and the evolution of Fennoscandia Subduction of continental crust, the ...

Journal of the Geological Society Subduction of continental crust, the origin of post-orogenic granitoids (and anorthosites?) and the evolution of Fennoscandia Hannes K. Brueckner Journal of the Geological Society 2009; v. 166; p. 753-762 doi:10.1144/0016-76492008-028

Email alerting service

click here to receive free email alerts when new articles cite this article

Permission request

click here to seek permission to re-use all or part of this article

Subscribe

click here to subscribe to Journal of the Geological Society or the Lyell Collection

Notes

Downloaded by

on 23 June 2009

© 2009 Geological Society of London

Journal of the Geological Society, London, Vol. 166, 2009, pp. 753–762. doi: 10.1144/0016-76492008-028.

Subduction of continental crust, the origin of post-orogenic granitoids (and anorthosites?) and the evolution of Fennoscandia H A N N E S K . B RU E C K N E R 1,2 * 1 Queens College and the Graduate Center of the City University of New York, Flushing, NY 11367, USA 2 Lamont–Doherty Earth Observatory of Columbia University, 212 Seismology, PO Box 1000, 61 Route 9W, Palisades, NY 10964, USA *Corresponding author (e-mail: [email protected]) Abstract: Slices of silica-rich continental crust subducted into the mantle during collision may undergo metamorphism and exhumation towards the surface as coherent high-pressure or ultrahigh-pressure (HP or UHP) terranes or, if stalled in the mantle, melting and return towards the surface as magmas, or a combination of these two processes. Some exposed HP or UHP terranes contain anatectic granitoids demonstrating that melting does occur during exhumation. Therefore crust trapped in the mantle should also melt when radioactive heating and/or conductive heating raise temperatures to the appropriate solidus. Terranes with hydrous phases will melt readily through hydrate-breakdown reactions. Terranes lacking hydrous phases may require adiabatic decompression to melt, possibly as heated quartz-rich crust becomes ductile and rises diapirically. The magmas generated will intrude the overlying plate to form late-, post- and possibly anorogenic granitoids, depending on the time required to reach solidus temperatures. Geochemical characteristics will depend on P–T conditions, the chemistry and mineralogy of the subducted terrane (especially the presence of hydrous phases), and the amount of melt interaction with the mantle. The removal of sialic upper crust may strand the denser mafic lower crust, which subsequently could melt to generate anorogenic anorthosites, Fe–K granitoids and related rocks. The evolution of the Fennoscandian Shield documents a change from slab melting in the Mesoproterozoic to combined melting and exhumation in the Neoproterozoic to intact exhumation without significant melting in the Palaeozoic.

A growing body of experimental and field evidence exists for melting of continental crust within the mantle during subduction and exhumation (Vrańa 1989; Patinõ-Douce & McCarthy 1998; Hermann & Green 2001; Parkinson & Kohn 2002; Schmidt et al. 2004; Patinõ-Douce 2005; Wallis et al. 2005; Auzanneau et al. 2006; Zhao et al. 2007). Crustal xenoliths within Miocene ultrapotassic lavas from the Pamir provide direct textural and compositional evidence of dehydration-driven partial melting while under HP–UHP conditions in the mantle (Hacker et al. 2005). Most studies emphasize the role of pressure changes to induce melting (i.e. adiabatic decompression). Here the emphasis is on terranes that stay at depth for considerable periods of geological time. This static configuration will, over time, cause a terrane to heat isobarically to temperatures that are not attained in exhumed terranes, resulting, in suitable circumstances, in melting. Although this model reduces the importance of decompression to induce melting for many situations, it will become clear that decreasing pressure may still be required to explain the melting of some terranes, probably through the buoyant upwelling of sialic diapirs (Whitney et al. 2004). Factors that might stall slices of continental crust in the mantle include the following: (1) eclogitization and/or early melt extraction, which results in the formation of mineral assemblages that are close to the density of the enclosing mantle and thus have reduced buoyancy (Massonne et al. 2007); (2) slices that are so thin that the body forces within them cannot overcome frictional resistance along the lower and upper boundaries; (3) a low initial subduction angle so that subsequent eduction is prevented by too much friction along the upper fault boundary for reverse motion;

