Exhumation of deeply buried material at plate convergence settings brings up fragments of oceanic crust and continental margin, and even larger segments of continental crust that equilibrated at high or ultrahigh pressure (mantle depths). Subduction channels are capable of cycling sizable crustal blocks from downward to upward trajectories, depending on the viscosity and buoyancy of material at the subduction interface and the geometric evolution of the channel. In collision zones, the exhumation of deeply buried continental crust also relies on the nature of the coupling between the subducted and overlying plates (if there is partial melting of subducted crust, hydration of mantle wedge, etc.), as well as the evolution of boundary conditions (e.g., switch to extension or transtension, break-off of mantle slab). Exhumation in mature collisional orogens is dominated by erosion of the thrust wedge at the front of the overthickened continental plateau. For the case of the Himalaya, debate exists on the nature and path of material that has fed the orogenic front and constructed the Greater Himalayan Sequence, particularly in the Miocene. The work by Grujic et al. (2011) adds significantly to this debate and documents mid-Miocene exhumation of rocks that were at the base of the south Tibetan crust in mid-Miocene time. Within only a few million years, these young eclogites were heated to granulite facies and rapidly exhumed along with their partially molten crustal host. This discovery offers new insight on the relative role of thrusting versus crustal flow in the construction of orogenic wedges; links mantle heat input, crustal melting, and crustal flow with the dynamics of the orogenic front; and provides renewed knowledge of lithospheric evolution in collisional orogens.


While it is well established that oceanic or continental margin crust can be subducted, one of the most exciting discoveries of the last decades is that large volumes of continental crust can also reach great depth and, within the same orogenic cycle, come back to the shallow Earth. Recognizing exhumed deep crust is important in geodynamics because it informs the magnitudes and rates involved in the transfer of mass and heat during orogeny. The paper by Grujic et al. (2011) in the October issue of Lithosphere is a well-documented example of how deep crustal conditions are evaluated in metamorphic rocks and how the rate of transfer from deep to shallow crust is measured using a variety of geochemical and isotopic methods (Rubatto, 2002; Kylander-Clark and Hacker, 2011; Wooden et al., 2006) that define P-T-t (pressure-temperature-time) paths more precisely than ever before. Grujic et al. (2011) document that deep-crustal eclogites found in Bhutan are not associated with initial subduction of India, but rather were exhumed very recently at plate-tectonic velocities as a result of crustal flow, which helps us better understand the dynamics and rheology of collision zones.

By the time rocks that equilibrated at high-pressure (HP) or ultrahigh-pressure (UHP) reach the surface, the record of such conditions is partially or totally erased. Fortunately, small bodies with appropriate compositions (typically isolated pods or boudinaged and dislocated layers) preserve eclogite-facies assemblages, and in some cases the minerals themselves, especially garnet and zircon, contain micro-inclusions of coesite or diamond, which are the signature of UHP metamorphism (Chopin, 1984; Smith, 1984; Sobolev and Shatsky, 1990; van Roermund et al., 2002). In most cases, the metamorphic pressures calculated from thermodynamics reflect lithostatic pressure, the effect of tectonic stresses being of second order (Li et al., 2010, and references therein).

Here, I summarize the paper by Grujic et al., in particular as it relates to exhumation mechanisms of HP (eclogite-facies) rocks, starting with oceanic subduction complexes, where the concept of geologic extraction of deep rocks was first established. I also discuss the subduction of continental margins, from which many exhumed eclogite terrains are derived, and the subduction and exhumation of continental crust, with a brief excursion to the Scandinavian Caledonides as an example. I will end with a discussion of Tibetan-type mature orogens, for which Grujic et al. have added valuable new contribution to the ongoing debate of possible exhumation mechanisms of HP rocks.


The work by Grujic et al. (2011) exemplifies a useful multipronged approach to the problem of exhumation of deep-seated rocks. (Various aspects of this recent work were also published by Warren et al., 2011a, b.) The authors document the formation and exhumation of a metamorphic terrain in the eastern Himalaya (NW Bhutan). This terrain comprises the northern and uppermost portion of the Greater Himalayan Sequence that is carried over the Main Central Thrust (MCT). The metamorphic rocks in NW Bhutan are involved in a south-verging, inclined to nearly recumbent antiform or dome that is bounded to the south by a thrust zone (Laya thrust) and to the north by the extensional South Tibetan detachment system.

