Buoyancy-driven, rapid exhumation of ultrahigh-pressure metamorphosed continental crust

Abstract

Preservation of ultrahigh-pressure (UHP) minerals formed at depths of 90–125 km require unusual conditions. Our subduction model involves underflow of a salient (250 ± 150 km wide, 90–125 km long) of continental crust embedded in cold, largely oceanic crust-capped lithosphere; loss of leading portions of the high-density oceanic lithosphere by slab break-off, as increasing volumes of microcontinental material enter the subduction zone; buoyancy-driven return toward midcrustal levels of a thin (2–15 km thick), low-density slice; finally, uplift, backfolding, normal faulting, and exposure of the UHP terrane. Sustained over ≈20 million years, rapid (≈5 mm/year) exhumation of the thin-aspect ratio UHP sialic sheet caught between cooler hanging-wall plate and refrigerating, downgoing lithosphere allows withdrawal of heat along both its upper and lower surfaces. The intracratonal position of most UHP complexes reflects consumption of an intervening ocean basin and introduction of a sialic promontory into the subduction zone. UHP metamorphic terranes consist chiefly of transformed, yet relatively low-density continental crust compared with displaced mantle material—otherwise such complexes could not return to shallow depths. Relatively rare metabasaltic, metagabbroic, and metacherty lithologies retain traces of phases characteristic of UHP conditions because they are massive, virtually impervious to fluids, and nearly anhydrous. In contrast, H₂O-rich quartzofeldspathic, gneissose/schistose, more permeable metasedimentary and metagranitic units have backreacted thoroughly, so coesite and other UHP silicates are exceedingly rare. Because of the initial presence of biogenic carbon, and its especially sluggish transformation rate, UHP paragneisses contain the most abundantly preserved crustal diamonds.

Previous workers (1) have demonstrated that deep subduction of continental crust is required to explain the generation of ultrahigh-pressure (UHP) terranes. An outstanding petrotectonic problem consists of elucidating the manner in which these complexes have returned to shallow levels while preserving intact relics of the UHP phase assemblages. The “two-way street” nature of subduction zones was recognized long ago (2–4). Briefly, salients or peninsulas of old continental crust, thoroughly embedded in chiefly cold, oceanic-crust-capped lithospheric plates, descend rapidly, generating the characteristic UHP mineralogy (5, 6). The UHP slabs will be subducted to depths where the buoyancy forces tending to drive them back upward are exactly balanced by the dynamic forces tending to subduct them still further. In a contrasting type of continental collision, where the sialic crust is weak and poorly coupled to the descending lithosphere, continental indentation would occur instead (7).

For the UHP case discussed here involving well-bonded crust plus mantle, entrance of increasing amounts of sialic material into the subduction zone enhances the braking effect of buoyancy; this in turn results in loss of the high-density lithospheric anchor leading the downgoing plate at intermediate upper mantle depths where the sinking lithosphere is in extension (8). Slab break-off (9, 10) enhances buoyancy further, and causes the sialic prong—or at least a slice thereof—to decouple from the descending but faltering lithospheric plate and move back up the subduction channel. Exhumation is aided in part by (i) progressive shallowing of the ruptured and now buoyant, rebounding continental-crust-capped lithosphere, and perhaps more importantly, (ii) due to reduction of the shear force acting along its base due to its increasingly ductile behavior as the slab gradually warms with depth in the deep upper mantle. Because of continued subduction-induced refrigeration tectonically beneath the rising UHP complex, and observed extensional faulting against the overlying, cooler hanging-wall plate, relatively thin slices of UHP terranes effectively lose heat along both upper and lower surfaces during ascent; thus, such complexes may nearly retrace the subduction-zone pressure–temperature (P–T) trajectory during decompression (11, 12).

Proposed relationships are shown diagrammatically in Fig. 1, and apply equally well to the exhumation of high-pressure (HP) and UHP terranes. An upper normal fault and a lower reverse fault bound the thin-aspect-ratio slab. Such shear senses seem required by structural relations, for example, in the Dora Maira Massif (13–15), and in the Dabie Shan (16). Yet another exhumation scenario involves the antithetic faulting characteristic of some compressional orogens, in which double vergence is produced during end stages of the collision and ascent of sialic crust (17).

Figure 1.

