The initiation of subduction is a process that cannot be observed directly but must be inferred from the rock record after subduction is well established. There are many approaches possible to infer the origin of subduction zones that are still active, but paleosubduction zones present special challenges, since their geodynamic setting can no longer be directly observed. In this study, we examine evidence for subduction initiation of the proto-Franciscan subduction zone along a transform fault, based on a subduction initiation origin for the Coast Range ophiolite, and on the Tehama-Colusa serpentinite mélange, which underlies the ophiolite and separates it from high-pressure/temperature metamorphic rocks of the Franciscan complex. The Coast Range ophiolite consists of volcanic, plutonic, and mantle components, each of which contains elements that reflect subduction initiation or hydrous melting within a subduction-zone setting. The volcanic assemblage includes forearc basalts and boninites, as well as more evolved calc-alkaline rocks; the plutonic complex contains intrusive suites that reflect this same range of parent magmas. Peridotites of the mantle section include both abyssal-like and refractory peridotites formed by hydrous decompression melting.

The Tehama-Colusa serpentinite mélange consists of blocks of basalt, chert, sedimentary rocks, and peridotite (harzburgite and lherzolite) in a sheared serpentinite matrix. The mélange matrix represents hydrated refractory peridotites with forearc affinities, and blocks within the mélange consist largely of upper-plate lithologies (harzburgite, arc volcanics, and arc-derived sediments). Lower-plate blocks within the mélange include oceanic basalts and chert with rare blueschist and amphibolite. The abyssal peridotites have low pyroxene equilibration temperatures that are consistent with formation in a fracture-zone setting. However, the current mélange reflects largely upper-plate lithologies in both its matrix and its constituent blocks.

We propose that the proto-Franciscan subduction zone nucleated on a large offset transform fault or fracture zone that evolved into a subduction-zone mélange complex. The nucleation of subduction zones along former transform boundaries has long been proposed for both modern arc systems and for the Franciscan–Coast Range ophiolite system. Our data support this interpretation and document more fully how this mechanism is expressed by mixing within the evolving serpentinite mélange.

The process of subduction initiation cannot be observed directly but must be inferred from the rock record after a subduction zone is established. “Induced” subduction initiation (Hall et al., 2003; Stern, 2004) results from far-field plate stresses that force convergence across a zone of weakness within a plate during plate-boundary reorganizations (e.g., the Maquarie-Puysegur-Fiordland system near New Zealand; Ruff et al., 1989) or in response to a collisional event between an existing subduction zone and continental or unusually thick oceanic crust (e.g., Ontong Java Plateau collision with the Solomon Islands Trench; Cooper and Taylor, 1985). In contrast, “spontaneous” subduction initiation occurs most commonly when stable ocean lithosphere, which forms by anhydrous decompression melting at mid-oceanic spreading centers and essentially “floats” on top of the underlying asthenosphere, becomes gravitationally unstable and begins to sink back into the mantle, either adjacent to gravitationally stable oceanic lithosphere, or adjacent to buoyant continental crust (e.g., Casey and Dewey, 1984; Leitch, 1984; Stern and Bloomer, 1992; Stern, 2004). In either case, the ultimate result is a subduction zone that underlies a volcanic arc, where the arc rests on older oceanic crust or continental crust (Tatsumi and Eggins, 1995). Investigations of early arc volcanism over the past two decades have shown that primitive early arc volcanics comprise the same rock assemblages as most ophiolites, and they are commonly preserved in highly extended forearc regions, which are interpreted to represent the products of spontaneous subduction initiation (Bloomer et al., 1995; Stern and Bloomer, 1992; Hawkins, 2003; Reagan et al., 2010).

Most extensive ophiolite terranes are believed, on the basis of their lavas and mantle residues, to have formed in, or passed through, a suprasubduction-zone environment (e.g., Miyashiro, 1973; Alabaster et al., 1982; Shervais, 1982; Pearce et al., 1984; Shervais and Kimbrough, 1985; Metcalf and Shervais, 2008). While the specific environment (backarc, arc, and forearc) is often debated, it is clear that the volcanic rock series found in ophiolites most closely corresponds to lavas now found within the forearc region of active arc terranes (e.g., Metcalf and Shervais, 2008). Further, it has been proposed that at their time of formation, there is little or no evidence for the existence of a volcanic arc (Stern and Bloomer, 1992; Pearce, 2003; Metcalf and Shervais, 2008).

Shervais (2001) showed that suprasubduction-zone ophiolites display a consistent sequence of events during their formation and evolution, suggesting that they form in response to processes that are common to all such ophiolites. This sequence includes: (1) birth—formation of the ophiolite above a nascent or reconfigured subduction zone; (2) youth—continued melting of refractory asthenosphere (depleted during birth) in response to fluid flux from the subducting slab; (3) maturity—onset of semistable arc volcanism; (4) death—the sudden demise of active spreading and ophiolite-related volcanism; and (5) resurrection—emplacement by obduction onto a passive margin or accretionary uplift with continued subduction. This sequence of events is similar to that inferred by Reagan et al. (2010) from their detailed studies of the Izu-Bonin-Mariana forearc, with tholeiitic “forearc basalts” preceding boninites, and with both forming during an episode of rapid extension in the forearc prior to establishment of the calc-alkaline volcanic arc.

