| Journal of Petrology | Pages |
© 1999 Oxford University Press |
|
|
This paper discusses the nature and significance of genetic relationships among seven units that define a discrete intrusion in the coastal region of the Peninsular Ranges batholith (PRb), the largest and most primitive segment of the Cordillera in southwestern North America (Larsen, 1948; Gastil et al., 1975). Despite subsequent faulting and volcanism associated with rifting in the adjacent Gulf of California (Fig. 1), western and eastern extents of the PRb have been interpreted as the completely preserved roots of adjacent oceanic island and continental margin arcs, respectively (e.g. Gastil, 1983; Lipman, 1992). Regional mapping and analysis throughout southern California and northwestern Mexico over the past 50 years has determined that the PRb is bimodal and intruded during a sustained (~50 my) magmatic event in the Mesozoic (e.g. Gastil, 1983; Silver & Chappell, 1988). Particularly remarkable are the spatial variations of granitoid compositions from west to east across the batholith, which systematically correspond to regional changes of pre-batholith country rocks (e.g. Baird et al., 1974; Todd & Shaw, 1985; Ague & Brimhall, 1988). Correlated behaviour of chemical components and both radiogenic and stable isotopic values appears to be largely independent of rock type inside exposed plutons, perhaps correlative with a transition from continental to oceanic lithosphere sources during partial melting (Gromet & Silver, 1987; Silver & Chappell, 1988). The low 87Sr/86Sri and high [epsilon]Nd isotopic compositions of westerly granitoids certainly support an oceanic arc genesis (DePaolo, 1981; Davidson, 1987). Figure
However, the western PRb possesses observed and inferred characteristics not generally typical of oceanic plate margin magmatism. Regardless of well-documented features such as magnetite-ilmenite and `I-type'-`S-type' lines similar to those in Japan, the crust is currently continent thickness and it contains extremely voluminous quantities of calcic tonalite (Todd & Shaw, 1985; Gromet & Silver, 1987; Gastil et al., 1990). Representative compositions for intrusions in the western region are biased towards unusually silicic averages (>63 wt % SiO2) relative to intermediate compositions from modern subduction-related environments, and they also contain subtle incompatible element enrichments (Silver &Chappell, 1988; Fig. 1). In this regard, more detailed interpretations of the PRb have obvious relevance for theoretical models describing net lithospheric growth and evolution (e.g. Arculus, 1981; Yanagi & Yamashita, 1994). Current models for the batholith predict only that most granitoids formed by partial melting of mafic protoliths at depths of at least 1 GPa (>30 km; Gromet & Silver, 1987). Absence of consistent chemical and mineralogical trends among contemporaneous gabbros and granitoids apparently precludes any simple origin by crystal fractionation (Walawender & Smith, 1980; Gromet & Silver, 1987). No explanations currently account for the behaviour of enriched trace element in PRb tonalites, and so the process(es) responsible for generating voluminous granitoid magmas are entirely unknown in the absence of terrigenous protoliths (e.g. Singer et al., 1992; Devine, 1995). Detailed petrogenetic evaluations of the mafic and granitoid rocks (including the potential for mutually comagmatic relationships) in the PRb have not been attempted because of strong cumulate signatures and a lack of compositions amenable to modelling (Walawender & Smith, 1980). Our approach differs from previous investigations because it focuses on a dominantly (>85%) mafic complex that contains intrusive relationships not evident in contemporary granitoid plutons. By considering a bimodal system exposed at the westernmost extent of the batholith, we sought to evaluate the origin of isotopically primitive constituents and explain apparent discrepancies relative to modern oceanic settings. With new chemical and isotopic data, we establish links between the gabbroic cumulates and relatively evolved granitoids, using fine-grained rocks as magma analogues. Our results advocate a more complex model that involves mantle sources controlling intratelluric volatiles and, ultimately, fractionation in one or more mafic magma chambers located in thick arc basement (e.g. Gust & Perfit, 1987; Kersting & Arculus, 1994). Lack of correspondence between chemical trends for the gabbros and tonalites can be explained by mixing between two silica-saturated magma types (Perfit et al., 1980; Beard & Lofgren, 1991; Tate et al., 1997). Regional extrapolation of a dual-magma mixing hypothesis explains the compositions of voluminous tonalite throughout the western PRb, and may apply elsewhere around the Pacific rim after consideration of the mineral assemblages in contemporaneous gabbros. The PRb trends northwest and crops out for ~1600 km from the Transverse Ranges of southern California (USA) to the southern tip of Baja California in Mexico (Fig. 1). It comprises several hundred plutons with mutually intrusive units of tonalite-trondhjemite (73%), granodiorite (23%), granite (sensu stricto, 2%) and/or gabbro (2%) (Gastil et al., 1975). The country rocks change from Palaeozoic, amphibolite-facies metasedimentary rocks with migmatite in the east, to dominantly Jurassic-Cretaceous volcanic and volcaniclastic rocks at upper greenschist-facies grades in the west (e.g. Gastil, 1983). An intensely deformed but discontinuous, kilometre-scale zone separates these assemblages and subdivides the region into discrete eastern and western belts, and magmatism occurred on both sides of this tectonic discontinuity between 130 and 80 Ma (Gastil, 1983; Todd & Shaw, 1985; Thompson & Girty, 1994). Pluton diameters, compositions and opaque mineral assemblages vary from west to east across the batholith, apparently with rapidly decreasing gabbro abundance and increasing depth of exposure in the crust (Walawender & Smith, 1980; Gastil et al., 1990). Isopleths for every measured chemical variable lie parallel to the axis of the batholith, and the most mafic values occur at the western (coastal) limit of intrusive rocks (Fig. 1; Silver & Chappell, 1988). Radiogenic and O isotopes also show variations, with the most mantle-like values in the west and hiatuses that coincide approximately with the location of the inferred tectonic discontinuity (Gromet & Silver, 1987; Ague & Brimhall, 1988). Much of what is currently known about the western belt of the batholith comes from exposures north of the Mexico-USA international border (Fig. 1). Southwestern California contains relatively large (1-10 km diameter) granitoid plutons with either normal or reversed, concentric zonation (Gastil et al., 1990). The country rocks include basalt-andesite-dacite volcanic and pyroclastic rocks of the Santiago Peak Formation, with paragneisses assigned to the Julian Schist present in some areas. Most of the tonalite plutons discordantly intrude adjacent mafic bodies, and contact aureole assemblages reflect magma emplacement at ~0·5 GPa (>17 km) depths (Smith et al., 1983). Plutons always lack zonation and instead contain randomly distributed gabbroic units with gradational contacts and cryptic mineralogical variations (M. Walawender, personal communication, 1997). Adjacent areas of Baja California differ substantially because the mafic rocks have more variable textures, and isolated ring complexes containing cone-sheets and structurally concordant granitoid units accompany massive granitoid plutons (Silver & Chappell, 1988; Johnson et al., 1999). Country rocks belong only to the Alisitos Formation metavolcanic sequence, which may be an equivalent facies of the Santiago Peak Formation to the north (Allison, 1955; Silver et al., 1979). Investigation of the Zarza Intrusive Complex (ZIC; Fig. 1) described below avoids previous problems with contamination by pelitic crust (Walawender & Smith, 1980), and takes advantage of some of the most intricate spatial relationships preserved in the western belt. The ZIC is a small (~7 km2), ellipsoidal ring complex that intruded dominantly andesitic pyroclastic horizons among a typical portion of the Alisitos Formation succession (Fig. 1). Textural variations and internal intrusive contacts allow subdivision of the constituent rocks into two cone-sheet assemblages (notated CS1 and CS2), four gabbroic bosses (notated G1-G4), an irregular granitoid body (notated T1), and a series of radially oriented diabase, aplite and epidotite dykes (Johnson et al., 1999). Figure 2 shows the relative locations and areal extents of the largest units identified. Regardless of their location in the complex, individual cone-sheets range widely in width from 0·1 to 5 m, and consist of fine- to medium-grained, sparsely feldspar-phyric diabase, with elongate phenocrysts oriented parallel to the sharp contacts. Typical examples from each assemblage dip steeply inwards at ~70° towards common focal points, and are laterally discontinuous over 10-500 m. Centimetric, quartzo-feldspathic veins that presumably represent late-stage magma (e.g. Wagner et al., 1995) dissect many cone-sheets, whereas isolated country rock screens (1-5 m wide) separate adjacent sheets. Screens become wider and increasingly abundant outwards, and effectively define a gradational contact at the margins of the complex. Country rocks in the exocontact region are pyroxene-bearing hornfelses that contain hornblende porphyroblasts and intrusive leucosomes, both of which grade into regional greenschist-facies metabasites over ~100 m. Figure
