Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

Earth’s crust is primarily generated in oceanic settings via divergent and convergent margin processes. In divergent margins, the crust is ultimately recycled back into the mantle, whereas at convergent margins, the resulting island arc crust becomes the buoyant nucleus of new continental crust. The collision of island arcs with continental margins is well accepted as a dominant process of post-Archean continental growth (e.g., Rudnick and Gao 2003). The goal of this paper is to describe what we know about island arc crust from exposed crustal sections, and summarize the inferred processes that produce this crust and its characteristics. From this, we can also better understand modern arcs, and the nature of continental crust.

The net magmatic input into island arc crust at convergent margins is dominated by fluid-fluxed melting of the mantle wedge to produce basalt (e.g., Tatsumi 2005). However, net magmatic output at arc volcanoes is commonly more silicic than basalt (andesite, e.g., Gill 1981). Thus, the arc crust acts as a distiller to modify mantle-derived magmas. The geochemistry and diversity of the crystal cargo of extrusive products of arc magmatism lead us to inferences about those distillation processes, but complementary evidence from plutonic exposures is usually lacking. In this paper, the exposed crustal sections of paleo-arcs provide both the intrusive and extrusive record of these processes.

In linking the formation of island arcs to the formation of continental crust, a few central points about continental crust are important to consider. Continental crust is high-standing and thick, with an average “andesitic” bulk composition (Rudnick and Gao 2003; Taylor and McLennan 1985, and references therein). The best estimates show that upper continental crust is granodioritic in composition (66.6 wt.% SiO2, 2.5 wt.% MgO), rich in incompatible elements, and depleted in compatible elements (Rudnick and Gao 2003). Their mid-crust estimate is intermediate in composition (63.5 wt% SiO2, 3.6 wt% MgO), and their lower crust is mafic (53.4 wt% SiO2, 7.2 wt% MgO). The weighted average produces an andesitic bulk composition of 60.6 wt% SiO2 and 4.7 wt% MgO. The bulk continental crust estimate is depleted in Nb, with a subchondritic Nb/Ta ratio (12.4; Rudnick and Gao 2003). Calculations using this Nb depletion suggest that at least 80% of the continental crust was generated in convergent margins (Barth et al. 2000; Plank and Langmuir 1998).

If we accept the compositional estimate of Rudnick and Gao (2003), as well as the convergent margin origin for continental crust, then study of exposed crustal sections of island arcs becomes even more relevant. First, the compositional stratification in continental crust is also present in exposed island arcs, and may be related to similar processes. Second, an andesitic bulk composition complicates the simple premise of one-stage melting of the mantle to produce continental crust. Melts of the mantle are basaltic, thus some return flow of mafic components to the mantle are required to balance the geochemical budget of continental crust formation. Several hypotheses have been proposed for this (see summary in Rudnick and Gao 2003). Exposed island arc crustal sections can be used to test these hypotheses, especially with respect to evidence for crustal foundering (delamination).

However, there are some important distinctions between modern island arcs and continental crust as shown by modern arc seismic signatures. Calvert (2011) shows that at depths >8–10 km, island arcs have higher seismic velocities (i.e., have more mafic compositions) than continental crust, whereas at depths <8–10 km, seismic velocities are similar.

From the arc crustal sections described in this paper, we discuss differentiation processes that may account for compositional stratification within arcs. We also discuss evidence from these sections that strongly support lower crustal foundering of mafic and ultramafic compositions to produce a more felsic arc crustal composition. Finally, we relate this information to interpretation of geophysical signatures in arcs.

2 Exposed Crust and Upper Mantle Sections of Mesozoic Arcs

We summarize four exposed cross sections of accreted arcs (Fig. 5.1), three of which are island arcs and one which was built on a newly accreted continental margin. Some sections are more complete than others, and each are different thicknesses. However, all arc sections share common mid-crustal lithologies, and two of the four share common upper mantle and lower crustal lithologies. All share common lithological stratification with depth. Taken together, the petrologic story that each of these sections tells corroborates the others, and provides us with an unprecedented view into the workings of magmatic arcs.

Fig. 5.1
figure 1

Schematic lithologic sections of the four accreted arcs discussed in the text. Colors are meant to be broad generalizations of lithologic types. Grey is volcanic rocks (all compositions), tan is intermediate to felsic plutonic rocks (53–72 wt% SiO2), red is gabbroic rock (typically cumulate) (45–52 wt% SiO2), purple is ultramafic rock (pyroxenite, wehrlite, dunite and harzburgite) (<52 wt% SiO2). Orange regions show crustal levels where partial melting has taken place. Green areas represent abundant metamorphic country rock that has not melted. The paleo-Moho is defined in this study as the transition between ultramafic rock (plagioclase absent) and gabbroic rock (plagioclase present).

Full size image

In this paper, we define the crust-mantle boundary as the boundary between plagioclase-bearing and plagioclase-free lithologies. Thus the “Moho” described in the following sections marks the contact between gabbroic rocks above and ultramafic rocks below. Gabbroic rocks in the lower crust include lithologies such as gabbronorite (variably called two-pyroxene gabbro or two-pyroxene granulite by different authors) and garnet gabbro (also called garnet granulite). These lower crustal gabbroic rocks are understood to be cumulate in nature. Ultramafic rocks below the Moho are comprised of varying proportions of olivine and pyroxene. They may be residual mantle that has interacted with percolating melts, or purely cumulate.

In this paper we also use the term felsic to describe rocks with >63 wt% SiO2 (extrusive dacite or intrusive tonalite), intermediate to describe rocks with ~52–63 wt% SiO2 (extrusive andesite or intrusive diorite), and mafic to describe rocks with ~48–52 wt% SiO2 (extrusive basalt to intrusive gabbro).

2.1 Talkeetna Arc, Alaska

The Jurassic Talkeetna island arc in south-central Alaska is an exhumed and tilted arc section where subarc mantle, lower crust, and middle and upper crust are now exposed at the surface (Burns 1985; DeBari and Coleman 1989). The arc is estimated to be exposed along at least 1,000 km of strike length (Plafker et al. 1989; Reed and Lanphere 1973) from the Chugach and Talkeetna Mountains in the east to the Lower Cook Inlet Region and Alaska Peninsula in the west (Fig. 5.2a).

Fig. 5.2
figure 2a

(a) Map of the location of the Talkeetna arc in Alaska from Greene et al. (2006). The arc extends for over 1,000 km in an east west direction. The easternmost segment of the arc exposes the deepest crustal levels, and that region is shown in the most detail. The paleo-Moho is exposed in the far eastern region on Bernard and Scarp Mountains. Thick sections of cumulate gabbroic rock are exposed in the eastern part of the arc. The western part of the arc on the Alaska Peninsula exposes a much thicker section of intermediate and felsic plutonic rocks (Johnsen 2007; Reed and Lanphere 1973). (b) Detailed crustal section for the Talkeetna arc based on stratigraphic thickness of the volcanic section (Clift et al. 2005a) and geobarometry of Hacker et al. (2008). The paleo-Moho is defined as in Fig. 5.1 and the Bernard Mountain Moho section is shown in detail. A simplified cross section from Greene et al. (2006) is also included.