According to the petrology of boudinaged mafic layers, rocks in this unit underwent eclogite-facies metamorphism at 50–60 km depth, followed by granulite-facies metamorphism, before being rapidly extruded over the Greater Himalayan Sequence and beneath the overlying detachment system. The metamorphic history of mafic layers indicates a rapid decompression event around 15 Ma that lasted only 1–2 m.y. and brought the rocks upward by 20–40 km. Upward motion was so rapid that heat was advected with the rising dome (near-isothermal decompression), resulting in a very high transient geotherm around the dome. Lower temperature thermochronology (< 600 °C) suggests that slowing exhumation led to rapid cooling by heat loss to the cooler surrounding rocks until ca. 10 Ma, when the dome rocks cooled below the argon closure temperature in mica. By then, rocks were likely at relatively shallow levels (∼15 km depth). For students of tectonics and geodynamics, this example is a striking illustration that cooling rates and exhumation rates cannot be used interchangeably!

In contrast to other regions where mafic rocks containing an eclogite-facies imprint are part of an older protolith, the Bhutan mafic rocks are tentatively interpreted to be Miocene mafic intrusives (now boudinaged dikes and sills). This interpretation is supported, among other arguments, by the lack of inherited zircon cores in the mafic rocks compared to the host rocks. A basaltic volcanic event of Miocene age (ca. 18–13 Ma) is documented on the Tibetan plateau, ∼250 km to the north of the Bhutan rocks. In a tantalizing interpretation Grujic et al. (2011) consider that the mafic granulites in Bhutan may represent the melt conduits, which were “frozen” at depth, of the Miocene Tibetan basalt. If so, this gives a glimpse of the magnitude of displacement of these rocks on their trajectories from the base of the thick Tibetan crust right underneath the basalt volcanic fields to their exhumation site beneath the South Tibetan detachment today.

Grujic et al. (2011) also proposed that the Ama Drime metamorphic terrain, located 200 km to the west of NW Bhutan, likely followed a similar history, and certainly the metamorphic and age data there are comparable to the NW Bhutan data (Leloup et al., 2010; Kali et al., 2010). Ama Drime was partially exhumed in the mid-Miocene, with renewed E-W extension in the last few million years (Jessup et al., 2008; Jessup and Cottle, 2010), forming a N-S oriented rift in southern Tibet and providing a window through the upper part of the Greater Himalayan Sequence. Mafic rocks show a very similar HP evolution (burial to ∼50 km, Groppo et al., 2007), and the timing of high-temperature overprint and subsequent cooling is similar to NW Bhutan (ca. 13–9 Ma; Kali et al., 2010), indicating that the metamorphic history uncovered in both of these areas may be projected over several hundred kilometers along strike and likely reflects the general exhumation history of the eastern Himalayan orogen.

This scenario contrasts with the exhumation of UHP rocks in Tso Morari in western Himalaya, Ladakh, where continental subduction is dated at 55 Ma, with return of continental material to ∼30 km depth by 47 Ma (de Sigoyer et al., 2000; Guillot et al., 2008). These contrasting examples, with Ladakh on the one hand and Ama Drime and NW Bhutan on the other, illustrate the selective preservation of geodynamic imprints in orogens, where pieces of the puzzle are fragmented spatially and temporally. These examples also highlight the diversity of processes that participate in the exhumation of HP and UHP rocks, from the early stages of continental subduction to more evolved stages of continental collision.


The study by Grujic et al. (2011), and many others of this type, address general questions regarding the exhumation of deep-seated crust that go beyond Tibetan/Himalayan tectonics. In the following, I address the issue of the return of HP and UHP rocks in several geodynamic settings. This is not meant to be an exhaustive review of exhumation mechanisms; the diversity and complexity of HP terrains tells us that there are probably many causes that control or trigger their return to shallow Earth. Nevertheless, a few examples may help illustrate some of the parameters that are critical in exhuming HP and UHP rocks.

Rapid Return of Subducted Oceanic Crust and Continental Margins

A good place to start is the question of exhumation of blueschist- and eclogite-facies rocks in subduction complexes, because this is historically where the return of HP rocks at convergent plate boundaries was first studied systematically (see review by Platt, 1993). It is also clear from this commonly subaqueous setting that surface erosion is not an option to drive exhumation. Therefore, the concept of a subduction channel (Cloos and Shreve, 1988) was developed in order to explain the rapid ascent of blueschist and eclogite relative to rocks above and below them. Filling of the subduction channel with oceanic and continental margin material generates some form of weakening at the roof of the system, against the overlying mantle wedge, which allows the rocks of the down-going plate at the base of the subduction channel to be transferred to the roof of the channel and begin an upward trajectory. Mechanical weakening at the roof of the channel may occur owing to the presence of wet sediments at shallow levels (Cloos and Shreve, 1988). Deeper in the subduction channel, the formation of serpentinite by hydration of the overlying mantle provides a low-viscosity surface that facilitates exhumation of the underlying subducted rocks (Blake et al., 1995). As was proposed for the exhumation of Tso Morari in NW Himalaya, serpentinite would also form a weak and buoyant unit (Guillot et al., 2000) that could take the underlying HP rocks for a “balloon ride” (Fig. 1A). The association between HP rocks and serpentinite along the Tethyan and other sutures suggests that the mechanical and buoyancy consequences of mantle wedge hydration is a necessary condition for exhumation of eclogite-facies terrains.