Gabbros are always more massive and leucocratic than the cone-sheets, and entirely crop out inside arcuate envelopes defined by cone-sheet assemblages (Fig. 2). Each of the four units shares similar inequigranular textures and consistently plagioclase-rich mineral assemblages. Obviously cumulate features, such as centimetre-scale layers marked by higher plagioclase concentrations, occur rarely. Boundaries between adjacent units appear to be steep, gradational, and define arcuate or sub-circular intrusive patterns that lie sub-parallel to the margins of the entire complex. A concentric magmatic flow foliation is defined by a strong alignment of tabular plagioclase laths in all gabbro units except G4, and it transects gabbro unit boundaries obliquely at several localities (Fig. 2). A rarely exposed, metre-wide contact zone separating the tonalite from adjacent gabbros contains medium- to coarse-grained, tonalitic dykes with fine-grained (presumably chilled) margins. Internally, the most felsic constituent has a coarse grain size and is texturally and modally heterogeneous; it contains rare enclaves of rounded-subangular microtonalite and lacks any consistent magmatic foliation. Altered areas contain centimetre-scale hornblende oikocrysts and cross-cutting epidotite veins, whereas otherwise fresh granitoid contains ovoid epidote patches that appear to be poorly formed miarolitic cavities. No macroscopic feature or textural heterogeneity shows any systematic distribution that can be correlated with the proximity of adjacent units or the contact, at least at the current exposure level. Lithological and structural characteristics of the cone-sheets described here strikingly resemble those of type examples exposed in the British Tertiary Volcanic Province (Holland & Brown, 1972; Bell et al., 1994), and they deserve recognition as among the best-preserved examples documented in North America. Ring complexes have particular petrogenetic significance because the relative order of intrusion is generally obvious, and the fine-grained compositions are potentially magma analogues (e.g. Turner & Bowden, 1979; Bell et al., 1994). We believe that ring complexes also preserve petrogenetic relationships applicable to (but no longer evident in) more massive plutons, at least in regions containing contemporaneous intrusions of both types. The ZIC preserves the following important information: (1) each cone-sheet assemblage converges on a different intrusive focus and contains a unique mode; (2) gabbro units have similar modes and magmatic foliation orientations to those of surrounding cone-sheets, and locally contain abundant cone-sheet xenoliths; (3) magmatic fabric intensity decreases southwards in the entire complex because gabbro G4 and tonalite T1 contain the weakest foliations. On this basis, we infer that magmatism produced two cone-sheet centres punctuated by emplacement of gabbros, which were followed at some later time by penecontemporaneous intrusion of gabbro and a relatively silicic magma. For purposes of convenience hereafter, the ZIC has been subdivided into northern, central and southern intrusive centres with progressively younger inferred relative ages (Fig. 2). Absolute ages were determined by U-Pb (zircon) geochronology of units G3 and T1 in the central and southern intrusive centres, respectively (Fig. 2). Zircon was concentrated from crushed samples using well-established heavy liquid and paramagnetic techniques for heavy mineral separation and concentration. Representative aliquots of crystals were mounted in epoxy resin with chips of reference zircon, and prepared as polished thin sections. Cathodoluminescence images were used to identify internal structures in the sectioned grains before analysis with SHRIMP II at the Australian National University. Each U-Pb analysis consists of six scans through the mass range, and data were reduced using methods similar to those of
Compston et al., (1992) and
Williams & Claesson, (1987); isotopic ratio uncertainties were augmented using software written by T. R. Ireland (see
Muir et al., 1996). The Pb-U ratios have been normalized relative to the AS3 standard zircon, which has a 206Pb/238U value of 0·1859 and an age of 1099·1 Ma (Paces & Miller, 1993). Uncertainties given for individual ratios and ages are quoted at 1[sgr], whereas weighted mean ages are reported at 2[sgr]. Results for 47 analyses of two samples appear in Table 1, whereas Fig. 3 contains the corresponding
Tera & Wasserburg, (1972) concordia curves in conjunction with measured 207Pb-206Pb and calibrated 238U-206Pb ratios uncorrected for common Pb. Table 1. U-Pb isotopic data determined with SHRIMP II for ZIC units G3 gabbro and T1 tonalite
Figure
Gabbro in the central intrusive centre yielded small quantities of zircon fragments that have equant shapes, no colour, and minimal turbidity, with few pyramidal terminations. Cathodoluminescence images show extremely strong oscillatory zonation for U and Th (58-4949 ppm) that is typical for gabbroic rocks. All of the 23 areas analysed plot very close to concordia and are, therefore, dominated by radiogenic Pb. A weighted mean of 206Pb/238U ratios for all 23 analyses yields an age of ~116 Ma, with excess scatter [mean square of weighted deviates (MSWD) 2·7] that is mostly attributable to an older analysis 7.1 and a younger analysis 22.1 (Table 1; Fig. 3). The latter analysis comes from low-U growth zones that contain cracks suggestive of radiogenic Pb loss. A weighted mean for the remaining 21 analyses has no excess scatter and provides an age of 116·2 ± 0·9 Ma (MSWD 1·45). The southern intrusive centre granitoid provided much higher yields of equant-elongate zircon that has common bipyramidal terminations and more moderate oscillatory zonation (U, Th 36-1372 ppm). A few grains also contain rounded (presumably resorbed) and apparently unzoned cores that have lower but still variable concentrations of U and Th (77-888 ppm), despite similar 206Pb-238U ratios (0·00171-0·00187) relative to rims from the same grain. A combined age for 25 analyses of both cores and rims lies close to concordia, with a weighted mean 206Pb/238U age of 114·5 ± 0·9 Ma (MSWD 1·02); a weighted mean for the 15 cores alone provides an age of 114·3 ± 1·4 Ma (MSWD 0·54). Oscillatory zonation in the zircon population confirms a magmatic origin for the analysed grains, and we interpret data for both intrusive centres as recording magmatic crystallization events. Both ages overlap slightly at the 95% confidence limits, and so the collective ZIC has an Early Cretaceous (Aptian) age of 115 ± 1 Ma. This determination should be treated as a minimum estimate of emplacement because the most northerly (and potentially the oldest) exposed units were not dated (Fig. 2). However, the small diameter of the complex as a whole appears to be consistent with rapid magma emplacement, and any periodicity reflected by repetitive emplacement of cone-sheets and massive gabbroic units probably occurred on a smaller scale than our analytical uncertainties. Compared with the range of U-Pb dates published for western belt PRb intrusions in general (130-105 Ma; Silver et al., 1979), the ZIC lies close to the median and overlaps with concordant 116 ± 2 Ma ages determined from multiple-grain zircon fractions in lavas at the base of the Alisitos Formation type section (D. L. Kimbrough, personal communication, 1997; Fig. 1; Allison, 1955). Both determinations agree with Aptian-Albian fossil ages determined from intra-formational limestones throughout northwestern Baja California, potentially suggesting that mafic components of the ZIC intruded comagmatic volcanic ejecta. Indistinguishable ages determined for inherited zircon cores in unit T1 potentially also implicate Alisitos Formation involvement during its genesis, and we will explore partial melting of Alisitos metabasite in a subsequent section. Holocrystalline mafic and intermediate rocks with a substantial variety of grain sizes, textures and modal mineral assemblages characterize the ZIC. Tables 2 and 3 summarize the lithological and modal characteristics of all seven intrusive units, and Table 4 contains representative averages of their mineral compositions; nomenclature throughout follows the modal IUGS scheme of
Streckeisen, (1976). In general, the gabbros have coarse grain sizes (2·0-6·0 mm), seriate textures, and either noritic anorthosite or hornblende anorthositic gabbro compositions. The cone-sheets have much finer grain sizes (0·3-1·5 mm) in conjunction with noritic or hornblende-bearing leuco-diabase compositions that overlap entirely with those of immediately adjacent gabbros (Fig. 1). The granitoid samples also range widely in grain size (2·0-5·0 mm) and share some modal and mineralogical features with all other rock types in the complex. Detailed summaries of textural and mineralogical characteristics for each of the seven intrusive units appear below, using cumulate terminology in the non-genetic sense advocated by
