Testing modes of exhumation in collisional orogens: Synconvergent channel flow in the southeastern Canadian Cordillera

The mechanisms that lead to the exposure of large areas of high-grade rock in the hinterland of collisional orogens are long-standing tectonic issues (e.g., Ring et al., 1999). The uplift of rock relative to Earth's surface is referred to as “exhumation” (cf. England and Molnar, 1990). It can occur during convergence or during postcollisional extension. It commonly involves a complex interaction among erosion, normal faulting, and underplating (e.g., Ring et al., 1999), and may involve crustal decoupling and ductile flow (e.g., Beaumont et al., 2001). It is possible to group the various exhumation processes into three main modes: postconvergence gravitational collapse (cf. Rey et al., 2001), synconvergent orogenic wedges (e.g., Platt, 1986; Dahlen, 1990), and synconvergent channel flow and ductile extrusion (e.g., Beaumont et al., 2001).

The southeastern Canadian Cordillera (Fig. 1), with its easy access and excellent exposure, is one of the most studied collisional orogens exposing a large tract of midcrustal migmatites. Although it has been cited as a type example of postconvergent gravitational collapse resulting in extensional metamorphic core complexes (Coney and Harms, 1984; Parrish et al., 1988; Vanderhaeghe and Teyssier, 1997; Teyssier and Whitney, 2002; Teyssier et al., 2005; Rey et al., 2009), other workers have argued that a significant portion of the exhumation path occurred during convergence in an orogenic wedge (Brown and Journeay, 1987; Parrish, 1995; Brown, 2004), or by ductile extrusion and channel flow (Scammell, 1993; Johnston et al., 2000; Williams and Jiang, 2005; Brown and Gibson, 2006; Glombick et al., 2006a; Kuiper et al., 2006). It therefore provides an excellent natural laboratory in which to devise and apply diagnostic tests for the different processes that may contribute to exhumation processes in collisional orogens.

In this contribution, we first devise such a test by extracting a set of diagnostic criteria from relevant numerical and analogue modeling studies. It is then applied to rocks surrounding the Monashee Complex in the southeastern Canadian Cordillera. Only one model passes the critical test.

Our diagnostic criteria were derived from a compilation of the main characteristics of 14 models that provide predictions about the evolution of structural and metamorphic patterns (Table 1). We found that the shape of pressure-temperature (P-T) paths and absolute peak P-T conditions are not sufficiently diagnostic of the processes involved. For example, near-isothermal decompression paths can be produced in all three modes, depending on modeling parameters. In contrast, the spatial distribution of finite-strain patterns, cooling ages, and P-T-time (t) paths, as well as metamorphic field gradients and the timing of motion along reverse and normal shear zones, are diagnostic. Along with the set of criteria, Figure 2 presents the three models from each category that are most consistent with the overall geometry of the area surrounding the Monashee Complex in the southeastern Canadian Cordillera, which is illustrated on the cross section of Figure 3A.

From the subduction to the collision phase, convergent orogens have a wedge shape that is consistent with the critical-taper theory (e.g., Platt, 1986; Dahlen, 1990). In thermomechanical models of continental collision, in which the mantle lithosphere on the prowedge side detaches and subducts beneath a stationary retromantle (e.g., Willett et al., 1993), orogens develop a bivergent wedge geometry bounded by two oppositely verging shear zones. In the absence of erosion, exhumation is limited, and the wedge maintains its critical taper by thickening and growing laterally (Willet, 1999a; Vanderhaeghe et al., 2003a). Erosion significantly modifies the wedge dynamics by inducing a continuous cycling of rock that promotes significant exhumation of midcrustal rocks. The wide range of possible parameters produces a similarly wide range of dynamics, but some characteristics are common to most wedges with active erosion. In the vast majority of models, rocks enter the wedge from the proside, are buried and heated, and then follow an exhumation path in the hanging wall of the retro-shear zone, which migrates retroward during the lateral growth of the wedge (Willett et al., 1993; Jamieson et al., 1998; Batt and Braun, 1999; Willett, 1999a; Jamieson et al., 2002). This cycle results in the general key characteristics listed in Figure 2A. Finally, numerical and analogue models demonstrate that extension is unlikely in orogenic wedges (Willett, 1999b; Vanderhaeghe et al., 2003b; Bonnet et al., 2008). Therefore, major normal faults (or shear zones) should be younger and not related to the formation of an orogenic wedge.

In this mode of exhumation, it is the progressive underthrusting analogue model with décollements and erosion that is most consistent with the overall geometry of the southeastern Canadian Cordillera (Fig. 2A; Malavieille, 2010). In this model, exhumation results from the combined effect of underplating at the base of the wedge and denudation at the surface, which give rise to an antiformal stack of duplexes surrounded by low-grade rocks. In contrast to most other orogenic wedge models, a normal-sense shear zone may develop by vertical shearing of the left side of the first (top) duplex as it is uplifted.

Wedge dynamics have been shown to prevail until a significant thickness of the lower crust reaches a critical viscosity below which it can no longer support the weight of the wedge (Vanderhaeghe et al., 2003a). From this point onward, the dynamics of the orogen are not controlled by the critical-taper theory, but rather by the channel-flow concept derived from fluid mechanics (Grujic, 2006). Driven by the pressure gradient between a thick orogenic core and a thin foreland, a Poiseuille-dominated flow can be produced in large, hot, orogen-scale thermomechanical models that incorporate a stepwise decrease in viscosity at high temperature to simulate partial melting (Fig. 2B; Beaumont et al., 2001; Jamieson et al., 2002, 2004, 2006; Beaumont et al., 2004, 2006). In these models, significant exhumation occurs only when active erosion at the mountain front leads to an upward bifurcation of the channel, which then flows toward the erosion front. Alternatively, underthrusting of a stiff indentor can expel overlying low-viscosity rocks, which are exhumed by the equivalent of the indentor thickness (Jamieson et al., 2007, 2010). In addition, underthrusting of an indentor beneath a flowing channel (Fig. 2B) promotes exhumation, first by doming the channel and inducing extension of the overlying upper crust (Beaumont et al., 2004), and then by squeezing out this early-formed dome above the indentor (Jamieson et al., 2006, 2007). In contrast to the other exhumation modes, the progressive exhumation of the channel leads to a lateral—rather than vertical—spatial gradient of cooling ages, which become older toward the rear of the channel (Fig. 2B). The other general key characteristics of this model are listed in Figure 2B.

In the channel-flow category, it is the HT111 model of Jamieson et al. (2006) that is most consistent with the overall geometry of the southeastern Canadian Cordillera (Fig. 2B). In this model, a dome forms by the underthrusting of a stiff basement indentor beneath a flowing midcrustal channel. Once thermally softened, the tip of the indentor is thrust back over the rear of the indentor.