This paper investigates the simple idea that quartz-rich continental crust subducted into the mantle can become trapped and, ultimately, melt. The idea that continental crust is subducted into the mantle to depths of at least 60–150 km is becoming accepted by the geological community as a mechanism for explaining the presence of high-pressure (HP) and ultrahigh-pressure (UHP) eclogite terranes in many collisional mountain systems (see reviews by Chopin 2003; Ernst 2005; Brueckner 2006). It also provides a simple model for the presence of mantle-derived ‘orogenic’ peridotite lenses in many HP or UHP terranes, as they could be derived from the mantle wedge that overlies the subducted crust (Brueckner & Medaris 2000). Exposed HP or UHP terranes with well-preserved eclogite-facies assemblages are generally assumed to have returned buoyantly towards the surface as coherent slabs because of their low density relative to the enclosing mantle (Chemenda et al. 1996; Ernst et al. 1997). Slices of continental crust that do not return as coherent slabs but instead stall indefinitely within the mantle will undergo a different history. Initially, they will undergo metamorphic recrystallization as pressure and temperature increase during subduction. However, only temperature will continue to rise when they are trapped in the mantle, presumably to levels at which partial melting occurs. Limited melting may not destroy the integrity of a terrane, allowing it to exhume eventually as a coherent mass, possibly as a diapir or plume (Gerya et al. 2006). Higher degrees of melting, however, would probably preclude the terrane from ever exhuming as a solid. Instead, granitic magmas will rise towards the surface leaving behind a dense residue that may linger indefinitely in the mantle. 753

754

H . K . B RU E C K N E R

(4) strong compression in the overlying mountain system that prevents the slabs from inserting themselves into high levels of the orogen. The ultimate melting of trapped continental crust may explain the origin of late orogenic and post-orogenic igneous complexes in several Proterozoic and Phanerozoic mountain systems, and may even apply to ‘A-type’ or anorogenic granite-rich systems with no obvious relationship to orogeny. If a subducted HP–UHP terrane melts instead of being exhumed as a solid, the mountain system could end up losing the evidence for HP–UHP metamorphism (Patinõ-Douce & McCarthy 1998) and instead would be characterized by voluminous igneous intrusions after subduction had ceased. Thus the apparent absence of HP–UHP terranes in some Proterozoic or Phanerozoic mountain systems does not preclude subduction of continental crust into the mantle during their evolution. The term ‘granitoid’ used throughout this paper should be understood as shorthand for suites that are dominated by granites, but can include a variety of other high-silica rock types. Late orogenic, post-orogenic and anorogenic granitoids (all henceforth referred to as ‘PO granitoids’ unless specificity is required) are mineralogically, chemically and petrogenetically complex. Space limitations preclude an extended discussion of these complexities. Instead, the emphasis is on a general tectonic framework for how PO granitoids might be generated.

Exhumed HP–UHP terranes All exposed HP–UHP terranes show evidence of at least some melting during exhumation, but the surface area of exposed granitoids ranges from relatively inconsequential (,5–10% granitoids; e.g. the Western Gneiss Region, Norway; the Dora Maira Massif, Italy and the North East Greenland Eclogite Province), to extensive (25–50% granitoids; e.g. Liverpool Land, Greenland; the Bohemian Massif, central Europe). Melting is invariably associated with a high-temperature–low-pressure (HT–LP) overprint suggesting that melting occurs during exhumation (Gerdes et al. 2000). The presence of a hydrous vapour phase would induce water-saturated melting at temperatures as low as 650–700 8C (Fig. 1). However, free H2 O-rich fluids are probably absent from subducted continental crust (Yardley & Valley 1997), so dehydration melting (also known as hydratebreakdown melting) is probably the most common melting mechanism when hydrous phases are present (Thompson 1982). Muscovite, biotite and amphibole at low to moderate pressures and phengite and zoisite at high and ultrahigh pressures (Fig. 1) are all capable of generating melts as a result of increased temperature or decreased pressure at constant temperature (Huang & Wyllie 1973; Thompson & Algor 1977; Patinõ Douce & McCarthy 1998; Hermann & Green 2001; Hermann 2002; Schmidt et al. 2004; Patinõ-Douce 2005; Auzanneau et al. 2006). Hence the abundance of hydrous minerals within a subducted terrane becomes a critical variable (Vielzeuf & Holloway 1988; Montel & Vielzeuf 1997). Terranes lacking hydrated minerals, such as granulite-facies crystalline basement complexes, cannot generate magma through dehydration reactions whereas terranes containing mica-rich schists and gneiss or amphibole-bearing metabasic rocks should melt readily, with the degree of melting dependent on the amount of hydrous phases present in the rock. The other important variable is the P–T path taken as a terrane subducts and then educts. Terranes that do so at fast rates (.10 mm a 1 ) will have less time to heat and should have ‘skinny hairpin’ P–T paths (‘S’ in Fig. 1) that are less likely to