Our primary focus here is on the first two stages proposed by Shervais (2001): (1) birth, characterized by forearc basalts (Reagan et al., 2010), represented by a variety of early arc tholeiites with strong mid-ocean-ridge basalt (MORB)–like characteristics, and (2) youth, characterized by high-Mg basalts and andesites of the boninite suite, which are among the most depleted volcanic rocks on Earth (e.g., Metcalf and Shervais, 2008). These stages are thought to represent the onset and development of subduction initiation, followed by the transition to stable subduction (stage three: maturity and calc-alkaline volcanism). Ophiolites that represent the first two stages in this progression (birth-youth) are interpreted to have formed during a subduction initiation event, regardless of their current apparent setting (Fig. 1). Further, these stages can be established by careful mapping within the plutonic series of the ophiolites (Shervais et al., 2004), and they can be inferred from highly refractory peridotite assemblages within the underlying mantle tectonites (e.g., Choi et al., 2008a, 2008b; Jean et al., 2010).

Although suprasubduction-zone ophiolites are commonly linked to subduction initiation (Stern and Bloomer, 1992; Shervais, 2001; Stern, 2004; Metcalf and Shervais, 2008; Whattam and Stern, 2011), the specific setting or circumstances of this event are commonly unclear. Our second focus here is on the evidence provided by serpentinite mélange zones, which are commonly associated with suprasubduction-zone ophiolites. The petrologic and geochemical character of blocks within these mélange zones provides clues regarding their origin, which may be unrelated to the adjacent ophiolite, or may reflect a linkage among the mélange, subduction initiation, and ophiolite formation.

We take as our case example the Coast Range ophiolite of California, which has been studied extensively for over three decades (Hopson et al., 1981, 2008; Shervais and Kimbrough, 1985; Shervais, 1990, 2001; Coleman, 2000; Shervais et al., 2005a, 2005b). The Coast Range ophiolite is closely linked to the Franciscan assemblage—one of the world’s most intensely studied subduction complexes and a model for convergent boundary plate processes (Bailey et al., 1964; Blake and Jones, 1974; Blake et al., 1982; Wakabayashi, 1999; Ernst, 1993). In northern California, the ophiolite is separated from younger rocks of the Franciscan assemblage by the Tehama-Colusa mélange—a serpentinite-matrix mélange containing blocks of peridotite, metabasalt, chert, and high-grade metamorphic rocks that has been interpreted as an oceanic fracture-zone assemblage (Hopson and Pessagno, 2005) linked to subduction initiation (Choi et al., 2008b).

The Coast Ranges of California comprise a complex orogen thought to have formed in the forearc region of the Sierra Nevada arc (or farther south, e.g., Wright and Wyld, 2007) during the Mesozoic and Paleogene (Fig. 2; Shervais et al., 2004; Hopson et al., 2008). This orogen is distinct from the classic collisional orogens (e.g., the Appalachians, Alps, or Himalayas) because the events preserved here occurred within an active convergent system that did not experience a major continental collision—all of the deformation and metamorphism found here formed entirely within a subduction-zone setting. This orogen documents the complexity that may arise within any active margin prior to a continent-continent collision (or arc-continent collision), and it shows that not all of the structures and metamorphic features found in collisional orogens are necessarily related to that collision (e.g., Yarlung-Tsangpo suture; Ratschbacher et al., 1992; Guilmette et al., 2008).

The Coast Range orogen of California is composed of three main elements: the Upper Jurassic to Upper Cretaceous Great Valley Group, deposited within a complex forearc basin; the Middle Jurassic Coast Range ophiolite, which forms the basement upon which the Great Valley Group was deposited; and the Franciscan assemblage, which represents the accretionary prism of the Cordilleran subduction zone (Fig. 3). Here, we discuss two of those elements—the Coast Range ophiolite and the Franciscan assemblage—and the Tehama-Colusa mélange, which separates these units in the northern Coast Ranges.

Coast Range Ophiolite

The Coast Range ophiolite is a linear belt of felsic, mafic, and ultramafic rocks exposed primarily along the eastern flank of the California Coast Ranges, but also as scattered locales west of the Salinia block in the western Transverse Ranges and farther north along the coast (Fig. 2). It is one of the most extensive ophiolite terranes in North America and has long been central to our understanding of Cordilleran tectonics. Nonetheless, its origin is controversial and three primary hypotheses have been advanced: (1) formation at a mid-ocean-ridge spreading center at low paleolatitudes, and subsequent rapid drift northward to collide with North America (Hopson et al., 1981, 2008; Pessagno et al., 2000); (2) formation as a backarc basin behind an east-facing volcanic arc that collided with North America during the Late Jurassic Nevadan orogeny (Godfrey and Klemperer, 1998; Ingersoll, 2000); and (3) formation by forearc or intra-arc rifting along the western margin of North America, in response to nascent or renewed subduction of oceanic plates beneath North America (Shervais and Kimbrough, 1985; Shervais, 1990, 2001; Stern and Bloomer, 1992). The consensus of most recent studies support the suprasubduction-zone model (Stern and Bloomer, 1992; Shervais et al., 2004, 2005a; Shervais, 2008; Choi et al., 2008a; Jean et al., 2010), but models based on backarc basins and oceanic spreading centers still persist (e.g., Hopson et al., 2008).

The Coast Range ophiolite is composed of three major components: the volcanic-hypabyssal upper crust and its overlying section of volcano-pelagic sediments, the plutonic lower crust, and the underlying refractory peridotites, which represent the upper mantle upon which the crust was built after the extraction of partial melts parental to the crust (Hopson et al., 1981, 2008; Shervais et al., 2004). The crustal components display the developmental cycle described by Shervais (2001) that has been interpreted to reflect a consistent sequence related to subduction initiation and the subsequent transition toward more normal arc activity. The mantle rocks preserve evidence for the early part of this cycle, as well as for events that predate ophiolite formation (Choi et al., 2008b).