Irvine, (1982). Mineral analyses were obtained with the Cameca SX-50 scanning electron microprobe at Macquarie University, which was operated in wavelength-dispersive mode with a 20 nA beam current, 15 kV accelerating voltage, 2-5 µm spot diameter and an integrated counting time of 40 s. Repeated determinations of internal geological standards analysed with the samples show that the results are better than ±5% for MnO and Na2O, and better than ±2% for all other oxides. Table 2. Lithological characterisitics and areal abundances of constituent ZIC units
Table 3. Modal mineralogical characteristics of constituent ZIC units
Table 4. Representative average mineral compositions and magmatic intensive variables for units from the ZIC
All gabbro units in the northern and central intrusive centres have pronounced orthocumulate textures. They contain elongate plagioclase (38-64 vol. %; An91-85) and hypidiomorphic augite (0·2-26%; Wo47En39Fs13) phenocrysts in an inequigranular groundmass of plagioclase (16-27%; An84-46) subophitically enclosed by oikocrysts of green-brown hornblende {<21%; mg-number [Mg/(Mg + Fe)] 41-64} and virtually colourless hypersthene (<12%; En64). Equant magnetite, minor quartz (usually <5%), and rare blebs of chalcopyrite occur interstitially with rare apatite and accessory zircon, and most samples contain little (<3%) deuteric alteration. Unit G1 differs slightly because it contains phenocrysts of both pyroxenes, in conjunction with some magnetite oikocrysts in the groundmass. Units G2 and G3 also contain minor (<2%) primocrysts of allotriomorphic and unzoned olivine (Fo54 and Fo72, respectively), surrounded by kelyphitic symplectites of what is now bronzite (En~70) and serpentine. Gabbro G4, from the southern intrusive centre, is a mesocumulate that primarily contains more sodic plagioclase (<83%; An61-31) intergrown with sub-poikilitic hornblende (mg-number 48-70). Panidiomorphic hornblende also occurs enclosing unzoned kernels of relatively iron-rich clinopyroxene and equant magnetite in some samples. A unique feature of G4 is the relatively abundant saussuritization of plagioclase zones and the presence of fibrous tremolite-actinolite rims surrounding hornblende ( <= 8%). Pervasive alteration generally corresponds to the coarsest and most feldspathic samples. The most abundant cone-sheets in unit CS1 resemble gabbros from the northern and central intrusive centres in terms of both modal constituents and mineral compositions. Individual sheets have intergranular textures and contain oscillatory zoned plagioclase (<79%; An90-45)and unzoned augite (<35%; Wo48En38Fs14) sub-poikilitically enclosed by hypersthene (0·2-12%; En69) and minor hornblende (0·1-30%; mg-number 65-50); titano-magnetite occurs as both equant and dendritic crystalssurrounded by interstitial quartz (~3%). Much rarer examples have quench textures with compositionally similar microlites (~0·05 mm) of cumulophyric augite (19-42%), hypersthene (0·2-1%), plagioclase (55-77%) and relatively magnesian olivine (trace; Fo75). Cone-sheets from the central intrusive centre resemble gabbro unit G4 most closely because they contain green-brown hornblende (<15%; mg-number 68-71), and minor allotriomorphic clinopyroxene (>4%; Wo47En40Fs13) with a variety of habits as the sole ferromagnesian mineral phases. Plagioclase phenocrysts (<2%; An43-27) have less calcic compositions than equivalents in unit CS1. They compositionally resemble groundmass crystals (46-81%) instead, which occur intergrown with dendritic titano-magnetite (~11%), relatively abundant quartz (<16%), and apatite crystals <= 1 mm long. The late-stage radial dykes have felty groundmass textures typical of chilled diabases, but otherwise share similar mineral assemblages, compositions, and mild saussuritization and sericitization with all of the CS2 cone-sheets sampled. Unit T1 has a medium-coarse (1-5 mm) grain size in conjunction with considerable textural, modal and mineralogical variety (e.g. `M' 3-16%). Most of the samples are tonalites that contain hiatal and oscillatory zoned plagioclase phenocrysts (36-47%; An61-24) distributed randomly in an equigranular groundmass containing green-brown hornblende (6-10%; mg-number 65-34), annitic biotite (<8%; mg-number 36), patchily zoned plagioclase (>17%; An47-19), quartz (<36%) and equant magnetite (0·3-4%). Plagioclase phenocrysts contain panidiomorphic hornblende and plagioclase microcrysts (<1 mm) in conjunction with equant magnetite inclusions, and have saussuritized zones that presumably correspond to regions of particularly calcic growth. Altered kernels of clinopyroxene with a fibrous appearance (<0·6%) also occur as inclusions in some groundmass hornblende crystals. More felsic samples lack phenocrysts, are classified as leuco-tonalites (trondhjemites), and consist dominantly of consertal, patchily zoned plagioclase (<73%; An<45), annitic biotite and hornblende (0·1-9%), interstitial quartz (<35%), and traces of magnetite and K-feldspar (<0·5%). Both of the textural varieties contain accessory zircon, titanite and apatite as rare inclusions in the groundmass minerals, and subcircular patches of sub-grained quartz, ~1·5 mm in diameter, also occur in some samples. These features appear to be recrystallized and perhaps resorbed magmatic phenocrysts, rather than xenocrysts derived externally from the country rocks or alteration-related features formed by replacement. Approximately 10 kg samples representative of textural and modal variation throughout the ZIC were collected during field seasons in 1995 (21 `bc' samples) and 1997 (24 `ZP' samples). Sampling of the coarse-grained rocks specifically avoided obvious enclaves, alteration and/or weathering. Cone-sheet coverage always focused on the freshest material with the lowest possible concentrations of phenocrysts and late-stage veins, most of which were too narrow and discontinuous to sample discretely. All of the samples were pulverized to <200 µm in C-steel (`ZP' only) or WC (`bc' only) shatter boxes, and large (~100 g) aliquots of powder were homogenized by inversion before analysis. Major and trace element chemical compositions of `bc' samples were determined with an automated Rigaku 3070 wavelength-dispersive X-ray fluorescence (XRF) spectrometer using fused discs and pressed powder briquettes (respectively) at the University of Southern California. Operational procedures were modified after
Norrish & Hutton, (1969), with LOI (loss on ignition) determined gravimetrically. Major elements only in `ZP' samples were analysed using fused discs in a Siemens SRS-1 XRF spectrometer at Macquarie University, using the methods of
Norrish & Chappell, (1967); ferrous iron, H2O+, H2O- and CO2 concentrations were determined by conventional titration and gravimetric adsorption, respectively. Duplicate analyses of the same international standards in both laboratories (Govindaraju, 1989) agree to better than ±1% for oxides, Rb, Sr, Zr and Cr, ±2% for Ba, Y, Nb, Zn and Cu, and ±5-10% for Ga, Pb, Ni, Th and U. For `ZP' samples alone, trace and rare earth elements (REE) were determined in duplicate with a Perkin Elmer Elan 6000 solution-source inductively coupled plasma mass spectrometer at Macquarie University using closed-beaker dissolution procedures outlined by
Norman et al., (1998). Trace element analyses were calibrated against the well-characterized BHVO-1 standard using values from
Eggins et al., (1997) with duplicates showing 1[sgr] precision better than 5% for all elements, except for those with Z < 85 (Rb-U) at <2%, over the range of concentrations encountered. Also for `ZP' samples, isotopic compositions of Rb-Sr and Sm-Nd were determined in automatic mode on the triple Re filament assembly of a VG Micromass 54E TIMS at the Australian Research Council-CSIRO Centre for Isotopic Studies in Sydney; Korsch & Gulson, (1986) have described the open beaker sample dissolution, column chemistry and mixed Sm-Nd spiking procedures employed. Values of 87Sr/86Sr and 143Nd/144Nd were normalized to 86Sr/87Sr = 0·1190 and 146Nd/144Nd = 0·7219, respectively, and repeated standard analyses recommend 2[sgr] uncertainties of ±3% or better. Duplicate sample powders were not analysed in both XRF laboratories, and so we cannot demonstrate internal consistency among the elemental analyses. Consequently, our interpretations below rely on `ZP' sample information, recognizing that inclusion of all compiled data in some of the figures changes the data density much more than the sample distribution. Table 5 contains chemical and Sr-Nd isotopic analyses for the ZIC, and focuses on samples with the largest number of analytes. Table 5. Representative whole-rock chemical, CIPW-normative, and Sr-Nd isotopic compositions for all ZIC units