Gravitational collapse of the thickened welt takes place when the convergence rate declines to a point at which the related traction force can no longer sustain the weight of the thickened orogen. Three main modes of gravitational collapse of thickened orogens can lead to the exhumation of midcrustal rocks (Rey et al., 2001). If the middle crust is stiff and refractory, extension would result in the formation of crustal-scale boudins cored by competent rocks such as older basement. This is akin to the heterogeneous crustal boudinage model proposed by Price and Monger (2003) for the formation of the domes of the southeastern Canadian Cordillera. On the contrary, a low-viscosity middle crust would flow. In the models considered herein, those with a precursor zone of weakness, such as upper-crustal faults or midcrustal magmatic sills, result in the formation of metamorphic core complexes formed by the flow of the ductile crust toward a necked upper crust (Tirel et al., 2006, 2009; Brun et al., 1994; Rey et al., 2009). Alternatively, models without such weaknesses result in proward midcrustal flow that induces thrusting in the foreland (Jamieson et al., 2002, 2010; Beaumont et al., 2010). Despite important differences, all these models share common characteristics (Fig. 2C). In general, reverse shear zones, which contributed to the thickening phase, should be older than the normal shear zones, which exhumed the complex. Better distinctive criteria predicted by this model are listed in Figure 2C. In the gravitational collapse category, it is the type 2, sequential core complex model (fig. 5D inTirel et al., 2009) that is most consistent with the overall geometry of the southeastern Canadian Cordillera.

In this model, the ductile middle crust flows toward the necked upper crust to form a first asymmetric dome that is then flattened as a second dome develops in its footwall. The early-formed dome is bounded by reverse and normal shear zones at its base and roof, respectively. This is one of the rare models of this exhumation mode resulting in coeval oppositely verging shear zones.

The ∼900-km-wide southern Canadian Cordillera can be subdivided into five morphogeological belts (Fig. 1A) that record a protracted tectonic history from the Early Jurassic to the Eocene (e.g., Evenchick et al., 2007). The Insular, Intermontane, and Foreland belts consist of rocks of mainly low metamorphic grade, whereas the Coast and Omineca belts consist of variously deformed metamorphic and plutonic rocks (Monger et al., 1982). The study area is located within the southern part of the Omineca belt (Fig. 1A).

Four main periods led to the formation of the Omineca belt (Evenchick et al., 2007): the Jurassic, 145–135 Ma, 115–90 Ma, and 74–59 Ma. The Jurassic was a period of final amalgamation of terranes that were thrust and obducted over North American pericratonic terranes and basement. The southwest-vergent deformation and exhumation that characterized the Jurassic (Colpron et al., 1996) was interpreted to have developed in a prowedge setting (Brown and Willett, 1993; Gibson et al., 2005). The two mid-Cretaceous periods involved thickening and regional uplift accompanied by the formation of zones of penetrative northeast-verging deformation at midcrustal levels (Evenchick et al., 2007). The Late Cretaceous period (74–59 Ma) is critical for this study because it is during this period that ∼200 of the ∼250 km of net shortening took place in the Foreland belt at this latitude (Price and Sears, 2000). It is commonly accepted that the Paleogene (60–48 Ma) was a period of extension and gravitational collapse of the thickened orogenic welt (e.g., Coney and Harms, 1984; Parrish et al., 1988; Liu, 2001; Teyssier et al., 2005), a view partially contested by the results of this study and that of Gervais et al. (2010).

In order to critically test the three modes of exhumation (Fig. 2) and to improve the subdivision scheme of the southeastern Canadian Cordillera, we compiled plutonic, structural, metamorphic, and geochronologic data from 52 published papers and unpublished theses (Table 2). The striking differences in terms of names and spatial distribution of tectonometamorphic units on regional maps published in the past 10–15 yr (cf. Parrish, 1995; Colpron et al., 1999; Teyssier et al., 2005; Brown and Gibson, 2006; Carr and Simony, 2006; Glombick et al., 2006a; Johnson, 2006) motivated us to propose an updated subdivision scheme. We build on Read and Brown (1981), who divided the crust of the southern Omineca belt into two main units: the Selkirk allochthon and the Monashee Complex. The Selkirk allochthon includes variously deformed and metamorphosed rocks from accreted terranes, as well as sediments and mafic sills deposited/intruded on the attenuated western margin of the North American craton. It was thrust onto basement rocks of the craton and its cover sequence, which together form the Monashee Complex. These two units are clearly distinguished by the timing of their respective plutonic events. The Selkirk allochthon is composed of various plutonic suites of Devonian–Mississippian, Jurassic, Early to Late Cretaceous, and Paleogene ages, whereas the underlying Monashee Complex contains only Paleoproterozoic (in the basement) and Paleogene intrusions (Table 2). Notwithstanding their distinct lithostratigraphic origins and tectonometamorphic histories (see following), their contrasting plutonic histories alone imply that the two units were not juxtaposed until the Late Cretaceous, at the earliest, and cannot be grouped into one unit, as has been done recently (e.g., Vanderhaeghe et al., 1999; Williams and Jiang, 2005; Teyssier et al., 2005; Glombick et al., 2006a). Our compilation further suggests that the Selkirk allochthon should be divided into an upper and a lower panel because they each have contrasting structural, metamorphic, and geochronologic characteristics. The upper Selkirk allochthon consists of greenschist- to lower-amphibolite–facies rocks deformed by SW-verging folds in the Jurassic, whereas the lower Selkirk allochthon consists of upper-amphibolite–facies rocks that were strongly deformed by top-to-the-E/NE shearing for a protracted period in the Cretaceous to Paleocene (Table 2). The new subdivision is presented on a map (Fig. 1B) and on a cross section (Fig. 3A). The reader is referred to Gervais (2009) for an extensive discussion about the locations of the boundaries and other specific information concerning this subdivision scheme. The architecture of the exposed crustal section thus consists of a stack of three tectonometamorphic units, similar to the three crustal zones of Carr (1991).

This section describes the salient structural, metamorphic, and geochronologic characteristics of each tectonometamorphic unit from lower to upper structural levels. The spatial distribution of finite strain pattern and metamorphic field gradients derived from this description will be used to test the tectonic models of exhumation. The sources of data and related references are given in Table 2, unless otherwise cited.

The deepest unit is the Monashee Complex, which consists of two domes, Frenchman Cap in the north and Thor-Odin in the south (Fig. 1B). On its east flank, the Monashee Complex is bounded by the east-dipping Columbia River fault, a brittle-ductile normal fault that was active between 49 and 47 Ma (Mulch et al., 2006). On its west flank, it is bounded by the westerly dipping, Monashee décollement (Brown et al., 1992; McNicoll and Brown, 1995), a major ductile reverse shear zone. Recent studies in the Frenchman Cap dome (Crowley et al., 2008; Gervais et al., 2010) revealed a stepwise disappearance of pervasive Cordilleran deformation downward, with a 4–5-km-thick section of felsic basement gneisses preserving a Paleoproterozoic migmatitic gneissosity, and with Cordilleran deformation limited to local upright folds and meter-scale shear zones. Above this orogenic base, an ∼5-km-thick rock package records high finite general shear-strain characterized by a transposition foliation and a well-developed W-plunging lineation (stretching and mineral), toward which minor fold axes have been rotated into near parallelism during top-to-the-E shearing. Peak metamorphic assemblages in metapelites reached the kyanite–K-feldspar field and were heterogeneously retrogressed in the sillimanite and andalusite fields. Ductile deformation and high-temperature metamorphism took place between 63 and 49 Ma, depending on structural level. Gervais et al. (2010) showed that the dome formed between 52 and 49 Ma synchronously with the waning stages of top-to-the-E crustal shearing and shortly after the underthrusting of a basement ramp.