Schematic diagram portraying the deep burial and thermal structure of a subducted microcontinent or continental salient (a), then decompression cooling of a rising slice of UHP quartzofeldspathic rock—not necessarily the complete section of sialic crust—accompanying steady-state subduction (b) [after Ernst and Peacock (12)]. During uplift of a thin UHP terrane (thickness somewhat exaggerated for clarity), cooling of the upper margin of the sheet takes place where it is juxtaposed against the shallower, lower temperature hanging-wall plate; cooling along the lower margin of the sheet takes place where it is juxtaposed against the lower temperature, subducting/refrigerating plate. Stages depicted are as follows: (a) prior to exhumation of the UHP complex; and (b) during exhumation of a thin (2–15 km thick) slice of the UHP complex. It is evident that tectonic exhumation of UHP continental slices requires erosive denudation and/or gravitational collapse as well as the presence of a sialic root still at depth. The resolution of forces acting on the sialic slab in stages a and b are discussed in the text. A, asthenosphere; L, lithosphere.

Where thin UHP slices are exhumed during continued subduction/refrigeration, the ascending complex will more-or-less follow the prograde metamorphic P–T path in reverse; this phenomenon has been documented, for instance, in HP/UHP terranes of the western Alps and the California Coast Ranges (18–20). For thick, more nearly equidimensional masses (>30 km thick?), the ratio of cooling surface to mass is low, and central portions are likely to remain sufficiently hot during decompression for the complete obliteration of all UHP relics, and perhaps even for partial melting to ensue; accordingly, such ascending, hot bodies retain none of the precursor UHP mineral assemblages.

Buoyancy Forces

The densities of unaltered oceanic crust (≈3.0), continental crust (≈2.7), and mantle materials (≈3.2) increase with elevated pressure, reflecting the progressive transformation of framework- and layer-silicates to chain- and ortho-silicates. Typical UHP mineralogic assemblages and computed rock densities appropriate for burial depths of ≈100 km are listed in Table 1 (21, 22). Coesite-bearing granitic gneiss is less dense than garnet (or spinel) lherzolite, whereas metabasaltic eclogite is more dense than the mantle. Accordingly, deeply subducted UHP sialic crustal sections will remain buoyant relative to the displaced mantle and will tend to rise, whereas eclogitized oceanic crust will become negatively buoyant and will tend to sink. Of course, if the conversion of crustal slices to UHP mineral assemblages is incomplete, continental crust should be even more buoyant than indicated in Table 1, whereas the oceanic crust would be less negatively (even positively) buoyant, depending on the extent of transformation to high-density phases.

Table 1.

Densities of rocks and minerals (21, 22) involved in UHP metamorphism

Rock type	Mineral	Mineral density	Mode, %	Rock density
Garnet lherzolite	Enstatite	3.21	30	3.24
	Diopside	3.25	15
	Olivine	3.22	50
	Pyropic garnet	3.67	5
Basaltic eclogite	Omphacite	3.34	55	3.74
	Garnet	4.23	40
	Rutile	4.25	5
Granitic gneiss	Jadeite	3.28	40	3.03
	Coesite	3.01	35
	K-feldspar	2.56	15
	Muscovite	2.85	10

The several forces acting upon a sheet of subducted sialic crust, illustrated schematically in Fig. 1, may be described as follows. (i) Subduction of a low-density sialic slab occurs provided shear forces caused by underflow (F_s) overcome the combined effects of buoyancy (F_b) and frictional resistance along the upper wall of the subduction channel (F_r). In this case, F_s > F_b sinΘ + F_r. (ii) Slab rise—not necessarily the complete section of sialic crust—occurs provided buoyancy is positive, and greater than the combined effects of shearing along its base and resistance to movement along its upper surface. For this situation, F_b sinΘ > F_s + F_r.

Kinematically, the illustrated process is rather similar to the “slab-extrusion” mechanism that Maruyama et al. (23) proposed specifically to account for the Dabie Shan UHP rocks of east-central China, as well as tectonic models advanced for the Himalayas and the Alps (24–27). Explicit in our scenario, however, is body–force propulsion of the UHP metamorphosed sheet of continental crust back up the subduction zone due to its overall buoyancy, in contrast to the mechanism of compressional extrusion (28–31). But are other observed features of HP and UHP complexes explicable as consequences of our formation and exhumation mechanism?

