Linking deep and shallow crustal processes during regional transtension in an exhumed continental arc, North Cascades, northwestern Cordillera (USA)

Collisional orogens and some contractional continental magmatic arcs have gross similarities in their structure, including thickened crust (50–70 km) and broad plateaus that have zones of partially molten lower to middle crust (e.g., Tibet and Altiplano-Puna; Nelson et al., 1996; Schilling et al., 2006; Ward et al., 2014). Studies of exhumed orogenic crust and models for lateral and vertical crustal flow have most commonly been formulated based on collisional orogens, such as the Himalayas (e.g., Law et al., 2006, and references therein). Much less is known about the deep levels of continental arcs, which contain the rocks that record the thermal, rheological, and mechanical transition of an orogen from contractional crustal thickening to extension and/or transtension. For example, in the North American Cordillera, more research has concentrated on deep crustal flow and exhumation of the hinterland metamorphic core complexes (e.g., MacCready et al., 1997; Vanderhaeghe and Teyssier, 1997; Teyssier et al., 2005; Gervais and Brown, 2011) than on the Mesozoic arc system to the west (Fig. 1). Similarly, lower to middle crustal flow is poorly documented in other exhumed continental magmatic arcs, with a few exceptions (e.g., Fiordland, New Zealand; Klepeis et al., 2004; Klepeis and King, 2009; Betka and Klepeis, 2013).

In this study we focus on the 96–45 Ma North Cascades magmatic arc of the State of Washington (USA) and southwest British Columbia (Canada), which is the southern continuation of the Coast Mountains batholith. This arc likely reached thicknesses of ≥55 km (Miller and Paterson, 2001) in the Late Cretaceous, and a wide range of Cretaceous crustal levels (0 to ≥40 km) is exposed. It is interpreted as the western margin of a proposed Late Cretaceous to Eocene orogenic plateau, the eastern margin of which is marked by metamorphic core complexes (Whitney et al., 2004). Eocene migmatization, plutonism, and exhumation of some of the deepest (8–12 kbar) rocks of the Cascades crystalline core were in part coeval with motion on dextral strike-slip and oblique-slip faults, formation of nonmarine, transtensional basins, and intrusion of extensive basaltic to rhyolitic dike swarms (e.g., Haugerud et al., 1991; Gordon et al., 2010a; Eddy et al., 2016). This orogen thus offers an exceptional opportunity to evaluate links between contemporaneous deep and shallow crustal processes during regional transtension. Furthermore, the record of how different crustal levels evolved provides a view of the construction and collapse of a thick continental arc that may be analogous in some respects to the Cenozoic central Andes (Western Cordillera and Altiplano) (e.g., Scheuber and Reutter, 1992; Beck and Zandt, 2002), where deep crust has not been exhumed.

We integrate a large body of different types of data to arrive at a synthesis of processes occurring from the middle crust to the surface during a time interval largely centered on the Eocene (60–45 Ma). We address the potential interplay of deformation, metamorphism, partial melting, and magmatism at different depths, including the temporal and dynamic relations of these processes to exhumation of arc orogenic crust, formation of nonmarine basins, and intrusion of dike swarms. Particular emphasis is placed on how strain was partitioned at different depths in the crust during transtension and whether the upper crust was decoupled from the deeper, ductilely flowing lower crust (e.g., Tibetan Plateau, Royden et al., 1997; Altiplano-Puna, Husson and Semere, 2003; Gerbault et al., 2005; Ouimet and Cook, 2010).

The Coast Plutonic Complex and its southeast extension, the crystalline core of the North Cascades (Cascades core), form a >1500-km-long plutonic and metamorphic belt that represents an exhumed Cretaceous and Paleogene magmatic arc (Figs. 1 and 2) (e.g., Armstrong, 1988; Tabor et al., 1989; Miller et al., 2009a). In the southern Coast Mountains and North Cascades, arc plutons intrude small late Paleozoic to Cretaceous oceanic and arc terranes (Journeay and Friedman, 1993). The Cascades core includes the Chelan Mountains terrane, Nason terrane, and Swakane terrane. The Chelan Mountains terrane consists of the Napeequa and Cascade River–Holden units, and metasupracrustal rocks in the Skagit Gneiss Complex are also commonly considered part of this terrane (Fig. 3) (Tabor et al., 1989). The Napeequa unit is dominated by amphibolite, siliceous schist, and quartzite (metachert), and contains widespread biotite schist and ultramafic rock, and minor marble and calc-silicate rock. It may correlate with low-grade Mississippian–Jurassic rocks northeast of the Cascades core (e.g., Monger, 1986; Miller et al., 1993b). The Cascade River–Holden unit is made up of hornblende-biotite schist and gneiss, amphibolite, calc-silicate rock, and metaconglomerate (e.g., Misch, 1966; Cater, 1982; Brown et al., 1994). This unit and approximately coeval orthogneisses represent a Late Triassic arc. Much of the southern part of the Cascades core consists of the dominantly metapelitic and metapsammitic Chiwaukum Schist of the Nason terrane (e.g., Plummer, 1980; Tabor et al., 1987b), and includes Early Cretaceous protoliths (Brown and Gehrels, 2007). The structurally deepest terrane is the Late Cretaceous metapsammitic Swakane Biotite Gneiss (Fig. 2) (Matzel et al., 2004; Paterson et al., 2004; Gatewood and Stowell, 2012).

Much of the Cascades core was intruded, metamorphosed, and ductilely shortened in the mid-Cretaceous (ca. 100–85 Ma) synchronous with regional thrusting and folding (Fig. 1) (e.g., Misch, 1966; Brown, 1987; McGroder, 1991). Crustal thickness in the core probably reached ≥55 km at ca. 90 Ma (Miller and Paterson, 2001). After the major shortening event, magmatism (79–45 Ma), metamorphism, and deformation were focused in the northeastern part of the Cascades core (Chelan block; Fig. 1). The Swakane Biotite Gneiss was buried to ∼35–40 km depth by 68 Ma (Matzel et al., 2004; Paterson et al., 2004; Gatewood and Stowell, 2012). High-angle Paleogene faults (Fig. 1) underwent variable amounts of dextral strike slip and dip slip, and a transition from components of reverse slip to normal slip on some of these faults likely coincided with a regional change from transpression to transtension (Miller and Bowring, 1990; Haugerud et al., 1991).

The youngest ⁴⁰Ar/³⁹Ar and K-Ar ages (e.g., Wernicke and Getty, 1997; Paterson et al., 2004) and ductile deformation are found in the two domains of deep crust that underwent substantial exhumation accompanied by Eocene near-isothermal decompression and subsequent rapid cooling (Fig. 2). These domains include the orthogneiss-dominated, partially migmatitic 8–10 kbar Skagit Gneiss Complex, and the largely nonmigmatitic 9–12 kbar Swakane Biotite Gneiss (Misch, 1968; Haugerud et al., 1991; Whitney, 1992b; Valley et al., 2003; Gordon et al., 2010a). Ductile deformation, late magmatism, partial melting of Skagit rocks, and exhumation of these domains overlapped with regional transtension, indicated in the shallow crust by coeval formation and rapid subsidence of Eocene nonmarine basins bounded by oblique-slip (dextral and normal) faults (e.g., Johnson, 1985; Evans, 1994; Eddy et al., 2016), and intrusion of abundant Eocene dikes. These basins accumulated clastic rocks that in the central Washington Cascades may have reached ≥12 km in thickness (Tabor et al., 1984; Evans, 1994).