Fig. 1. Schematic P–T diagram showing possible evolution of a slice of continental crust that is stalled in the mantle. The diagonally shaded area represents the stability field of most HP–UHP eclogite-facies terranes. In the absence of free H2 O, the phengite-out reaction (continuous curve 8), will result in hydrate-breakdown melting at depths .100 km as a terrane heats by conduction (wide arrow A) to the ambient continental gradient (dotted line; from Pollack & Chapman 1977). Anhdrous terranes can melt at the ‘dry’ solidus (dashed line 2) either by the addition of radioactive or advective heat (wide arrow B) or by rising towards the surface at constant temperature (wide arrow C). Depths ,100 km require the introduction of a very large amount of heat and/or H2 O to melt a stationary terrane. However, upwelling as diapirs at constant temperature (wide arrow C) would result in decompression melting. Dashed lines are the wet (1) and dry (2) solidus for granite (Huang & Wyllie 1973). The quartz–coesite transition is from Bohlen & Boettcher (1982). The bold kinked line delineates the vapour-absent solidus for a tonalite (PatinõDouce 2005, fig. 11). The phases that control the solidus are amphibole + quartz (3 and 5), biotite + plagioclase + quartz (4), zoisite + phengite + quartz (6), zoisite + K-feldspar + quartz (7) and phengite + quartz– coesite + clinopyroxene (8). The open (green) arrows represent ‘fat’ (‘F) and ‘skinny’ (‘S’) hairpin P–T paths.

intersect hydrate-breakdown melting curves as they returns towards the surface (Patinõ-Douce 2005). Presumably these terranes will return as intact slabs without a significant HT–LP overprint and without melting. Terranes that subduct and educt at slow rates (,1 mm a 1 ) are more likely to warm up as heat is conducted in from the enclosing mantle and/or generated by radioactive decay and would undergo a ‘fat hairpin’ P–T evolution (arrow ‘F’, Fig. 1) resulting in high temperatures during exhumation. These terranes will be characterized by recrystallization of the HP–UHP assemblages to HT–LP amphibolite or lower facies assemblages and are likely to partially melt as their P–T path crosses dehydration melting curves (Fig. 1) potentially producing volumetrically significant granitoid igneous complexes. These granitoids will be largely ‘syntectonic’ if they do not leave the slab during exhumation; that is, they will develop fabrics (foliations, lineations, boudins) during and after crystallization associated with the deformation that accompanies exhumation. However, magmas may also be sheared out of the educting slab to rise through the upper shear boundary into the overlying plate (Brown & Solar 1998). If they enter an upper plate undergoing little or no deformation they will develop few structural elements and will appear to be PO granitoids even as isochronous granites in the exhuming slab will appear to be synorogenic. If, however, the overlying plate is actively short-

C O N T I N E N TA L C RU S T A N D T H E O R I G I N O F G R A N I T E S

755

ening, the granitic melts will provide the heat to generate the ‘pluton-driven’ metamorphism and deformation that appears to characterize, for example, terranes dominated by gneiss domes and migmatite complexes (Whitney et al. 2004).

Trapped terranes with hydrous assemblages A subducted HP–UHP terrane may become trapped in the mantle at its deepest penetration or at some intermediate level during exhumation. Even terranes that rise to the top of the mantle may stall at or near the Moho (Walsh & Hacker 2004; Schneider et al. 2007). Given sufficient time, these stalled terranes will become heated to the ambient temperature as depressed isotherms slowly but inevitably relax. Melting is a function of temperature if the terrane is truly stalled in place: adiabatic decompression would play no role. Ignoring heat generated by radioactive decay, the attained temperature in a trapped slab will be a function of depth and the regional geothermal gradient that develops after active subduction has ceased. A simplified case assuming a relatively low post-orogenic geothermal gradient (i.e. a ‘shield’ gradient) is shown by arrow ‘A’ in Fig. 1 (adapted from fig. 5.3 of Hacker & Peacock 1995), which shows isobaric heating towards a linear thermal gradient of 10 8C km 1 so that a terrane at c. 100 km depth will achieve a temperature of c. 1000 8C. This temperature is not high enough to reach the ‘dry’ solidus for granite, but is close to or above the dehydration melting curves of zoisite and phengite-bearing assemblages (curves 7 and 8). Subducted crust containing a significant volume of shale, greywacke or other mica-rich rocks will form phengite and zoisite at high pressures (.20 kbar) and therefore will initiate dehydration melting reactions deep in the mantle. Partial melting becomes more likely at higher pressures as the geotherm temperature exceeds the stability fields of phengite and zoisite by increasing amounts. In short, deeply trapped terranes with hydrous minerals could generate granitoid melts by conductive heating alone.