Volcanic-Hypabyssal Upper Crust and Volcano-Pelagic Sediments

The upper crust at most Coast Range ophiolite localities consists of mafic volcanic rocks, which commonly transition upward into more-evolved intermediate-felsic volcanics, and an overlying cover of ash-rich radiolarian cherts and tuffs. The volcanic rocks are dominated by volcanic arc tholeiite basalts (similar to MORB in trace-element composition), overlain by boninites, or boninitic basalts (which also form dikes that crosscut the earlier tholeiites). The MORB-like arc tholeiites are similar to the forearc basalts of Reagan et al. (2010). Calc-alkaline volcanics, which occur at many locales (e.g., Del Puerto, Llanada, Elder Creek), form a fractionation series of basalt-andesite-dacite-rhyolite (Fig. 4). At Del Puerto, these include a younger series of hornblende-bearing volcaniclastics called the Lotta Creek tuff (Evarts et al., 1999), while at Elder Creek, the calc-alkaline volcanics are found only in a sedimentary breccia called the Crowfoot Point Breccia, which overlies the ophiolite disconformably (Hopson et al., 1981, 2008; Robertson, 1990; Shervais et al., 2004). Relationships at Llanada are similar to Del Puerto, with extrusive volcaniclastics and intrusive keratophyre sills (Giaramita et al., 1998; Hopson et al., 2008).

Plutonic Lower Crust

The lower crust of the Coast Range ophiolite can be divided into three suites: (1) a mafic cumulate series of layered gabbros with associated dunites and isotropic gabbros, (2) an intrusive series of ultramafic-mafic cumulates made up of wehrlite, olivine clinopyroxenite, and isotropic gabbro (±dunite), and (3) a later intrusive series of tonalite-trondhjemite-diorite (TTD) that postdates the first two plutonic suites. Cumulate gabbros of suite 1 are dominant and typically form over 60% of the lower-crustal section. These rocks are typical of ophiolite layered gabbro sequences, with ∼60%–70% modal plagioclase and 30%–40% modal pyroxene or pyroxene and amphibole; olivine is relatively rare. Augite is the dominant pyroxene, although hypersthene may occur as well in place of hornblende. Suite 2 wehrlites, clinopyroxenites, and isotropic gabbros are intrusive into the layered gabbros of suite 1, documented by massive wehrlite-pyroxenite dikes that crosscut layering in the gabbro, and by xenoliths of layered gabbro in wehrlite and suite 2 isotropic gabbro (Hopson et al., 1981; Shervais, 2001; Shervais et al., 2004). The TTD suite (suite 3) forms small intrusive bodies, thick (10–15 m) dikes, and flat-lying, kilometer-scale plutons that intrude all of the older plutonic suites and parts of the overlying volcanic-hypabyssal complex (Shervais, 2001, 2008; Shervais et al., 2004). Radiometric U-Pb zircon ages for the Coast Range ophiolite represent this final intrusive suite; 238U-206Pb ages typically range from ca. 172 Ma to ca. 162 Ma (Shervais et al., 2005a; Hopson et al., 2008; Mattinson and Hopson, 2008), although some locales contain a younger suite of hornblende-bearing quartz diorites and keratophyres with 238U-206Pb and 40Ar/39Ar ages of ca. 150 Ma (Evarts et al., 1999; Hopson et al., 2008).

Subcrustal Mantle Peridotites

The crustal series described here is underlain structurally by mantle peridotites—including lherzolite, harzburgite, and dunite—that reflect increasing melt extraction and progressively more refractory compositions (Choi et al., 2008a; Jean et al., 2010). Contacts with the overlying crustal section are rarely exposed and, where seen, are generally faulted. In the northern Coast Ranges, these blocks are typically associated with the underlying Tehama-Colusa serpentinite mélange, whereas in the Diablo Range and west of the Sur-Nacimiento fault, the peridotites directly underlie the crustal sequence. Harzburgites, with lesser dunite, are the most common lithologies at all locations where peridotites crop out; however, lherzolite is found locally at Cuesta Ridge and forms a large, kilometer-scale block near Stonyford (Fig. 2). Jean et al. (2010) have shown that the harzburgites represent up to 23% melt extraction under decompressing pressure conditions, first in the garnet field and then in the spinel facies. The high fraction of melting and the modal compositions are consistent with hydrous melting in the mantle wedge of a subduction zone, a conclusion supported by the high fluid mobile element concentrations and evolved isotopic compositions of clean clinopyroxene mineral separates (Choi et al., 2008a, 2008b). In contrast, the lherzolites have mineral compositions similar to abyssal peridotites formed at mid-ocean-ridge spreading centers, but they are still characterized by high fluid mobile element concentrations and evolved isotopic compositions (Choi et al., 2008a, 2008b; Jean et al., 2010).