Similar structural and metamorphic features were documented in the southern Thor-Odin dome. Top-to-the-NE noncoaxial finite strain is ubiquitous, and near parallelism between WSW-ENE–stretching lineations and minor fold axes is also common. Peak metamorphic conditions reached conditions similar to those of the Frenchman Cap dome, but more melt was produced as a result of crossing the low-pressure biotite breakdown reaction, which produces cordierite and garnet as peritectic products (Norlander et al., 2002). The timing of top-to-the-NE shearing and high-temperature metamorphism is also the same as that of the Frenchman Cap dome and occurred between 56 and 51 Ma. The main difference between the two domes is that the base of Cordilleran strain has not been identified in the Thor-Odin dome and that a body of Cordilleran diatexite with a subvertical lineation and a dome-up sense of shear was mapped (Vanderhaeghe et al., 1999). The presence of this diatexite body, combined with the isothermal decompression path documented nearby, constituted the basic arguments in favor of the widely publicized model of doming by vertical flow (Vanderhaeghe and Teyssier, 1997; Vanderhaeghe et al., 1999; Norlander et al., 2002; Teyssier and Whitney, 2002; Whitney et al., 2004; Teyssier et al., 2005). However, the absence of a Cordilleran penetrative fabric in the core of the Frenchman Cap dome combined with the continuity of lithologic markers across the domes (Read, 1980) rule out the vertical-flow model for the formation of the domes of the Monashee Complex (Gervais et al., 2010).

Overlying the Monashee Complex, there is the 8–10-km-thick lower Selkirk allochthon. All features described here are common to the continuous exposure of the lower Selkirk allochthon from the western flank of the Monashee Complex to the northern Monashee Mountains, where isograds and lithological contacts are continuous toward the SE in the Selkirk Mountains (Fig. 1B). Along cross-section A–B (Fig. 3A), the panel is exposed in the footwall of the top-to-the-W Okanagan Valley fault system (OVfs; as per Johnson, 2006) to the west of the Monashee Complex; it is domed above the complex, down-faulted under the surface by the Columbia River fault to the east of the complex, and surfaces again in the footwall of the Selkirk detachment system in the Selkirk Mountains. As suggested by Brown and Gibson (2006), the Okanagan Valley fault system and the Selkirk detachment system likely formed a single structure before being cut by the Columbia River fault.

The lower Selkirk allochthon is characterized by high finite general shear-strain like the upper part of the Monashee Complex. It has a transposition foliation dipping shallowly to the W-SW and a pervasive E-W to WSW-ENE stretching lineation, toward which minor fold axes have been substantially rotated during top-to-the-E shearing. Because such rotation by noncoaxial shearing alone requires a very high amount of strain, Scammell (1993) concluded that deformation possessed a component of coaxial strain (i.e., general shear). The mixture of noncoaxial and coaxial structures observed in the structural center of the lower Selkirk allochthon supports this conclusion (Journeay, 1986; Scammell, 1993; Johnson, 1994; Glombick, 2005). In the southern part of the map (Fig. 1B), the lower Selkirk allochthon is bounded at its roof by the top-to-the-W Okanagan Valley fault system, and at its base by the top-to-the-E/NE Monashee décollement. These two oppositely verging shear zones are several hundreds of meters thick and consist of mylonite intruded by large volumes of sheared leucogranite (Journeay, 1986; Scammell, 1993; Johnson, 1994; Glombick et al., 2006a). In the northern part of the map (Fig. 1B), the boundary between the lower Selkirk allochthon and the upper Selkirk allochthon was mapped as a gradational, SW-dipping metamorphic gradient (Simony et al., 1980; Raeside and Simony, 1983). In contrast to the southern part of the map, the lower Selkirk allochthon is therefore not bounded by a normal shear zone in the northern part of the study area (Fig. 1B).

The lower Selkirk allochthon is at the upper amphibolite facies. Rocks are highly migmatitic and intruded by leucogranite sills, dikes, and plutons that together form between 30% and 60% of any outcrop (Journeay, 1986; Carr, 1991; Scammell, 1993; Glombick et al., 2006a; Johnson, 2006). Metapelitic schists contain sillimanite and K-feldspar, except near the boundaries with the upper Selkirk allochthon, where muscovite is commonly stable (Fig. 1B). At the latitude of the Thor-Odin dome, the muscovite-out and the kyanite-out isograds are exposed within an attenuated gradient less than ∼3 km thick in the footwall of the Okanagan Valley fault system (Carr, 1991; Bardoux, 1993; Lemieux, 2006; Glombick et al., 2006a), whereas in the northern Monashee and Selkirk Mountains, these two isograds are also exposed, but within a thickness in excess of 7 km structurally above the upper Selkirk allochthon–lower Selkirk allochthon boundary (Simony et al., 1980; Raeside and Simony, 1983; Digel et al., 1998).

A key characteristic of the lower Selkirk allochthon is a protracted history of high-temperature metamorphism and deformation. The emplacement of numerous synkinematic leucogranites between 105 and 90 Ma throughout the lower Selkirk allochthon implies that it was deforming at high temperature at this time (Scammell, 1993; Johnson, 1994; Parrish, 1995; Crowley et al., 2003), but crosscutting relationships of other leucogranite dikes further suggest that a substantial amount of strain was produced before ca. 135 Ma in the northern part of the area (between A and B in Fig. 1B; Scammell, 1993) and after 70 Ma in the southern part of the area. Furthermore, a major thermal event near area A (Fig. 1B) between 105 and 90 Ma is indicated by the cluster of monazite dates in migmatitic schists and of zircon ages in leucogranite. The temperature was also high enough for protracted growth of zircon in leucosomes in the Early Cretaceous and from 120 to 65 Ma in the hanging wall of the Monashee décollement near area C (Fig. 1B), and in the Jurassic and from 100 to 57 Ma in the footwall of the Okanagan Valley fault system near Sicamous and Vernon (Fig. 1B). These observations suggest that the thermal peak was reached in the mid-Cretaceous throughout, but that the temperature remained high until the Paleocene in lower Selkirk allochthon rocks cropping out west of the Monashee Complex.