The kinematics and dynamics of the subduction channel have been studied through numerical modeling that takes into account boundary conditions and material rheology (Gerya et al., 2002), in particular the presence of a weak layer (serpentinite) at the subduction interface. To a first order, downgoing particles experience rapid return within the channel, while subduction is ongoing, and the idea of a subduction channel appears physically sound. In comparison, field studies of the structure and kinematics of subduction channels lag behind, in part because the rocks are typically very highly strained and broken up by the time they reach the surface (hence the notion of mélange). However, with progress in kinematic and microfabric analysis, especially by electron backscatter diffraction (EBSD), a renaissance in the study of HP terrains gives a chance to retrieve structural, kinematic, and rheologic information from channel rocks. Eclogite and blueschist terrains typically consist of a suite of rocks that have recorded variable conditions of metamorphism from peak pressure to lower pressure conditions during ascent, and have therefore locked in strain, fabric patterns, and anisotropies that represent different points in the exhuming channel (Teyssier et al., 2010; Abalos et al., 2011).

Exhumation of Subducted Continent

Following consumption (and partial “regurgitation,” Ernst, 2001) of a continental margin, subduction zones may see the arrival of continental lithosphere, carrying an old craton or shield. This phenomenon is active today at various plate margins, where seismology confirms the presence of continental crust at mantle depths (Australia/Indonesia: Elburg et al., 2002; Elburg and Kamenetsky, 2008; Central Asia: Searle et al., 2001; Lister et al., 2008; Aegean: van Hinsbergen et al., 2005). Prior to arrival of the continent, the subduction was typically Andean- or Cordilleran-type; once the oceanic lithosphere is totally consumed between continental masses, continental collision occurs. This is a case exemplified by the Scandinavian Caledonides, where a pile of east-directed nappes (Caledonian allochthons of continental and oceanic affinities) and a steep metamorphic pressure gradient, particularly in the Western Gneiss Region of Norway, imply west-directed oceanic subduction followed by subduction of Baltica crust beneath Laurentia (Griffin et al., 1985; Andersen et al., 1991; Fossen, 2010).

The Western Gneiss Region represents one of the largest continuous UHP terrains in the world (60,000 km2), and mafic and ultramafic bodies indicate pressures up to 3–5 GPa (Smith, 1984; Carswell et al., 1999; Cuthbert et al. 2000; van Roermund et al. 2001; Scambelluri et al. 2008). The eclogite bodies are hosted in migmatitic gneisses that record polyphase metamorphism, significant melting, and intense shearing (e.g., Tucker et al. 1990; Andersen et al., 1991; Hacker et al., 2003, 2010; Terry and Robinson 2003, 2004; Brueckner and van Roermund 2004; Engvik et al. 2007; Fossen, 2010); recent studies summarize the timing of HP and UHP metamorphism and exhumation (Carswell et al. 2006; Kylander-Clark et al., 2008; Spengler et al. 2009). These studies show that UHP rocks are located on culminations in the corrugated Western Gneiss Region, that the gneisses are deformed to very high strain, and that the dominant kinematics are top-to-the-west, not to the east, signifying unroofing of the UHP domains. Once again, this strain localization points to weakening and intense shearing of the subduction interface.

Devonian basins in the upper plate that overlies the Western Gneiss units were deposited shortly after the last recorded UHP rocks were formed (ca. 400 Ma), and while most UHP rocks were being exhumed and partial melt in felsic gneisses was crystallizing at depth (Fossen, 2010; Labrousse et al., 2004). Large extensional displacement was accommodated at the basal contact of the basins and also in a thick west-directed shear zone. The underlying Western Gneiss units are also extremely laminated, with top-to-the-west sense of shear, in general, which indicates that substantial shearing and thinning took place both above and within the HP and UHP units. The extensional shear zone and underlying units are corrugated by E-W upright folds of ∼10–20 km wavelength and high amplitude; these folds are a characteristic of SW Norway, with Devonian basins occupying some of the synforms, while UHP rocks are found in the core of antiformal culminations.

In a large-scale reconstruction of the Scandinavian, Greenland, and Scottish Caledonides, Fossen (2010), following Dewey and Strachan (2003), proposed that Devonian extension is located in a lozenge-shape region that was characterized by transtension. In this view, the large corrugations in SW Norway may have initiated as transtensional folds, which have been documented in analog experiments (Venkat-Ramani and Tikoff, 2002) and in other oblique tectonic settings (Teyssier and Tikoff, 1998). Because transtension results in focused extension, exhumation may be very effective under this regime. The Western Gneiss Region can therefore be understood as a zone of localized exhumation of UHP rocks owing to the combination of flow of soft material (hot and partially molten crust) and very high strain, possibly in transtension.