Franciscan Assemblage

The Franciscan assemblage formed during prolonged subduction of oceanic lithosphere beneath the western margin of North America in the late Mesozoic and early Tertiary (Bailey et al., 1964; Hsü, 1968; Ernst, 1993; Wakabayashi, 1999). It extends from Eureka to the southern Diablo Range east of the San Andreas fault system, and from Big Sur to the western Transverse Ranges west of the Sur-Nacimiento fault, encompassing an area ∼700 km long and up to 200 km wide (Fig. 2). The Franciscan assemblage includes three tectonic belts that generally become progressively younger (structural and metamorphic ages) from east to west: the Eastern belt, the Central belt, and the Coastal belt (Bailey et al., 1964; Berkland et al., 1972). The early to mid-Cretaceous Eastern belt is subdivided into the Pickett Peak terrane (South Fork Mountain Schist, Valentine Spring Formation; Worrall, 1981) and the subjacent Yolla Bolly terrane (Blake et al., 1982). The Pickett Peak terrane consists of blueschist-facies mudstone and basalt (South Fork Mountain Schist) and mudstone-graywacke (Valentine Spring Formation), forming coherent sheets up to hundreds of meters thick transposed along low-angle thrust faults (Blake et al., 1982). The Yolla Bolly terrane consists of blueschist-facies mudstone and graywacke, with minor basalt and chert, in coherent thrust sheets, including metamorphosed mélange (Fig. 3; Bailey et al., 1964; Blake and Jones, 1974; Wakabayashi, 1992, 1999; Ernst, 1993). The Coastal belt, which ranges in age from mid-Cretaceous through Miocene, consists of west-verging thrust slices that get younger to the west, comprising both coherent graywacke and broken formations of arkose, mudstone, and conglomerate (Fig. 3; McLaughlin et al., 1982; Blake et al., 1985). The Central belt mélange contains blocks of graywacke, greenstone, serpentinite, chert, limestone, blueschist, eclogite, and garnet amphibolite, in a matrix of finely comminuted micrograywacke siltstone/shale (Bailey et al., 1964; Blake and Jones, 1974; Blake and Wentworth, 1999). Large tracts of arkosic wackes interpreted as slope basin deposits form mappable units within the mélange (Becker and Cloos, 1985).

Tehama-Colusa Mélange

The Tehama-Colusa serpentinite mélange is restricted to the northern Coast Ranges, where it separates the Middle Jurassic Coast Range ophiolite from Lower Cretaceous rocks of the Franciscan assemblage Eastern belt (Huot and Maury, 2002; Hopson and Pessagno, 2005; Shervais et al., 2011). The Tehama-Colusa mélange extends for over 160 km from Wilbur Springs to the North Fork of Elder Creek (Fig. 2). The mélange consists of a sheared serpentinite matrix with tectonic blocks of abyssal and refractory peridotite, lower-crustal plutonic rocks (cumulate gabbro, wehrlite, and diorite), low-grade metavolcanic rocks, high-grade metamorphic blocks of blueschist, amphibolite, and garnet amphibolite, volcaniclastic sandstones of uncertain provenance, and foliated metasediments (Jayko and Blake, 1986; Hopson and Pessagno, 2005; Shervais et al., 2011). Many of these blocks were derived from the overlying Coast Range ophiolite, e.g., the lower-crustal gabbros, wehrlites, and diorites (which have U-Pb zircon ages identical to diorites of the Coast Range ophiolite; Shervais et al., 2005a), and many of the low-grade metavolcanic blocks, which are identical in composition and grade to volcanic rocks in the ophiolite. It is inferred that the volcaniclastic sandstones were derived from the Coast Range ophiolite as well, and that they may be correlated with the Crowfoot Point Breccia (Seymore-Simpson, 1999), whereas the foliated metasediments have been correlated with the Galice Formation (Jayko and Blake, 1986). These rocks document a close association between the mélange and the overlying ophiolite, which requires that the mélange and ophiolite formed within the same tectonic realm and that mélange formation continued after formation of the ophiolite crustal assemblages.

Peridotite blocks of the Tehama-Colusa mélange can be classified into two groups, as discussed previously for the entire Coast Range ophiolite. The dominant group is highly refractory (harzburgite and dunite) and reflects extensive melt extraction in a hydrous melting environment (Huot and Maury, 2002; Choi et al., 2008a, 2008b; Jean et al., 2010). The Black Diamond Ridge massif near Stonyford consists of abyssal-like lherzolite that has shown limited melt extraction (Choi et al., 2008a, 2008b; Jean et al., 2010). The refractory harzburgites have compositions consistent with extraction of partial melts to form the overlying ophiolite massif; the provenance of the fertile lherzolites is uncertain, but their trace-element and isotopic compositions suggest residence within a subduction-zone environment for at least part of its existence. Finally, the mélange matrix has been shown to contain relict spinel grains with a wide range of Cr/Al ratios, consistent with a mixed provenance of highly refractory harzburgite-dunite and more fertile lherzolite (Huot and Maury, 2002; Shervais et al., 2011).

Volcanic rock compositions were compiled from published sources and plotted on standard petrogenetic diagrams to allow comparison with other settings. We also present new secondary ion mass spectrometry (SIMS) analyses of in situ clinopyroxene grains from gabbros, wehrlites, and pyroxenites of the Elder Creek ophiolite, along with calculated equilibrium magma compositions based on pyroxene-melt partition coefficients. The SIMS analyses were carried out on the Cameca 4f ion microprobe at the University of New Mexico. Finally, we compiled published data for mantle peridotites that underlie the ophiolite, or that are part of the Tehama-Colusa mélange, and separate it from the Franciscan assemblage, in order to deduce the original setting that led to subduction initiation.