Full size image
Fig. 5.2
figure 2

(Continued)

Full size image

The arc is interpreted to have formed in an oceanic subduction zone setting as a result of the northward dipping subduction (present-day coordinates) of the Farallon plate beneath the Peninsular terrane (Clift et al. 2005b; Plafker et al. 1989). Talkeetna arc magmatism is believed to have been active from ~200 Ma to at least 160 Ma (Plafker et al. 1989; Rioux et al. 2007, 2010). A trondhjemite pluton, dated at ~153 Ma, intrudes the central Talkeeetna Mountains, possibly marking the end of arc plutonism (Rioux et al. 2007). Cessation of magmatism, along with evidence for deformation, exhumation, and sedimentation at 160–150 Ma, is thought to reflect the collision of the Talkeetna arc with a tectonic block to the north (Clift et al. 2005b; Trop et al. 2005). It is unclear if the collision was between the Talkeetna arc and another terrane (Wrangellia) or the accretion of the entire amalgamated terrane onto the margin of North America.

Clift et al. (2005b) interpret northward migration of magmatism in the arc to be due to tectonic erosion of the forearc while the arc was active. They interpret the juxtaposition of younger accreted trench sedimentary rocks (Chugach terrane) against the base of the Talkeetna arc sequence as a change from a state of tectonic erosion to accretion after collision-induced increase of sedimentary flux at ~160 Ma.

The arc crustal section, with an inferred original thickness of ~35 km (DeBari and Sleep 1991; Hacker et al. 2008) sits above a residual mantle section that includes ultramafic cumulate rocks. Pyroxenites and garnet-bearing gabbros are characteristic of the lowermost crust, layered gabbronorite in the lower and middle crust, gabbroic rocks and intermediate-felsic plutonic rocks in the middle-upper crust, and volcanic rocks in the uppermost crust. The arc is not physically exposed as a contiguous section, but most stratigraphic levels are present in plutonic and volcanic exposures in the Chugach and Talkeetna Mountains. There are no exposures of remnant oceanic crust upon which the arc was built. This led Clift et al. (2005a, b) to assume that the arc was extensional.

The upper crust of the arc includes a 7-km-thick volcanic section (Clift et al. 2005a) intruded by felsic to intermediate tonalites and quartz diorites. Hacker et al. (2008) infer that these upper crustal plutonic rocks crystallized at 0.13–0.27 GPa (5–9 km), which agrees with the observed stratigraphic thickness of the volcanic rocks (Fig. 5.2b). These volcanic and shallow plutonic rocks are exposed over an enormous area of the Talkeetna Mountains and the Alaska Peninsula.

The middle crust of the arc is dominantly hornblende gabbronorite in its lower part, but upper middle crust lithology (9–15 km) is not well constrained due to a gap in geobarometric data (Hacker et al. 2008). In contrast, rocks of the lower middle crust and lower crust are layered gabbronorites (2-pyroxene gabbro) that are exposed semi-continuously for approximately 150 km in an E–W direction and 5–18 km in a N–S direction (Fig. 5.2a). These rocks were described by Greene et al. (2006), who evaluated their role as cumulates in arc differentiation. Hacker et al. (2008) infer that these plutonic rocks crystallized and equilibrated at 0.43–0.72 GPa (15–24 km) and 700–900°C (Fig. 5.2b). A klippe to the south of the main body of the arc consists of diorites with igneous garnet that crystallized at 0.7–1.0 GPa (24–34 km depth).

The lowermost crust and Moho of the arc is exposed in the far eastern part of the arc near Tonsina at Bernard and Scarp Mountains (Fig. 5.2a). The lithologies consist of gabbronorites ± hornblende ± garnet. Anhydrous assemblages give way to more abundant hornblende gabbronorites upward in the section. These lowermost crustal rocks yield equilibration pressures of 0.9–1.2 GPa (Hacker et al. 2008).

The Moho of the arc is best exposed on Bernard Mountain as a transitional boundary between garnet gabbros above (up to 10 volume % garnet) and cumulate pyroxenites below. These garnet gabbros crystallized at ~1.0 GPa and 875–1,000°C (30–35 km, Fig. 5.2b) (DeBari and Coleman 1989; Hacker et al. 2008). Beneath the paleo-Moho, 50–200 m of websterite and olivine clinopyroxenite overlie an ~100 m transition zone of coarse chromite-bearing dunite. Beneath that is ~2,500 m of residual spinel harzburgite with interlayered dunite. Residual mantle rocks have Moho-parallel foliation and display stretching lineations indicative of flow paralled to the arc axis (Mehl et al. 2003).

Studies focusing on the geochemistry of the Talkeetna arc have shown that lower crustal gabbronorites and upper crustal volcanic rocks can be related to a single type of parent magma, each group of rocks forming as a result of simple fractional crystallization (DeBari and Sleep 1991; Greene et al. 2006; Kelemen et al. 2003). Gabbronorites represent the crystallized cumulate pile and erupted volcanic rocks represent the remaining liquid following differentiation. Fractional crystallization was also used to explain the formation of middle-upper crustal intermediate and felsic plutonic rocks (Rioux et al. 2007, 2010; Johnsen 2007). Isotopic signatures from these intermediate-felsic plutonic rocks are mostly intra-oceanic with little or no involvement of continental crustal material (Amato et al. 2007; Rioux et al. 2007, 2010). However, in the westernmost end of the Talkeetna arc on the Alaska Peninsula, Johnsen (2007) showed that the youngest felsic plutonic rocks were probably produced by variable mixing of an andesitic parent magma with a partial melt of arc lower crust.

The bulk composition of the Talkeetna arc was originally calculated to be basaltic by DeBari and Sleep (1991) (51 wt% SiO2 and 11 wt% MgO). However, their mass balance calculations may have over-estimated the ultramafic cumulate component. Greene et al. (2006) calculate a “primary magma” composition for the arc that would be in equilibrium with the mantle, based on the most primitive compositions erupted. This primary magma composition is also basaltic, and is similar to the bulk composition of DeBari and Sleep (1991). Hacker et al. (2008) also calculated a bulk composition based on their geobarometric reconstruction of the arc. They calculated a potential range of compositions from basalt to andesite (51–58 wt% SiO2) and 4–11 wt% MgO. The wide range is due to extrapolations on a “missing” mid crustal component.

2.2 Kohistan Arc, Pakistan

The Kohistan arc of NE Pakistan represents an island arc that was obducted between the collision of the Indian and Eurasian plates. The arc extends for at least 300 km in an east–west direction, from the Nanga Parbat massif to northeastern Afghanistan. The arc is juxtaposed against the Asian plate to the north along the Karakorum-Kohistan Suture, and against the Indian plate to the south along the Indus Suture.