Coarse-grained gabbros (44·5-52·7 wt % SiO2) represent the most mafic rocks currently exposed in the ZIC. Despite somewhat variable compositions, they characteristically contain high Al2O3 (16·6-28·6%) and CaO (10·8-14·9%) concentrations, in conjunction with moderate to low MgO (0·4-8·0%) and extremely low K2O (<0·4%). Cross-Iddings-Pirsson-Washington (CIPW)-normative mineral assemblages were calculated using a ferric-ferrous iron ratio of 0·15 to potentially alleviate effects related to post-emplacement oxidation (Brooks, 1976; Table 5). Most units contain normative olivine (Ol 2·2-16·8%) and high concentrations of normative anorthite (An 37-63%) relative to hypersthene (Hy 0·8-13·3%), except for some G1 samples that are weakly saturated with respect to silica (Qtz < 3%). Strontium (243-463 ppm) also exists at high concentrations relative to other large-ion lithophile elements (LILE; notably Ba 23-91 ppm and Rb 1-8 ppm), which have abundances similar to most of the high field strength elements (HFSE; e.g. Zr 10-58 ppm, and Th and Nb <8 ppm). Individual gabbro compositions overlap substantially and define scattered patterns on variation diagrams involving major and trace elements (Figs 4 and 5). The chondrite-normalized (CN) REE abundances shown in Fig. 6a and b are more consistent and dominated by large, positive Eu anomalies (Eu/Eu* 1·1-2·2), smaller negative Ce anomalies (Ce/Ce* ~0·6), and variable but modest fractionation of light REE (LREE) relative to heavy REE (HREE) (La/LuCN 0·8-1·8). The 87Sr/86Sri (0·7032-0·7034) and [epsilon]Nd values (+7·0 to +7·5) are similar and primitive for all units (Fig. 7). Figure
Figure
Figure
Figure
Cone-sheets from both the northern and central intrusive centres can be subdivided into two groups using only silica content and texture. Mafic samples containing <55% SiO2 have the coarsest grain sizes, in conjunction with aluminous and extremely calcic compositions that plot with the gabbros for all elements except TiO2 and (to a smaller extent) Cr (Figs 4a and 5f). They also contain variably high concentrations of most transition series metals (e.g. Ni, Sc 25-35 ppm), low concentrations of REE [sum of 13 elements ([Sigma]13REE) 16-29], and characteristically gabbroic REE patterns with pronounced positive Eu anomalies (Eu/Eu* 1·1-1·5; Fig. 6a and b). Intermediate sheets with SiO2 > 55% have more consistent compositions that define a diffuse cluster with sample ZP-16 from the radial dykes (Figs 4 and 5). Collectively, these rocks contain similar Al2O3-CaOratios to those of the gabbros (1·3-2·0), albeit with much lower abundances of all major elements except Na2O (4-5·6%). In contrast, Zr, Nb, Th, Ta, Y, REE and (to some extent) Hf exist at much higher concentrations (e.g. Zr 71-150 ppm; [Sigma]13REE 52-78; Table 5). Chondrite-normalized REE patterns are dominantly flat and unfractionated (La/LuCN < 1·6) with ubiquitous depletion of Ce, but lacking any discernible Eu anomaly (Fig. 6). No consistent chemical or isotopic differences distinguish samples from the two intrusive centres, which both show relatively wide and overlapping ranges of 87Sr/86Sri values (0·7033-0·7039) that plot at the right of the mantle array with [epsilon]Nd values equivalent to those in the gabbros (+7·0 to +7·4; Fig. 7). The tonalite-trondhjemite unit has the most evolved compositions (SiO2 63·9-74·2%) and the most diverse compositional range of any ZIC unit. All of the rocks collected contain variable concentrations of silica and most major elements, particularly MgO (0·5-1·5%; Fig. 4d), regardless of modal classification. Except for the alkalis and Y (Figs 4f and Fig. 5c), elements define relatively coherent trends that project through the intermediate cone-sheet compositions before ultimately intersecting the coarse-grained gabbro field (e.g. Figs 4a-c and 5a-c). Despite scatter for mobile elements, values for (for example) Na2O/K2O (17·6-19·3) and K/Rb (290-935) consistently overlap with those calculated for more mafic units. Tonalites have exclusively peraluminous (A/CNK 1·01-1·13) compositions and ~0·3% normative corundum, whereas the trondhjemites have metaluminous A/CNK values of 0·98-1·01; these differences reflect diagnostic hornblende compositions and hornblende-biotite ratios (Tables 3 and 4). Non-porphyritic trondhjemites also have variably elevated concentrations of K2O (2·4-4·2%), Ba (449-583 ppm), Rb (52-82 ppm), Th (7-25 ppm) and, in some rocks, REE ([Sigma]13REE 45-151), relative to tonalites with similar silica contents (e.g. Figs 5b-d and 6; Table 5). All of the patterns show relative middle REE (MREE) depletion (La/SmCN 1·7-2·6), albeit with shapes and Eu anomalies that do not vary systematically with silica (Fig. 6). Onetonalite sample also contains low [epsilon]Nd and high 87Sr/86Sri (+6·5 and 0·7036, respectively) relative to a trondhjemitic sample nearby (+7·1 and 0·7022; Fig. 7). Chemical and CIPW-normative compositions of the gabbros described above typify those of sub-alkaline basalts belonging to the high-Al (and relatively low-Mg) tholeiitic magma series, a significant component of early magmatic activity in oceanic island arcs (Fig. 4f; Gill, 1981; Brophy & Marsh, 1986; Crawford et al., 1987). As is expected of arc tholeiites, all of the coarse-grained samples define a weak trend of moderate Fe enrichment (Fig. 8a), and they characteristically have high and variable FeO-MgOratios (Fig. 8b). Relative to modern orogenic lavas with equivalent SiO2 contents, the ZIC gabbros are skewed towards the most primitive end of the high-Al basaltic spectrum (Basaltic Volcanism Study Project, 1981), and they have extremely pristine Sr-Nd isotope characteristics that preclude obvious involvement of continental crust (Fig. 7; Davidson, 1987). However, whole-rock compositions of each unit should not be considered as reliable indicators of tectonic environment without prior investigation of the differentiation processes involved, particularly given their coarse grain sizes and variable mineral compositions (Table 4). Some samples show compositional characteristics atypical of lavas in oceanic arcs (particularly TiO2 <= 2·2%; Fig. 4a), and they have concentrations of other elements more appropriate for cumulates than liquids (e.g. V < 490 ppm; Fig. 5e). Considering the dominance of ortho- and meso-cumulate textures, cause(s) of compositional variations need not reflect a simple liquid-line-of-descent, even under closed-system conditions. Figure
Walawender & Smith, (1980) and
Smith et al., (1983) recognized the high-Al compositions of PRb gabbros in southern California, and distinguished discrete series using textural and modal criteria: (1) olivine- and two-pyroxene-bearing gabbros with interstitial opaque minerals, assigned to the olivine-clinopyroxene subseries; (2) two-pyroxene gabbros containing prismatic hypersthene that belong to the amphibole-orthopyroxene subseries; (3) amphibole-series gabbros that contain subequal proportions of hornblende and plagioclase (bojites in the IUGS scheme). They interpreted each series as having accumulated from compositionally similar but heterogeneous magmas differentiating close to the ~0·5 GPa (~17 km) depths now exposed in each intrusion. Extrapolation from type intrusions (Fig. 1) to the ZIC shows that: (1) unit G1 belongs to the amphibole-orthopyroxene subseries; (2) units G2 and G3 represent the olivine-clinopyroxene subseries; (3) unit G4 has mostly amphibole gabbro series affinities (Table 2). Reaction between olivine and liquid to form hypersthene rather than pigeonite can be explained if ZIC magmas were primitive (refuted below) and/or emplaced shallowly relative to counterparts from the USA (e.g. Grove & Baker, 1984; Ariskin et al., 1993). Preservation of cone-sheets and the existence of miarolitic cavities in the tonalite are consistent with subvolcanic emplacement, an inference supported by barometric estimates of <0·23 GPa (~8 km) based on Al in hornblende (Johnson et al., 1999) and in hypersthene (Longhi et al., 1993; Holland & Blundy, 1994; Table 4). Gabbros in the ZIC all contain overlapping modal proportions (Table 3) in conjunction with similar ranges of mineral compositions (Table 4), which initially suggest a comagmatic relationship between the four units (as their similar inferred ages and close spatial relationships also suggest). Normal zonation patterns in augite, hornblende and plagioclase from all of the gabbroic samples argue for relatively simple nucleation histories without the influence of open-system processes such as contamination or magma mixing (e.g. Kersting & Arculus, 1994; Wagner et al., 1995), and their similar and primitive Sr-Nd isotopic compositions permit this conclusion (Fig. 7). Consistently low H2O+ and CO2 concentrations (<0·9%), and lack of any obvious correlations between volatile and other elemental abundances or isotopic characteristics, also suggest that the hydrous minerals formed mainly from late-stage magma under deuteric conditions. Absence of chemical trends defined by samples from the same unit is best explained by effectively random crystal accumulation. Consequently, the major and trace element compositions of each gabbro generally cluster together within polygons that have influential mineral compositions at their apices (Figs 4 and 5). Apart from high TiO2 and V that probably reflect abundant Fe-Ti oxides in unit G3 (Figs 4a and 5e; Table 4), relative proportions of the silicate minerals apparently determined most compositions. Particularly prominent are the anorthositic modes with non-cotectic proportions of feldspar relative to ferromagnesian constituents (Table 3). Positive Eu anomalies always characterize plagioclase-cumulates under conditions sufficiently reducing to stabilize divalent Eu cations, and ZIC gabbros are no exception (Fig. 6a and b). Relatively strong, positive correlations also always involve chemical constituents and plagioclase (Fig. 9), and so particularly feldspathic samples contain elevated concentrations of Al2O3, CaO and Sr, the most fractionated REE abundances, and the lowest concentrations of incompatible elements such as Rb, Y and Zr (Gromet & Silver, 1987). Volumetrically minor olivine has little discernible effect on whole-rock compositions (Figs 4c and d, and 5f), whereas the pyroxenes potentially account for variable FeO, Cr and Sc abundances in conjunction with low whole-rock CaO/Sr values (Table 3). Prior crystallization of pyroxene could also explain the less calcic plagioclase in unit G4, variably Fe-rich olivine throughout units G2 and G3, and the absence of olivine from G1 samples (Barnes, 1986). Wide but overlapping ranges of cumulus mineral compositions in the different rock types initially support the previous suggestion of compositionally similar, heterogeneous parents for each series (Smith et al., 1983). However, cumulates reflect the compositions of primocrysts combined with an unknown proportion of interstitial magma in closed compositional systems (e.g. McBirney & Hunter, 1995). Consequently, the compositions of ZIC units cannot be used to reliably infer parental characteristics until after the effects of late-stage processes have been isolated and considered. Figure
Peritectic reaction jackets around olivine, markedly different olivine compositions between units, lack of detectable zonation inside the most mafic minerals (olivine and orthopyroxene), and the subsolidus temperatures estimated with two-pyroxene thermometry (Table 4), all suggest some degree of re-equilibration with late-stage magma. Incompatible element concentrations vary by up to one order of magnitude between samples from the same unit (e.g. P2O5 0·02-0·11 wt %, Y 7-17 ppm and Zr 10-54 ppm; Table 5), which presumably represents mobilization of residual liquids on a scale larger than our samples are representative of the cumulate pile. Somewhat scattered alkalis (Fig. 4f) and variable FeOT-MgO ratios (Fig. 8b) may also reflect late-stage interactions. However, the absolute chemical effects attributable solely to primary porosity are indeterminate because the proportion of entirely interstitial crystals is impossible to estimate texturally in these rocks (e.g. Miller & Wieblen, 1990). Lack of obvious correspondence between mineralogical variables such as Fo-number in olivine and the An content of plagioclase indicates to us that trapped liquid effects are not necessarily profound in each sample (Chalokwu & Grant, 1987). Recognizing that ratios of incompatible elements such as Ba/Rb (14-44), Th/Yb (0·1-0·7) and Zr/Hf (32-101) do not reliably distinguish any individual unit, we suggest that variable contributions from trapped liquid exist but do not account for individuality among the olivine-pyroxene- and amphibole-bearing magma series. In summary, the compositions of gabbroic units in the ZIC primarily reflect variable proportions of plagioclase primocrysts relative to ferromagnesian constituents. Reaction textures and lack of ferromagnesian mineral zonation in some samples probably reflect only minor re-equilibration between cumulus minerals and interstitial liquid. On this basis, mineralogical and whole-rock chemical similarities between members of the olivine- and hornblende-bearing series do not rule out previous suggestions of similar parents. On the basis of cumulate gabbro data we cannot comment on the likelihood of heterogeneous magmas, but realize that characteristic plagioclase compositions in G4 (Table 4) are likely to reflect different magmatic intensive variables relative to olivine-bearing equivalents. Recognizable intrusive contacts imply that each unit accumulated from a discrete magma batch in a chamber close to the current exposure level. Common magmatic foliation orientations in such a complex intrusion, and the general absence of modal or cryptic stratification on all scales, could also reflect emplacement of each unit as a crystalline suspension that differentiated during ascent and after emplacement, a model applied successfully to lunar and terrestrial anorthosites (e.g. Miller & Wieblen, 1990). Perhaps more importantly, relatively shallow intrusive depths exposed in the ZIC imply that associated eruptions may have generated local units in the Alisitos Formation and its stratigraphic equivalents (Walawender & Smith, 1980). Similar relationships may also have existed in other regions of the western PRb. Gabbro and cone-sheet relationships Minerals in the gabbros have compositions that typify the products of magmatism beneath modern arcs where cumulates occur (e.g. Erikson, 1977; Perfit et al., 1980; Beard, 1986). In conjunction with subordinate Or components in all of the plagioclase analyses, low alkali and TiO2 contents in hornblende and pyroxenes (
Table 4) are compatible with a tholeiitic or primitive calc-alkaline basalt with Na2O > K2O and low Ba, Rb and U concentrations (Erikson, 1977; Arculus & Wills, 1981; Sisson & Grove, 1993). At least in the ZIC, trace element information can be retrieved approximately by attempting to compensate for the effect of cumulus mineral compositions on gabbro whole-rock data. Assuming a precursor crystallizing plagioclase and pyroxenes in respective upper-crustal, cotectic proportions of 65% and 35%, Fig. 9 shows that the parent would have contained ~320-350 ppm Sr at 16-19% Al2O3 if plagioclase accumulated alone (Grove & Baker, 1984; Wagner et al., 1995). Extrapolation of this procedure to other elements strongly compatible or incompatible in plagioclase (not shown) appears to be consistent with an andesite containing 11·5-12·5% CaO, 50-70 ppm Ba, 1·5-2·5 ppm Rb, 0·05-0·3 ppm Th and U, 15-35 ppm Y, 30-60 ppm Cr and Sc and 12-25 ppm Ni. Allowing for the fact that cumulus pyroxene will reduce transition series metal estimates slightly, and that hydrothermal alteration potentially accounts for some scatter among labile LILE, these results strikingly resemble those of
Walawender & Smith, (1980), who calculated distribution coefficients from XRF analyses of mineral separates. We assume that this `accumulation-corrected' parent has at least partial geological relevance because it classifies as a basaltic andesite in the scheme of
Le Maitre et al., (1989), and consistently matches high-Al compositions that
Crawford et al., (1987) considered to represent differentiated liquids. Gabbroic samples with the lowest Al2O3, CaO and Al2O3-CaO ratios (presumably those with the smallest quantities of cumulus minerals) in the ZIC also approximate a basaltic andesite (Fig. 8b; Le Maitre et al., 1989; Bartels et al., 1991). Another compelling relationship is almost complete correspondence of the most siliceous (>55% SiO2) cone-sheets from both the northern and central intrusive centres, suggesting that they may be representatives of this potentially parental material (notated `PM' in Figs 4 and 5). Figure 10 (a and b) contains mid-ocean ridge basalt (MORB)-normalized and CN spiderdiagram patterns for trace elements in the gabbros, the average northern intrusive centre cone-sheet, and a plausible parent calculated independently using the equilibrium distribution method that
Bédard, (1994) described in detail. These supposed parents overlap with modern andesites and match the average cone-sheet pattern with low residuals for all immobile elements ([Sigma]R2 < 0·35), assuming 30-40% trapped melt (as seems likely). Such high intercumulus liquid concentrations typify orthocumulate anorthosites and probably account for the absence of intra-crystalline deformation attributable to foliation development in the ZIC (Miller & Wieblen, 1990; Johnson et al., 1999). Figure