Volcanic Rock Compositions

Shervais (2001) divided volcanic rocks of the Coast Range ophiolite into four main stages, based on composition, relative age, and occurrence. The first stage (birth) includes volcanic arc tholeiites with limited compositional range (basalt, basaltic andesite, and andesite; Fig. 4) that have major- and trace-element characteristics transitional to MORB (Fig. 5). These volcanic rocks are equivalent to the so-called “forearc basalts” of the Mariana forearc (Reagan et al., 2010), and they occupy a similar setting—they predate all other volcanic arc basalts and underlie the later boninitic lavas (e.g., Shervais, 2001). The forearc basalts are followed by boninites (extremely depleted high-Mg andesites) and boninitic lavas (somewhat less-depleted or more-evolved volcanics directly related to boninite magmatism). The boninites and boninitic lavas correspond to stage two (youth) of Shervais (2001). These magmas form in response to extremely high melt fractions—a direct result of continued decompression melting in the presence of high fluid phase fluxes. They are characterized by higher MgO and Cr contents, and by lower concentrations of Ti, Y, and other moderately incompatible elements (Figs. 4 and 5).

Stage three lavas are calc-alkaline rocks thought to be related to the onset of stable subduction (maturity), but the more magnesian members of this suite may form in response to fractionation of primitive boninitic magmas. Finally, in the Coast Range ophiolite, stage four lavas are MORBs formed in response to propagation of a backarc spreading center into an arc (e.g., Josephine ophiolite; Harper, 2003) or collision of the subduction zone with an actively spreading oceanic ridge (death; Shervais, 2001).

Plutonic Rock Parent Magmas

Detailed mapping and field work carried out over the past 20 yr have shown that many ophiolite plutonic suites—once thought to represent simple cumulate piles—are in fact polymagmatic complexes composed of at least two or three distinct magmatic suites (Juteau et al., 1988; Hebert and Laurent, 1990; Laurent, 1992). The most common suites are similar to those found in the Coast Range ophiolite: (1) a primary suite of cumulate dunite and layered gabbros composed of plagioclase and clinopyroxene, with less common olivine, orthopyroxene, or hornblende; (2) a later intrusive suite of wehrlites, pyroxenites, and primitive isotropic gabbros; and (3) a final suite of calc-alkaline dikes and plutons composed of tonalite-trondhjemite-quartz diorite, which may grade into isotropic gabbro as modal quartz declines (Shervais, 2001, 2008; Shervais et al., 2004).

Crosscutting relationships within the plutonic complex clearly establish the order of intrusion, which is consistent from one ophiolite to another, and this sequence of intrusion is similar to the eruptive order noted in the volcanic rocks. This implies that the early cumulate gabbros are related to the forearc basalts, that the wehrlite-pyroxenite suite is derived from the boninitic suite magma, and that the tonalite-trondhjemite-quartz diorite suite is derived from the same magmas as the calc-alkaline volcanic rocks. We can assess these potential relationships by calculating the rare earth element (REE) concentration of magmas in equilibrium with cumulate pyroxene in the gabbros and wehrlites, using pyroxene SIMS analyses (Table 1). In Figure 6, we compare these equilibrium melts with the forearc basalts and boninites of Reagan et al. (2010), and with two isotropic gabbros related to the stage-one layered gabbros. As can be seen in the figure, the isotropic gabbros and the equilibrium melts with pyroxenes from the layered gabbros display REE concentrations similar to forearc basalts. In contrast, pyroxenes from the wehrlites and clinopyroxenites are in equilibrium with melts that have REE concentrations similar to boninitic melts—the overall REE concentrations are lower, but they are enriched in light (L) REE relative to heavy (H) REEs, probably as a result of slab-derived fluid metasomatism (Fig. 6).

Subcrustal Mantle Peridotites and the Tehama-Colusa Mélange

The mixing of volcanic, plutonic, and sedimentary rocks of Coast Range ophiolite (upper plate) provenance with volcanic and sedimentary rocks of oceanic (lower plate) provenance has been well documented in the Tehama-Colusa mélange (Shervais et al., 2011). We focus here on peridotite blocks within the serpentinite matrix mélange and on massive peridotites that directly underlie the ophiolite—which also exhibit mixed provenance behavior.

Peridotites associated with the Coast Range ophiolite define two distinct groups. The dominant group found at all locations includes highly refractory harzburgite and dunite, often cut by veins or dikes of orthopyroxenite (Choi et al., 2008a). Pyroxenes in the refractory harzburgites have spoon-shaped REE patterns (strongly depleted middle REEs with slightly enriched LREEs) and high concentrations of fluid mobile incompatible trace elements (Jean et al., 2010). Spinels in these harzburgites, dunites, and orthopyroxenites have high Cr#’s (100 × Cr/[Cr + Al]) that correspond to the compositional range of forearc peridotites (Fig. 7). The less common group found at a few locations is made up of relatively fertile lherzolite and clinopyroxene-harzburgite with aluminous spinels (Fig. 7; Choi et al., 2008A; Jean et al., 2010). Spinels recovered from the scaly serpentinite matrix of the mélange (Huot and Maury, 2002) are mostly refractory Cr-rich spinels, with a few more aluminous spinels, showing that the matrix is formed largely by the shearing of upper plate (forearc) peridotite (e.g., Shervais et al., 2011).