The arc developed in response to northward-directed subduction of the Tethyan lithosphere at the leading edge of the Indian plate during latest Jurassic to Cretaceous times (the oldest pluton is 154 ± 0.6 Ma, Schaltegger et al. 2003). The arc underwent a rifting episode at ~85 Ma to produce the voluminous Chilas mafic intrusion (Khan et al. 1989; Schaltegger et al. 2002; Burg et al. 2006). Continued subduction resulted in complete consumption of the leading oceanic edge of the Indian plate, resulting in obduction of the arc onto the Indian continent at the site of the Indus suture (~60–45 Ma) (Burg (2011) and references therein). The timing of collision of the Kohistan arc on its northern end with the Karakorum (Eurasian) plate is controversial. It has long been thought that this collision pre-dated Kohistan arc rifting (e.g., Petterson and Windley 1985) such that after 100 Ma the Kohistan arc was an Andean-type continental arc. However, more recent work suggests that this northern collision post-dates collision of Kohistan with India (e.g., Khan et al. 2009, Burg 2011). It is unclear whether the Kohistan arc was ever Andean in nature.

The arc was tilted to the north during final docking of terranes. Subsequent uplift and erosion have exposed a cross section that consists of rocks from the upper mantle through the lower, middle and upper crust (Fig. 5.3). As with the Talkeetna arc, forearc sequences in the Kohistan arc are rare, suggesting tectonic erosion of the forearc may have occurred during arc magmatism or subsequent collision.

Fig. 5.3
figure 3

(a) Simplified cross section through the Kohistan arc from Burg et al. (2006) and Burg (2011). The main units, the Southern Amphibolites, the Chilas Complex, and the Kohistan Batholith are described in the text. (b) Detail of the Moho section (south of the Southern Amphibolites) from Burg (2011) and Dhuime et al. (2007). GH is garnet-hornblendite/pyroxenite and hornblende-garnetite rocks. The inset map shows the location of the Kohistan arc along the suture between the Indian and Asian plates.

Full size image

A detailed description of the exposed crustal levels can be found in Burg (2011). These are briefly reviewed below.

The shallowest part of the arc section consists of volcanic, sedimentary, and shallow-level granitic rocks in the northern part of the exposure. The volcanic rocks record all phases of arc evolution from intraoceanic arc, continental arc, and arc rifting (Petterson and Treloar 2004). The volcanic and sedimentary rocks are intruded by granitoids of the Kohistan batholith, the oldest of which predate suturing of Kohistan to Eurasia (154 ± 0.6 Ma, Schaltegger et al. 2003). Arc-related plutons of the batholith span ages from 112 to 38 Ma (Jagoutz et al. 2009; Petterson and Windley 1985, 1991; Schaltegger et al. 2003).

The middle crust of the arc is separated into a northern section (middle crust represented by calcalkaline plutons of the Kohistan batholith) and a southern section (middle crust represented by metadiorites, metagabbros, metasediments, and metavolcanics of the Southern Amphibolite Belt) (Fig. 5.3a). The intervening Chilas Complex is a massive body of layered gabbroic rocks up to 50 km wide (N–S) and 300 km long (E–W). This Complex is interpreted to be an intrusive body related to rifting of the Kohistan arc at ~85 Ma (Burg et al. 2006; Jagoutz et al. 2006, 2007; Khan et al. 1989) and thus is not strictly part of the arc crustal section. The Southern Amphibolite belt includes the Kamila amphibolite, and is 15–35 km thick. Some metabasalts within this belt are interpreted to be the oceanic remnant oceanic crust upon which the arc was built (Jan 1988; Khan et al. 1993), whereas others are related to arc magmatism (Burg 2011). Partial melting within this belt is interpreted to have occurred at ~97 Ma based on the presence of a granitic migmatite (Schaltegger et al. 2002).

The lowermost crust of the arc is exposed to the south of the Southern Amphibolite belt and consists of garnet-gabbros (granulites) that overlie and intrude the residual/cumulate rocks of the subarc mantle (Burg et al. 1998). These garnet-gabbros are ~5–6 km thick in the Jijal Complex (Fig. 5.3b). They consist of ~20–30 vol% garnet, plus plagioclase, clinopyroxene, amphibole, and oxides. Rare two-pyroxene gabbros from this section were dated at 118 ± 12 Ma by a Sm–Nd mineral isochron (Yamamoto and Nakamura 2000). They are overlain by metagabbro (98.9 ± 4 Ma, Schaltegger et al. 2002) that forms the southernmost part of the Southern Amphibolite belt.

The 2–3-km thick ultramafic section of the Kohistan arc is found in the Jijal and Sapat Complexes along the southernmost margin of the arc (Fig. 5.3b). Dunite, wehrlite, olivine-clinopyroxenite, and websterite lithologies are the most abundant lithologies in this basal section. Pyroxenite increases in abundance up section, as does the presence of amphibole. Pyroxene-rich ultramafic rocks grade into hornblende-, clinopyroxene-, and garnet-rich rocks. Modal abundances vary dramatically in short distances, with complex relationships between hornblendite, garnetite, and pyroxenite. Above these rocks, plagioclase appears in gabbronorites with Mg# <0.60 (Garrido et al. 2006). Clinopyroxene from the ultramafic rocks yields a Sm–Nd isochron of 117 ± 7 Ma, interpreted as a crystallization age (Dhuime et al. 2007). This section has some residual harzburgite (Bouilhol et al. 2009; Dhuime et al. 2007; Jan and Howie 1981; Miller et al. 1991; Burg 2011) and is interpreted as a melt-mantle reaction zone beneath the base of the arc (e.g., Burg et al. 1998; Dhuime et al. 2007).

The Kohistan arc preserves a geochemical record that spans arc inception, arc thickening, and subsequent uplift (Dhuime et al. 2007; Yoshino and Okudaira 2004). Metamorphic P-T-t paths of lower crustal rocks clearly indicate arc thickening due to magmatic input (Yoshino and Okudaira 2004; Yoshino et al. 1998). However, there is no consensus as to whether the arc underwent extensive intracrustal differentiation through melting of the lower crust (e.g., Garrido et al. 2006) or whether voluminous intermediate to felsic magmas of the upper crust (Kohistan batholith) are dominantly a result of fractional crystallization processes (Jagoutz 2010). In either case, more mafic rocks of the Kohistan batholith preserve geochemical evidence for evolution throughout the history of the collisional origin (increasing crustal component, Petterson et al. 1993; Petterson and Windley 1991).

2.3 Bonanza Arc, Vancouver Island

The Bonanza arc (500 km strike length) is a Jurassic arc crustal section (middle to upper crust) exposed on Vancouver Island and the Queen Charlotte Islands, Canada (DeBari et al. 1999) (Fig. 5.4). The arc is part of the much larger Wrangellia terrane that was accreted to western North America during the latest Jurassic to early Cretaceous. The arc is thought to be the southern extension of the Talkeetna arc described in Sect. 5.2.1 above (Plafker et al. 1989; Clift et al. 2005b).