Distribution of the data in Figs 4 and 5 shows that mineral compositions used to define the shaded polygons readily accommodate our calculated parent at the most silica-rich apex while enclosing almost all of the gabbroic samples. Perhaps more significant is the observation that some cone-sheet samples approach andesitic melt compositions, which are described only uncommonly in arcs (e.g. Brophy, 1989a; Bartels et al., 1991; Romick et al., 1992). Quenched sheets, in particular, have extremely homogeneous textures in conjunction with low concentrations of phenocrysts (<2%) and near-cotectic modes (Fig. 9; Bell et al., 1994). Their feldspar phenocryst compositions overlap with groundmass grains in the same thin section, and all constituent silicate minerals except olivine are indistinguishable relative to counterparts in associated gabbros (Luhr, 1992; Table 4). Even the olivine compositions match consistently after an appropriate correction for interaction with the 30-40% intercumulus liquid estimated above (Chalokwu & Grant, 1987). Most of the siliceous cone-sheet group cannot be explained as residual liquids produced by extensive plagioclase accumulation in the gabbros because of their relatively low FeOT contents, high Sr, and unfractionated REE patterns (Fig. 6). Their characteristics should be more reliable than most of the published estimates for high-Al magmas, which typically contain coarse grains and/or show chemical heterogeneities suggestive of differential movement between coexisting phenocrysts and melt (Crawford et al., 1987). Diversification in the northern intrusive centre Atmospheric experiments dictate that high-Al basalts and andesites with at least 1% H2O are multiply saturated with (in relative order of appearance) olivine, calcic plagioclase, augite and titano-magnetite, over much of the magmatic temperature range at subvolcanic depths (Grove & Baker, 1984; Gust & Perfit, 1987; Sisson & Grove, 1993). Under conditions surrounding the Ni-NiO (NNO) buffer, the olivine phase field contracts at depths shallower than 0·5 GPa (~17 km), which causes reaction between olivine and coexisting liquid to form orthopyroxene (Huebner & Sato, 1970; Draper & Johnston, 1992). Some of the CS1 cone-sheets project on volatile-undersaturated, olivine-plagioclase cotectics near the augite piercing point at atmospheric pressure (designated `PP' at 0·1 GPa in Fig. 11a), a relationship which advocates accumulation of northern gabbros from spatially associated cone-sheet magmas, and supports augite removal as an explanation for differences between units G2 and G3. Modestly increased aH2O levels inhibit olivine stability further while promoting plagioclase crystallization at higher An contents (Longhi et al., 1993; Sisson & Grove, 1993). Thus, absence of olivine in unit G2, and variable displacement of the two remaining cone-sheets towards the plagioclase phase field (Fig. 11a), may be explained by different volatile contents during magma emplacement (Baker & Eggler, 1987; Sisson & Grove, 1993). All of the CS2 compositions lack prerequisite mineral assemblages for this approach, and their compositions will be discussed separately below. Figure
Assuming that individual cone-sheets differentiated under fO2 conditions analogous to those in northern intrusive centre gabbros, their dominantly unfractionated REE profiles cannot reflect closed-system magmatic evolution in equilibrium with plagioclase at very shallow depths (Fig. 6a and b). However, analyses of the unzoned olivines in Table 4 are more Fe rich than upper-mantle counterparts (Fo>88), and they would coexist in equilibrium with an evolved (mg-number 49-54; ~1000°C) basalt at shallow depths (Roeder & Emslie, 1970). Transition series element concentrations also lie below those expected for primary upper-mantle melts (Ni 150 ppm; Cr 300-500 ppm), and so moderate amounts of ferromagnesian mineral fractionation appear to be necessary unless the source itself contained unusual characteristics (Gust & Perfit, 1987). Olivine-plagioclase-pyroxene assemblages remain stable in H2O-undersaturated, aluminous basalts at pressures much greater than 0·5 GPa, but increasing depth and aH2O progressively shrink the phase fields for olivine and plagioclase (Grove & Baker, 1984; Baker & Eggler, 1987). Consequently, wet magmas should be more basic with dominant pyroxene phenocrysts at pressures <= 1·5 GPa (~50 km) during ascent (Fig. 11a; Bartels et al., 1991; Draper & Johnston, 1992). Removal of an augite-rich assemblage from high-Mg basalt at 0·5-1·5 GPa depths is consistent with the cone-sheet compositions if intratelluric water mediates any reduction of Al2O3 LREE and Eu2+ concentrations by plagioclase fractionation (Baker & Eggler, 1987; Sisson & Grove, 1993). Low Cr and Sc concentrations in the cone-sheets can accommodate extensive augite fractionation, low Ni also suggests removal of at least some olivine, and relatively high CaO and V contents implicate some combination of hypersthene and augite (Gust & Perfit, 1987). However, the scattered cluster of major and trace element contents (Figs 4 and 5), including the variable total REE contents (Table 5) and variable 87Sr/86Sri relative to [epsilon]Nd in Fig. 7, are unlikely to be artefacts of closed-system evolution. The complete lack of statistically significant correlations between Sr isotopic ratios and either silica or common Sr (r < -0·4) obviates extensive involvement of sediment unless it resembled the cone-sheets isotopically (Davidson, 1987). Recently published petrogenetic models for andesites instead focus on open-system fractionation in both deep and shallow magma chambers (Grove & Baker, 1984; Bartels et al., 1991; Devine, 1995). Kersting & Arculus, (1994) described a likely situation where relatively magnesian basalts fractionated high-Al products in a high-flux magma chamber located close to the Moho. Applied to the ZIC, recharge towards the base of thick arc crust at ~0·8 GPa (~28 km depth) favours nucleation of olivine and pyroxenes relative to plagioclase, which provides an alternative explanation for the lack of obvious plagioclase involvement (Fig. 11a). The fact that xenoliths and xenocrysts from greater depths were not observed in any cone-sheet may also reflect dissolution in response to a high-temperature, recharged regime (Baker & Eggler, 1987; Gust & Perfit, 1987). Source components and the amphibole gabbro series Appropriate sources for calc-alkaline basalts and andesites have been extensively debated over the last decade, as summarized recently by
Hawkesworth et al., (1993) and
Pearce & Peate, (1995). Two main end-member scenarios exist, namely: (1) extensive melting of oceanic lithosphere metamorphosed to quartz eclogitic assemblages during subduction; (2) modest partial melting of spinel lherzolite at shallower depths in the mantle wedge overlying a subducted slab, followed by extensive fractionation (e.g. Pearce, 1983; Davidson, 1987; Thirlwall et al., 1994). In the case of parents for the olivine-bearing cone-sheets, almost complete melting must have consumed residual garnet to produce primary magmas with flat REE profiles (La/LuCN < 2), which probably necessitates involvement of pelagic sediment or some type of enriched mantle to elevate abundances of most incompatible elements [Brophy & Marsh, (1986) and references therein]. Lack of REE fractionation and [epsilon]Nd values >+7 argue for the second alternative in the northern intrusive centre (Fig. 6; Crawford et al., 1987; Brophy, 1989a). The most substantial argument against an upper-mantle source relates to the absence of olivine from the 0·8 GPa liquidus of a high-Mg basalt (Draper & Johnston, 1992). More recent experiments have shown that equilibrium with upper-mantle peridotite depends critically on MgO-FeO ratios, H2O saturation and oxidation state (Bartels et al., 1991; Sisson & Grove, 1993; Wagner et al., 1995). Such factors vary widely in natural liquids and are entirely unconstrained beneath the ZIC. The northern intrusive centre cone-sheets preserve no direct evidence for deep-level fractionation, and so the early magmatic history of suitable parents must be inferred qualitatively. We assume necessarily that fractionation of a mixed pyroxene-plagioclase assemblage cannot significantly change element profiles on normalized plots with logarithmic scales (Pearce, 1983). Figure 10a shows that the cone-sheets contain approximately MORB-like levels of elements between P and Yb, in conjunction with pronounced relative depletion at Ta and Nb and 1·5-8 times enrichments of Ce and most incompatible elements between Sr and Th. Crystal fractionation and/or accumulation should not account for peaks and troughs because each of the four gabbro units also plot with similar shapes on this diagram; only elevated levels of Sr and Ti can be attributed to multiple saturation of plagioclase, pyroxene and perhaps Fe-Ti oxides during gabbro accumulation (Grove & Baker, 1984; Gust & Perfit, 1987). In conjunction with modest relative fractionation of individual HFSE, these apparently contradictory characteristics typify oceanic magmas derived from a MORB-source metasomatized by incompatible elements and some radiogenic Sr (e.g. Pearce, 1983; Ewart & Hawksworth, 1987; Figs 7 and 10a). The relative degrees of partial melting, metasomatism and fractionation required depend critically on the depth of equilibration between parental melt and the upper-mantle peridotite, which is impossible to constrain using currently exposed ZIC units (Kelemen et al., 1990; Pearce & Peate, 1995). Origins for multiple components present in the source during melting can be approximated because trace element ratios only change rapidly at small melt fractions not obviously applicable to the CS1 cone-sheet compositions [e.g. MORB-normalized (MN) Zr and Hf 0·8-1·3MN; Gd-Lu >17CN]. Values of Zr/Yb at 25-30, Zr/Hf at 45-56 and La/Nb > 0·9 in Table 5 initially support a suggestion of depleted lherzolite in preference to most compositional estimates for