Figure 8 shows the Cr# of clinopyroxene as a function of equilibrium two-pyroxene temperatures of the Coast Range ophiolite peridotites calculated using the geothermometer of Brey and Köhler (1990). Information also shown for comparison includes the available data for abyssal peridotites associated with major fracture zones (Vema, Owen, Romanche) and normal, small-offset transforms. Major fracture-zone peridotites are characterized by lower equilibration temperatures at a given Cr# than normal, small-offset transform peridotites (Fig. 8). All Coast Range ophiolite peridotite pyroxenes have low equilibration temperatures. The refractory Coast Range ophiolite peridotites fall outside the field for all abyssal peridotites, whereas the fertile Coast Range ophiolite peridotites (lherzolites) have temperatures that correspond to large-offset transform faults/fracture zones (Choi et al., 2008b). The occurrence of low equilibration temperatures in mantle associated with large-offset fracture zones may reflect hydrothermal cooling of the crust and mantle to deeper levels along transform systems. This suggests that fertile peridotites of the Coast Range ophiolite may be remnants that formed originally within a large-offset transform system. If this interpretation is correct, it has significant implications for subduction initiation.

Ophiolites and Subduction Initiation

The progression from early MORB-like arc tholeiites (both as volcanic rocks and cumulate gabbros with forearc basalt parent magmas), followed by boninitic lavas and related intrusions (wehrlite-pyroxenite suite) represents the first two stages in the “life cycle” of suprasubduction-zone ophiolites recognized by Shervais (2001). This progression is characteristic of most suprasubduction-zone ophiolites and has long been correlated with forearc or intra-arc rifting (e.g., Pearce et al., 1981, 1984; Shervais, 1982; Shervais and Kimbrough, 1985; Juteau et al., 1988; Crawford et al., 1989). The correlation of this progression with subduction initiation and nascent arc volcanism was recognized more recently (e.g., Shervais, 1990, 2001; Stern and Bloomer, 1992; Pearce, 2003; Stern, 2004; Whattam and Stern, 2011).

The transition to calc-alkaline suite rocks (volcanic rocks and plutonics of the TTD suite) in the Coast Range and other suprasubduction-zone ophiolites is thought to represent the onset of stable subduction (e.g., Shervais, 2001; Stern, 2004), but these rocks may represent in part evolved members of the preceding volcanic arc basalt-boninite suites (e.g., Shervais, 2008). This latter view is supported by the whole-rock chemistry of the TTD suite—with relatively high MgO and Cr at high silica contents (Shervais, 2008)—and by their U-Pb zircon ages, which are nearly coincident with subduction initiation at 172–166 Ma (Shervais et al., 2005a).

The connection between ophiolites and subduction initiation has been strengthened by the recent discovery of MORB-like forearc basalts as a dominant volcanic assemblage in the Mariana forearc (Reagan et al., 2010). In the Marianas, this suite underlies the distinctive boninite suite and documents a period of nearly anhydrous decompression melting during the early stages of forearc spreading. The significance of this discovery is enormous, since it has commonly been assumed that the MORB-like early volcanic rocks of many ophiolites (and their cumulate gabbro plutonic equivalents) represented true MORB formed at a mid-ocean-ridge spreading center, and this led to geodynamically suspect models of ridge-centered thrusting to initiate subduction (e.g., Oman; Boudier et al., 1988). These models were reinforced by the short elapsed time between formation of the high-grade metamorphic soles and ophiolite formation ages (e.g., Hacker, 1994). The recognition of slightly hydrous MORB-like basalts that form in a forearc setting resolves these contradictions and finally permits a unified model of subduction initiation that is consistent both with ophiolite stratigraphy and with the volcanic evolution of in situ forearc crust (Reagan et al., 2010).

Serpentinite Mélange and Subduction Initiation along a Transform

The connection between ophiolite formation and subduction initiation is now a robust and well-established paradigm. It is more difficult, however, to document the exact nature of subduction initiation itself for any specific example—Is it truly spontaneous in response to simple gravitational forces, or induced by the compressional forces of plate dynamics (e.g., Stern, 2004)? Does it nucleate on a fracture zone/transform, or on some other zone of weakness? To answer these questions, we need to examine the lithologic units that comprise the subduction assemblage and the hanging wall of the subduction complex.

The former hanging wall of the proto-Franciscan subduction zone appears to be preserved in the northern Coast Ranges as the Tehama-Colusa mélange (Jayko and Blake, 1986; Hopson and Pessagno, 2005; Shervais et al., 2011). This serpentinite-matrix mélange contains elements of the overlying forearc (suprasubduction) ophiolite complex, as well as rocks derived from the underlying subducting oceanic plate (Shervais et al., 2011). Upper-plate (hanging-wall) rocks include volcanic and plutonic rocks derived from the ophiolite, sedimentary rocks from its overlying volcano-sedimentary cover sequence, and highly refractory peridotites (harzburgites and dunites) that formed by hydrous melting in the mantle wedge of the subduction zone (Choi et al., 2008a, 2008b; Jean et al., 2010). Lower-plate (footwall) rocks include the high-grade metamorphic blocks (amphibolite, garnet amphibolite, and high-grade blueschist with oceanic basalt affinity) interpreted to have formed during subduction initiation, when the subducting oceanic crust was juxtaposed against hot hanging-wall peridotites (e.g., Wakabayashi, 1990). Fertile peridotites with compositions similar to abyssal peridotites are also found (Choi et al., 2008a, 2008b; Jean et al., 2010). While these have affinities to the mantle lithosphere that underlies the footwall, their elevated fluid-mobile element concentrations require prolonged residence within the upper plate, where hydrous flux from the subducting slab is rich in fluid mobile elements (e.g., Jean et al., 2010).

Parkinson and Pearce (1998) have shown that fertile peridotites similar to abyssal peridotites are commonly found within forearc regions, where they coexist with highly refractory peridotites. The fertile peridotites may represent former oceanic lithosphere that was trapped in the forearc region after subduction initiation, e.g., after subduction initiation along a fracture zone. Choi et al. (2008b) have proposed that this is the case, based on the low pyroxene equilibration temperatures that characterize these peridotites, and a comparison of these temperatures with fracture-zone and non-fracture-zone abyssal peridotites. These blocks must have been located sufficiently far from the actual subduction interface to allow them to persist and become mixed with the refractory peridotites, rather than be removed by subduction erosion.