Fig. 5.4
figure 4

(a) Map of Vancouver Island showing the main units of the Jurassic Bonanza arc (from DeBari et al. 1999). The deepest intrusive level of the arc is represented by the Westcoast Complex, the intermediate level is represented by the Island Intrusions, and the shallowest level is represented by the Bonanza Volcanics. (b) Schematic crustal section of the arc showing intrusive relationships between plutons and pre-existing country rock.

Full size image

The Bonanza arc was built on a thick oceanic basement comprising distinctive mid-Paleozoic arc volcanic rocks (Sicker Group), Pennsylvanian-Permian clastic rocks and limestones, and a remarkably uniform 6,000-m-thick sequence of Late Triassic tholeiitic flood basalts of the Karmutsen Formation (e.g., Greene et al. 2009 and references therein). These basalts are in turn overlain by Late Triassic-Early Jurassic shallow to deep-water carbonates and shales. The arc developed into and on this basement in response to eastward-directed subduction of Pacific ocean lithosphere during Early to Middle Jurassic times (Armstrong 1988).

Unroofing of the arc began with deposition of detritus derived from Bonanza Group volcanic rocks to form Middle Jurassic (Callovian) and Lower Cretaceous (Valanginian) conglomerates. This unroofing continued with Late Cretaceous granitoid-volcanic conglomerate deposited directly on unroofed Middle Jurassic Island Intrusions (Muller et al. 1974). Accretion to the North American margin may have occurred at this time (Monger et al. 1994). Afterwards, subduction continued under an Andean-type continental margin, and the locus of magmatism swept eastward to form the Late Jurassic to Cretaceous Coast Plutonic Complex. Late Cretaceous and younger contraction and possible sinistral transcurrent faulting along the continental margin further dismembered the arc (Monger et al. 1994; Plafker et al. 1989).

The exposed arc can be divided into discrete stratigraphic levels (DeBari et al. 1999). The Westcoast Complex represents the deepest exposed levels of the arc and displays a wide variety of lithologic types, including metamorphosed and migmatized segments of pre-existing crust (Sicker Group rocks) and mafic to intermediate plutonic rocks (190.3–186.6 Ma) representative of mantle-derived magmas (diorites to gabbros) (DeBari et al. 1999). The unit is composed of a heterogenous mixture of multiple magma types, partial melts, and restites. Stratigraphy of the pre-existing crust suggests that the deepest levels of the exposed arc represent depths of ~15–20 km (DeBari et al. 1999). This unit may have acted as a deep crustal “filter” where mantle-derived magmas stalled, fractionated, and interacted with older crust.

The higher levels of the arc are represented by the plutons of the Island Intrusions, whose ages overlap those of the Westcoast Complex (DeBari et al. 1999). These plutonic rocks are more texturally homogeneous than the deeper Westcoast Complex. They include diorite, quartz diorite, and granodiorite that intrude the Triassic Karmutsen Formation and overlying sediments. The lack of migmatized country rock and the general homogeneity of the plutons indicate that the mixing and homogenization processes were more or less complete before migration of magmas to these shallower crustal levels.

The Bonanza Group volcanic rocks record effusive magmatism and mark the highest crustal level of the arc. They are a thick sequence (up to 2,500 m) of lava flows, pyroclastic deposits, and thin interbedded sedimentary rocks. They conformably overlie Late Triassic-Early Cretaceous sediments and are intruded at their deepest levels by plutons of the Island Intrusions (Fig. 5.4).

Migmatitic partial melts of amphibolite country rock in the deep crust of the arc are tonalitic to trondhjemitic in composition. They have a very distinctive chemical signature, with depletions of Nb, Y, Zr, and the intermediate- and heavy-REE relative to the typical plutonic rocks of the Island Intrusions or WCC. These tonalites and trondhjemites were able to coalesce and migrate from their site of generation in the deep crust to shallower levels in the crust.

2.4 Southern Coast Plutonic Complex: WA Cascades

The >1,500-km-long Coast Plutonic Complex of British Columbia, and its southern extension in the crystalline core of the North Cascades, represents a voluminous Cretaceous to Paleogene magmatic arc (Armstrong 1988; Tabor et al. 1989) (Fig. 5.5). In Washington State, exposed plutons (96–45 Ma) represent various crustal levels of the arc from deep (~20–35 km) to shallow (~7 km) (Miller and Paterson 2001) (Fig. 5.5d). The base of the exposed section represents paleodepths of ~40 km; but the Cretaceous Moho was thought to be at depths >55 km as a result of arc contraction (Miller and Paterson 2001). Regional shortening resulted in crustal thickening and burial of supracrustal rocks to great depths of 25 to >40 km (Matzel et al. 2004; Valley et al. 2003; Whitney et al. 1999).

Fig. 5.5
figure 5

Maps (a–c) showing region and detail of the southern Coast Plutonic Complex in Washington State after Miller et al. (2009). Schematic crustal section in (d) is from Miller et al. (2009) and is based on structural thicknesses and geobarometric constraints.

Full size image

Arc plutons that range in age from 96 to 45 Ma intrude oceanic terranes of varying affinity that had only recently been accreted to each other (by 96 Ma) and to the margin of North America (Umhoefer and Miller 1996). From upper to lower paleodepths these include the Ingalls Ophiolite, the Chiwaukum schist (accretionary complex), and the Napeequa Complex (ocean floor assemblage) (Fig. 5.5d). Based on equilibration pressures, the Swakane gneiss is at the bottom of the stack (0.9–1.2 GPa; Matzel et al. 2004), but it has not been intruded by arc plutons. Matzel et al. (2004) conclude that Swakane protolith was arc-derived clastic sediments that were buried and metamorphosed after the most voluminous pulse of plutonism.

The base of the arc section at ~35 km comprises intermediate (tonalitic) plutons that intrude amphibolite facies (<700°C) country rock [e.g., (Miller et al. 2009)] (Fig. 5.5d). Miller and Paterson (2001) interpret low temperatures at such deep levels (in contrast to thinner arcs) to be a result of rapid crustal thickening. These low temperatures are responsible for the lack of migmatization observed in host rocks at the 30–35 km level. However, cryptic evidence for host rock melting at deeper, currently unexposed levels of the arc is preserved in geochemical signatures in the tonalitic rocks (Dawes 1993; Parent 1999; Miller et al. in preparation).

Tonalite is the most abundant plutonic lithology at all levels of the crustal section (10–30 km). However, deeper plutons (>0.6 GPa, ~20 km) commonly display evidence for broadly synchronous mafic magmatism (Miller and Paterson 2001). These mafic compositions are most abundant as sheeted zones at the margins of the plutons and suggest that mafic magmas, with minor crustal component, predominated at the early stages of pluton emplacement. As the plutons evolved, magma volume generated by partial melting at deep levels in the crust increased (Miller et al. in preparation). These crustal melts mixed with the mantle-derived magmas, producing hybrid compositions. Sheet-like emplacement may have continued, but the volume of magma increased and partially erased its record. Throughout the history of these plutons, mafic magmas and tonalitic hybrid magmas were injected simultaneously, as shown by intermingled mafic enclaves and mafic sheets (see later discussion).