Easternmost Java Trench: A Modern Collision Zone

A modern geologic analogue of the model shown in Fig. 1 is represented by the eastern portion of the Indonesian arc (32–34). The strongly curved portion of the Australian–Eurasian collisional suture zone between Timor and Seram is illustrated in Fig. 2a) (33, 35, 36), with the geographic distribution of uplifted HP blueschists indicated. The driving force for wedge “extrusion” is derived from decoupling and accelerated sinking of the oceanic-crust-capped lithospheric slab diagrammed in Fig. 2b, and the consequent reduction in shear stress along the base of the buoyant, ductile sialic material as it warms within the upper mantle. This lithospheric detachment has been documented seismically by Osada and Abe (35) and by Widiyantoro and van der Hilst (37). Loss of the dense, leading, oceanic portion of the Australian plate may be partly responsible for a shallowing of the angle of northward subduction, thus aiding in the exhumation of continental crust. Most importantly, highly ductile, relatively low-density quartzofeldspathic material is sandwiched between dense, relatively slowly deforming, more nearly rigid mantle peridotite constituting both hanging-wall and foot-wall blocks; hence buoyancy-driven ascent of the sialic material as a ≈10-km-thick wedge may take place along the subduction channel (the plate junction) with walls acting as stress guides (38).

Figure 2.

Modern analogue of continental collision and exhumation of a type A blueschist belt, modified after Osada and Abe (35), Charlton (33), and Maruyama et al. (36). The plate-tectonic setting and partly exposed HP blueschist belt (shaded pattern) are shown in a. The Australian continental crust is being subducted beneath the Timor–Seram segment of the Indonesian suture zone. The interpretive cross-section along line N-S is depicted in b. Geophysically documented slab break-off is shown as localized at the continent–ocean crustal interface (reasonable but not essential). The hypothesized buoyant exhumation of a slice of profoundly subducted continental crust (pattern of crosses) is a consequence of the contrasting densities of surrounding mantle peridotite above and below, and medial, sandwiched quartzofeldspathic sheet (Table 1). Shown are two possible contributions to the exhumation of UHP crustal rocks: (i) rebound of the subducted, ruptured lithosphere due to removal of load through decoupling of the dense, oceanic crust-capped (block pattern) lithospheric anchor; and (ii) buoyancy-driven return flow back up the subduction channel, reflecting increase in ductility and reduction in the relative importance of subduction-related shear forces. RF, reverse fault; NF, normal fault.

Brief Summary of Several Exhumed Sialic UHP Complexes

A small number of sialic crustal complexes, chiefly located within Eurasia, exhibit rare, scattered effects of UHP recrystallization (20, 39). Well-recognized tracts include the Qinling–Dabie–Sulu belt of east-central China, the Kokchetav Massif of northern Kazakhstan, the Maksyutov Complex of the southern Urals, the Dora Maira Massif of the western Alps, and the Western Gneiss Region of Norway. Each complex contains relict phases indicative of production at mantle depths approaching—and in some cases substantially exceeding—100 km (40, 41). Mineralogic evidence includes the preservation of trace amounts of minerals and assemblages such as coesite, diamond, K-rich clinopyroxene, pyropic garnet, and volatile-bearing assemblages including magnesite and diopside, coesite and dolomite, talc and kyanite, and/or phengite, ellenbergerite, lawsonite, zoisite, and Na-amphibole (42). According to numerous, multi-laboratory thermobarometric measurements, attending temperatures were ≈700–900°C at confining pressures approaching 28–40 kbar. Prograde (subduction) and retrograde (exhumation) P–T trajectories for several well-studied UHP and somewhat comparable HP complexes are illustrated in Fig. 3.

Figure 3.

Stability fields for blueschists (shaded pattern), eclogites (stippled pattern), and adjacent metamorphic facies (unpatterned) [after Maruyama et al. (36)]. Light lines provide equilibrium curves for several univariant reactions, mid-weight lines indicate metamorphic–facies boundaries (actually multivariant zones of appreciable P—T width), and heavy lines show generalized P—T trajectories for specific, well-documented HP/UHP terranes. Light dashed line indicates the minimum melting curve for H₂O-saturated granite. P–T trajectories for exhumation of the preserved UHP relics may be uncommon, and according to our tectonic scheme, would be more tightly constrained to retrace the prograde path in reverse (see text for discussion). Facies abbreviations: BS, blueschist; Zeo, zeolite; PP, prehnite-pumpellyite; PrA, prehnite-actinolite; GS, greenschist; AP, actinolite-calcic plagioclase; EA, epidote-amphibolite; AM, amphibolite; HGR, HP granulite; GR, granulite/pyroxene-hornfels; EC, eclogite; PG, pumpellyite-glaucophane subfacies; LG, lawsonite-glaucophane subfacies; EG, epidote-glaucophane subfacies. Phase abbreviations: Jd, jadeite; Qz, quartz; LAb, low albite; HAb, high albite.