The upper Selkirk allochthon includes all rocks deformed and metamorphosed prior to ca. 120 Ma. Most of these rocks are at the garnet grade or lower, but local areas reached the sillimanite-muscovite grade (Fig. 1B). Except for rocks west of Blue River (Fig. 1B), sillimanite is post-tectonic and occurs in pluton aureoles or in the hanging wall of the Selkirk detachment system (Fig. 1B). Similarly, most of the upper Selkirk allochthon was deformed by W- to SW-vergent Jurassic folds, but it locally records NE-vergent Jurassic structures, mostly in the NE part of the area shown in Figure 1B. These rocks cooled below the hornblende and mica closure temperature (350–550 °C) before ca. 120 Ma, except near the lower Selkirk allochthon–upper Selkirk allochthon boundary, where Late Cretaceous ages are recorded (Fig. 3B). Consequently, rocks of the upper Selkirk allochthon had been exhumed above the brittle-ductile transition (300–400 °C) by the mid-Cretaceous and thus formed the upper-crustal lid of the southeastern Canadian Cordillera in the Late Cretaceous to Paleogene.

Several studies have documented the distribution of cooling ages throughout the lower Selkirk allochthon. In our compilation (Table 2; Fig. 3B), we retained the original authors’ interpretations of U-Pb dates, but only retained ⁴⁰Ar/³⁹Ar hornblende, biotite, and muscovite plateau ages derived from >80% of the released ³⁹Ar gas to avoid problems associated with excess argon and complex cooling/reheating histories. Several interesting features are derived from Figure 3B, which shows all cooling ages of the lower Selkirk allochthon projected onto cross-section A–B. First, it highlights the contrast between the upper Selkirk allochthon, which records mainly Jurassic ages, and the lower Selkirk allochthon, which records ages varying between 95 and 55 Ma. Rocks of the upper Selkirk allochthon located near the contact of the lower Selkirk allochthon, however, record younger hornblende Ar/Ar ages in the range 89–110 Ma, which coincide with the thermal peak recorded in the lower Selkirk allochthon (Table 2). Second, there is a trend of titanite and hornblende cooling ages that become younger southwestward in lower Selkirk allochthon rocks cropping out NE of the Monashee Complex (Figs. 3A and 3B), but not in those cropping out west of the complex. Titanite ages decrease from 90–98 Ma at location A (Fig. 1B) to ca. 74 Ma at location B, to ca. 62 Ma at location C (see Table 2 and following section for data) and are then constant at 62–55 Ma throughout the lower Selkirk allochthon west of the Monashee Complex. Similarly, hornblende cooling ages decrease from ca. 75 to ca. 62 Ma between locations A and B, and are then constant at 57–54 Ma throughout the lower Selkirk allochthon west of the Monashee Complex. As noted earlier, this trend is also reflected in U-Pb zircon ages from leucosomes that range in age between 100 and 90 Ma NE of the Monashee Complex, whereas those west of the complex extend to ages as young as ca. 57 Ma. Third, mica cooling ages are slightly younger NE of the Monashee Complex, where they range between 56 and 53 Ma, than west of it, where they cluster at 50–47 Ma. Consequently, there is no clear trend of cooling ages across structural levels of the lower Selkirk allochthon (as proposed by Parrish, 1995), but there is a lateral NE-SW trend. Because this trend is perpendicular to the axis of the autochthonous basement of the Monashee Complex (Armstrong et al., 1991; Crowley et al., 2001; Gervais et al., 2010), it implies that rocks of the lower Selkirk allochthon that were thrust above the complex cooled prior to rocks that were not.

In this section, P-T-t paths are presented for different localities of the lower Selkirk allochthon and of the Monashee Complex. Robust paths are available for the lower Selkirk allochthon NE of the Monashee Complex (locations A and B in Fig. 1B) and from the Frenchman Cap and Thor-Odin domes in the complex itself. However, no such path is available for the lower Selkirk allochthon west of the complex (Table 2).

Two samples, less than 20 m apart, were collected at area C (Fig. 1B) to construct a P-T-t-deformation (P-T-t-d) path. Rocks at this locality consist of interlayered semipelitic schists, quartzofeldspathic gneisses, and boudinaged amphibolitic layers. As for all rocks of the ∼1-km-thick Monashee décollement, they are migmatitic, mylonitic, and record top-to-the-E sense of shear in a variety of kinematic indicators (Fig. 4). Many observations indicate that melt was present during deformation (e.g., Fig. 4C). Migmatitic schists contain more than 40% of leucocratic material in the form of leucosome (15%–30%) and leucogranite intrusions (25%–40%). The minimum age of deformation is provided by the 57.0 ± 0.2 Ma age of a suite of leucogranite that crosscuts the mylonitic foliation at right angles (Scammell, 1993).

Conventional garnet-plagioclase-sillimanite-quartz (GASP) barometry yields an estimate of the pressure peak reached by sample Fg-462a. The uniform plagioclase and grossular composition in this sample justify the assumption of equilibrium necessary for the GASP calculations. However, garnet-biotite thermometry could not be used because it yielded P-T conditions below the muscovite-dehydration melting reaction, likely because garnet and biotite composition were modified by Fe-Mg exchange reaction and by R1 (Kohn and Spear, 2000). The temperature input for the GASP calculations were therefore estimated through experimental and numerical phase equilibria modeling of the muscovite-out and biotite-out reactions for metapelitic schists (Vielzeuf and Schmidt, 2001; White et al., 2007). Results (Table 4; Fig. 6) indicate a P-T field located immediately below the kyanite-sillimanite reaction between temperatures of 750 °C and 860 °C at pressures of 1200–800 MPa.

The second sample, Fg-462d, is a leucosome cored by a clinopyroxene-rich aggregate (Fig. 4D). The aggregate is an intergrowth of coarse-grained and euhedral to subhedral grains of clinopyroxene, titanite, plagioclase, K-feldspar, and quartz, whereas the host rock consists of aligned, medium-grained and recrystallized feldspar, quartz, biotite, and hornblende grains (Fig. 5B). The intergrowth texture (Fig. 5C), the euhedral shape of grains (Fig. 5C), and the sharp adjacent crystal faces (Fig. 5D) are interpreted as evidence for growth of clinopyroxene and titanite in the presence of melt, whereas the microstructures in the host gneiss indicate solid-state deformation (cf. Vernon, 2004). The concentration of leucosome into the asymmetric pressure shadows of the aggregate (Fig. 4D) further suggests migration of the melt phase during top-to-the-E shearing, which promoted the pseudomorphism of clinopyroxene by hornblende (Fig. 5D).

Conventional thermobarometry was conducted on sample Fg-462d. Clinopyroxene, hornblende, and plagioclase are unzoned, and their compositions do not vary with textural setting (Table 3). Table 4 and Figure 6 show the results of thermobarometry for sample Fg-462d using the average-pressure method of Thermocalc version 3.1 (Powell and Holland, 1994), the SCAn barometer (McCarthy and Patiño-Douce, 1998), and the hornblende-plagioclase thermometer (Holland and Blundy, 1994). The uniform compositions of clinopyroxene, plagioclase, and hornblende in specimen Fg-462d, combined with the excellent σ_fit = 1.0–1.1 for three independent reactions obtained by Thermocalc (see Powell and Holland, 1994), indicate that equilibrium between these three minerals was reached. Inasmuch as hornblende is interpreted to have formed during crystallization of a melt phase, the quartz-saturated hornblende-plagioclase thermometer (Fig. 6; Holland and Blundy, 1994) is preferred because it yields a temperature closer to the wet-granite solidus (Fig. 6). The average pressure of ∼270 MPa at 700 °C calculated with Thermocalc (Fig. 6) is consistent with the pressure of ∼285 MPa and 700 °C calculated by the SCAn barometer (Fig. 6). The preferred P-T conditions of equilibrium during hornblende formation in sample Fg-462d are thus estimated at ∼700 °C and 300 MPa (Fig. 6).