quartz eclogite (Brophy & Marsh, 1986; Bartels et al., 1991; Thirlwall et al., 1994). In contrast, relatively high ratios of elements with increased ionic potential such as Ba-La and Rb-Y (15-20) correlate positively with 87Sr/86Sri, a relationship most consistent with contributions from subducted lithosphere (Fig. 7; Davidson, 1987; Pearce & Peate, 1995). Low values for Ba/Zr and Th/U between 0·4 and 1·3 probably require aqueous fluids as the main metasomatic agents (Bartels et al., 1991; Thirlwall et al., 1994). If so, then equilibration of LREE between melts and aqueous fluids in the source provides the most reliable explanation for the existence of negative Ce anomalies (Bohrson & Reid, 1997); the magnitude of these features does not appreciably change with reliable indicators of fractionation such as SiO2, mg-number and Zr in the cone-sheets (Smith & Leeman, 1987). Depleted mantle model ages of ~250 Ma pre-date even the oldest intrusion in the western PRb and do not appear to be geologically meaningful, despite minimal evidence for pelagic sediment involvement (Fig. 7; Silver et al., 1979; DePaolo, 1988). Relative to olivine-bearing cone-sheets in the northern intrusive centre, unit CS2 cone-sheet compositions with similar silica contents contain slightly increased Al2O3, Na2O (not shown), Ba and Sr, and somewhat decreased concentrations of MREE (Figs 4, 5 and 6a and b). In conjunction with measurably decreased An in plagioclase (Table 4), these features probably reflect prior hornblende fractionation, which ultimately requires more hydrous conditions in the central intrusive centre magmas (Grove & Baker, 1984; Romick et al., 1992); most experiments require >4% intratelluric volatiles for hornblende stability in andesites (Grove & Baker, 1984; Sisson & Grove, 1993). Abundant deuteric alteration effects in CS2 and G4 can lend support to this hypothesis, although similar ratios among the HFSE in Table 5 require the existence of otherwise similar magmas before fractionation. Hygrometric calculations using the method of
Housh & Luhr, (1991) in conjunction with the temperature estimates in Table 4 predict 6·2-7·3% intratelluric water in unit CS2, relative to 3·0-4·5% in CS1 contemporaries with equivalent silica contents. These estimates also match experiments described by
Gust & Perfit, (1987), and so the CS2 cone-sheets potentially define water-saturated cotectics at shallow depths in Fig. 11a (Sisson & Grove, 1993). In the absence of crustal contamination, their substantially increased Ba-La, Ba-Zr and Zr-Hf ratios (>1·4, 30·9 and >89, respectively) can only reflect metasomatic equilibration or variable fluid fluxes in the source during melting (e.g. Davidson, 1987; Luhr, 1992). In summary, all of the intermediate cone-sheets indirectly record a petrogenetic history that typifies high-Al basalts and andesites from most modern arcs. What are now quenched cone-sheet magmas probably formed originally from a moderately melted and depleted spinel lherzolite in the upper mantle. Source metasomatism by aqueous fluids during melting presumably transported incompatible elements and radiogenic Sr from subducted ocean crust. Different degrees of equilibration between fluids and peridotite best explain the coexistence of polymorphic (containing either olivine or hornblende) magma series. After melting, high-Mg basaltic parents probably fractionated ferromagnesian minerals at ~0·8 GPa, and produced relatively Mg-rich, high-Al andesitic differentiates in at least one deep magma chamber. After ascent to subvolcanic levels close to the current exposure level, compositionally similar gabbroic units accumulated from magma batches analogous to cone-sheets in the northern and central intrusive centres (Fig. 2). Multiply intrusive (replenished) mafic plutons from the PRb in California apparently represent appropriate magma chambers located at depths equivalent to 0·2-0·5 GPa (<17 km), as Brophy, (1989b) envisaged beneath the Aleutian arc. Lithostatic pressures, in conjunction with 3-7% intratelluric water, probably suppressed plagioclase nucleation until shallow depths, where feldspars accumulated in response to marginal crystallization and perhaps other physical mechanisms for concentrating phenocrysts. Closed-system fractionation models Sodic and intermediate compositions dominate granitoids throughout oceanic regions, regardless of the prevailing tectonic regime (e.g. Erikson, 1977; Nakada et al., 1994). Compared with a compilation of data for tonalites and trondhjemites exposed in various modern settings, the ZIC tonalite and trondhjemite samples lie compositionally equidistant between abyssal plagiogranites and back-arc dacite lavas for all elements except Hf and Yb (Pearce et al., 1984; Figs 8b and 10c). Spatial (and presumably temporal) association of unit T1 with G4 gabbro in the southern intrusive centre (Fig. 2) also initially favours some form of genetic relationship with a more mafic parent (e.g. Spulber & Rutherford, 1983; Smith & Leeman, 1987; Romick et al., 1992). Both units contain similar modal minerals (Table 3), albeit in different proportions, and they have textural commonalities in the form of hornblende glomerophenocrysts and relatively elongate plagioclase euhedra. The most primitive mineral compositions in unit T1 consistently overlap with the most evolved G4 equivalents, and so the hottest thermometric estimates in Table 4 also correspond. Hornblende and plagioclase phenocrysts always show continuous optical zonation, and progressively evolved samples contain elevated pargasitic substitution in hornblende, more edenitic biotite and increasingly sodic plagioclase (Walawender & Smith, 1980; Smith et al., 1983). These relationships apparently favour a closed compositional system with an andesitic precursor similar to the CS2 cone-sheets in the central intrusive centre (Sisson & Grove, 1993). Intermediate rocks account for 11% (by area) of all exposed intrusive rocks in the ZIC (Table 2), which is typical of intrusions thought to preserve extensive paragenetic sequences (e.g. Erikson, 1977; Perfit et al., 1980). Simple mass balance arguments cannot refute the suggestion of a genetic relationship unless the tonalite-trondhjemite becomes substantially more voluminous at depth, the likelihood of which we question but cannot evaluate. Instead, overwhelmingly curvilinear trends for most of the major element oxides and Sr support at least some fractionation relationship with hornblende-bearing cone-sheet magmas (Figs 4 and 5a). Ratios involving the alkalis, Zr, Th, Hf, Nb and Ta in every combination always overlap among samples representing both units, despite generally higher absolute concentrations of most HFSE in the tonalite-trondhjemite. Ubiquitously low concentrations of Cr and Ni, in conjunction with primitive [epsilon]Nd values and low 87Sr/86Sri (Fig. 7), could also be characteristics inherited from an andesitic parent that ultimately represents the product of extensive ferromagnesian mineral fractionation (Gust & Perfit, 1987; Table 5). On this basis, progressively decreasing Al2O3, CaO, FeO, MgO and Sr, in conjunction with increasing alkalis, Rb, Ba, Th and Hf, can all be explained by removal of feldspar and hornblende phenocrysts in preference to K-feldspar and biotite (Arth & Barker, 1976; Romick et al., 1992; Figs 4 and 5). MREE depletion occurs generally in the supposed fractionates (Fig. 6), and the only positive Eu anomaly coincides with elevated abundances of Al2O3, CaO and Na2O that implicate plagioclase accumulation (Table 5). Forward fractional crystallization models describing the evolution of an andesitic magma towards dacite under atmospheric conditions were calculated using the `BIGD' program of
Nielsen, (1992) and
Ariskin et al., (1993). We justify results for general comparisons because of the subvolcanic emplacement of ZIC gabbros (inferred above), in conjunction with miarolitic cavities, quartz phenocrysts, and somewhat Fe-rich biotite in the tonalite (Tuttle & Bowen, 1958; Table 4). Applicability of dry models to these variably hydrous rocks requires an assumption that hornblende and augite share similar bulk element distribution coefficients, as actually confirmed for all constituents other than TiO2 and Nb in PRb gabbros (Gromet & Silver, 1987). Increasing concentrations of P2O5, Zr, U and Nd show that apatite, zircon, and/or titanite did not effectively buffer the T1 compositions (Table 5), and so the models also ignore accessory mineral involvement. Scattered trace elements preclude any unique solution and our preferred result involves removal of plagioclase and hornblende in subequal proportions, which satisfies mineral fractionation vectors calculated independently using microprobe data (Fig. 4; Perfit et al., 1980). The corresponding calculated liquid line of descent reproduces T1 compositions with excellent residuals ([Sigma]R2 < 0·1; Table 6) for all major and trace elements except K2O, Ba, Rb and Th ([Sigma]R2 < 0·8-24·9; Fig. 5). Individual REE patterns are not always mirror images of those in the gabbros, and so open-system behaviour is also apparently required (Fig. 6b and c; Smith & Leeman, 1987).
