Catalina Schist terrane, California (abstr.). Eos, Trans. Am. Geo- Acknowledgements phys. Union, 68: 1525. Barton, M.D., Ilchik, R.P. and Marikos, M.A., 1991. Me- tasomatism. In: D.M. Kerrick (Editor), Contact This research was funded by grants to MDB Metamorphism. Rev. Mineral., 26: 321-350. by the National Science Foundation (EAR86- Basaltic Volcanism Study Project, 1981. Basaltic Volca- 07542 and OCE90-2215) and the American nism on the Terrestrial Planets. Pergamon Press, New York, NY, 1286 pp. Chemical Society Petroleum Research Fund Bebout, G.E., 1989. Geological and geochemical investi- (20067-AC2 to MDB), and by grants to GEB gations of fluid flow and mass transfer during subduc- from the American Chemical Society Petro- tion-zone metamorphism. Ph.D. Thesis, Univ. Calif., leum Research Fund (25246-G2), the Geolog- Los Angeles, CA, 370 pp. ical Society of America, and Sigma Xi. GEB's Bebout, G.E., 199 la. Field-based evidence for devolatilization in subduction zones: implications for arc mag- research was supported in part by the Carnegie matism. Science, 251: 413-416. Institution of Washington (Geophysical Lab- Beboul, G.E., 1991b. Geometry and mechanisms of fluid oratory, ). A large part of this research was con- flow at 15 to 45 kilometer depths of an early Creta- ducted while both authors were at the Univer- ceous aecretionary complex. Geophys. Res. Lett., 18: 923-926. sity of California, Los Angeles. We extend Bebout, G.E., 1992. Nitrogen isotope signatures of high- special thanks to the Santa Catalina Conser- temperature fluid-rock interactions (abstr.). Geol. vancy for its support of the field research. We Soc. Am. Abstr. Progr., 24: 251. thank M. Grove, S. Sorensen, and A.E. Bebout Bebout, G.E. and Barton, M.D., 1988. Field evidence for for helpful discussions, and A.E. Bebout, S. partial melting of sediment and oceanic crust in a subduction complex: Catalina Schist Terrane, California Peacock, P. Philippot, and J. Touret for help- (abstr.). Eos, Trans. Am. Geophys. Union, 69: 505. ful reviews. Bebout, G.E. and Barton, M.D., 1989a. Fluid flow and metasomatism in a subduction zone hydrothermal system: Catalina Schist Terrane, California. Geology, References 17: 976-980. Bebout, G.E. and Barton, M.D., 1989b. Fluid-rock inter- Agrinier, P., Javoy, M., Smith, D.C. and Pineau, F., 1985. action and carbon recycling in subduction zones: evi- Carbon and oxygen isotopes in eclogites, amphibo- dence from stable isotope systematics in a high-pres- lites, veins and marbles from the Western Gneiss Re- sure metamorphic complex, Catalina Schist, California gion, Norway. Chem. Geol. Isot. Geosci., 52: 145-162. (ext. abstr.). Proc., 1989 Int. Geol. Congr., Washing- Alt, J.C., Honnorez, J., Laverne, C. and Emmermann, R., ton, DC, pp. 1-108-1-109. 1986. Hydrothermal alteration of a 1 km section Bebout, G.E. and Fogel, M.L., 1992. Nitrogen-isotope through the upper oceanic crust, Deep Sea Drilling compositions of metasedimentary rocks in the Catal- Project Hole 504B: mineralogy, chemistry, and evolu- ina Schist, California: implications for metamorphic tion of seawater-basalt interactions. J. Geophys. Res., devolatilization history. Geochim. Cosmochim. Acta., 91:10 309-10 335. 56: 2139-2149. Anderson, R.N., DeLong, S.E. and Schwarz, W.M., 1978, Bebout, G.E., Ryan, J.G. and Leeman, W.P., t993a. B- 90 G.E. BEBOUT AND M.D. BARTON