Twenty-six U-Pb analyses were obtained from 18 zircon grains extracted from sample Fg-462d (for analytical methods, see Appendices). Zircon commonly consists of rounded cores surrounded by euhedral and elongated rims with sharp crystal faces (Figs. 7A–7C). Cathodoluminescence (CL) images reveal a sector to faint oscillatory zoning (Figs. 7A–7C; Corfu et al., 2003). Grain L02 (CL image in Fig. 7A) is interesting because it exhibits an euhedral core overgrown by an euhedral rim separated by a dark thin band spatially associated with minor resorption textures. U-Pb dates of rounded cores range in age from Archean to Devonian (Table 5) and are interpreted to reflect xenocrystic inheritance. U-Pb dates of zircon rims that fall on, or near, concordia yield dates of ca. 120, 107, 83, and 62 Ma, with two clusters at ca. 83 Ma and ca. 62 Ma. The older cluster of rim dates, which represents four analyses from three grains including grain L02, yields a weighted mean average ²⁰⁶Pb/²³⁸U age of 82 ± 4 Ma (mean square of weighted deviates [MSWD] = 1.6; probability of fit (prob.) = 0.19; Fig. 7D). The younger cluster of rim dates, which represents five analyses from four grains including grain L02, yields a weighted mean average ²⁰⁶Pb/²³⁸U age of 62 ± 2 Ma (MSWD = 3.7; prob. fit = 0.01; Fig. 7D). The morphology of zircon grains and the overgrowth relationship of grain L02, which is formed by a ca. 82 Ma core and a ca. 62 Ma rim, suggest the presence of a melt phase (Corfu et al., 2003) during the two main periods of zircon growth.

Titanite in sample Fg-462d is ∼300 × 600 μm, euhedral, and in textural equilibrium with clinopyroxene and plagioclase (Figs. 5C and 5D). Titanite has dark cores surrounded by light rims when examined on backscattered electron images (Figs. 8A–8D). Regression of 21 of the 22 uncorrected analyses on a Tera-Wasserburg plot intersects concordia at 58 ± 9 Ma (MSWD = 18) and yields a y-axis ²⁰⁷Pb/²⁰⁶Pb intercept of 0.65 ± 0.13 (Fig. 8E). Using the y-axis for common Pb correction (Storey et al., 2006; Aleinikoff et al., 2007), we find a corrected weighted mean average ²⁰⁶Pb/²³⁸U age of 57 ± 2 Ma (MSWD = 1.00; prob. = 0.46; Fig. 8F; Table 6). There is no difference in age between cores and rims, even if treated separately for common Pb correction. We interpret this ca. 57 Ma age as the time at which the sample cooled below the closure temperature for Pb diffusion in titanite (600–680 °C, depending on the cooling rate for grains of this size; Cherniak, 1993).

The data presented here allow the construction of the retrograde P-T-t path followed by rocks of the northern segment of the Monashee décollement. The record of zircon growth in leucosome sample Fg-462d (Fig. 7D) indicates protracted high-temperature metamorphism between 120 and 62 Ma, typical of lower Selkirk allochthon rocks (Table 2). Peak metamorphic conditions were reached some time in this period and are estimated at 800–1200 MPa and 750–860 °C from the migmatitic pelitic schist sample Fg-462a (Fig. 6). Thermobarometric results indicate that the synkinematic clinopyroxene-bearing leucosome sample Fg-462d equilibrated at P-T conditions of ∼300 MPa and 700 °C (Fig. 6; Table 4). The retrograde P-T path, therefore, involved >500 MPa of decompression at high temperature. Top-to-the-E shearing was taking place during decompression because the shear fabric is formed by the product of the retrograde reaction R1. A key to deriving the timing of this path is the presence of a melt phase in sample Fg-462d. Indeed, evidence presented here indicates the presence of melt during: (1) the growth of zircon at ca. 82 and ca. 62 Ma (Fig. 7D); (2) the growth of the synkinematic clinopyroxene-titanite aggregate (Figs. 4D, 5C, and 5D); and (3) the retrogression of clinopyroxene to hornblende in the top-to-the-E asymmetric pressure shadow of the aggregate that equilibrated at ∼300 MPa and ∼700 °C (Figs. 4D and 5D). However, melt was most likely crystallized at ca. 57 Ma when Pb ceased to diffuse in titanite. Hence, the estimated P-T of equilibration (Fig. 6) is tightly bracketed between 62 and 57 Ma (Fig. 6). This is consistent with the ca. 57 Ma age of the leucogranite suite crosscutting the foliation at this outcrop (Scammell, 1993). Consequently, the northern segment of the Monashee décollement was shearing along a high-temperature decompression path before ca. 62 Ma and the last increments of top-to-the-E shearing occurred between 62 and 57 Ma at P-T conditions of ∼300 MPa and 700 °C.

Scammell (1993) constructed a P-T-t path for lower Selkirk allochthon rocks between locations A and B (Fig. 1B). Scammell was the first to point out that a large amount of top-to-the-E shear strain accumulated along a cooling path by R1 because the foliation, lineation, and many kinematic indicators are formed by its products. From a detailed analysis taking into account the effect of Fe-Mg exchange reactions between biotite and garnet, as well as the effect of R1, Scammell (1993) estimated peak metamorphic conditions at 700–800 °C and 700–900 MPa and retrograde conditions at ∼450 MPa and 600 °C. This 300–500 MPa of decompression at high temperature is consistent with our results for location C. The timing of this synshearing exhumation episode was bracketed between ca. 100 Ma, the U-Pb ages of transposed leucogranites and U-Pb monazite ages in schist, and ca. 90 Ma, the oldest U-Pb titanite age at location A (Fig. 1B). Scammell used this latter age rather than the ca. 74 Ma titanite age obtained at location B because he assumed that cooling was synchronous throughout the entire area. However, based on the trend of cooling ages reported in Figure 3B here, we reinterpret these two different ages as indicative of this trend. Inasmuch as similar retrograde P-T paths and timing of high-temperature metamorphism were recorded at locations A, B, and C, the trend of cooling ages implies successive high-temperature decompression paths becoming younger southwestward, as shown in Figure 9.