Several studies have argued that ridge subduction occurred at ca. 50 Ma at the latitude of northern Washington (Thorkelson and Taylor, 1989; Cowan, 2003; Haeussler et al., 2003; Madsen et al., 2006) and created a slab window beneath the Cascades at that time. In the Cascades core, magmatism changed from dominantly tonalite at 96–60 Ma to mainly granodiorite at 50–45 Ma (Misch, 1966; Miller et al., 2009a). Most of the widespread dikes are inferred to be ca. 49–45 Ma (Tabor et al., 1984; Eddy et al., 2016). Dikes occur in granitic swarms, some of which are probably related to shallow plutons in the Cascades core; the most voluminous are in a mafic swarm that intruded the Eocene Swauk basin, and others are isolated bodies that intruded parts of the core. All magmatism, metamorphism, and ductile deformation ended by ca. 43 Ma, when the crystalline core rocks were exhumed to the near surface, as recorded by ⁴⁰Ar/³⁹Ar and K-Ar hornblende and biotite dates (Engels et al., 1976; Wernicke and Getty, 1997; Tabor et al., 2003; Matzel, 2004). Subsequent subduction of the Farallon plate resulted in the modern Cascades arc (e.g., Tepper, 1996) starting ca. 40 Ma (du Bray and John, 2011).

In the following we integrate new data from our multifaceted research with those of previous workers to provide a view of the Eocene crustal architecture and processes operating throughout the crustal section. These data include material in theses by some of the authors (Michels, 2008; Shea, 2008; Doran, 2009; McLean, 2012), publications arising from group research (Miller and Bowring, 1990; Gordon et al., 2010a, 2010b; Wintzer, 2012), and other publications, notably Misch (1968), Haugerud et al. (1991), Paterson et al. (2004), and Eddy et al. (2016). The main new data sets presented herein include high-precision U-Pb (chemical abrasion–isotope dilution–thermal ionization mass spectrometry, CA- ID-TIMS) zircon dating of 10 orthogneiss bodies in the Skagit Gneiss Complex (Fig. 3; Table 1; see Supplemental Table 1¹ and http://geochron.org/dataset/html/geochron_dataset_2016_01_22_kCnxS for details of dates); mapping and structural analysis of the central and southern parts of the gneiss complex and the upper contact of the complex; structural fabrics in Eocene plutons; and orientations and extension directions of Eocene dikes in the shallow crust. The data description begins with the deep crust, focusing on the P-T-t-d (pressure-temperature-time-deformation) history of the Skagit Gneiss Complex and to a lesser extent the Swakane Biotite Gneiss, followed by analysis of fabrics and regional strain in Eocene plutons, and then on high-angle faults, nonmarine basins, and dike swarms in the shallow crust.

The Skagit Gneiss Complex extends for >100 km along strike, and is as much as 30 km wide (Misch, 1966, 1968; Haugerud et al., 1991) (Figs. 2 and 3). It is bounded on the northeast by the Ross Lake fault zone and is intruded on the southwest by Eocene plutons or is in contact with the Napeequa unit, Late Triassic arc rocks, or Cretaceous orthogneiss (Fig. 3). An intrusive contact with the ca. 48 Ma Cooper Mountain batholith marks the southern boundary and the complex is truncated by the intersection of the Ross Lake and Straight Creek–Fraser faults to the north (Figs. 2 and 3). The Skagit Gneiss Complex consists of orthogneiss, “banded gneiss,” which is strongly foliated and commonly migmatitic gneiss of uncertain protoliths, and metasupracrustal rocks (Misch, 1968). Orthogneiss makes up ≥75% (Haugerud et al., 1991) of the complex, and on the basis of our mapping, probably >90% of its central and southern parts. Much of the northern half is migmatitic (Misch, 1968).

The dominant protoliths of Skagit orthogneisses are biotite tonalite and leucotonalite; hornblende-bearing tonalite and biotite granodiorite are moderately abundant, and granite to quartz monzonite and diorite orthogneiss occur locally (Misch, 1968; Haugerud et al., 1991; Miller et al., 1994; our mapping). Crystallization ages of the orthogneiss protoliths range from ca. 89 to 45 Ma, and the largest volume of orthogneiss probably intruded between ca. 73 and 59 Ma (Fig. 3; U-Pb CA-ID-TIMS zircon data in Table 1; other ID-TIMS zircon dates from Mattinson, 1972; Miller et al., 1989; Miller and Bowring, 1990; Haugerud et al., 1991). Most of these orthogneisses are sheeted, or heterogeneous on the outcrop to 100 m scale, and are intimately intruded by 10-cm-thick to 10-m-thick leucocratic, commonly pegmatitic sheets. An apparent intrusive hiatus occurred between the ca. 59 and 50–45 Ma orthogneisses (Fig. 3). Distinctive tonalitic orthogneiss sheets (most <20 m thick) that occur as late intrusive bodies scattered throughout the gneiss complex contain plagioclase phenocrysts that are commonly cored by igneous-appearing sphene. One crystallized at 47.192 ± 0.002 Ma (FL, Fig. 3; Table 1). The youngest intrusions are ca. 45 Ma (44.857 ± 0.023 Ma; DL, Fig. 3; Table 1; also a less precise age in Haugerud et al., 1991) bodies, ranging from <1 m to ∼750 m across, of biotite granodiorite to granite of the Diablo Lake orthogneiss. These bodies discordantly cut foliation and contacts in the gneiss complex and have a weak or no foliation, but have a strong mineral lineation, which is parallel to that in the older gneisses (Misch, 1968, 1977a; Haugerud et al., 1991; Wintzer, 2012). This stretching lineation is marked best by biotite and quartz aggregrates that have undergone grain-size reduction.

Metasedimentary and metavolcanic rocks commonly occur as inclusions within, and/or rafts between, orthogneiss bodies (Misch, 1968). Most are <25 m wide, but zones rich in metasupracrustal rocks reach ∼3 km across (Tabor et al., 2003). Amphibolite and biotite schist dominate, and lithologies overall are most similar to those in the Napeequa unit, except in the southwest, where hornblende-biotite schist resembles the Cascade River–Holden unit (Miller et al., 1994).

Amphibolite facies assemblages occur throughout the Skagit Gneiss Complex (Misch, 1968). Sillimanite and, to a lesser extent, kyanite occur in local metapelite, mainly in the northern part of the complex (Misch, 1968). These metapelites record maximum conditions of 8–10 kbar at 650–725 °C (Whitney, 1992b; Gordon et al., 2010a) that were followed by a near-isothermal decompression path to P < 5 kbar (Whitney, 1992b).

The northern half of the complex also contains migmatites with low-K (trondhjemitic), layer-parallel to discordant leucosomes (Fig. 4A) that likely formed under a combination of subsolidus processes (Misch, 1968; Yardley, 1978; Babcock and Misch, 1989) and anatexis driven by water-saturated melting and decompression (Whitney, 1992a; Whitney and Irving, 1994; Gordon et al., 2010a). Metamorphism and melt crystallization in the northern Skagit Gneiss Complex likely occurred (possibly intermittently) from 71 to 47 Ma, on the basis of U-Pb CA-ID-TIMS zircon and monazite dates from three leucosomes, interpreted to have formed under anatectic conditions, biotite gneiss, and metapelite in an ∼7 km, across-strike transect in the Highway 20 corridor (Gordon et al., 2010a); these crystallization ages overlap with those of the orthogneisses.