Plutons that crystallized at shallower levels (< 0.6 GPa) are more homogeneous, presumably because hybridization was completed at greater crustal depth (Miller and Paterson 2001). These shallower plutons, and their erupted counterparts, may not record the cryptic history and variety of sources necessary for their genesis.

3 Depth-Specific Processes: Pre-collision

The sections described above expose multiple levels of island arc crust. Some expose only parts of the section, some expose more complete traverses. But taken together, these four sections show very clear commonalities and provide a fairly complete picture of arc crust from top to bottom (Fig. 5.1).

3.1 Crystal Fractionation and Production of Pyroxenite and Px-Gabbro Assemblages Near the Moho

In both the Kohistan arc and the Talkeetna arc, the crustal section includes paleo-Moho exposures. In the field, the Moho is clearly observed as a relatively sharp (10–100s of meters) boundary between plagioclase-bearing and plagioclase free lithologies [e.g., see photo in Fig. 10.10 of Burg (2011)]. It is a high temperature contact, where fabric on both sides of the Moho displays evidence for high-temperature flow (e.g., Burg et al. 2005; Mehl et al. 2003)

In exposures beneath the paleo-Moho, minor dunites and abundant pyroxenites (Fig. 5.6), indicate early crystal fractionation of olivine and pyroxene. At both Kohistan and Talkeetna, pyroxenite veins, dikes, and layers increase in abundance up section, such that cumulate pyroxenite is the dominant lithology near the Moho. Cumulate pyroxenite is ~500 m thick near the paleo-Moho exposures in Talkeetna (Fig. 5.2b), and several km thick in Kohistan (Fig. 5.3b).

Fig. 5.6
figure 6

Photographs of pyroxenite layers in the mantle sections of the Talkeetna and Kohistan arcs. (a) Shows sub-parallel layers of pyroxenite (dipping ~70°) in dunite at Talkeetna, with one layer adjacent to the hammer. (b) Shows sub-parallel pyroxenite layers and one cross-cutting pyroxenite layer within dunite at Kohistan. In the latter example, the pyroxenite stands out from the dunite background.

Full size image

Above the Moho, in the lower and mid crust of the Kohistan and Talkeetna arc sections, the plagioclase-bearing crystal assemblage is a cumulate consisting of plagioclase, clinopyroxene, orthopyroxene, oxide (Mg–Al spinel or Fe–Ti oxide), ± amphibole, ± garnet. Note that this assemblage is olivine free, indicative of the relatively high pressures at which this crystal fractionation occurred (~0.5–1.0 GPa, Hacker et al. 2008, Yoshino et al. 1998). However, some pyroxene + spinel-rich assemblages in the lower crust of the Talkeetna arc are thought to have formed from olivine + plagioclase assemblages crystallized at 0.1–0.6 GPa that later underwent pressure increase to ~1 GPa (Hacker et al. 2008).

In both arc sections, fractionation of the observed cumulate rocks was dominantly responsible for producing the more differentiated compositions in the upper sections of the respective arcs (Greene et al. 2006; Jagoutz 2010). Qualitatively, this can be seen in Fig. 5.7a, where positive Eu and Ti abundances in the cumulate rocks (from accumulation of plagioclase and oxides) are mirrored by low abundances of these elements in the more differentiated volcanic rocks. Ratios of Zr/Sm, V/Ti, Sr/Y, and Ti/Zr in the cumulate rocks are matched by reciprocal ratios of these elements in the volcanic rocks with increasing SiO2 (Fig. 5.7b)

Fig. 5.7
figure 7

(a) Normalization diagrams for volcanic rocks (purple), intermediate-felsic plutonic rocks (green), chilled mafic rocks (blue), and cumulate gabbronorite (red) from Talkeetna. Note that where some positive spikes (enrichment) occurs in cumulate gabbronorite (e.g., Eu, Ti), corresponding negative spikes (depletion) occurs in more differentiated rocks in the upper crust. Corresponding enrichment and depletion is shown more quantitatively for trace element ratios in (b). Figures are from Greene et al. (2006).

Full size image

In the Talkeetna arc, Greene et al. (2006) quantitatively modeled the fractionation process using observed cumulate assemblages. The rare earth element (REE) abundances in clinopyroxene in the cumulate rocks were used to calculate REE abundances of liquids in equilibrium with those cumulates. The calculated equilibrium liquids closely match observed compositions in the volcanic rocks in the upper crust of the arc. Greene et al. (2006) also modeled the differentiation trend of the upper crustal volcanic rocks from basalt to andesite to predict calculated modes and compositions of fractionated minerals. Those predicted modes and compositions were a striking match to the observed cumulates in the lower crust. The basalt to andesite trend in the volcanic rocks was reproduced by ~73% fractionation of the observed lower crustal cumulate rocks.

Jagoutz (2010) modeled the fractionation process in the Kohistan arc by mass balance using observed volumes of cumulate rocks (dunite, wehrlite, pyroxenite, gabbro). He was able to produce the bulk of the upper crustal granitoids after 80% fractionation of the cumulate rocks, with some assimilation of leucocratic rock produced by deep crustal melting.

One critical outcome of the Greene et al. (2006) modeling in the Talkeetna arc is that a substantial section of cumulate pyroxenites must be missing from the base of the arc (Kelemen et al. 2003). These results are supported by the thermodynamic modeling of Behn and Kelemen (2006). In order to bring the most mafic Talkeetna liquid composition in Fe/Mg equilibrium with the mantle, a much larger, and more Mg-rich section of cumulate pyroxenites would be necessary. Even the most Mg-rich cumulates currently exposed in the arc were fractioned from liquids that had already themselves been fractionated. Thus, some fractionated cumulate material should be present beneath the exposed crustal section. Fractionation of the parental basaltic magma could have occurred at deeper levels in the mantle than exposed. However, modeling by Greene et al. (2006) shows that any mantle-derived parental basaltic magma must have fractionated ~25% pyroxene before it reached the paleo-Moho. Given that cumulate pyroxenites increase upward (toward the Moho) in the observed mantle section, most pyroxenites were more likely fractionated near the Moho, and should be exposed there. As discussed below, it is likely (and physically reasonable) that missing pyroxenite cumulates were removed by lower crustal foundering (delamination).