Foster et al. (2004) derived a prograde P-T-t path for a migmatitic kyanite-bearing schist collected by Gibson et al. (1999) in the footwall of the Monashee décollement near location D in the northern Frenchman Cap dome (Fig. 1B). The path involves heating and burial in the kyanite field between ca. 75 and 56 Ma and a peak pressure estimate of ∼1000 MPa. These results are consistent with the ca. 58 Ma estimate for the timing at which migmatitic schists from a similar structural level, but on the west flank of the dome, reached the kyanite + melt field (Crowley and Parrish, 1999; Crowley et al., 2001). The end of ductile deformation is constrained at ca. 55 Ma from the age of a crosscutting pegmatite and from a hornblende ⁴⁰Ar/³⁹Ar plateau age (Crowley and Parrish, 1999). Because kyanite in the ca. 58 Ma leucosomes is pseudomorphed by andalusite and because synkinematic andalusite has been observed nearby (Journeay, 1986; Scammell, 1986), there must have been significant decompression between ca. 58 and 55 Ma (Fig. 3C). Although slightly older, these results are consistent with the near-isothermal, high-temperature decompression path between 56 and 51 Ma documented in the Thor-Odin dome (Norlander et al., 2002; Hinchey et al., 2006).

A key criterion for testing tectonic models of exhumation is the timing of shearing along reverse and normal shear zones (Fig. 2). The existence of the Monashee décollement has been questioned by Vanderhaeghe et al. (1999), Williams and Jiang (2005), and Kruse and Williams (2007). However, the contrast in plutonic history (Table 2) and the contrast in depth-time paths (Fig. 9) between the lower Selkirk allochthon and the Monashee Complex clearly demonstrate that the two units were not juxtaposed until the Eocene. Furthermore, results presented herein for the northern segment agree with those obtained previously for the southern segment (Carr, 1992, 1995) and confirm that the last increments of top-to-the-E shearing along the reverse shear zone took place between 62 and 57 Ma along its entire length. The contrast in lithologic units may be less striking in the south than in the north (but see McNicoll and Brown, 1995), but the similarities in structure, metamorphism, and timing of shearing demonstrate that the Monashee décollement bounds the west flank of the Monashee Complex along its entire length. It is not possible to determine if shearing started earlier than ca. 62 Ma, but the protracted high-temperature conditions recorded by zircon growth in leucosomes and sheared leucogranites of this shear zone (continuous record from 100 to 58 Ma; Table 2) suggest that it did.

A general P-T-t trend for the Okanagan Valley fault system can be estimated from the metamorphic paragenesis in top-to-the-W microstructures combined with the timing of synkinematic intrusions and cooling ages. Top-to-the-W kinematic indicators are formed by: (1) the peak metamorphic minerals biotite–sillimanite-garnet-K-feldspar; (2) biotite-sillimanite folia in pressure shadows of resorbed garnet (Bardoux, 1993; Glombick et al., 2006a; Johnson, 2006); (3) retrograde minerals muscovite and chlorite; and (4) brittle structures (Bardoux, 1993; Johnson, 2006). This sequence indicates that shearing started at the thermal peak, continued along a cooling path, first at high temperature during garnet breakdown by R1, and then at lower temperature during crystallization of hydrous minerals, and finally ended in the brittle field. As for the Monashee décollement, the record of zircon growth in leucosomes from 100 to 57 Ma indicates protracted high-temperature conditions. Furthermore, the overlap between the intrusion of the synkinematic Ladybird leucogranite and hornblende Ar/Ar cooling ages implies that the last increments of top-to-the-W ductile shearing along the Okanagan Valley fault system took place between 60 and 56 Ma (Table 2; Carr, 1992; Johnson, 1994; Vanderhaeghe et al., 1999). It is not possible to ascertain if shearing started earlier, but the protracted record of zircon growth in sheared leucosome between 100 and 57 Ma suggests that it did.

The two shear zones bounding the lower Selkirk allochthon therefore share the same structural, metamorphic, and geochronological characteristics (Table 7). Combined with the high-temperature decompression path documented previously (Fig. 6), these similarities imply that the lower Selkirk allochthon was approaching the surface of Earth by coeval shearing along the top-to-the-E Monashee décollement at its base and the top-to-the-W Okanagan Valley fault system at its roof in the Paleocene.

In this section, the newly devised test of exhumation processes in collisional orogens (Fig. 2D) is applied to the area surrounding the Monashee Complex in the southeastern Canadian Cordillera. The three main modes of exhumation are ranked from least to most plausible based on a comparison of the data set described herein with the predictions of analogue and numerical models of collisional orogens (Fig. 2D).

Most studies have attributed the exhumation of midcrustal rocks in the southeastern Canadian Cordillera to the gravitational collapse of a thickened welt following a reduction of convergence rate and/or delamination of the mantle lithosphere in the Eocene (Coney and Harms, 1984; Tempelman-Kluit and Parkinson, 1986; Parrish et al., 1988; Carr, 1992; Bardoux and Mareschal, 1994; Vanderhaeghe and Teyssier, 1997; Liu, 2001; Norlander et al., 2002; Teyssier and Whitney, 2002; Price and Monger, 2003; Teyssier et al., 2005; Kruse and Williams, 2007; Rey et al., 2009). Our test could be partially compatible with the gravitational collapse model (Fig. 2D) if applied to the area south of the Frenchman Cap dome, where most of the aforementioned studies were conducted, but it would not be compatible if applied to the area surrounding the dome (e.g., cross section of Fig. 3A). However, the continuity of the lower Selkirk allochthon from west of the Thor-Odin dome to the northern Monashee Mountains (Fig. 1), which is well marked by similar lithological, structural, metamorphic, and geochronological characteristics, requires the incorporation of the entire area into the test. Hence, the absence of a major normal shear zone in the northern Monashee Mountains (Figs. 1 and 3A), the prevalence of finite strain indicative of general shear rather than horizontal flattening, and moreover the synchronous exhumation of the lower Selkirk allochthon and burial of the Monashee Complex (Fig. 9) rule out gravitational collapse as the preferred exhumation mode of the lower Selkirk allochthon. As for the Monashee Complex, the absence of a penetrative fabric of Cordilleran age, the preservation of Paleoproterozoic Pb in titanite, as well as the 52–49 Ma contractional structures in felsic gneisses located 5 to 10 km structurally below the kyanite–K-feldspar migmatites of the upper part of the complex imply late basement underthrusting (cf. Gervais et al., 2010). This, in turn, implies that the exhumation of these kyanite-migmatites to the andalusite field between 59 and 55 Ma occurred during convergence. A synconvergent mode of exhumation is therefore required for both the lower Selkirk allochthon and the upper part of the Monashee Complex.