The structure of the Skagit Gneiss Complex varies along strike and can be considered in terms of three domains (shown in Figs. 3 and 5) that are distinguished by different foliation orientations. Foliation is generally defined by aligned biotite, quartz, and quartz-plagioclase aggregates, and in some rocks by hornblende. Most of the complex is foliated and lineated, but fabric intensity varies considerably. The weakest fabrics are in orthogneiss and particularly in the younger (ca. 50–48 Ma) orthogneiss in the central and southern domains. In addition, constrictional fabrics are common in some areas, and are recognized qualitatively by stronger lineation than foliation, and by strain markers, such as in metamorphosed fragmental rocks and quartz-plagioclase aggregates in orthogneisses (Misch, 1968; our observations). They are most notable in the young (ca. 45 Ma) Diablo Lake orthogneiss (Misch, 1968; Haugerud et al., 1991; Wintzer, 2012) and parts of the stepover in the Ross Lake fault zone.

In much of the gneiss complex, foliation strikes northwest (Fig. 5A), and on a regional scale, the complex forms a northwest-southeast–trending anticlinorium (Tabor et al., 1989). Mineral lineation typically plunges gently to the northwest or southeast (e.g., Brown and Talbot, 1989), but is domainal in orientation (Fig. 5B). Outcrop-scale folds are common in well-layered gneiss and migmatite in the northern domain. Earliest structures are tight to isoclinal, gently inclined to recumbent folds that typically have wavelengths of <25 cm (Misch, 1977a; Wintzer, 2012). Locally, the dominant foliation is axial-planar to folded leucosomes and compositional layers, but it is generally folded by the earliest structures. Hinge lines are variably oriented reflecting subsequent refolding, and the average orientation of axial surfaces is approximately parallel to foliation (Wintzer, 2012; our observations). The dominant mesoscopic folds are upright to inclined structures that range from gentle to tight, and have wavelengths of ∼1 cm to 20 m (Fig. 4A) (Misch, 1968; Haugerud et al., 1991; Wintzer, 2012). These folds are probably parasitic to the upright map-scale folds (Fig. 5C). Hinge lines mostly plunge southeast, are subparallel to mineral lineations, and are commonly stretched parallel to lineation (Fig. 4B).

Map-scale folds of foliation and unit contacts are mainly upright to steeply inclined structures (Figs. 6A, 6B; e.g., Misch, 1966; Tabor and Haugerud, 1999; Tabor et al., 2003; Miller et al., 2009b; Wintzer, 2012). They most commonly plunge gently southeast or northwest, but in parts of the complex the axial traces are curved. In the south, symmetric gentle to open map-scale folds of foliation have axial traces that trend west-northwest near the Cooper Mountain batholith, and curve to a northwest trend to the north (Fig. 5C). Gently plunging mineral lineation similarly swings from east-southeast near the batholith to northwest farther to the north, and to northwest-southeast and northeast in the northern part of the complex (Fig. 5B). Most map-scale folds die out northward from the Cooper Mountain batholith, whereas the longest (8 km) wavelength and westernmost antiform continues for ≥30 km. In the central domain, foliation defines a broadly fan-like antiform; in the west, it strikes northwest and dips southwest, and in the east, it strikes north-northeast and dips east. Lineation shows a similar swing (Adams, 1961; our data). In the northern domain, gently southeast-plunging map-scale folds have wavelengths of a few kilometers (Fig. 6A, A-A′). The northwest-striking foliation of this corridor changes in the east near the stepover in the Ross Lake fault zone where strikes rotate to northeast and east-west (Fig. 5A), and dips are lower compared to the south. Mineral lineation here mostly plunges gently to the south, or to the south-southwest and north-northeast, in contrast to the gentle southeast plunge of lineation away from the stepover.

The latest folds are defined by the large swings in the regional trends of foliation, lineation, and hinge lines (Fig. 5), and probably plunge moderately to steeply. These open folds record approximately orogen-parallel shortening that deformed foliation in rocks as young as the ca. 47.2 Ma sheet (FL, Fig. 3; Table 1) in the northern domain.

Noncoaxial flow is recognized in some areas at the outcrop scale to microscale by asymmetric structures, but well-defined ductile shear zones are uncommon at the 10 m scale or larger. These asymmetric structures, which are used to determine shear sense, include S-C relations, shear bands (C′ surfaces), oblique quartz foliation, asymmetric plagioclase porphyroclast systems (Fig. 4C), mica fish, asymmetric leucosomes and boudins, and foliation curvature. These features are concentrated in weaker micaceous layers, particularly in well-layered gneisses next to orthogneiss bodies and pegmatitic sheets. Shear zones are generally subparallel to the dominant foliation and compositional layering. Widespread, but less common, generally ≤15-cm-thick shear zones cut foliation at moderate to high angles, are generally steeper than the foliation-parallel ones, and are highly variable in strike. Lineations in the late zones also show considerable scatter, but the majority plunges gently to moderately to the east-southeast and southeast. Similar kinematic indicators occur in both types of shear zones.

In a few outcrops, asymmetric fabrics in foliation-parallel shear zones show a reversal in shear sense from sinistral to dextral across fold hinges. The best-documented cases have top-to-the-northwest shear in both limbs and in the hinge. This reversal and the shear direction (mineral lineation) do not result from flexural shear, as the lineations and hinge lines are subparallel (cf. Goscombe and Trouw, 1999). On the basis of these observations, and following Wintzer (2012), we reevaluated kinematic patterns for the foliation-parallel shear zones after unfolding the upright, gently plunging folds (such as in Fig. 4A) and restoring foliations to an approximately subhorizontal orientation.

In the southern domain, after unfolding we found examples of the top-to-the-southeast and top-to-the-northwest shear. In the central part of the gneiss complex, we rarely observed shear-sense indicators outside of the dextral Ross Lake fault zone (see following). In the northern domain, shear-sense indicators are widespread (Wintzer, 2012; our data), including in orthogneiss as young as ca. 48–47 Ma, but are not recognized in the ca. 45 Ma Diablo Lake orthogneiss bodies. Both top-to-the-northwest (mostly dextral) and top-to-the-southeast (mostly sinistral) shear are recorded throughout the corridor. Noncoaxial shear initiated at amphibolite facies conditions, as cummingtonite and sillimanite are in C-surfaces and plagioclase is recrystallized in foliation-subparallel zones (Gordon et al., 2010a), and some of the shear zones at moderate angles to foliation in the southern domain appear to have been filled with melt (Fig. 4D). Shear continued during cooling to greenschist facies conditions in both types of shear zones, as in some zones quartz shows basal <a> slip, bulging recrystallization, and considerable unrecovered strain, and plagioclase is microfractured and not recrystallized (cf. Gordon et al., 2010a, 2010b; Wintzer, 2012).

Overall patterns of noncoaxial flow are clearly complicated in the Skagit Gneiss Complex. We thus further refined our analyses by looking at shear sense in some of the youngest dated orthogneisses, including the 47.192 Ma ± 0.002 Ma, plagioclase-phyric, sphene-cored, orthogneiss sheet and a 48.490 Ma ± 0.027 Ma (PO, Fig. 3; Table 1) protomylonitic orthogneiss body. There appears to be an association of the youngest greenschist facies microstructures and top-to-the-northwest shear, but we also found examples with top-to-the-southeast shear.

We estimated the minimum shortening of the gneiss complex recorded by map-scale upright folds in two cross sections (Figs. 6A, 6B) by assuming flexural slip and using simple line-length balancing. Results are ≥33% in the northern domain and ≥25% in the southern domain. The magnitude of shortening by foliation development and the homogeneous shortening indicated by the thickening of layers in fold hinges is unknown.