3.2 Densification of Crust and Lower Crustal Foundering at the Moho

In both the Talkeetna arc and the Kohistan arc, assemblages near the paleo-Moho are garnet bearing. These include garnet-pyroxenites as well as garnet gabbros (referred to as granulites by some authors) (Fig. 5.8). The garnet gabbros in both arcs form irregular veins and lenses within gabbronorite, and garnet percentages range from 1 to 40 volume %. Kohistan has a much thicker preserved garnet-bearing section than Talkeetna (~5 km vs. ≪1 km), but as discussed below, the Talkeetna arc may once have had a garnet-bearing section of comparable thickness.

Fig. 5.8
figure 8

Garnet-bearing lithologies from the Kohistan arc (a and b) and Talkeetna arc (c). The photo in (a) shows Kohistan garnet-hornblende pyroxenite in mantle rocks, whereas (b) and (c) show lower crustal garnet gabbros from Kohistan and Talkeetna, respectively. Reddish mineral grains are garnet, plagioclase is white, hornblende and pyroxene are black.

Full size image

The Jijal Complex at the base of the Kohistan arc includes a 5-km thick section of garnet gabbros equilibrated at 700–950°C and >1 GPa (Yamamoto 1993). There is no consensus on whether these garnet gabbros formed during igneous cooling (Ringuette et al. 1999), postmagmatic heating and crustal thickening (Yamamoto and Yoshino 1998; Yoshino and Okudaira 2004; Yoshino et al. 1998) or dehydration melting of hornblende (Garrido et al. 2006). In the Talkeetna arc, the garnet-bearing gabbros at the base of the section (at Bernard Mountain) formed through subsolidus cooling (DeBari and Coleman 1989). They record conditions at the base of the arc section of 900°C–1,000°C and ~1 GPa (Hacker et al. 2008).

The presence of garnet bearing lithologies at the Moho in both crustal sections has important implications. At pressures of 1–1.2 GPa, the slope of the garnet-in reaction for gabbronorites such as those in Kohistan and Talkeetna is nearly parallel to likely arc geotherms, making the stability of garnet highly sensitive to small changes in temperature (Fig. 5.9) (Behn and Kelemen 2006; Jull and Kelemen 2001). A shift from 1,000 to 800°C at 1 GPa will result in a density increase of more than 250 kg/m3 due to the formation of garnet at lower temperature (Behn and Kelemen 2006). Thus as 2-pyroxene gabbroic rocks cool at 1 GPa, or if the arc thickens, formation of garnet should be a common process. Jull and Kelemen (2001) show that Talkeetna garnet gabbros will be ~100 kg/m3 denser than the underlying mantle at 800°C. The density contrast will be even greater for gabbronorites with Mg# >80 (Behn and Kelemen 2006). In Kohistan, no primitive gabbronorites with Mg# >80 are observed in the arc section, providing evidence that those rocks may have been lost to the mantle (Behn and Kelemen 2006).

Fig. 5.9
figure 9

Calculated density contrast (Δρ, kg/m3) from Jull and Kelemen (2001) between crustal rock and pyrolite mantle. The colored regions are those in which the density contrast is positive, and hence the rock would be denser than the underlying mantle. In (a) the crustal rock is olivine clinopyroxenite and in (b) the crustal rock is arc gabbronorite. Both lithologies are typical of Talkeetna arc rocks near the Moho (1 GPa, 800–1,000°C). At those P-T conditions, these arc cumulates would have been denser than their underying mantle.

Full size image

The presence of pyroxenite near the Moho also has important implications for lower crustal stability. Pyroxenite, even without garnet, is predicted to be denser than residual peridotite when the proportion of plagioclase is less than 6–8% (Behn and Kelemen 2006; Jull and Kelemen 2001). Thus the density contrast between pyroxenite and residual peridotite is less sensitive to temperature. Consequently, while pyroxenite is denser than residual mantle at 1,000°C and 1 GPa, gabbronorite will be less dense than mantle at these higher temperatures. Jull and Kelemen (2001) show that a 10 km thick ultramafic cumulate layer at the base of the crust at ~1,000°C would need 10 Ma to delaminate. But if temperatures or strain rates are higher, then ultramafic cumulate layers a few hundred meters thick may become unstable as quickly as they form. Arc gabbronorites on the other hand, need lower temperatures (~800°C).

Thus, the discrepancy between the observed proportions of pyroxenites near the Moho in the Talkeetna arc (<5% of the arc section) and the proportion required by crystal fractionation modeling described above (>25%) can be explained by gravitational instability. Dense ultramafic cumulates, probably together with dense garnet gabbros, may have foundered into the underlying mantle during the time that the Talkeetna arc was magmatically active, or in the initial phases of slow cooling (and sub-solidus garnet growth). Given the “missing” pyroxenites, the Talkeetna arc lower crust was interpreted by Behn and Kelemen (2006) to be an equilibrium configuration that was convectively stable relative to the underlying mantle. The denser, more primitive cumulates may have been removed via foundering into the asthenospheric mantle.

In the Kohistan arc, the Moho is also interpreted to be a site of active underplating, granulite facies metamorphism, and efficient recycling of ultramafic and garnet-bearing lower crust into the asthenospheric mantle (Dhuime et al. 2009; Garrido et al. 2007). This occurred during the main stage of oceanic arc development (105–91 Ma), prior to accretion to the continent. These authors use similar reasoning to Greene et al. (2006) – the evolved character of underplated magmas (Mg# <59) implies a thick cumulate sequence that is no longer preserved in the crustal section.

3.3 Crustal Melting

Three of the four crustal sections described above show clear evidence for crustal melting to produce intermediate compositions (shown in orange in Fig. 5.1). One section shows cryptic geochemical evidence for crustal melting at levels deeper than those exposed (southern Coast Plutonic Complex), whereas two sections (Kohistan and Bonanza) display clear field and geochemical evidence for partial melting of country rock at the level of exposure. These partial melts exist in situ in migmatite complexes and as extracted magmas in pods or stocks at shallower crustal levels (e.g., DeBari et al. 1999; Dhuime et al. 2009) (Fig. 5.10). In the Talkeetna arc, cryptic geochemical evidence for crustal melting occurs only in the far western part of the arc in the latest stages of magmatism (Johnsen 2007).

Fig. 5.10
figure 10

Field examples of migmatitic rocks where crustal melting has taken place. (a) Is in the Southern Amphibolites of the Kohistan arc and (b) is in the Westcoast Complex of the Bonanza arc. Black rocks are restite, white rocks are tonalitic partial melts.

Full size image

Interestingly, these migmatite complexes do not exist right at the Moho. Moho sections are generally anhydrous assemblages, with evidence for only late stage, crosscutting amphibole-bearing lithologies. In Kohistan, migmatites occur within the Southern Amphibolite unit at pressures 0.8–1.0 GPa (Fig. 5.3). In the Bonanza arc, migmatites are within the Westcoast Complex at >0.6 GPa (Fig. 5.4) (DeBari et al. 1999).