As with other relatively HP metamorphic terranes, these remarkably high pressures and moderate temperatures can be generated through the profound subduction of coherent tracts of cold, sialic-crust-capped lithosphere, reflecting the fact that geologic materials are poor thermal conductors (5, 43). Return toward the surface is also understandable, based on the buoyancy of UHP continental crust, once it has decoupled from the downgoing slab (6, 38). Because purely adiabatic decompression would result in the transit of rising subduction complexes through P–T regimes (700–800°C, 2–10 kbar) appropriate to the granulite and pyroxene-hornfels metamorphic facies, complete overprinting of the earlier UHP assemblages would be expected, as is evident from Fig. 3. Although this recrystallization may not run to completion under very dry conditions (44–47), the rare and fragmentary preservation of UHP minerals, chiefly as microscopic inclusions in tough, refractory host minerals (garnet, zircon, clinopyroxene, etc.) suggests that these UHP phases retrogressed rather thoroughly on return toward the Earth’s surface. A more plausible conclusion than that involving an adiabat is that complexes retaining traces of UHP phases must have more-or-less retraced the prograde subduction P–T path in reverse.

Alternatively, because of the apparent short duration of individual UHP events, on the order of a few million years, it is apparent that in at least a few cases, metastable lower pressure precursor mineral assemblages failed to react during prograde metamorphism to produce the stable UHP configuration except in kinetically favorable metamorphic environments (44, 48–52). Because H₂O catalyzes transformations through a solution-redeposition mechanism, massive, coarse-grained, impermeable, anhydrous protoliths would be expected to retain original lower pressure mineral assemblages temporarily even at ultrahigh pressures. UHP assemblages, therefore, are not likely to be formed if H₂O is absent during relatively brief periods of prograde metamorphism, or preserved if H₂O is present during later retrogression. Optimal conditions are doubtless only imperfectly achieved at best, and could be an important reason for the extreme rarity of UHP relics.

Summaries of the geologic setting, mineral parageneses, tectonic evolution, and age constraints for five UHP terranes were presented previously (53). Pertinent data and tentative conclusions are summarized in Table 2 (12, 54). As evident from this compilation, well-studied Eurasian UHP metamorphic complexes consist of relatively thin slabs or sheets 2–15 km thick; these are composed dominantly of sialic crust, and those that retain UHP relics appear to have been exhumed rapidly to midcrustal levels—approximately 5 mm/year—maintained over an interval of roughly 20 million years. Cooling from above along the upper, bounding normal fault, and from below along the lower, bounding reverse fault could account for the preserved UHP mineralogic relics, as diagrammed in Fig. 1. Similar geologic relationships have been summarized recently for HP blueschist belts worldwide by Maruyama et al. (36). Now we turn to an enumeration of the special characteristics of these relatively unusual UHP terranes.

Table 2.

Metamorphic-tectonic summary of ultrahigh-pressure metamorphic complexes, modified after Ernst Peacock (12) and Beane et al. (54)

	Qinling–Dabie–Sulu belt, coesite-eclogite	Kokchetav Massif, unit I	Maksyutov Complex, unit 1	Doar Maira Massif, lower Venasca	Western Gneiss Region, Fjordane complex
Protolith formation age	1.3–2.9 Ga	2.2–2.3 Ga	1.2 Ga	303 Ma	1.6–1.8 Ga
Temperature of UHP metamorphism, °C	750 ± 75	900 ± 75	625 ± 50	725 ± 50	825 ± 75
Depth of UHP metamorphism, km	90–120	125	90	90–120	90–125
Time of UHP metamorphism, Ma	210–220	530–540	375–380	35–40	420–440
Midcrustal annealing, Ma	180–200	515–517	360 ± 5 (?)	15–25	375
Rise time to midcrust, Ma	25 ± 10	20 ± 5	15–25 (?)	20 ± 5	55 ± 5
Exhumation rate, mm/yr	3–4	5–6	3–5	4–5	1–2
Coesite included	Relatively abundant	Rare	Very rare (?)	Relatively abundant	Very rare
Diamond included	Very rare	Relatively abundant	Questionable	Absent	Very rare
Blueschists	Present	Rare (?)	Present	Present	Absent
Areal extent, km	400 × 75	≈100 × 15	120 × 8	225 × 60	350 × 70
Maximum thickness of complex, km	10	5–10	4–6	1–2	10–15 (?)