The Skagit Gneiss Complex underwent a protracted history of metamorphism and deformation (Table 2). The overprinting by the extensive ca. 73–45 Ma magmatism, metamorphism, and deformation makes it difficult to correlate any structures within the complex to the mid-Cretaceous (ca. 100–80 Ma) regional shortening that is well documented in the Wenatchee block of the Cascades core and in low-grade rocks adjacent to the core. Similarly, the high-P metamorphism that initiated by 75 Ma in the core (Miller et al., 1993a, 1993b; Brown et al., 1994) has been overprinted.

Orthogneisses that yield inferred crystallization ages as young as 47.2 Ma (FL, Fig. 3; Table 1), ca. 69–51 Ma leucosomes, and metasupracrustal rocks with monazite dates as young as 47 Ma (Gordon et al., 2010a) display the dominant foliation and lineation in the Skagit Gneiss Complex. In two of the dated migmatite localities, a 65–64 Ma and a 53 Ma discordant leucosome cut the foliation of the host metapelite and biotite gneiss, indicating that foliation initiated by 65 Ma in at least part of the gneiss complex. Both discordant leucosomes were later deformed under greenschist-facies conditions (Gordon et al., 2010a).

Fabric development was in part time transgressive. Solid-state foliation and lineation in orthogneisses in the southern part of the complex, and folding of the 49.484 ± 0.026 Ma Sunrise Lake tonalite (SL, Fig. 3; Table 1) predated intrusion of the ca. 48 Ma Cooper Mountain batholith. A 49.198 ± 0.043 Ma felsic sheet (SL1, Fig. 3; Table 1) that cuts foliation in the tonalite may better define the timing of deformation. In comparison, the 48.5–47.2 Ma orthogneisses (FG, PO, Fig. 3; Table 1) in the northern part of the complex record significant solid-state deformation. Moreover, ca. 45 Ma Diablo Lake orthogneiss bodies crosscut foliation, although these orthogneisses have strong solid-state lineation (Misch, 1968; Haugerud et al., 1991; Wintzer, 2012). All ductile deformation ended before intrusion of the 34 Ma phase of the Chilliwack batholith (Fig. 3).

Most K-Ar and ⁴⁰Ar/³⁹Ar biotite and hornblende dates from the Skagit Gneiss Complex range from ca. 50 to 44 Ma (Engels et al., 1976; Wernicke and Getty, 1997; Tabor et al., 2003; Gordon et al., 2010b). A sample from near (∼2.5–5 km) dated migmatites gives ⁴⁰Ar/³⁹Ar hornblende and biotite dates of 47.1 ± 1.3 Ma and 45.2 ± 0.2 Ma, respectively (Wernicke and Getty, 1997), and muscovite from two structurally higher pegmatite samples on Ruby Mountain (Fig. 3) yield muscovite ⁴⁰Ar/³⁹Ar dates of 47.1 ± 0.3 Ma and 46.8 ± 0.3 Ma (Gordon et al., 2010b). Overall, the temporal overlap of the youngest zircon (ca. 47 Ma) and monazite (ca. 49–46 Ma) dates from migmatites and orthogneisses with the ⁴⁰Ar/³⁹Ar and K-Ar dates indicates rapid cooling at 47–45 Ma.

The relationships of the Skagit Gneiss Complex to adjacent, generally lower grade rocks are important for models of crustal flow and partitioning of deformation at different crustal levels. Lengthy segments of the western and southern boundaries are intrusive contacts with younger weakly deformed plutons. Where the gneiss complex is in contact with older rocks on the west (Fig. 3), foliation and lineation are concordant across the contact (Cater and Wright, 1967; Miller, 1987; Tabor et al., 2003; our mapping). The northeast boundary in the Ross Lake fault zone and the upper contact with the Napeequa unit are more complicated.

The ≥10-km-wide Ross Lake fault system (Misch, 1966; Miller, 1994) is part of a 500-km-long zone of Paleogene, northwest-striking high-angle faults (Fig. 1) (e.g., Monger, 1986). The northernmost strand of the Ross Lake fault is a vertical mylonite zone in British Columbia that separates upper-amphibolite-facies rocks of the Skagit Gneiss Complex from sub-greenschist-facies rocks to the east (McTaggart and Thompson, 1967). Dextral shear and 6–12 km of northeast-side-down normal slip occurred from ca. 50(?) to after 45 Ma (Haugerud, 1985) and predated Oligocene magmatic rocks to the south. In Washington, the fault is intruded by the ca. 48 Ma Ruby Creek plutonic belt, and steps westward 7 km across a gently to moderately dipping stepover zone to the Gabriel Peak tectonic belt (Figs. 3 and 4E) (Miller et al., 1994; Gordon et al., 2010b). This mylonitic, moderately to steeply northeast-dipping belt records dextral shear in the north and reverse shear farther south (Fig. 3). Transpressional deformation in the belt occurred from 65 Ma (and earlier?) to ca. 58 Ma (Miller and Bowring, 1990; Miller et al., 1994). The tectonic belt merges to the south with the North Creek fault to form the Foggy Dew fault zone (Fig. 3), which separates sillimanite-bearing mylonites from greenschist facies rocks to the east, and underwent oblique dextral-normal slip (down to the east) between ca. 50 and 48 Ma (Miller and Bowring, 1990).

The gently to moderately northeast-dipping upper contact of the Skagit Gneiss Complex is best exposed on Ruby Mountain (Fig. 3), where we analyzed the structure. There, orthogneiss and local pelitic schist are structurally below amphibolite, quartzite, and metaperidotite of the Napeequa unit, and Skagit and Napeequa rocks are both intruded by trondhjemitic pegmatites. This contact is part of a shear zone marked by a strain gradient with pronounced intensification of lineation and the formation of constrictional (L >> S) fabrics. Protomylonitic gneiss near the contact has moderate- to high-temperature microstructures, marked by recrystallization of hornblende, plagioclase, and quartz, and is cut by widespread contact-parallel ductile shear zones. In addition, kinematic indicators give conflicting senses of noncoaxial shear from outcrop to outcrop. In one scenario, the intense fabrics and juxtaposition of different rock types represent a detachment zone; however, the inconsistent sense of shear, similar P-T conditions of 8–10 kbar at 650 °C, and muscovite ⁴⁰Ar/³⁹Ar cooling ages (ca. 47 Ma) on both sides of the contact argue against this interpretation (Gordon et al., 2010b). Thus, the Skagit-Napeeqa contact was overprinted by a high-temperature shear zone that was localized at this contact. Localization probably resulted from the rheological contrast between the Skagit orthogneiss and Napeequa rocks, possibly reactivating an older shear zone, and/or deformation may have transposed an intrusive contact.

The second major domain of rocks with Eocene ⁴⁰Ar/³⁹Ar cooling ages and ductile deformation is south of the Skagit Gneiss Complex within the Cretaceous Swakane Biotite Gneiss. The Dinkelman décollement separates the main body of Swakane Biotite Gneiss from the structurally overlying Napeequa unit and 91–72 Ma arc plutons (Fig. 2) (Matzel et al., 2004; Paterson et al., 2004). A narrow belt of Swakane Biotite Gneiss is also exposed southwest of the Entiat fault (Fig. 2). In the following we briefly summarize the major features of the gneiss that are most relevant to the Eocene evolution of the rocks, using previously published data.