Where observed in the field, the partial melts are tonalitic, produced by partial melting of metabasalts and metagabbros. In the Kohistan and Bonanza arcs, the partial melts have REE patterns distinctive of hornblende-bearing sources (Dy/Yb < 3). In contrast, in the Coast Plutonic Complex, inferred partial melts from pressures >1.5 GPa (below the level of exposure) have steep REE patterns distinctive of garnet-bearing sources (Dy/Yb > 3).

3.4 Mingling, Mixing, and Homogenization: “MASH” Zone

One commonality between all crustal sections is the heterogeneous nature of the middle crust. All crustal sections at this level display complex relationships between multiple lithologies, at all scales, and over many kilometers of section (Fig. 5.11). In the Westcoast Complex of the Bonanza arc and the Southern Amphibolite belt of the Kohistan arc, outcrops over many square kilometers show complex relationships between amphibolite facies host rock and multiple magma types (including partial melts). In the southern Coast Plutonic complex, where host rock abundance is higher, these complex relationships are evident in km-wide sheeted zones at pluton margins. In the mid crust of the Talkeetna arc, multiple magma types are complexly intermingled (Fig. 5.11), but without the presence of intermixed host rock.

Fig. 5.11
figure 11

Examples of magma mixing/mingling relationships between mafic and felsic magmas in the Coast Plutonic Complex, Bonanza arc, and Talkeetna arc. Flow textures in the mafic (dark) rocks suggest that all compositions were molten at the same time. (a) Shows physical mixing between basaltic compositions and plagioclase rich tonalites from the Tenpeak pluton in the Coast Plutonic Complex. Photos in (b) through (d) show basaltic dikes that were intruded into a more felsic magma chamber with various amounts of mixing and homogenization (b and c are from the Bonanaza arc and d is from the Talkeetna arc.

Full size image

These observed heterogeneous zones in all the crustal sections are a good match for the proposed “MASH” zone of Hildreth and Moorbath (1988). These authors proposed a zone in the deep arc crust where crustal melting, assimilation of host rock, stagnation, and homogenization takes place. However, the major difference between their model and the representations in crustal sections is that this MASH zone does not occur right at the Moho. The Moho is instead dominated by the cumulate products of crystal fractionation, corroborating the deep crustal “hot zone” model of Annen et al. (2006). The MASH zone occurs above that cumulate-dominated zone, presumably where magmas that had been variably differentiated by removal of crystal cumulates coalesce and mingle with host rocks and any partial melts.

4 Post-collision Geochemical Changes

In two of the arc crustal sections, Talkeetna and the southern Coast Plutonic Complex, arc magmatism continued after accretion to a continental margin (Clift et al. 2005b; Rioux et al. 2007; Tabor et al. 1989), or in the case of Talkeetna perhaps to another large terrane (Rioux et al. 2007). In both cases, this produced a thickened arc whose geochemical signature begins to reflect some incorporation of new basement rock, intracrustal melting and the segregation of garnet (either through partial melting or deep seated crystallization). In the eastern Talkeetna arc, Rioux et al. (2007) interpret plutonic rocks with more evolved initial isotopic ratios and the presence of xenocrystic zircons as reflecting docking to either the Wrangellia terrane or to North America. In the western Talkeetna arc, Johnsen (2007) correlates changing isotopic and trace element ratios to this same process.

Thickening of the arc due to the contractional processes is what is responsible for the generation of a more strongly calcalkaline, “continent-like” geochemical signature. For example, Johnsen (2007) shows that late-stage plutons in the western Talkeetna arc have much steeper rare earth element patterns ([La/Yb]N > 5) than the older plutons (Fig. 5.12). He interprets this as change from a pure fractionation origin (fractionation of pyroxene, plagioclase, hornblende, and oxides from an andesitic parental magma) to a more complex origin involving fractionation combined with partial melting of overthickened crust. Similar interpretations are proposed for plutonic rocks of the southern Coast Plutonic Complex (crust >55 km thick) where trace element patterns are very steep ([La/Yb]N > 10) and a role for garnet in magma genesis is required (DeBari et al. 1998, Miller et al. in preparation).

Fig. 5.12
figure 12

REE patterns of Type 1 plutonic rocks from the western Talkeetna arc (green) compared to Type 2 plutonic rocks (blue). Type 1 patterns are from older plutons produced by crystal fractionation of pyroxene, plagioclase, hornblende, and magnetite. Type 2 patterns are found in younger plutons, and require some crustal melting and assimilation, and possibly a garnet component. Data from Johnsen (2007).138.

Full size image

5 Relationship Between Inferred Magmatic Processes, Lithological “Stratification”, and Geophysical Signature

In general, seismic velocity profiles in modern arcs show an increase of P-wave velocities (V p) with depth (Fig. 5.13) with a clear segmentation into upper, middle, and lower crustal regions. Calvert (2011) reports a typical ratio of three parts upper crust to four parts middle crust to five parts lower crust. The depth of these segments depends on total thickness of the arc. Figure 5.13 shows a comparison of V p profiles for several modern arcs, compared to a V p profile calculated in this study for the lithologies present in the Talkeetna crustal section.

Fig. 5.13
figure 13

Seismic velocity profiles of various modern island arcs and continental arcs compared to the Talkeetna crustal section. The Izu-Bonin arc is from Suyehiro et al. (1996), the Aleutian arc is from Shillington et al. (2004), the Tonga arc is from Crawford et al. (2003), the Kurile arc is from Nakanishi et al. (2009), NE Honshu is from Iwasaki et al. (2001), and the Cascade arc is from Parsons et al. (1998).

Full size image

Intriguingly, some modern arcs (e.g., Kurile, Izu Bonin, Mariana) have a thick layer of low-velocity upper middle crust (V p = 6.0–6.5 km/s) that has been interpreted to be intermediate or felsic intrusive rock (e.g., Kodaira et al. 2007; Nakanishi et al. 2009; Takahashi et al. 2007). There has been much discussion about the presence of this layer in arcs as the unsubductable nucleus of continental crust (e.g., Tatsumi et al. 2008). Field analogues to this layer are observed in all crustal sections described above. Tonalite and diorite (the light blue fields in Fig. 5.1) are common upper middle crust lithologies in these arc sections. In contrast, the Aleutian arc middle crust has a V p =6.5–6.9 km/s (Shillington et al. 2004), which is higher than the western Pacific arcs. Shillington et al. (2004) interpret this mid crustal layer to be dominantly gabbroic, with some proportion of more felsic rocks. Clearly more felsic rocks are produced in the Aleutians, given the composition of felsic plutons and volcanic rocks exposed at the surface (e.g., Romick et al. 1992).

The Honshu and Cascade arcs are continental, but both are built on juvenile continental crust. They both also have a middle crust with Vp = 6.0–6.5 km/s. In the Cascades, Parsons et al. (1998) interpret the 10 km-thick layer with Vp = 6.0–6.5 km/s to contain felsic intrusive rocks similar to plutonic inclusions in Mt. St. Helens’ volcanic rocks. Laboratory experiments on P-wave velocities support this interpretation (Parsons et al. 1998).