Be systematics in subduction-related metamorphic blueschists, North Cascades, Washington: variably rocks: characterization of the subducted component. fractionated and altered metabasalts of oceanic affin- Geochim. Cosmochim. Acta, 57: 2227-2237. ity. Contrib. Mineral. Petrol., 82: 131-146. Bebout, G.E., Ryan, J.G., Leeman, W.P. and Bebout, A.E., Ellam, R.M. and Hawkesworth, ,C.J., 1988. Elemental and 1993b. Fractionation of trace element and stable iso- isotopic variations in subduction related basalts: evi- tope signatures during subduction-zone metamor- dence for a three component model. Contrib. Mineral. phism (abstr.). Los, Trans. Am. Geophys. Union, 74: Petrol., 98: 72-80. 331. Engebretson, D.C., Cox, A. and Gordon, R.G., 1985. Rel- Bodnar, R.J. and Costain, J.K., 1991. Effect of varying ative motions between oceanic and continental plates fluid composition on mass and energy transport in the in the Pacific Basin. Geol. Soc. Am. Spec. Pap., 206, Earth's crust. Geophys. Res. Lett.o 18: 983-986. 59 pp. Brenan, J., 1991. Development and maintenance of met- Ernst, W.G., 1965. Mineral parageneses in Franciscan amorphic permeability: implications for fluid trans- metamorphic rocks, Panoche Pass, California. Geol. port, Metasomatism. In D.M. Kerrick (Editor), Con- Soc. Am. Bull., 76: 879-914. tact Metamorphism. Rev. Mineral., 26:291-319. Ernst, W.G., 1988. Tectonic history of subduction zones Carswell, D.A., Curtis, C.D. and Kanaris-Sotiriou, R., inferred from retrograde blueschist P-T paths. Geol- 1974. Vein metasomatism in peridotite at Kalskaret ogy, 16: 1081-1084. near Tafjord, South Norway. J. Petrol., 15: 383-402. Ernst, W.G., Seki, Y., Onuki, H. and Gilbert, M.C., 1970. Clayton, R.N. and Kieffer, S.W., 1992. Oxygen isotope Comparative study of low-grade metamorphism in the thermometer calibrations. In: H.P. Taylor, Jr., J.R. California Coast Ranges and the Outer Metamorphic O'Neill and 1.R. Kaplan (Editors), Stable Isotope Belt of Japan. Geol. Soc. Am. Mem., 124,276 pp. Geochemistry: A Tribute to Samuel E~)stein. Geo- Erzinger, J., 1989. Chemical alteration of the oceanic crust. chem. Soc. Spec. Publ., 3: 3-10. Geol. Rundsch., 78:731-740. Cloos, M., 1986. Blueschists in the Franciscan Complex Essene, E.J. and Fyfe, W.S., 1967. Omphacite in Califor- of California: petrotectonic constraints on uplift nian rocks. Contrib. Mineral. Petrol., 15: 1-23. mechanisms. Geol. Soc. Am. Mem., 164: 77-93. Etheridge, M.A., Wall, V.J., Cox, S.F. and Vernon, R.H., Cloos, M. and Shreve, R.L., 1988. Subduction-channel 1984. High fluid pressures cluring regional metamor- model of prism accretion, melange formation, sedi- phism and deformation: implications for mass trans- ment subduction and subduction erosion at conver- port and deformation mechanisms. J. Geophys. Res., gent plate margins: 2. Implications and discussion. Pure 89: 4344-4358. Appl. Geophys., 128: 501-545. Fisher, D. and Byrne. T., 1990. The character and distri- Criss, R.E. and Hofmeister, A.M., 1991. Application of bution of mineralized fractures in the Kodiak Forma- fluid dynamics principles in tilted permeable media to tion, Alaska: implications fi~r fluid flow in an under- terrestrial hydrothermal systems. Geophys. Res. Lett., thrust sequence. J. Geophys. Res., 95: 9069-9080. 