Exhumation rates were estimated by dividing depth of UHP metamorphism by rise time to near-surface of midcrustal levels of ≈15 km (20, 52). 

Petrotectonic Features of UHP Metamorphosed Continental Crust

Any exhumation scenario proposed to explain the origin, plate-tectonic setting, and P–T evolution of UHP metamorphosed sialic complexes must account for the following general observations regarding such terranes (42, 55).

(i) They are developed within old (= cold) sialic crust and are now confined principally to intracratonal collisional sutures.

(ii) Quartzofeldspathic rocks (ortho- and paragneisses/paraschists, migmatites, and metagranitoids) dominate the section, followed by metapelitic and metacalcareous strata. Mafic and ultramafic rocks are volumetrically minor.

(iii) Massive metamafic rocks and/or siliceous schists typically contain most of the unambiguous relict UHP silicate phases. Metacarbonates carry very rare UHP minerals; schistose metapelites and gneissose quartzofeldspathic lithologies generally lack UHP silicates (exceedingly rare UHP micro-inclusions in garnet and/or zircon have been described), but do carry evidence of neoblastic diamond in a few occurrences.

(iv) Postmetamorphic products of erosion are present but are not necessarily voluminous.

(v) Well-developed coeval calc-alkaline volcanic/plutonic belts are not generally associated with these UHP terranes.

(vi) Insofar as known, the lateral dimensions of UHP belts developed within continental crust are relatively modest (100–400 km in length, 8–75 km in breadth).

Circumpacific convergence zones are characterized by widespread blueschists, tectonic melanges, and dismembered ophiolite complexes, reflecting the underflow of thousands of kilometers of oceanic-crust-capped lithosphere, an environment termed type B subduction by Bally (56); in contrast, type A zones of continental collision typically involve the consumption of intervening ocean basins of more modest size. Table 3 lists aspects of the contrasting natures of type A and type B subduction-zone assemblages. Metamorphic prograde and retrograde P–T paths followed by such terranes during subduction and subsequent exhumation are topologically similar to one another, but type A continental collisional complexes retain relics of UHP metamorphism (P_max = 28–40 kbar) whereas type B circumpacific-style underflow characteristically generates only HP belts (P_max = ≈12–15 kbar).

Table 3.

Generalized lithologic comparison of type A and type B subduction-zone tectonic regimes [after Maryuama et al. (36)]

	Type A (collisional)	Type B (circumpacific)	UHP	MP
Protoliths
Shallow-marine sediments	Platform carbonates	Reefal limestones
Clastic wedges	Multicycle orthoquartzites, peraluminous shales	First-cycle graywackes, olistostromes
Deep-sea sediments	Uncommon	Bedded cherts, Mn nodules
Igneous rocks	Bimodal, basalts + dacites	MORBs, seamounts (OIBs)
Continental basement	Granitic gneiss complexes	Absent
Ore deposits	Kuroko-type (massive sulfides)	Mid-ocean ridge origin
Petrology
Typical maximum pressure, kbar			28–40	≈12–15
Associated calc-alkaline belt			Rare or absent	Huge
Degree of retrogression			Nearly complete	Incomplete
Mantle fragments			Garnet lherzolite	Spinel, plagioclase lherzolite
Coeval paired belts			Absent	Present

MORB, mid-oceanic ridge basalt; OIB, oceanic island basalt. 

Discussion

How do these characteristic petrotectonic features of sialic UHP complexes relate to the dynamic generation/exhumation history previously proposed? We discuss them below in the order just enumerated.

(i) Mechanical analyses suggest that continental crust several kilometers or more thick may well be subducted if it enters a convergent plate junction (17, 57, 58). A favorable geologic environment would involve entrance into the subduction zone of a narrow salient or prong of continental crust as an integral part of an old, thermally relaxed, largely oceanic crust-capped slab (6). Although very rare, the existence of UHP metagranitoids and quartzofeldspathic gneisses (59, 60) demonstrates that such subduction does occur, and that buoyant sialic material may be carried down to depths of at least 90–125 km. UHP metamorphism is a predictable consequence of this process and is independent of lithology. In our model, however, the presence of buoyant sialic rocks is an absolutely essential prerequisite for the subsequent uplift/exhumation of the deeply buried terrane. Large masses of eclogitized oceanic crust would remain negatively buoyant, and hence would continue descent into the deep mantle. Accordingly, resurrected UHP complexes are restricted to zones of continental (and microcontinental) collision, the environment of type A subduction (56).