Lower crustal P-wave velocities in modern arcs are high (>6.9 km/s, Fig. 5.13), indicative of rocks in dry “granulite facies” conditions (e.g., Rudnick and Fountain 1995). These granulite facies conditions are preserved in rocks from the lower crust of both Talkeetna and Kohistan. However, seismic velocities will be slower if the lithologies are hydrous (i.e., amphibole bearing). For example, P-wave velocities are lower in the lower crust of northeast Japan than in other arcs (V p = 6.5–7.0 km/s) (Iwasaki et al. 2001). Nishimoto et al. (2005) interpret this lower velocity to be due to an extensively hydrated lower crust where hornblende, plagioclase, and magnetite are dominant, as represented by lower crustal xenoliths. This hydration may be due to infiltration of H2O-rich magmas, a process locally documented in the deep crust of the Kohistan arc during decompression (Yamamoto and Yoshino 1998) and in the Talkeetna arc (DeBari and Coleman 1989).

Behn and Kelemen (2006) conclude that P wave velocities >7.4 km/s are indicative of lower crust that is denser than underlying mantle peridotite (e.g., the garnet-bearing gabbros or pyroxenites from Talkeetna), and can become convectively unstable. Interestingly, seismic refraction studies in modern arcs typically find lower crustal V p < 7.4 km/s. This implies that, like the Talkeetna arc, either (1) gravitationally unstable material founders rapidly on geologic timescales, or (2) the missing high V p cumulates crystallize beneath the Moho. Preservation of such a thick section of these dense rocks in Kohistan is surprising. However, their low Mg # (<0.60; Garrido et al. 2006; Jagoutz 2010) may suggest a greater density stability for these rocks (cf. Behn and Kelemen 2006).

The paleo-Moho in Kohistan and the Talkeetna arc sections is a sharp boundary between ultramafic rocks and plagioclase-abundant gabbroic rocks (generally <100 m transition zone). Both density and P-wave velocity would change rapidly over this type of transition. Thus if these exposed sections are representative of modern arcs, seismic profiles should produce a sharp Moho. An interesting implication of the Moho as being a plagioclase-saturation boundary is that much of the “igneous” part of the arc (ultramafic cumulates) is not part of the crust. Experiments by Müntener et al. (2001) show that up to 50% of primary mantle-derive liquid could crystallize as ultramafic plutonic rocks before plagioclase saturation. Thus the arc crust is necessarily more felsic than the parental magma extracted from the mantle.

Many geophysical studies seek to link geophysical profiles from modern arcs with magmatic processes (e.g., Calvert et al. 2008; Nakanishi et al. 2009; Nishimoto et al. 2005; Shillington et al. 2004; Suyehiro et al. 1996). One particularly elegant model (Tatsumi et al. 2008) integrates petrologic modeling with seismic features of the Izu-Bonin-Mariana (IBM) arc. They conclude that the IBM arc has gone through two stages of arc development. Stage 1 is the initial growth of the arc, where crystal fractionation is dominant, and a mafic-intermediate crust is created. The bulk composition is basaltic, with differentiated magmas in the upper crust and cumulates in the lower crust/upper mantle. In Stage 2, successive basaltic underplating causes partial melting of Stage 1 crust to produce more differentiated (andesitic) magmas. This can take place either by moderate degrees of melting of basaltic arc crust (which produces andesite and complementary mafic restite), or by smaller percent melting that produces felsic magma that mixes with differentiated basaltic magma to produce andesite. This latter process also leaves a restite.

These two stages (crystal fractionation dominated and crustal melting dominated) are clearly represented in the crustal sections described above (Fig. 5.14). Much of the exposed crustal section in the eastern Talkeetna arc (where the deepest exposures are) is a final product of a Stage 1 arc, even with a 30 km crustal thickness. Intermediate and felsic plutons are exposed, but they are all clearly related by fractional crystallization. The seismic profile of the Mariana modern arc (Calvert et al. 2008) suggests that this arc is also in Stage 1. A Stage 2 arc, where crustal melting is important, is represented by the crustal sections from Kohistan, the Bonanza arc, and the Coast Plutonic Complex. Crustal melting must also have occurred locally in the later stages of the Talkeetna arc (Rioux et al. 2007; Johnsen 2007).

Fig. 5.14
figure 14

Evolution of an arc through (a) Stage 1 and (b) Stage 2 phases, modified after Garrido et al. (2006). A Stage 1 arc is dominated by crystal fractionation and mingling/mixing processes, with arc crust generally <30 km thick. This stage occurs during early stages of arc development, but as indicated by the Talkeetna arc, this stage may persist for tens of millions of years. A Stage 2 arc has been thickened extensively, typically by arc contraction, and may be indicative of collision with another terrane. During this stage, crustal melting can be extensive. Delamination of dense cumulates as discussed in the text occrs during both of these stages.

Full size image

From the petrologic modeling, Tatsumi et al. (2008) calculate that much of the cumulate and restitic rock produced in the fractionation and partial melting processes must be below the Moho. This is required because the lower crustal thickness is too thin to accommodate the required rock material. This hypothesis of mass transfer across the Moho is strongly corroborated in the arc crustal sections (Fig. 5.14).

6 Conclusion

The arc-crustal sections described in this paper show that mantle-derived magmas pond and fractionate at the base of the crust, producing thick swaths of cumulate rocks. The seismic Moho is within these cumulate rocks, with plagioclase-bearing rocks (gabbros) above and plagioclase-absent rocks (pyroxenites and dunites) below. It is a relatively sharp boundary. As shown in both Kohistan and the Talkeetna arc, as the arc thickens through magma input (or by contraction) to >30 km, gabbroic rocks may become garnet-bearing either through crystallization of igneous garnet, by subsolidus garnet growth, or by growth of garnet during partial melting. Pyroxenites and garnet gabbros are denser than underlying mantle and are very likely removed via crustal foundering. Missing pyroxenites from the base of Talkeetna arc crust, coupled with high temperature flow fabrics in the underlying mantle, and a thin layer of garnet gabbros, provide strong evidence that the Moho can be a delamination feature. Missing ultramafic cumulates in the Kohistan arc also corroborate this hypothesis.

The lower middle crust of these arc sections is a zone of extreme inhomogeneity. This is the part of the crust where there is the best evidence for crustal melting, and where mantle-derived magmas at variable stage of differentiation mix with each other and with these crustal melts. The middle and upper crust is more homogeneous, more felsic, and much less dense than the rocks of the lower crust. This then becomes the unsubductable nucleus of continental crust, with compositions that range from diorite to quartz diorite to tonalite. Finally, the uppermost crust contains the volcanic pile that is variably intruded by upper crustal plutonic rocks.

As arcs thicken over time, especially after accretion, the development of garnet in the lower crust and the greater role for deep crustal melting has profound impacts on arc chemistry. These magmas become more felsic, and develop steeper rare-earth element patterns that more closely resemble continental crust.