18: 199-202. Frey, F.A., Bryan, W.B. and Thompson, G., 1974. Atlan- Davidson, J.P., 1987. Crustal contamination versus sub- tic ocean floor: geochemistry and petrology of basalts duction zone enrichment: examples from the Lesser from Legs 2 and 3 of the Deep-Sea Drilling Project. J. Antilles and implications for mantle source composi- Geophys. Res., 79: 5507-5527. tions of island arc volcanic rocks. Geochim. Cosmo- Fyfe, W.S. and Zardini, R., 1967. Metaconglomerate in chim. Acta, 51: 2185-2198. the Franciscan Formation near Pacheco Pass, Califor- Delany. J.M. and Helgeson, H.C., 1978. Calculation of the nia. Am. J. Sci., 265: 819-8130. thermodynamic consequences of dehydration in sub- Fyson, W.K., 1987. A succession of quartz veins in Ar- ducting oceanic crust to 100 kb and > 800= C. Am. J. chaen metaturbidites, Yellowknife Bay, Slave Prov- Sci., 278: 638-686. ince. Can. J. Earth Sci., 24: 698-710. Donnelly, T.W., Thompson, G. and Salisbury, M.H., 1980. Gill, J.B., 1981. Orogenic andesites and plate tectonics, The chemistry of altered basalts at site 417, DSDP Leg Springer-Verlag, Berlin, 390, pp. 51: in Donnelly, T., Francheteau, J., Bryan, W., Ro- Gillet, P. and Goffe, B., 1988. On the significance ofara- binson, P., Flower, M. and Salisbury, M., et al. (eds.), gonite occurrences in the Western Alps. Contrib. Min- lnit. Rep. Deep Sea Drill. Proj., 51, U.S. Gov. Print. eral. Petrol., 99: 70-81. Off. Washington, D.C., pp. 1319-1330. Grove, M. and Harrison, T.M., 1992. Significance of 4°Ar/ Dumitru, T.A., 1991. Effects of subduction parameters on 39Ar results from the Catalina Schist (abstr.). Eos, geothermal gradients in forearcs, with an application Trans. Am. Geophys. Union, 73: 327. to Franciscan subduction in California. J. Geophys. Hart, R., 1976. Chemical variance in deep ocean basalts. Res., 96: 621-641. In: S.R. Yeats et al. (Editors), Init. Rep. Deep Sea Drill. Dungan, M.A., Vance, J.A. and Blanchard, D.P., 1983. Proj., 34. U.S. Gov. Print. Off., Washington, DC, pp. Geochemistry, of the Shuksan greenschists and 301-335. METASOMATISM DURING SUBDUCTION: PRODUCTS AND POSSIBLE PATHS IN CATALINA SCHIST, CA 91

Hart, S.R., Erlank, A.J. and Kable, E.J.D., 1974. Sea floor low-temperature Franciscan metabasites: a new ap- basalt alteration: some chemical and Sr isotopic ef- proach using the U-Pb system. Geol. Soc. Am. Mem., fects. Contrib. Mineral. Petrol., 44:219-230. 164: 95-105. Helgeson, H.C., Delany, J.M., Nesbitt, H.W. and Bird, Moore, D.E., 1984. Metamorphic history of a high-grade D.K., 1978. Summary and critique of the thermody- blueschist exotic block from the Franciscan Complex, namic properties of rock-forming minerals. Am. J. Sci., California. J. Petrol., 25:126-150. 278-A, 229 pp. Moore, D.E. and Eiou, J.G., 1979. Chessboard-twinned Hofmann, A.W., 1988. Chemical differentiation of the albite from Franciscan metaconglomerates of the Dia- Earth: the relationship between mantle, continental blo Range, California. Am. Mineral., 64: 329-336. crust, and oceanic crust. Earth Planet. Sci. Lett., 90: Moore, J.C. and Vrolijk, P., 1992. Fluids in accretionary 297-314. prisms. Rev. Geophys., 30: 113-135. Holland, H.D. and Malinin, S.D., 1979. The solubility and Moore, D.E., Liou, J.G. and King, B.-S., 1981. Chemical occurrence of non-ore minerals. In: J.L. Barnes (Edi- modifications accompanying blueschist facies meta- tor), Geochemistry of Hydrothermal Ore Deposits morphism of Franciscan conglomerates, Diablo Range, (2rid edition ). Wiley, Chichester, pp. 461-508. California. Chem. Geol.. 33: 237-263. Humphris, S.E. and Thompson, G., 1978. Hydrothermal Moran, A.E., Sisson, V.B. and Leeman. W.P., 1992. Bo- alteration of oceanic basalts by seawater. Geochim. ron depletion during progressive metamorphism: im- Cosmochim. Acta, 42: 107-125. plications for subduction processes. Earth Planet. Sci. Ito, E., Harris, D.M. and Anderson, A.T., Jr., 