Early History of the Proto-Franciscan Convergent Margin

Metasedimentary rocks of the Franciscan Eastern belt have Early to mid-Cretaceous depositional ages and mid- to Late Cretaceous metamorphic ages (e.g., Blake et al., 1982; Wakabayashi and Dumitru, 2007; Dumitru et al., 2010). These depositional and metamorphic ages are significantly younger than the overlying Coast Range ophiolite (Shervais et al., 2005a; Hopson et al., 2008; Mattinson and Hopson, 2008). High-grade metamorphic blocks and rare coherent slabs with ages of 146–169 Ma (Lanphere et al., 1978; McDowell et al., 1984; Mattinson, 1986; Ross and Sharp, 1988; Catlos and Sorensen, 2003; Anczkiewicz et al., 2004; Wakabayashi and Dumitru, 2007; Shervais et al., 2011) overlap with ages for the Coast Range ophiolite and imply that the oldest blocks formed in the same nascent subduction zone as the ophiolite.

Based on the oldest noninherited U-Pb zircon ages in the ophiolite (Kimbrough, inShervais et al., 2005a; Hopson et al., 2008; Mattinson and Hopson, 2008), and the upper age range of high-grade metamorphic blocks in the Franciscan assemblage (Ross and Sharp, 1988; Catlos and Sorensen, 2003; Anczkiewicz, et al., 2004), we infer that proto-Franciscan subduction initiated ca. 169–172 Ma (Aalenian-Bajocian). This age range shortly predates the oldest preserved radiolarian faunal assemblages in the ophiolite (Bajocian: Murchey, inShervais et al., 2005a) and postdates compressive deformation in the Sierra Foothills by 5–20 m.y. (see discussion in Shervais et al., 2005a). In contrast, Dumitru et al. (2010) documented deposition and metamorphism of the South Fork Mountain Schist at ca. 123 Ma using detrital zircon and white mica ages, followed shortly by the Valentine Spring Formation and Yolla Bolly terrane. They concluded that this rapid onset of accretionary subduction followed a prolonged period of nonaccretionary subduction that lasted some 40 m.y. (Dumitru et al., 2010). Many modern arcs are nonaccretionary, and, in many cases, the forearc region has been subject to subduction erosion (von Huene and Scholl, 1991; Scholl and von Huene, 2007). The significance of this is that under these circumstances, fragments of the original subduction interface may be preserved—albeit highly deformed and mixed with younger rocks.

Spontaneous Subduction Initiation

The most commonly applied model for spontaneous subduction initiation involves sinking of older denser lithosphere along a transform-fracture system (e.g., Karig, 1982; Casey and Dewey, 1984; Leitch, 1984; Stern and Bloomer, 1992). Our model builds on these previous efforts and on later work (e.g., Bloomer et al., 1995; Shervais, 2001; Stern, 2004; Metcalf and Shervais, 2008). The initial condition is a large-offset transform fault/fracture zone plate boundary with lithosphere of strongly contrasting age on each side (Fig. 9A). As discussed by Stern (2004), spontaneous subduction initiation begins when the older lithosphere begins to sink gravitationally into the asthenosphere (Fig. 9B); this may be induced in part by other factors affecting relative plate motions. As shallow MORB-source asthenosphere flows upward to fill the void left by the sinking lithosphere, it undergoes decompression melting—analogous to the process at true mid-ocean ridges. These melts contain a minor amount of hydrous fluids from the sinking slab but are otherwise analogous to normal mid-ocean-ridge basalts (Stern and Bloomer, 1992; Shervais, 2001; Metcalf and Shervais, 2008; Reagan et al., 2010). The melts generated during this stage (“birth”) have MORB-like arc tholeiite chemistry, especially in the trace elements; silica extends to slightly higher concentrations in response to the limited fluid influx (Fig. 9B). In the lower crust, these MORB-like arc tholeiites form layered gabbros and gabbronorites. In the Coast Range ophiolite, these rocks include the early arc tholeiites and main layered gabbro series.

Continued sinking of the dense lithospheric slab leads to catastrophic dehydration of the low-temperature hydrous phases and flooding of the overlying mantle wedge with an aqueous flux that lowers the melting temperature of the already depleted (by stage-one decompression melting) asthenosphere. The resulting melts are boninitic magmas with highly refractory major-element compositions and “U-shaped” or “spoon-shaped” trace-element concentrations (on mantle-normalized multi-element plots) that reflect extreme melt depletion and fluid-mobile element re-enrichment (e.g., Metcalf and Shervais, 2008; Whattam and Stern, 2011). Intrusion of the wehrlite-pyroxenite suite into the layered gabbro section reflects this same hydrous melting event (Fig. 9C). Continued extension of the nascent forearc crust during this stage leads to extensional plastic deformation of the layered gabbros, which enhances intrusion of the less-deformed wehrlite-pyroxenite suite (e.g., Juteau et al., 1988; Hebert and Laurent, 1990; Shervais et al., 2004). The depleted harzburgites and dunites found in the mantle section of most ophiolites reflect this stage of melting.