The orogenic wedge mode was proposed mainly based on data collected in and north of the Frenchman Cap dome (Brown and Journeay, 1987; Parrish, 1995; Brown, 2004) and has not been tested against predictions from modeling studies (Fig. 2A). The first prediction common to all models is that of reverse shearing localized at the retro-shear zone and steep flattening in the wedge interior. In contrast, a shallowly dipping transposition foliation formed by reverse shear strain is distributed through the 16–20-km-thick package of high-grade rocks consisting of the lower Selkirk allochthon and Monashee Complex. The sharp transition from migmatites to greenschist-facies rocks across the Okanagan Valley fault system, as well as the absence of a cooling age gradient in the lower Selkirk allochthon west of the Monashee Complex are both not compatible with this exhumation mode. Furthermore, the coeval motion of major normal (Okanagan Valley fault system) and reverse (Monashee décollement) shear zones bounding the lower Selkirk allochthon is difficult to reconcile with an orogenic wedge model. The model of vertical normal shearing at the top of an uplifting duplex stack (Fig. 2A; Malavieille, 2010) is not applicable either, because this model requires a systematic gradient in peak metamorphic and cooling ages across structural levels. Finally, theoretical considerations suggest that wedge geometry could not have been maintained for a low-viscosity middle crust inferred from the large volume of migmatitic rocks in the lower Selkirk allochthon (Vanderhaeghe et al., 2003a). Therefore, another synconvergent model is required.

The first model proposing that part of the exhumation of the lower Selkirk allochthon occurred during convergence and involved crustal decoupling and ductile flow was the “dynamic spreading” model of Scammell (1993). A ductile extrusion model was later proposed by Johnston et al. (2000), and arguments in favor (Brown and Gibson, 2006; Glombick et al., 2006a; Kuiper et al., 2006) and against (Carr and Simony, 2006) the channel-flow concept have recently been put forward. The coeval, but opposite motion of two shear zones bounding the migmatitic lower Selkirk allochthon (Fig. 3A; Table 7), the lateral trend of cooling ages (Figs. 3B and 9), the protracted residence time at high temperature of the lower Selkirk allochthon, and the synchronous exhumation of the lower Selkirk allochthon and burial of the Monashee Complex in its footwall (Figs. 3C and 9) are predicted features common to all channel-flow models (Fig. 2B; Table 2). The sharp transition from migmatitic to greenschist-facies rocks in the hanging wall of the Okanagan Valley fault system and the absence of a sharp metamorphic field gradient in the footwall of the Monashee décollement are not. However, Jamieson et al. (2004) acknowledged that the incorporation of strain localization in the numerical code could lead to a sharper gradient in the hanging wall of a hot channel, which could be an explanation for the observations. The absence of a gradient in the footwall of the lower Selkirk allochthon is explained by the downward migration of flow, as explained in the following. In contrast to the other models, most of the requirements of this exhumation mode are thus met (Fig. 2B), and we argue that a channel-flow process similar to the HT111 model of Jamieson et al. (2006; Fig. 2B) led to the exhumation the lower Selkirk allochthon in the Late Cretaceous to Paleocene.

The extensive multidisciplinary data set compiled herein allows us to derive an internally consistent tectonic model (Fig. 10). Building on the model of Gibson et al. (2008), we propose that rocks of the Selkirk allochthon (both the upper Selkirk allochthon and lower Selkirk allochthon) were deformed and metamorphosed in a SW-verging prowedge in the Jurassic. Whereas rocks of the upper Selkirk allochthon were exhumed to upper-crustal levels (<10–15 km depth) in Jurassic to Early Cretaceous time, rocks of the lower Selkirk allochthon remained at depth and were transported in a retrowedge setting where they continued to be thickened and heated. This reconciles the record of Jurassic metamorphic ages in both the upper and lower panels, whereas Jurassic cooling ages are recorded only in the upper Selkirk allochthon (Table 2; Gibson et al., 2008). The thermal peak was reached at ca. 100–90 Ma, when a significant portion of the middle crust reached partial melting conditions, as recorded by numerous zircon and monazite U-Pb ages in sheared leucosome, leucogranite, and schists of the lower Selkirk allochthon. Building on the models of Scammell (1993), Brown and Gibson (2006), and Glombick et al. (2006a), we propose that, at this moment, wedge dynamics ceased, and Poiseuille flow started to prevail in an ∼10-km-thick channel that decoupled from its lid and base and flowed toward the foreland. Between 90 and 70 Ma (Figs. 10A and 10B), the frontal part of the channel (location A on Figs. 1B and 3) cooled below the titanite closure temperature (600–650 °C) at pressures of 400–600 MPa as it flowed above an underthrusting basement ramp. Flow was accommodated by internal ductile deformation within the channel and by top-to-the-E and top-to-the-W shearing along the Monashee décollement and the Okanagan Valley fault system, respectively. The southwest lateral trend of cooling ages (Figs. 3B and 9) suggests that basement underthrusting under a flowing channel continued for at least 30 m.y., between 90 and 60 Ma, while rocks of the Monashee Complex were progressively buried to a depth >35 km (1000 MPa; Fig. 3C). As mentioned already here and in Gervais et al. (2010), data from the Frenchman Cap dome indicate that basement underthrusting took place until 52–59 Ma. We propose a model similar to that of numerical model HT111 (Fig. 2B; Jamieson et al., 2006), in which the leading edge of the underthrusting indentor is expelled back above its stiffer part once it has been thermally softened. In the Monashee Complex, this would have occurred at ca. 58 Ma, when underthrusted cover sequence and basement rocks reached partial melting conditions (Figs. 3C and 10C). This model provides a mechanism to juxtapose rocks of the Monashee Complex and the lower Selkirk allochthon, which were then separated by more than 25 vertical km (compare Figs. 3C and 6). Postorogenic extension occurred at 50–47 Ma and was first accommodated in the brittle-ductile regime by the development of the Columbia River fault and the reactivation of the Okanagan Valley fault system (Fig. 10D), and then in the brittle regime by the formation of N-S–striking normal faults and the intrusion of lamprophyre dikes. Regional extension is well recorded by the cluster of mica cooling ages (Fig. 3B) and the timing of late shearing along the two main normal brittle-ductile shear zones, Columbia River fault and Okanagan Valley fault system (Table 2). It is noteworthy that this model is compatible with the palinspastic restoration of Johnson and Brown (1996), which implies significant exhumation of the lower Selkirk allochthon and Monashee Complex prior to 20%–25% extension.

In contrast with channel-flow models (Beaumont et al., 2001, 2004), erosion was most likely not a major driving force in the southeastern Canadian Cordillera. Low denudation rates in the Late Cretaceous are indicated by the study of Sears (2001), who coupled low-temperature thermochronology with palinspastic restoration, and are indicated by the large gap of ∼20 m.y. between hornblende and mica cooling ages in the northern Monashee Mountains. We rather suggest that most of the exhumation was due to structural decoupling at the top of a channel flowing up an incline plane, and that part of the space was created by shortening in the foreland belt (Figs. 10B and 10C). This is supported by the synchronicity of the main shortening phase in the foreland belt and the exhumation of the channel between 75 and 59 Ma (Price and Sears, 2000; this study). With low denudation rates, we postulate that the main driving force was tectonic forcing by the underthrusting basement indentor, as reproduced in numerical models (Jamieson et al., 2006, 2010; Beaumont et al., 2010).

Consequently, a model of synconvergent channel flow and ductile intrusion above an underthrusting basement ramp reconciles all structural, metamorphic, and geochronological data available at the latitude of the Monashee Complex (Fig. 10). This region of the southeastern Canadian Cordillera thus constitutes a compelling natural analogue to the thermomechanical numerical models of channel flow (Beaumont et al., 2001, 2004, 2006; Jamieson et al., 2004, 2006, 2007, 2010).