1983. Alter- Left., 111 : 331-349. ation of oceanic crust and geologic cycling of chlorine Morris, J.D., Leeman, W.P and Tera, F., 1990. The sub- and water. Geochim. Cosmochim. Acta, 47: 1613- ducted component in island arc lavas: constraints from 1624. Be isotopes and B-Be systematics. Nature, 344: 31- Johnson, J.W., Oelkers, E.H. and Shock, E.L., 1991. De- 36. scription and analysis of fluid-mineral equilibria us- Nadeau, S., Philippot, P. and Pineau, F., 1993. Fluid in- ing the SUPCRT92 Software Package. Short Course clusion and mineral isotopic compositions (H-C-O) Manual, Annu. Meet. Geol. Soc. Am., San Diego, CA, in eclogitic rocks as tracers of local fluid migration 1991. during high-pressure metamorphism. Earth Planet. Sci. Langseth, M.G. and Moore, J.C., 1990. Introduction to Left., 114: 431-448. Nelson, B.K., 1991. Sediment-derived fluids in subduc- special section on the role of fluids in sediment accretion zones: isotopic evidence from veins in blueschist tion, deformation, diagenesis, and metamorphism in and eclogite of the Franciscan Complex, California. subduction zones. J. Geophys. Res., 95: 8737-8741. Geology, 19: 1033-1036. Liegeois, J.-P. and Duchesne, J.-C., 1981. The Lac Cornu Okay, A.I., 1982. Incipient blueschist metamorphism and retrograded eclogites (Aiguilles Rouges massif, West- metasomatism in the Tavsanli Region, Northwest ern Alps, France): evidence of crustal origin and me- Turkey. Contrib. Mineral. Petrol., 79: 36tl-367. tasomalic alteration. Lithos, 14: 35-48. O'Neil, J.R., 1986. Theoretical and experimental aspects Ludden, J.N. and Thompson, G., 1979. An evaluation of of isotopic fractionation, Rev. Mineral., 16: 445-489. the behavior of the rare-earth elements during weath- Owen, C., 1989. Magmatic differentiation and alteration ering of sea-floor basalts. Earth Planet. Sci. Lett., 43: in isofacial greenschists and blueschists, Shuksan Suite, 85-92. Washington: Statistical analysis of major-element Magaritz, M. and Taylor, H.P., Jr., 1976. Oxygen, hydro- variation. J. Petrol., 30: 739-761. gen and carbon isotope studies of the Franciscan for- Peacock, S.M., 1990. Fluid processes in subduction zones. mation, Coast Ranges, California. Geochim. Cosmo- Science, 248: 329-337. chim. Acta, 40: 215-234. Pfeifer, H.-R., 1987. A model for fluids in metamor- Maruyama, S. and Liou, J.G., 1987. Clinopyroxene -- a phosed ultramafic rocks: IV. Metasomatic veins in mineral telescoped through the processes ofblueschist metaharzburgites of Cima Di Gagnone, Valle Ver- facies metamorphism. J. Metamorph. Geol., 5: 529- zasca, Switzerland. In: H.C. Helgeson (Editor), 552. Chemical Transport in Metasomatic Processes. Rei- Maruyama, S. and Liou, J.G., 1988. Petrology of Francis- del, Dordrecht, pp. 591-632. can metabasites along the jadeite-glaucophane type Philippot, P. and Selverstone, J., 1991. Trace-element-rich facies series, Cazadero, California. J. Petrol., 29: 1-37. brines in eclogitic veins: implications for fluid com- Matlhews, A. and Schliestedt, M., 1984. Evolution of the position and transport during subduction. Contrib. blueschist and greenschist facies rocks of Sifnos, Cy- Mineral. Petrol., 106:417-430. clades, Greece: a stable isotope study of subduction- Platt, J.P., 1975. Metamorphic and deformational pro- related metamorphism. Contrib. Mineral. Petrol., 88: cesses in the Franciscan Complex, California: Some 150-163. insights from the Catalina Schist terrain. Geol. Soc. Mattinson, J.M., 1986. Geochronology of high-pressure- Am. Bull., 86: 1337-1347. 92 G,E. BEBOUT AND M.D. BARTON