The dominantly arkosic Cretaceous clastic protoliths of the Swakane Biotite Gneiss were buried and metamorphosed at peak conditions of ∼640–750 °C at 9–12 kbar (Valley et al., 2003; Gatewood and Stowell, 2012). Sm-Nd garnet dates of 73.5 ± 1.2 to 66.8 ± 0.7 Ma from the gneiss (Gatewood and Stowell, 2012) and a U-Pb zircon age of 68.36 ± 0.07 Ma from a syntectonic leucogranite sheet (Matzel et al., 2004) are interpreted to date the amphibolite facies metamorphism. Hornblende ⁴⁰Ar/³⁹Ar and K-Ar dates of 57.9 ± 0.5 Ma and 50.8 ± 1.4 Ma, respectively, and biotite dates ranging from 49.5 ± 0.3 to 46.2 ± 1.5 Ma record the cooling of the Swakane Biotite Gneiss in the Chelan block (Tabor et al., 1987a: Matzel, 2004; Paterson et al., 2004). Some of the gneiss was at the surface by ca. 48 Ma, as it is basement to part of the Eocene Chumstick basin (e.g., Gresens et al., 1981; Evans, 1994).

Structures associated with burial were probably largely obscured by pervasive ductile, noncoaxial shear that occurred at decreasing temperatures, presumably during exhumation (Paterson et al., 2004). Gently dipping foliations and the Dinkelman décollement are folded into a ≥10 km wavelength, gently northwest-plunging antiform. Lineation plunges gently to the north to north-northeast (Paterson et al., 2004; Miller et al., 2006). Mesoscopic fold hinge lines are more scattered, ranging from northwest to north-south, to south-southwest. Abundant kinematic indicators show that the gneiss records dextral shear on the southwest limb and sinistral shear on the northeast limb of the antiform (Paterson et al., 2004). Restoration of foliation to shallower dip by unfolding of the antiform results in uniform top-to-the-north to top-to-the-north-northeast shear (Paterson et al., 2004). Latest major motion on the Dinkelman detachment is also top-to-the-north, oblique to the strike of the orogen (∼315°–320°). Differences in ⁴⁰Ar/³⁹Ar and K-Ar cooling ages between the upper and lower plates of ∼10–13 and 10–16 m.y. for hornblende and biotite, respectively, suggest that late motion was extensional and resulted in excision of ≥8 km of crust (Matzel, 2004; Paterson et al., 2004). Undeformed dikes intrude both the hanging wall and footwall of the décollement, although we have not been able to trace any single dike across the structure. The two dated dikes in the footwall close to the décollement are 48.4 ± 2.2 Ma (K-Ar hornblende) and 47.8 ± 1.9 Ma (K-Ar biotite) (Tabor et al., 1987a). Displacement on the décollement probably occurred until ca. 48–47 Ma, as indicated by ⁴⁰Ar/³⁹Ar and K-Ar cooling ages.

Shallow to middle crustal Eocene (ca. 48–45 Ma) plutons intrude, or are near, the Skagit Gneiss Complex, and include the Cooper Mountain, Duncan Hill, and Railroad Creek plutons (Figs. 2 and 3). These intrusions generally display weak solid-state deformation (our mapping), but overlap temporally with late ductile deformation of the complex. Orientations (Figs. 5A, 5B) of magmatic foliation and lineation, defined best by biotite and to a lesser extent plagioclase and hornblende in these plutons, may provide “snapshots” of the regional strain field during emplacement (e.g., Paterson et al., 1998). Foliations in the intrusions are in places discordant to pluton contacts and subparallel to host-rock structures and the regional structural grain (Figs. 5A, 5B), and/or intensify in strength independent of contacts. Thus, these fabrics are interpreted to record tectonic strain and compliment results from the Skagit Gneiss Complex and Swakane Biotite Gneiss.

On the southwest side of the Skagit Gneiss Complex are the markedly elongate, ca. 46–45 Ma granodioritic Duncan Hill and Railroad Creek plutons (Fig. 3) (Cater, 1982; Dellinger, 1996). The tilted Duncan Hill pluton (∼2–13 km paleodepth) has a northwest-striking, vertical to steeply northeast-dipping, moderate to strong magmatic to solid-state foliation and northwest-trending lineation in its deeper northwest end (Figs. 5A, 5B) (Cater, 1982; Dellinger, 1996; our data) that has been interpreted to record regional strain by Haugerud et al. (1991). In comparison, our structural mapping indicates that the Railroad Creek pluton also has variable fabric intensity. In the south, the eastern margin of the pluton has a very weak magmatic foliation. A moderately strong magmatic fabric that is variably overprinted by solid-state deformation is found to the west, from the pluton interior to the margin. Foliation strikes north to northwest and dips steeply, and lineation trends northwest or southeast, and plunges gently (Figs. 5A, 5B). Lineation is stronger than foliation near the western contact. The strain gradient and solid-state deformation in the interior of the pluton are compatible with an origin by regional tectonic strain.

The western part of the ca. 48 Ma Cooper Mountain batholith (Shea, 2008) has a weak northwest-striking foliation where we have mapped it (Fig. 5A). Anisotropy of magnetic susceptibility data from the batholith generally show northwest-striking magnetic foliation and gently to moderately northwest- or southeast-plunging magnetic lineation, which has been interpreted to record regional tectonic strain (Fawcett et al., 2003).

In summary, magmatic and mostly weak solid-state foliation in Eocene plutons on average strike northwest, are steep, and probably record northeast-southwest shortening. Lineations are dominantly subhorizontal and northwest or southeast trending, and likely indicate weak northwest-southeast tectonic stretch.

Most major map-scale Eocene faults cutting and bounding the Cascades core and Eocene basins strike north to northwest and are steep. The most prominent is the right-lateral Straight Creek–Fraser fault, which forms the western boundary of the Cascades core (Figs. 1 and 2) and has displaced core rocks ∼110–160 km southward from equivalent rocks in southern British Columbia (e.g., Misch, 1977b; Vance and Miller, 1981; Tabor et al., 1984; Monger, 1986). Fault motion initiated by 48 Ma and ended by 34 Ma (Tabor et al., 1984; Umhoefer and Miller, 1996).

The northwest-striking Entiat fault is the eastern boundary of the Eocene Chumstick basin, separating the basin from the Swakane Biotite Gneiss (Fig. 7). To the northwest, the clastic rocks pinch out and the fault separates the Wenatchee block on the southwest, which contains mid-Cretaceous plutons and has largely Late Cretaceous ⁴⁰Ar/³⁹Ar and K-Ar biotite cooling ages, from the Chelan block on the northeast, which is intruded by abundant 79–45 Ma plutonic rocks and generally has Eocene biotite cooling ages (Fig. 1) (e.g., Tabor et al., 1989). The southeast segment of the Entiat fault has down-to-the-southwest normal separation and was active during Eocene deposition in the Chumstick basin (Gresens et al., 1981; Evans, 1994). The fault probably has a major strike-slip component given its straightness in map view, and Tabor et al. (1987b) suggested 30–40 km of dextral strike slip, using a tenuous piercing point, the hinge line of the folded contact between the Swakane Biotite Gneiss and Napeequa unit (Fig. 2).

The steep north-northwest–striking Leavenworth fault forms the western boundary of the Chumstick basin and separates it from the Eocene Swauk Formation and the Cascades core (Figs. 2 and 7). The fault consists of numerous segments that range in strike from ∼315° to 000° (Tabor et al., 1982, 1987a). It has been interpreted as a dextral strike-slip fault largely on the basis of its length, stratigraphic relations, and rapid depositional rates of the Chumstick basin (e.g., Johnson, 1985; Evans, 1994). Cheney and Hayman (2009) suggested that steep reverse faults are localized in and near the fault zone.