During this early extensional phase, the transform/fracture-zone assemblage on the upper plate forms the leading edge of the nascent suprasubduction crust/lithosphere and remains adjacent to the plate boundary as the lithosphere behind it extends over the sinking slab, and the hinge line retreats away from the trench (Figs. 9B and 9C). This allows the transform/fracture-zone assemblage to remain relatively intact, even as the former oceanic crust behind it becomes subsumed by the newly formed forearc tholeiite/boninite crust (including their plutonic equivalents in the lower crust). The transform/fracture-zone assemblage is thus preserved in a frontal position, trenchward of the forearc basin, where it can later interact with both upper-plate ophiolite assemblages and lower-plate ocean crust rocks.

Continued sinking of the slab eventually slows as a more stable subduction geometry is achieved and hinge rollback slows (Fig. 9D). At this time, the plate boundary may become accretionary (adding new crust through accretion of trench sediments) or nonaccretionary—where accretion is suppressed and erosion of the leading edge of the upper plate occurs (Scholl and von Huene, 2007). During this phase, the relict transform/fracture-zone assemblage becomes broken up and mixed with upper-plate lithologies (volcanics, plutonics, peridotites) and lower-plate lithologies (mostly volcanics and sediments). The transform/fracture-zone assemblage also may be partially or wholly removed during this phase.

Subduction Erosion of the Forearc

In most nascent arc systems, nonaccretionary or erosive subduction is the rule, because the lack of an emergent arc limits the potential sediment load available to the trench (Scholl and von Huene, 2007); this was apparently the situation during the early evolution of the proto-Franciscan subduction system (Dumitru et al., 2010). Over time, however, normal subduction leads to the formation of an emergent arc that can supply abundant sediment to the trench, which drives a transition from nonaccretionary to accretionary subduction; this occurred in the proto-Franciscan subduction system around 123 Ma (Dumitru et al., 2010). Continued evolution of the system could lead to arc rifting and the formation of a backarc basin, which would begin with suprasubduction-type magmas and evolve over time to a wider basin with MORB-like volcanics.

Reconciliation of the Subduction Initiation Model with Evidence for Prior Continental Margin Subduction

One of the central conundrums of Cordilleran geology is the conflict between a subduction initiation model for the widely distributed Middle Jurassic ophiolites of California, Oregon, and Washington, and the occurrence of continental margin arc volcanism along a NW-SE–trending volcanic arc that began in the Early Triassic and continued into the Early Jurassic (e.g., Saleeby and Busby-Spera, 1992). Saleeby (2011) investigated oceanic rocks of the Sierra Foothills belt and documented a long history of ocean floor volcanism (Early Ordovician–Pennsylvanian) adjacent to large-offset transforms that truncated the southwest margin of Laurentia by sinistral shear. Subduction initiation occurred in the Late Permian along this truncated margin (Saleeby, 2011). There is also clear evidence for an Early Jurassic deformation event, which may have occurred in response to passage of the Insular composite arc terrane (Saleeby, 2011).

There are two possible explanations for the onset of a Middle Jurassic subduction initiation event to form the Coast Range ophiolite and other Middle Jurassic ophiolites of the western Cordillera. First, the ophiolite may have formed elsewhere and been translated into its current position later (e.g., Wright and Wyld, 2006, 2007). This proposal is based in part on the lack of Nevadan orogeny deformation within the ophiolite and on detrital zircon ages from the Great Valley Group. It is also consistent with changes in radiolarian faunal assemblages up-section. Second, the Middle Jurassic subduction initiation event posited here occurred after the J1 cusp in the apparent polar wander path of North America (Beck and Housen, 2003). The J1 cusp represents a complete reversal in the absolute motion vector for North America and corresponds to the initial opening of the North Atlantic. This event would change both the direction and velocity of relative motion between North America and adjacent oceanic plates in the Pacific Basin, which was apparently sufficient to reorient convergence vectors and create a new subduction zone strongly oblique to the old Triassic–Early Jurassic system. It also remains possible that a combination of both explanations occurred: The ophiolite formed elsewhere in response to J1 cusp–induced plate boundary reorganization and was then translated into its current position.

The preservation of transform/fracture-zone assemblages at the leading edge of the extending forearc lithosphere is possible and may constitute our best evidence for subduction initiation along a large-offset transform/fracture-zone boundary. Nonetheless, preservation of this assemblage also depends on the continued history of the plate boundary, the extent of subduction erosion before accretion begins, and the extent of mixing with upper-plate lithologies.

Our data show that the proto-Franciscan subduction zone formed in response to subduction initiation ca. 172 Ma, most likely along an oceanic fracture zone–transform fault system. This conclusion is supported by lithologic assemblages within the Coast Range ophiolite—a suprasubduction-zone ophiolite complex characterized by forearc basalts, volcanic arc basalts, and boninites, and by plutonic complexes related to the same parent magma suites as the volcanic rocks. This assemblage correlates with the first two stages of the ophiolite life cycle (birth-youth) proposed by Shervais (2001) that are characteristic of subduction initiation, and which establish the Coast Range ophiolite as a subduction initiation indicator. Further, the observation of mixed provenance in the underlying Tehama-Colusa mélange, and the low pyroxene equilibration temperatures of the peridotites are consistent with formation along an active oceanic fracture zone–transform fault system that became the leading edge of the extending upper plate.

We wish to thank our colleagues for stimulating and insightful discussions—including Robert Stern, Rodney Metcalf, Henry Dick, John Wakabayashi, Sam Mukasa, and Marlon Jean—and cogent reviews by Jason Saleeby and Trevor Dumitru. This work was supported in part by National Science Foundation award EAR-44025 to Shervais.