1. We derived a test to distinguish among the main modes of exhumation in collisional orogens: orogenic wedge, channel flow, and gravitational collapse. The spatial distribution of finite strain patterns, cooling ages, and P-T-t paths as well as metamorphic field gradients and the timing of motion along reverse and normal shear zones can be used as a set of diagnostic criteria, whereas the shape of pressure-temperature (P-T) paths and absolute peak P-T conditions are not distinctive (Fig. 2).

2. Based on a new compilation of field-based data (Table 2), we propose a tectonometamorphic subdivision for studies focusing on high-grade rocks of the southeastern Canadian Cordillera. Hence, the crustal architecture consists of three main units stacked over each other (Figs. 1B and 3A) and includes, from upper to lower structural levels: the upper and lower Selkirk allochthon, and the Monashee Complex.

3. The commonly held view that exhumation of the high-grade core occurred by postconvergent gravitational collapse (Coney and Harms, 1984; Parrish et al., 1988; Vanderhaeghe and Teyssier, 1997; Teyssier and Whitney, 2002; Teyssier et al., 2005; Rey et al., 2009) is rejected because it fails our test. Notably, the prevalence of reverse shear strain rather than the predicted horizontal flattening (Fig. 3A), the exhumation of the lower Selkirk allochthon coeval with burial of the Monashee Complex in the Late Cretaceous to Paleocene (Fig. 9), the presence of 52–49 Ma contractional structures in the lower part of the Monashee Complex, and the absence of a penetrative foliation in the core of complex (Fig. 3A; Gervais et al., 2010) are strong arguments against this exhumation mode (Fig. 2C).

4. The absence of a steep flattening fabric (Fig. 3A), the sharp metamorphic field gradient between the lower Selkirk allochthon and the upper Selkirk allochthon across the Okanagan Valley fault system, and the coeval motion of major normal and reverse shear zones are characteristics difficult to reconcile with the orogenic wedge mode of exhumation (Fig. 2A).

5. In contrast, all the key requirements of the synconvergent model of channel flow are fulfilled (Fig. 2B). Notably: (1) the lower Selkirk allochthon is an 8–10-km-thick panel consisting of >30%–50% leucocratic material (leucosome + sheared leucogranite) bounded by reverse-sense and normal-sense shear zones at its base and roof, respectively, both of which were active together from at least the Paleocene (62–57 Ma) and probably since ca. 100–90 Ma (Table 7); (2) rocks of the upper Selkirk allochthon that surround the lower Selkirk allochthon had been exhumed to upper-crustal levels by the Late Cretaceous (Fig. 3B); (3) the lower Selkirk allochthon records a lateral gradient of cooling ages that become younger from the front to the rear of the proposed channel (Fig. 3B); and (4) footwall rocks of the proposed channel (i.e., the Monashee Complex) were being buried during synkinematic exhumation of the lower Selkirk allochthon (Fig. 9).

6. The database presented in Table 2 is internally consistent with a model of exhumation by channel flow above an underthrusting basement (Fig. 10).

Zircon and titanite grains were isolated using standard mineral separation techniques. Grains were mounted in epoxy and polished to reveal their centers. Cathodoluminescence (CL) imaging was conducted at Carleton University with an Electron Optics Services system interfaced to the Camebax microprobe, consisting of a high-sensitivity photo-multiplier (PM) tube connected to a Baush and Lomb CL amplifier. Images were collected at 15 kV accelerating potential and 15 nA beam current. Backscattered electron (BSE) images of the polished titanite were acquired with an electron microprobe (EMP) at the Massachusetts Institute of Technology with a JEOL 733 Superprobe, an accelerating voltage of 15 KeV, and beam current of 30 nA.

Isotopic data were acquired by laser ablation–multiple collector–inductively coupled plasma–mass spectrometry (LA-MC-ICP-MS) at the radiogenic isotope facility (RIF) of the University of Alberta, Edmonton. Instrumentation includes a Nu Plasma LA-MC-ICP-MS, coupled to a Nd:YAG UP213 laser ablation system. The LA-MC-ICP-MS instrument is equipped with a modified collector block containing 12 Faraday collectors and three ion counters (Simonetti et al., 2005). Raw data were normalized against the zircon standard FC-1 (ca. 1098 Ma) and titanite standard Khan (ca. 521 Ma; Simonetti et al., 2006). Additional analytical protocol and instrumentation are described in Simonetti et al. (2006). Errors are given at 2σ. Plots were made with Isoplot 3 (Ludwig, 2008). Using a 40 × 40 μm laser spot typically provided space for one or two analyses per grain. For a few analyses of small areas, labeled with asterisks in Tables 3 and 4, a 20 μm laser spot was used.

Because few studies have used LA-MC-ICP-MS to date titanite, it is important to evaluate the accuracy and precision obtained on the standard BLR-1, dated by Aleinikoff et al. (2007) at 1047.1 ± 0.4 Ma, which was routinely analyzed along with the unknowns. Three methods exist to correct for the large amount of common Pb typically present in titanite (Aleinikoff et al., 2004; Storey et al., 2006). The first regresses the uncorrected data on a Tera-Wasserburg semi–total Pb plot. If the data form an isochron, the intercept on the y-axis is assumed to yield the initial common Pb ratio, whereas the lower intercept on the concordia curve yields the age of the titanite. The second method calculates the weighted mean average of individual ²⁰⁶Pb/²³⁸U dates corrected with the Age7corr algorithm in Isoplot 3 (Ludwig, 2008), using the y-intercept on the Tera-Wasserburg plot and its uncertainty for the common Pb correction. If the analyses do not yield an isochron, the common Pb can be estimated from Stacey and Kramers (1975). The second and third methods have the disadvantage of assuming concordancy of the analyses, but they yield better precision. After discarding one of the 11 analyses of BLR-1 standard, which was clearly off the regression line on the Tera-Wasserburg plot, we found an age of 1041 ± 110 Ma (MSWD = 0.79; prob. = 0.62) using the first method and a weighted mean average of 1043 ± 35 Ma (MSWD = 0.09; prob. = 1.00) using the second method, both of which are in agreement with the true age.

This work is part of a Ph.D. thesis conducted by FG under the supervision of RLB. The project was financed by a Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery grant held by RLB. A. Simonetti of the Radiogenic Isotopic Facility of University of Alberta is thanked for his assistance during the acquisition of U-Pb LA-MC-ICP-MS data. We are grateful to S. Bowring from the Massachusetts Institute of Technology for providing access to his mineral separation laboratory, and particularly to J. Crowley, whose help in the field and in the geochronological labs was invaluable. D. Vachon is thanked for excellent field assistance. Numerous drafts of this manuscript were carefully edited and considerably improved by S. Carr and J. Percival. Critical constructive reviews by one anonymous reviewer and editorial handling by R. Russo significantly improved this manuscript. Two other anonymous reviewers are also thanked.