Platt, J.P., 1976. The petrology, structure, and geologic oceanic rocks. Earth Planet. Sci. Lett., 57: 285-304. history of the Catalina Schist terrain, southern Califor- Suppe, J. and Armstrong, R.L., 1972. Potassium-argon nia. Univ. Calif. Publ. Geol. Sci., 112, 111 pp. dating of Franciscan metamorphic rocks. Am. J. Sci., Rumble, D., 111 and Spear, F.S., 1983. Oxygen-isotope 272:217-233. equilibration and permeability enhancement during Tatsumi, Y., Hamilton, D.L. and Nesbitt, R.W., 1986. regional metamorphism. J. Geol. Soc. London, 140: Chemical characteristics of fluid phase released from 619-628. a subducted lithosphere and origin of arc magmas: evi- Selverstone, J., Franz, G., Thomas, S. and Getty, S., 1992. dence from high-pressure experiments and natural Fluid variability in 2 GPa eclogites as an indicator of rocks. J. Volcanol. Geotherm. Res., 29: 293-309. fluid behavior during subduction. Contrib. Mineral. Taylor, H.P., Jr. and Coleman, R.G., 1968. O18/O ~6 ra- Petrol., 112: 341-357. tios of coexisting minerals in glaucophane-bearing Seyfried, W.E., Jr., Mottl, M.J. and Bischoff, J.L., 1978. metamorphic rocks. Geol. Soc. Am. Bull., 79: 1727- Seawater/basalt ratio effects on the chemistry and 1756. mineralogy of spilites from the ocean floor. Nature, Tenore-Nortrup, J. and Bebout, G.E., 1993. Metasoma- 275: 211-213. tism of gabbroic and dioritic cobbles in blueschist fa- Sorensen, S.S., 1984. Petrology of basement rocks of the cies metaconglomerates: sources and sinks for high-P/ California Continental Borderland and the Los Ange- 1 metamorphic fluids (abstr.). Eos, Trans. Am. Geo- les Basin. Ph.D. Thesis, Univ. Calif., Los Angeles, CA, phys. Union, 74: 331. 423 pp. Vrolijk, P., 1987. Paleohydrogeology and fluid evolution Sorensen, S.S., 1986. Petrologic and geochemical comparison of the blueschist and greenschist units of the Ca- of the Kodiak accretionary complex, Alaska. Ph.D. talina Schist terrane, southern California. Geol. Soc. Thesis, Univ. Calif., Santa Cruz, CA. Am. Mem., 164: 59-75. Vrolijk, P., Myers, G. and Moore, J.C., 1988. Warm fluid Sorensen, S.S., 1988. Petrology ofamphibolite-facies mafic migration along tectonic melanges in the Kodiak ac- and ultramafic rocks from the Catalina Schist, south- cretionary complex, Alaska. J. Geophys. Res., 93: ern California: Metasomatism and migmatization in a 10313-10324. subduction zone metamorphic setting. J. Metamorph. Walther, J.V. and Helgeson, H.C., 1977. Calculation of Geol., 6: 405-435. the thermodynamic properties of aqueous silica and the Sorensen, S.S. and Barton, M.D,, 1987. Metasomatism and solubility of quartz and its polymorphs at high pres- partial melting in a subduction complex, Catalina sures and temperatures. Am. J. Sci,, 277:1315-1351. Schist, southern California. Geology, 15:115-118. Wyllie, P.J. and Sekine, T., 1982. The formation of man- Sorensen, S.S. and Grossman, J.N., 1989. Enrichment of tle phlogopite in subduction zone hybridization. Con- trace elements in garnet amphibolites from a paleo- trib. Mineral. Petrol., 79: 375-380. subduction zone: Catalina Schist, southern California. Yardley, B.W.D., 1986. Fluid migration and veining in Geochim. Cosmochim. Acta, 53:3155-3178. the Connemara Schists, Ireland. In: J.V. Walther and Stakes. D.S. and O'Neil, J.R., 1982. Mineralogy and sta- B.J. Wood (Editors), Fluid-Rock Interactions during ble isotope geochemistry of hydrothermally altered Metamorphism. Springer-Verlag, pp, 109-131.