The Eocene nonmarine basins (Figs. 2 and 7) that flank the Cascades core consist of very thick sequences of sandstone and mudstone, with variable amounts of conglomerate and local tuff (e.g., Johnson, 1984). Most important for comparison to the processes occurring in the deeper crust are the Chuckanut, Swauk, and Chumstick sequences (Fig. 2), which overlap in age with the Eocene crystalline rocks.

The Chuckanut Formation is west of the Skagit Gneiss Complex, but before dextral slip on the Straight Creek fault the formation was farther south, west of the Swauk Formation (e.g., Vance and Miller, 1981). A tuff deep in the Chuckanut section is 56.835 ± 0.087 Ma, and an intercalated felsic tuff (Silver Pass Volcanic Member) in the Swauk Formation is 51.364 ± 0.067 Ma (U-Pb zircon, CA-ID-TIMS; Eddy et al., 2016).

The Swauk and Chuckanut strata were openly to tightly folded. Folds in the Swauk Formation mostly have wavelengths of 100 m to 2 km and steep axial planes (Fig. 6C) (Tabor et al., 1982, 2000; Doran, 2009). Axial traces of these folds change from northwest in the west, to mainly east-southeast or west-northwest in the central region, to largely northwest in the easternmost part of the Swauk Formation (Tabor et al., 1982; our mapping; Fig. 7).

The Swauk Formation is overlain via an angular unconformity by the dominantly basaltic Teanaway Formation and is intruded by the voluminous mafic Teanaway dikes (Foster, 1958; Tabor et al., 1984), both of which help bracket the timing of Swauk folding. An intercalated rhyolite flow near the base of the Teanaway Formation gives a U-Pb zircon age of 49.341 ± 0.070 Ma (Eddy et al., 2016). Conformably overlying the Teanaway lavas are clastic rocks of the mid(?) to late Eocene Roslyn Formation.

The Swauk basin is juxtaposed on the east against the Chumstick basin by the Leavenworth fault, and the Entiat fault forms the northeast boundary of the Chumstick basin (Fig. 7). These two faults have components of dextral slip (Tabor et al., 1984). The Chumstick Formation contains numerous tuffs (McClincy, 1986); one from near the base of the section is 49.147 ± 0.073 Ma and another ∼6000 m higher in the lower-middle part of the section is 47.981 ± 0.059 Ma (U-Pb zircon, CA-ID-TIMS; Eddy et al., 2016), indicating very high sedimentation rates of ∼6–7 m/k.y. during the interval between eruption of these tuffs. This depositional rate, the large thickness of the lower half of the formation, migrating depocenters, and monolithologic breccias containing large (averaging >1 m) clasts, some of which are >25 km from their source, provide strong evidence that the Chumstick Formation accumulated in a transtensional basin (Evans, 1994). Eddy et al. (2016) showed that the onset of rapid sedimentation in the basin corresponds to a region-wide reorganization of sedimentation patterns and the onset or acceleration of movement on the regional strike-slip faults.

Dike swarms give insights into the regional strain field in the shallow crust. Felsic to mafic Eocene dikes widely intrude Cascades core rocks and the Eocene Swauk Formation.

The best studied, most abundant, and southernmost dikes are in the ca. 49.3 Ma (U-Pb date for Teanaway lavas) mafic Teanaway dike swarm, which primarily intrudes the Swauk Formation (Fig. 8; Foster, 1958; Tabor et al., 1984; Doran, 2009). This swarm extends ∼75 km from east to west and a maximum of 18 km from north to south (Fig. 7). We measured orientations and thicknesses of 392 Teanaway dikes throughout the basin. In 4 detailed transects in the western and central parts of the swarm (Fig. 7), average strikes range from 036° to 017° (Table 3), dips range from 72° to 62°, and thicknesses are from 20 to 12 m. By comparison, 145 dikes intruding the eastern part of the basin have a mean strike of 040°, dip of 78°, and thickness of 14 m (Mendoza, 2008). Average subhorizontal extension directions inferred for the dikes range from 306°–126° to 287°–107° in the western and central regions, to 310°–130° in the east. Minimum extensions of ∼16% and 43% were determined from 2 of the strike-normal transects across the western and central Swauk basin (Table 3). These extension values were calculated from the number and thickness of dikes over the length of a transect.

Orientations of Eocene dikes (Fig. 9) that intrude the Cascades core are less well documented. The few dated dikes that intrude the Chelan block are ca. 48–46 Ma (K-Ar; Tabor et al., 1987a). Granite and rhyolite dikes next to the shallow part of the ca. 46 Ma Duncan Hill pluton commonly strike northeast (Tabor et al., 1987a; Dellinger, 1996; Bryant and Miller, 2014). Northwest-striking dikes have also been reported from southeast of the pluton (Waters, 1926; Tabor et al., 1987a; Sylva and Miller, 2014). Granitic dikes next to the southern margin of the ca. 48 Ma Cooper Mountain batholith on average strike 005° (Raviola, 1988). Our study of dikes scattered throughout the Skagit Gneiss Complex indicates that they have a broad range of modal compositions and textures, are typically 10 cm to 2 m thick, and some are younger than ca. 48.2 Ma orthogneisses (RF, Fig. 3; Table 1) that they cut, but ages are poorly known. Those dikes that intruded the southern part of the gneiss complex have an average strike of 032° (Shea, 2008), whereas dikes in the central part of the complex on average strike 092° (Michels, 2008). Ductile stretching recorded by lineation trends in the latter area are also anomalous, averaging 010°, and are compatible with the extension direction from the dikes there (Fig. 9).

On a regional scale, dike extension directions, from those intruding the Swauk Formation in the south to the Skagit Gneiss Complex in the north, range from 275°–095° to 002°–182°, to northeast-southwest south of the Duncan Hill pluton (Fig. 9). Extension was thus highly variable, but a significant majority of the dikes, including the voluminous Teanaway swarm, indicate an extension direction that is considerably more consistent (275°–095° to 310°–130°) (Fig. 9). On average, extension was broadly arc parallel to arc oblique, rather than arc normal (Figs. 9 and 10). The wide range in extension directions may be explained by complex spatial variations of the strain field, a rapidly changing strain field during emplacement of dikes of different ages, and/or later vertical axis rotations. There are insufficient age data for the dikes with different orientations to choose among these three interpretations.

The North Cascades expose a >40 km crustal section, allowing evaluation of processes occurring in the deep to shallow crust. We use our new results and previously published data to summarize these processes, and their temporal, kinematic, and other links of tectonic processes at different crustal levels in the Eocene.

In the North Cascades, a relatively short time interval (ca. 53–45 Ma) was marked by (1) plutonism, metamorphism, partial melting, ductile flow, and rapid cooling of metamorphic rocks; (2) rapid subsidence and deposition in nonmarine basins; (3) brittle faulting; and (4) extensive dike intrusion in the shallow crust (Fig. 10; Table 2). In contrast to some orogens (e.g., Tauern window; Selverstone, 1988; Axen et al., 1998), where the upper rigid crust was not strained while the middle to lower crust underwent ductile flow, different crustal levels of the North Cascades underwent coeval deformation, suggesting a degree of coupling between the crustal layers.

The presence of both Eocene brittle and ductile structures allows further evaluation of structural patterns and the degree of strain coupling between different crustal levels during regional transtension (Fig. 10). In particular, our discussion focuses on deformation styles, extension directions, and noncoaxial shear at different depths, and the Eocene folding that involved the Swauk basin and the Cascades crystalline core.