(ii) Lithospheric slabs descend to depths of at least 650–700 km along inclined subduction channels, so where underflow exceeds a few centimeters per year, UHP metamorphism is inevitable (5). However, oceanic crust transformed to eclogite facies assemblages is denser than mantle peridotite and of course, garnet lherzolite is slightly denser than plagioclase- and spinel-bearing analogues, so the plate will continue sinking after phase transformations occur. Only where a sufficiently large volume of low-density continental material is subducted will body forces (buoyancy) overcome frictional resistance and permit its ascent after disengagement from the downgoing lithosphere. Accordingly, recovered UHP complexes, and circumpacific HP terranes as well, consist dominantly of sialic phase assemblages, possessing an aggregate density less than that of the mantle material they have dynamically displaced during subduction. Metabasaltic/metagabbroic and/or metaperidotitic complexes cannot represent more than a volumetrically minor portion of HP/UHP terranes, or the latter would not be sufficiently buoyant to rise back to crustal levels.

(iii) Foliated rocks of quartzofeldspathic and pelitic bulk-rock compositions characteristically contain appreciable H₂O bound as structural OH⁻ in the constituent mineral assemblages, whereas massive cherts, carbonates, and mafic igneous rocks are more nearly anhydrous. The former are relatively more permeable to aqueous fluids than the latter; thus it is possible that the retention of UHP relics is kinetically favored in anhydrous eclogites, siliceous schists, and some marbles, disfavored in more “juicy” quartzofeldspathic and pelitic compositions. The absence of H₂O and closed-system recrystallization is corroborated by the anomalously low δ¹⁸O values measured in garnet and omphacite from Sulu belt UHP eclogites by Yui et al. (61); it is also suggested by experimental rate studies of the transformation of coesite to quartz which indicate that even H₂O contents of 400–500 ppm in the silica are sufficient to cause backreaction to the low-pressure polymorph on decompression (62).

Carbon is present in important concentrations only in metasedimentary rocks, so it is no surprise that neoblastic diamond and diamond pseudomorphs in crustal rocks are virtually confined to paragneisses and paraschists; in such occurrences it is isotopically light, indicating a biological origin, according to Sobolev and Sobolev (63) and Lennykh et al. (64). Silicates in general tend to transform at faster rates than that which characterizes the diamond to graphite transition, so in some occurrences relict diamond may be preserved even where the UHP silicate assemblages have been completely obliterated.

(iv) To elucidate possible retrograde P–T trajectories allowing preservation of UHP relics, the exhumation of such terranes has been modelled (12) as extensional along the upper bounding surface of a thin sheet whereas subduction–refrigeration continues along the lower bounding surface (Fig. 1). The process, therefore, does not require the wholesale uplift of a lithospheric plate, or even the full thickness of continental crust. For this reason, postmetamorphic erosional debris, while considerable, need not be especially voluminous.

(v) Circumpacific subduction is a long-continued process, so sufficient time is available for the development and maturation of the typical calc-alkaline magmatic plumbing system. In contrast, continental collision in many cases involves relatively short-lived underflow of an old, thermally relaxed oceanic crust-capped lithospheric section; thus an andesitic volcanic arc/granitic plutonic belt may not have had time to develop on the stable, nonsubducted lithospheric plate because of volumetrically limited generation of calc-alkaline melts.

(vi) UHP sialic complexes possess modest along-strike dimensions, on the order of 100–400 km, and recovered depths of underflow of about 90–125 km; accordingly, they have dimensions compatible with a conjectural promontory or salient on the leading edge of the subducting microcontinental fragment or continental margin.

In conclusion, the described characteristic petrotectonic features appear to be compatible with the proposed scenario for the formation, thermal evolution, and ultimate tectonic exhumation and exposure through erosion and/or gravitational collapse of UHP terranes (65–67). However, few mineralogic, radiometric, tectonic, and geologic constraints are available for such UHP complexes; hence future work may well require revision or abandonment of this model. In particular, the rapid ascent rates of ≈5 mm/year averaged over more than 20 million years and seemingly required by UHP geochronologic date (Table 2) substantially exceed currently observed uplift and erosion rates in the world’s highest, most active mountain belt (68, 69), but are in agreement with rates of exhumation of ≈4 mm/year calculated by Genser et al. (70) for the Eastern Alps. Yet higher ascent rates have been proposed for the Dabie–Sulu and Kokchetav complexes by Liou and Zhang (71) and Dobretsov et al. (72).

ABBREVIATIONS

UHP

ultrahigh pressure

HP

high pressure

P–T

pressure–temperature