Geologic relations show that subduction of cold Baltica produced UHP metamorphism but also significant partial melting of the felsic gneisses around eclogite bodies. The depth and precise timing of partial melting are still poorly known, but the gap between the age of UHP metamorphism and the age of leucosome zircons is closing (Gordon et al., 2011). Partial melting at the depth of UHP rocks, particularly along the subduction interface, is physically possible (Whitney et al., 2009) and petrologically realistic (Labrousse et al., 2011). Melting should weaken both the subducted crust and the subduction interface, which may trigger the switch from subduction to exhumation (Labrousse et al., 2011) and allow the overriding plate to extend more freely (Fig. 1B).

Exhumation of Mature Orogenic Crust

After several tens of millions of years of continental collision under continued plate convergence, an orogen grows wide and thick and defines a high plateau, such as Tibet (Vanderhaeghe et al., 2003). Exhumation concentrates at the edge of the plateau, particularly the Himalayan range, as reflected in the sediments that are transported to the large sedimentary basins on either side of India. Because the height of the plateau depends on both crust and lithosphere thicknesses, the mechanical stability of the plateau can change dramatically if, for example, it loses parts of its lithosphere (review by Molnar et al., 1993). This change can in turn generate high elevation, a steep mountain front, and a monsoon effect, which modifies the erosional budget.

At the eastern and western ends of the Himalaya, rocks are eroded fastest where the two main rivers that drain southern Tibet are capable of transporting the sediment load rapidly; the positive feedback between erosion and exhumation is referred to as a tectonic aneurysm (Zeitler et al., 2001). For the rest of the Himalayan range, significant debate exists as to what drives the exhumation of the crust, and particularly the Greater Himalayan Sequence, the crystalline complex that makes up the front of the Himalaya and is bounded by the MCT at the base and the extensional South Tibetan detachment (STD) at the top.

Although there are many variations in the interpretation of Miocene Himalayan tectonics, two main views emerge (Fig. 1C). One favors a direct implication of flowing crust in Himalayan erosional forcing; in this model, a channel of weak mid- to lower crust flowing southward focused denudation at the Himalayan erosion front in the mid-Miocene (Beaumont et al., 2001; Hodges et al., 2001, and see the Special Publication on channel flow, Law et al., 2006). Channel flow is considered to be a consequence of overthickened, weak, and partially molten crust. The other interpretation proposes that exhumation of the Greater Himalayan crystalline complex is related to thrust-wedge dynamics, with material accreted from the down-going Indian crust leading to a condition of tectonic underplating that “jacks up” the overlying material. Variants of this view include a wedge extrusion model (Grujic et al., 1996) and a tectonic wedging model (Yin, 2006; Webb et al., 2007).

Resolution of the debate relies in part on the timing and kinematics of the MCT and its foreland migration toward the Main Boundary Thrust and on the timing and magnitude of activity on the STD, which ceased before the late Miocene. In this context, the work of Grujic et al. (2011) and colleagues in NW Bhutan, together with the work of Leloup et al. (2010), Kali et al. (2010), and others in the Ama Drime region, is of particular relevance. Considering that partially molten crust containing eclogite was rapidly exhumed in both regions, the debate is shifting to the dynamic interaction that must have taken place between the flow of Tibetan crust, heating by Tibetan subcrustal mantle, thrust-wedge evolution at the Himalayan front, and erosion.

This is particular true for the mid-Miocene, when a number of potentially interconnected phenomena occurred: slowing of plate convergence; emplacement of basalt in southern Tibet; migration of thrusting on the Himalayan front; cessation of slip on the STD; formation of N-S oriented extensional shear zones that may reflect a first period of E-W extension; rapid cooling of South Tibetan domes such as the Kangmar dome; 1.5 GPa rocks that underwent granulite metamorphism and exhumation within a very short time; and rapid sediment accumulation in the Bengal fan. If, as Grujic et al. (2011) propose, the granulitized eclogites exposed in NW Bhutan are indeed derived from 250 km farther north, some form of flow in a weak layer, in which material can be delivered rapidly (>1 cm/yr) to the Himalayan front, must be considered. This means that the cycling of Indian crust from accretion into the thrust wedge to exhumation at the top of the Greater Himalayan Sequence may involve a journey that reaches the base of the crust far into the hinterland under Tibet. For this reason, the current work provides vital clues to the time and space scales of geodynamic processes operating during collision.

I thank Donna Whitney and John Goodge for their help with this research focus piece, as well as the many students and colleagues associated with our Minnesota STAMP group.