Partial melting and western belt granitoids Central to our understanding of these siliceous differentiates are the coupled enrichments of incompatible elements and (to a smaller extent) [Sigma]13REE relative to our modelled extracts (Smith & Leeman, 1987; Figs 4, 5, 7 and 10c). Despite consistently hygromagmatophile behaviour, Th should not be labile in aqueous solutions, and processes such as hydrothermal alteration and volatile complexing alone cannot explain such selective enrichments. Our sample collection procedures actually biased against altered samples, and all of the silica-saturated compositions now contain extremely low H2O+ (<0·7%) and CO2 (<0·1%) concentrations (Tables 2 and 5). Consequently, the absence of progressively increasing total lanthanide concentrations and Eu anomalies with silica, and the LREE and HREE concentrations elevated above any of the cone-sheets (Fig. 6b and c), probably require a magmatic explanation. Where data exist, the Sr-Nd isotopic compositions of tonalite samples closely resemble mafic units in the ZIC, apparently consistent with a fractionation origin involving an andesitic precursor close to the current exposure level (Fig. 7). However, ternary projection into the haplogranite system clearly distinguishes two compositional groups and shows that the trondhjemites resemble quartzo-feldspathic minimum melts produced at shallow depths between 0·05 and 0·5 GPa (>1·5 km; Fig. 11b). Their relatively low CaO and Sr concentrations probably reflect lower concentrations of more sodic plagioclase phenocrysts, and hence much drier conditions during melting (Fig. 7; Helz, 1975; Sisson & Grove, 1993). Most Phanerozoic trondhjemites contain Al2O3 < 15%, in conjunction with low K2O concentrations, Y > 30 ppm and flat (La/LuCN < 2 and La/SmCN < 5) REE profiles, which have been used to suggest amphibolitic sources at depths shallower than the garnet stability field (e.g. Arth & Barker, 1976; Barker, 1979; Wolf & Wyllie, 1994). The ZIC is typical in this respect, and the presence of inherited zircon cores in unit T1 strongly favours local anatexis of metabasite country rocks in the Alisitos Formation. Accessory titanite and Eu anomalies in the trondhjemites collectively indicate conditions more oxidizing than NNO (Nakada, 1991), and thermometric estimates based on plagioclase compositions produce 750-880°C melt temperatures (Holland & Blundy, 1994; Table 4). With these constraints, variable (8-19%) degrees of dehydration melting with <1 wt % H2O account for (1) the range of trondhjemite compositions and phenocryst contents (Fig. 11b), and (2) the pyroxene-bearing residues associated with leucosomes throughout the contact aureole (~0·2 GPa). This prediction simultaneously agrees with conditions calculated using whole-rock compositions of FeOT/MgO > 3 and Al2O3/Na2O ~ 5 at 63% SiO2 (Fig. 8b; Table 5) in conjunction with the method of
Winther, (1996). The most extreme decoupling of incompatible elements also always corresponds to the most siliceous trondhjemites, as would be expected of melts produced by dehydration of arc volcanic rocks at shallow depths (Helz, 1975; Spulber & Rutherford, 1983; Beard & Lofgren, 1991). Most of the trondhjemites have aplitic textures that strikingly resemble those of leucosomes produced under similar conditions adjacent to a Cordilleran mafic intrusion (Perfit et al., 1980; Beard, 1990). They also crop out as dykes that invade the outer margin of the T1 unit (Fig. 2), apparently suggesting that partial melts from the contact aureole exploited magma conduits within the complex during ascent. Identical subvolcanic conditions promoted hybridization between two magmas coexisting in an African ring complex (Moreau et al., 1987), and we favour the same explanation for the ZIC. Incorporation of trondhjemitic partial melts into a fractionated tonalite should not disturb phase relationships in the latter, and the very different temperatures and water contents of each magma favour hybridization through mutual exchange of crystals, rather than by homogenization of two liquids (Tate et al., 1997). Consistently positive correlations do exist among all of the incompatible elements and HREE (r2 0·896-0·987; Fig. 6c), but features such as reversed zonation and linear trends would not necessarily form under these conditions. Closer examination of the data for unit T1 instead reveals several small SiO2 gaps (e.g. Figs. 4f, and 5c and g) in addition to the presence of tonalite samples with transitional trondhjemitic characteristics (e.g. ZP-35 in Table 5). Absence of an Sr-Nd mixing line in Fig. 7 may be a function of the incomplete mixing, isotopic disequilibrium at the low degrees of partial melting involved with trondhjemite genesis, or a combination of these options. In summary, generation of the intermediate granitoids probably involved incomplete mixing between discrete tonalitic and trondhjemitic magmas ascending contemporaneously at subvolcanic depths ~0·2 GPa (<8 km). Dominantly peraluminous tonalites probably reflect hornblende and plagioclase fractionation from parents represented by the amphibole-bearing cone-sheets. Coupled enrichments of K2O, Ba, Rb, Th and (sometimes) LREE and HREE in the exclusively metaluminous trondhjemites, instead appear to be suggestive of low degree (<20%) partial melting during dehydration and contact metamorphism of the Alisitos Formation. Absence of obvious mixing trends prevents any quantitative treatment of the fractionated and partially molten end-members, and a lack of published analyses for the PRb limits comparisons with granitoids exposed elsewhere in the western belt. However, we notice that the ZIC tonalite-trondhjemites have shallowly sloping, unfractionated REE patterns (Table 5 and Fig. 6c) and Sr-Nd isotopic compositions that overlap entirely with the most silica-poor and isotopically primitive granitoid compositions currently recognized in the batholith (Fig. 7; Gromet & Silver, 1987; Silver & Chappell, 1988). Of equal significance, the averaged western belt granitoid composition contains coupled enrichments of the same incompatible elements mentioned in our interpretation of ZIC unit T1 above (bold elements in Fig. 1). Consequently, a tonalite-trondhjemite mixing model potentially also explains the formation of western belt tonalites on a more regional scale. The multiply intrusive ZIC preserves a unique association of coarse-grained gabbros, quenched andesitic cone-sheets, and a compositionally variable tonalite-trondhjemite unit, all of which lie at the primitive end of the compositional spectrum for western belt PRb granitoids. Petrogenetic models require a complex history involving fractionation, magma mixing and polymorphic parents. Major influences on its development include (in relative order of involvement): (1) removal of dominantly ferromagnesian minerals from two near-primary batches of high-Al basalt, probably at >28 km depth; (2) accumulation of a plagioclase-dominated assemblage from andesitic differentiates at <8 km depth; (3) formation of granitoid at subvolcanic depths by mixing between fractionated andesite (i.e. tonalite) and trondhjemitic partial melts of the metabasite country rocks. These compositions and inferred combinations of processes closely resemble those typical of most intra-oceanic environments, which therefore confirms an island arc tectonic setting for the western PRb (Davidson, 1987; Yanagi & Yamashita, 1994). Relevant modern analogues appear to be the Aleutian and Lesser Antilles arcs, both of which contain identical mafic and intermediate compositions preserved simultaneously among the volcanic and plutonic assemblages. These areas also contain cumulate xenoliths indicative of two contemporaneous mafic magma series (e.g. Perfit et al., 1980; Arculus & Wills, 1981; Devine, 1995). In the case of the PRb, this history potentially recommends a new comagmatic relationship between the gabbros and granitoids. Suitably detailed explanations for granitoid magmatism in the western PRb must account for the occurrence and characteristics of abundant (and presumably hot) tonalite, within constraints imposed by arc lithospheric architecture and known magmatic processes. Solely on the basis of chemistry, we tentatively suggest that tonalite-trondhjemites elsewhere in the batholith may have formed as two-component mixtures similar to those recognized here. Using averages for the ZIC tonalite and trondhjemite as hypothetical end-members, least-squares material balance calculations predict that approximately 75% partial melting and 25% fractionation can account for the averaged western belt PRb granitoid composition at subvolcanic depths, at least in terms of recognizably enriched trace elements (Fig. 1). Compositional variations among discrete intrusions in the western belt can be accommodated by assuming that the hybrids in each pluton involved the same end-members in different proportions from those preserved in the ZIC. However, such a simplified model ignores the likely possibility of partial melting to form granitoids at various depths, which also changes the proportions of each end-member required. Mafic protolith composition has limited effect on the characteristics of metaluminous melts, and pressure progressively increases the primary Al2O3 and CaO concentrations of tonalitic liquids (Helz, 1975; Spulber & Rutherford, 1983; Wolf & Wyllie, 1994). Thus, involvement of fractionated material may be less necessary for tonalite generation at middle- or lower-crustal levels beneath the PRb. A dual magma mixing interpretation requires that the multiply intrusive nature of bimodal ring complexes preserves combinations of processes applicable to contemporaneous granitoid plutons at shallow depths; this is an intriguing possibility that deserves verification elsewhere. Despite being a dominantly mafic entity, we believe that the ZIC does represent a microcosm of PRb granitoid generation. It contains both mafic and granitoid units that have compositions indistinguishable from those of isotopically primitive plutons throughout the western belt. Existence of two silica-saturated magmas with somewhat similar compositions also explains the lack of consistent chemical behaviour among silicic samples, a well-established and particularly enigmatic feature of the batholith (Gromet & Silver, 1987). Fractionation and partial melting of mafic sources both produce relatively small volumes of tonalite individually, and so a combination of two processes readily accounts for the existence of large and homogeneous PRb plutons. If so, equilibration between subduction-related fluids and the upper mantle must have been a fundamental control that determined the volatile contents and liquidus temperatures of intraplate magmas, and, ultimately affected the quantity of fractionated granitoid produced (Perfit et al., 1980; Singer et al., 1992; Pearce & Peate, 1995). The degree to which similar scenarios can explain tonalites in other Cordilleran-type terranes depends on the nature and abundance of hornblende-bearing mafic intrusions contemporaneous with the granitoids. We thank Mike Walawender and David Kimbrough (San Diego State University) for discussing gabbro petrogenesis and for providing otherwise unpublished U-Pb ages for the Alisitos Formation, respectively. Carol Lawson and Norm Pearson (Macquarie University) performed the XRF analyses and/or helped with operation of the electron microprobe, Keegan Schmidt (University of Southern California) stained intermediate samples with sodium cobaltinitrite, and Gabriella Mora-Klepeis (CSIRO) conducted the isotopic analyses. Scott Paterson (University of Southern California) and Ron Vernon (Macquarie University) kindly commented on an earlier version, whereas Mike Walawender and Dick Arculus reviewed the manuscript. Combined funding from the Australian Research Council, Macquarie University, and the Mexican Concejo Nacional de Ciencias y Tecnologia supported most of our endeavours in the PRb. This contribution is a publication of the Key Centre for GEMOC at Macquarie University.INTRODUCTION
GEOLOGICAL SETTING
FIELD RELATIONSHIPS
SHRIMP GEOCHRONOLOGY
PETROGRAPHY AND MINERAL CHEMISTRY
Massive gabbroic units
Cone-sheets and radial dykes
Tonalite-trondhjemite
WHOLE-ROCK CHEMICAL COMPOSITIONS
Sampling strategy and analytical techniques
Massive gabbroic units
Cone-sheets and radial dykes
Tonalite-trondhjemite
PETROGENESIS OF MAFIC AND INTERMEDIATE UNITS FROM THE ZIC
Accumulation of the massive gabbros
Parent magmas and upper-mantle melting
Origins of the tonalite and trondhjemite
CONCLUSIONS
ACKNOWLEDGEMENTS
REFERENCES