Mesoscopic structures in some of the Skagit rocks are likely composite, as foliation in some places predates 65 Ma leucosomes and elsewhere deforms widespread 49–45 Ma orthogneisses (see preceding discussion). The concordance of fabrics in Eocene orthogneisses with those in older orthogneisses implies that Eocene fabrics are widespread. Fabrics reflecting ductile transtension include the combination of gently to moderately dipping foliations, gently plunging lineations, L > S fabrics in sizable (>10 km²) domains, and strong stretching of fold hinges parallel to lineation (Fig. 4B) (e.g., Robin and Cruden, 1994; Tikoff and Teyssier, 1994; Dewey, 2002; Fossen et al., 2013), most prominently in the ca. 45 Ma Diablo Lake orthogneiss bodies. The extension direction in the Skagit Gneiss Complex is largely northwest-southeast (∼330°–150°), except near segments of the Ross Lake fault zone.

Noncoaxial ductile shear in the Skagit Gneiss Complex is localized in weaker metasedimentary rocks, as described above. Both top-to-the-northwest and top-to-the-southeast shear occur in the northern domain with top-to-the-northwest shear more common, particularly in dated Eocene orthogneisses. This shear likely predated upright folding, and thus originally probably occurred on gently dipping surfaces. Overall, we infer that on the map scale, deformation was general shear marked by a component of top-to-the-northwest noncoaxial shear.

The ductile ca. 50–48 Ma strain of the Skagit Gneiss Complex was broadly coeval with ductile dextral-normal shear in the Ross Lake fault zone, which was top-to-the-northeast in the north and top-to-the-east-southeast in the south (Table 2) (Miller and Bowring, 1990; Haugerud et al., 1991), and the approximately north-south stretching in the dilational stepover of the Ross Lake fault. Eocene deformation may have overlapped earliest movement on the Straight Creek fault (Table 2).

The Swakane Biotite Gneiss is characterized by pervasive, top-to-the-north to north-northeast noncoaxial shear and north-south to north-northeast–south-southwest subhorizontal to moderately plunging stretching (Figs. 9 and 10). The fabrics thus indicate largely subvertical shortening and top-to-the-north subhorizontal flow prior to the late upright folding; some of the flow overlapped with the Eocene deformation in the Skagit rocks (Table 2).

The two deep-crustal domains in the Chelan block were exhumed in similar manners. The rapid cooling and, presumably, exhumation of the Skagit Gneiss Complex are compatible with vertical thinning associated with subhorizontal stretching and constrictional strain, and a component of normal slip on the dextral Ross Lake fault system. Similarly, exhumation of the Swakane Biotite Gneiss was broadly synchronous with vertical ductile shortening, and probably with dip slip on the Entiat fault.

The overall pattern of ductile Eocene deformation in the Cascades core thus includes subvertical shortening and subhorizontal northwest-southeast to north-south stretching in the Skagit and Swakane rocks; top-to-the-the-north to north-northeast noncoaxial shear in the Swakane Biotite Gneiss, and less prominent top-to-the-northwest shear in the Skagit Gneiss Complex; and dextral-normal shear in the Ross Lake fault zone (Figs. 9 and 10). The displacement direction in the fault zone is generally oblique to stretching in the gneisses.

The orientation of the stretching lineation relative to the Ross Lake fault zone and the overall trend of the orogen suggest that a simple transtensional model cannot be applied. In a dextral wrench mode, stretching should be oriented ∼45° from the strike of the fault zone, at ∼095°–275°, whereas in partitioned transtension it should be broadly perpendicular, ∼050°–230° (e.g., Tikoff and Teyssier, 1994). Near a major strike-slip zone, high strains may rotate fabrics into subparallelism with the zone, but this seems unlikely to affect patterns at distances of kilometers from the zone. The stretching direction must be influenced by other poorly understood factors, such as along-strike changes in crustal thickness (see following).

Transtension in the brittle crust was manifested by dextral strike-slip faults, widespread dike swarms, and subsidence and thick accumulation of clastic sediments in nonmarine basins (Figs. 9 and 10). Orientations of structures indicate complex partitioning of deformation. Dextral strike slip on the northwest-striking Ross Lake fault zone (Foggy Dew, Ross Lake faults) probably overlapped with motion of the similarly oriented Entiat fault and the north-south–striking Straight Creek–Fraser fault (Fig. 1; Table 1). The northwest-striking faults have components of down-to-the-southwest (Entiat fault) and down-to-the-northeast to east (Ross Lake and Foggy Dew faults) displacement. Orientations of the Eocene dikes indicate that extension was dominantly west-northwest–east-southeast to northwest-southeast, but in detail markedly variable (Fig. 9). The average strike of the Teanaway dike swarm is ∼30° to that of the Straight Creek fault, which is an appropriate orientation for the strain field associated with a transtensional dextral wrench fault (e.g., Sanderson and Marchini, 1984) (Fig. 7).

The magmatic fabrics in Eocene plutons that are interpreted to record regional strain formed at depths intermediate between those of the metamorphic rocks and the dikes. Magmatic lineations trend northwest or southeast and are gently to moderately plunging. These orientations are broadly similar to those in the ductile crust.

Extension directions from ductile and brittle structures and the sense of noncoaxial shear in different areas define a complex Eocene strain field (Figs. 9 and 10). Overall, the ductilely deformed rocks underwent largely orogen-parallel to orogen-oblique, subhorizontal flow. The shallow crustal extension direction inferred from dikes is also oblique to the strike of the orogen and to the flow direction in deeper crustal rocks. Differences in extension between the dikes and Skagit and Swakane rocks average ∼20°–25°and 65°–70°, respectively (Fig. 9).

A distinctive component of the deformation history is the analogous geometry and timing of Eocene folds of the Swauk Formation and Skagit Gneiss Complex that may also be in part synchronous with upright folding of foliation in the Swakane Biotite Gneiss and the Dinkelman décollement. Folding is bracketed between ca. 51 and 49 Ma in the Swauk Formation, and the latest folding occurred ca. 49–48 Ma in the southern part of the gneiss complex and continued until at least 48 Ma (ending by 45 Ma) in the north. Fold geometries are broadly analogous in the Swauk and Skagit rocks (Fig. 6). In both settings, axial traces of the larger folds curve considerably (Figs. 5C and 7). In the Swauk rocks, the trend may result in part from the orientation of boundaries of the basin, and in the west, from rotation due to strike slip on the Straight Creek fault. In the Skagit rocks, the late moderately to steeply plunging folds of foliation described here are difficult to explain. Such large folds with steep axes are commonly thought to not extend into the deep crust (cf. Johnston and Acton, 2003) and may be truncated by a detachment.

The origin of this short-lived folding is uncertain. Folding may occur during transtension, but the associated shortening is limited (Fletcher and Bartley, 1994; Fossen et al., 2013). If the folds are related to transtension, the low angle (<20°) of many of the hinge lines to the northwest-trending, dextral strike-slip faults (Fig. 5C) that involve the Cascades core suggests that the divergence angle is low (cf. Fossen et al., 2013). The folding more likely reflects a short-lived episode of regional shortening that interrupted the overall transtensional regime. In one interpretation, folding results from collision and accretion of the Siletz-Crescent terrane (Siletzia) (Miller and Umhoefer, 2013; Eddy et al., 2016), which is a large igneous province west of the North Cascades (Fig. 2). This collision has been inferred to account for Eocene deformation in southwest Oregon and Vancouver Island (Fig. 2), where the terrane boundary is marked by thrusting and folding (Johnston and Acton, 2003; Wells et al., 2014). Collision in southwest Oregon occurred between ca. 51 and 47 Ma.

In the Paleogene, ca. 55 Ma, there was a fundamental change from regional transpression to transtension in the western Cordillera at the latitude of the Coast Mountains and North Cascades (e.g., Parrish et al., 1988; Miller and Bowring, 1990). This swing in deformation regimes is interpreted as a response to a change in the motion between the North American and Farallon/Kula plates (e.g., Engebretson et al., 1985). In addition, a slab window may have formed during postulated subduction of the Farallon-Kula ridge or Farallon-Resurrection ridge beneath the North American plate ca. 50 Ma (e.g., Breitsprecher et al., 2003; Madsen et al., 2006).

The orogen-parallel to orogen-oblique extension in the brittle and ductile levels of the Washington Cascades indicates that plate motion was not simply partitioned into dextral strike slip parallel to the northwest-trending plate boundary and extension normal to the boundary. Furthermore, orogenic collapse did not result in extension normal (northeast-southwest) to the strike of the arc, as in the metamorphic core complexes to the east.

We postulate that orogen-parallel to orogen-oblique ductile stretching was driven by a combination of dextral shear along the plate margin and along-strike differences in crustal thickness and temperature (cf. Paterson et al., 2004). Crustal thickness of the Cascades core may have decreased from north to south as a result of differences in the magnitudes of mid-Cretaceous shortening. Shortening was driven by eastward underthrusting of the rigid Insular superterrane (McGroder, 1991). In addition, the contractional belt probably terminates at the southern end of the Coast Mountains–North Cascades arc (cf. Monger et al., 1982, 1994), leading to less thickening at this end of the orogen. Much of the southern part of the Cascades core, represented by the Nason terrane, had cooled below biotite K-Ar and ⁴⁰Ar/³⁹Ar closure temperatures by 80 Ma (e.g., Tabor et al., 1982, 1987a; Matzel, 2004), in marked contrast to the Skagit Gneiss Complex, where upwelling of hot asthenosphere in the inferred slab window may have helped localize late metamorphism, magmatism, and partial melting. The top-to-the-north shear in the Swakane Biotite Gneiss and top-to-the-northwest flow in some of the Skagit Gneiss Complex suggests that during at least part of this deformation, the hot and weak middle to deep crust flowed southward from the region of 79–45 Ma magmatism, metamorphism, and partial melting in the Skagit Gneiss Complex toward the region of thinner and cooler crust in the south. We speculate that this crustal flow is broadly similar to the central Andes where orogen-parallel lower crustal flow is interpreted to have occurred in response to differences in upper crustal shortening in the Altiplano and Puna Plateau (e.g., Hindle et al., 2005; Ouimet and Cook, 2010). Similarly, deep crustal flow discordant to shallower level extension is also thought to have been driven by differences in crustal thickness in the Ruby Mountains metamorphic core complex of Nevada (MacCready et al., 1997).

The North Cascades and Omineca metamorphic core complexes to the east were sites of thick (>55 km) crust in the Late Cretaceous–Paleogene, and underwent extension, cooling, and exhumation in the Eocene (e.g., Parrish et al., 1988; Miller and Paterson, 2001; Paterson et al., 2004). Both regions also contain migmatites with similar Eocene melt-crystallization ages (e.g., Vanderhaeghe et al., 1999; Crowley et al., 2001; Hinchey et al., 2006; Kruckenberg et al., 2008; Gordon et al., 2008, 2010a). The two areas are currently connected by a regionally flat Moho, with crustal thicknesses of ∼30 km (Cook et al., 1992). Based on these characteristics, Whitney et al. (2004) postulated that the Omineca and North Cascades metamorphic and plutonic belts were dynamically coupled and represent the exhumed eastern and western margins, respectively, of an orogenic plateau.

Pervasive Eocene flow and kinematics recorded in gneisses and mylonites associated with the major detachment of the Shuswap metamorphic core complex of the Omineca belt (Fig. 11) indicate largely east-west extension (e.g., Read and Brown, 1981; Brown and Journeay, 1987; Carr et al., 1987; Parrish et al., 1988; Teyssier et al., 2005). In the southwestern part of this complex, however, the Okanogan dome records west-northwest–east-southeast to northwest-southeast extension (Kruckenberg et al., 2008). In comparison, ductile flow was more northwest-southeast to north-south in the Cascades core. We infer that these differences in part reflect the position of the thick crust relative to the plate margin. In much of the Shuswap complex, extension was probably driven by orogenic collapse in a largely extensional regime (e.g., Teyssier et al., 2005), whereas in the western part of the Okanogan dome and North Cascades, dextral shear associated with the obliquity of the plate boundary modified the strain field and resulted in transtensional flow. Any linked deep crustal flow beneath the hypothesized orogenic plateau must thus be marked by a major swing in the direction of flow. In the Omineca metamorphic core complexes, a broad zone of ductilely deformed rocks has been kinematically and thermally linked for at least part, if not all, of the deformation history to the structurally higher detachment faults during exhumation of the footwall. This relationship may also apply to the Swakane Biotite Gneiss, but it is not apparent for the Skagit Gneiss Complex. Instead, exhumation and ductile shear in the gneiss complex were in part coeval with the dextral-normal shear in the Ross Lake fault system.

The North Cascades orogen exposes a wide range of crustal levels that recorded Eocene transtension differently. The largest domain of deep crust exhumed in the Eocene, the Skagit Gneiss Complex, reveals a long record of magmatism, metamorphism, and partial melting. The complex has gently to moderately dipping foliation, gently northwest- and southeast-plunging lineation, and sizable constrictional domains, all of which formed in part during dextral-normal ductile and brittle shear in the bounding Ross Lake fault system. The other deep-seated domain, the Swakane Biotite Gneiss, records pervasive Eocene top-to-the-north to north-northeast noncoaxial shear. Strain patterns in the Swakane and Skagit rocks are compatible with transtension and orogen-parallel to orogen-oblique subhorizontal stretching.

The shallow crust in the Eocene was marked by dextral and normal slip on regional northwest- and north-south–striking high-angle faults, basin subsidence, rapid deposition, and widespread intrusion of basaltic to rhyolitic dikes. On a regional scale, extension directions from dikes are highly variable, but are mostly oblique to the trend of the orogen.

The brittle and ductile levels of the North Cascades crust were deformed coevally, but record different strain geometries and extension directions. These differences illustrate the problems with extrapolating strain patterns from one crustal level to another level and using these patterns to interpret ancient plate motions. These differences may reflect the complex regional and local boundary conditions. Ductile flow of the hot and weak Skagit Gneiss Complex and Swakane Biotite Gneiss was probably in response to regional strike slip and extension, vertical shortening and collapse, and along-strike differences in crustal thicknesses in the Cascades core. In particular, the variation in crustal thickness and the presence of thick, hot and weak lower crust in the northern part of the core may have led to the differences in extension directions.

We thank Scott Paterson, Christian Teyssier, Mike Eddy, Niki Wintzer, Paul Umhoefer, Jamie Clay Johnson, and Jahan Ramezani for thoughtful discussions. Miller acknowledges Art Snoke for his major contributions to our understanding of crustal sections and Cordilleran tectonics, and his exemplary service to the geologic community. Reviews by Margi Rusmore, an anonymous reviewer, and editors Shan de Silva and Josh Schwartz were very helpful. The U.S. National Park Service generously helped with logistics. This research was supported by National Science Foundation grants EAR-0511062, EAR-1119358, and EAR-1419787 to Miller, EAR-0510591 to Bowring, and EAR-0510326 to Whitney, and U.S. Geological Survey EDMAP grants 07HQAG00125 and 08HQAG0044 to Miller.