Trapped terranes without hydrous minerals Subducted terranes that lack hydrous phases are harder to melt and therefore more likely to return towards the surface as coherent HP–UHP terranes. Indeed, many now exposed UHP terranes were largely old, crystalline, granitic, basement complexes that lacked significant amounts of hydrous phases at the time they were subducted into the mantle (Ernst 2005; Brueckner 2006). Subsequent retrograde metamorphism during exhumation turned many of these terranes into gneisses with hydrous minerals, making it difficult to assess how ‘wet’ or ‘dry’ these terranes originally were. Regardless, even anhydrous terranes should melt if they stall in the mantle, even at depths where the geotherm does not intersect the dry solidus. The depressed isotherms typical of actively subducting slabs (i.e. Peacock 1990) will migrate ‘up slab’ with time when subduction ends, the mantle becomes static, and heat begins to be introduced conductively from above and below (Fig. 2a). Uncertainties in the thickness and initial temperature of the subducted continental crust, the angle and depth of subduction, and the temperature gradients below and above the initially cold crust complicate detailed modelling of the amount of time it would take to heat the slab to melting temperatures through conduction. However, if the ‘cold subduction’ model of Peacock & Wang (1999) is taken as analogous to a recently stalled slab of subducted continental crust, mantle at a temperature of c. 1250 8C will be 10–20 km above the coldest part (c. 250 8C) of

Fig. 2. Diapir model for melting a slice of continental crust (diagonal shading) trapped in the upper mantle. (a) Isotherms (dashed lines, loosely adapted from Peacock 1990) move ‘up-slab’ (thin dark arrows) as heat is introduced from the mantle and generated by radioactive decay within the crust. The resultant ductility causes the slab to develop Rayleigh– Taylor instabilities and rise as diapirs (b; wide dark arrows). The domes ultimately undergo decompression melting in either the mantle (left diapir) or the crust (middle diapir), depending in part on the presence or absence of hydrous minerals. Loosely adapted from Gerya et al. (2006).

the subducted slab at a depth of c. 100 km. A thermal diffusivity of the mantle of 10 6 m2 s 1 will cause the temperature of the slab to reach 1000 8C in 20–60 Ma, depending on distance (equation (4-113) of Turcotte & Schubert 1982). These times are minimum values, as the calculation assumes that the hot end of the thermal gradient maintains a constant temperature, whereas in a static mantle it presumably cools as heat migrates away. In any case, conductive heating alone (arrow ‘A’ in Fig. 1) is not sufficient to elevate temperatures to the dry granite solidus if a relatively low geothermal gradient is established after subduction ends. However, additional heat will be generated, after a suitable ‘incubation period’ (Sylvester 1998), by radioactive decay. The heat generated through the decay of K, U and Th is generally discounted as a thermal source for deep-level melting because the parent isotopes are incompatible elements that are concentrated in the upper crust. The thinness of the crust and the high temperature gradient within it results in most of the generated heat being released into the atmosphere rather than remaining trapped within the crust. Suggested mechanisms for trapping more heat include thickening of the continental crust through thrusting (e.g. England & Thompson 1984; Chamberlain & Sonder 1990; Goffe´ et al. 2003) or by thrusting and subsequent thinning (Gerdes et al. 2000). These mechanisms theoretically can induce melting at the base of the crust, but they are unnecessary if a slice of crust becomes embedded deep within

756


the mantle through subduction. In effect, the crust will be within a thermal pocket or sheath with relatively hot mantle both above and below. This configuration reduces or even eliminates the loss of heat from the crustal slab. Radioactive heat production will cause the initially depressed isotherms within the slab to migrate further up slab (Fig. 2a) as time passes, with the ultimate result that the slab will be hotter than the ambient geotherm of the mantle alone (arrow ‘B’ in Fig. 1). A rise of c. 100–300 8C would be enough to induce melting in even the driest terrane imaginable, depending on depth. The typical granite generates 3.4 3 10 5 J g 1 a 1 by radioactive decay. A granite requires 0.8 J g 1 to raise its temperature by 1 8C, which would occur in c. 23 ka. Thus roughly 10 Ma are required to raise the temperature of 1 g of granite by 500 8C. Melting requires even more heat; c. 340 J g 1 , which means that another 10 Ma or so would be required to melt 1 g of rock. As was the case with conduction, the surprisingly short time (c. 20 Ma) needed to heat and melt a rock should be regarded as a minimum, as subducted terranes are not simple slices of granite. Also, early melting will partition U, Th and K into the melt and remove it from the system, leaving behind a residue that will generate less heat per unit time. These processes do not necessarily eliminate the traditional process of adding heat to the crust by the upwelling of the hot asthenosphere (.1200 8C). This mechanism is usually invoked to melt the base of overthickened continental crust (e.g. Sylvester 1998; Duchesne et al. 1999). However, the subducted slab mechanism works even better, as slab breakoff (Brueckner & Medaris 2000; Parkinson & Kohn 2002) provides a straightforward mechanism for allowing the asthenosphere to well up into the mantle wedge and adjacent to the subducted slab of continental crust. The introduction of heat through this mechanism could be relatively rapid, depending on how much time has passed before breakoff occurs. It seems inevitable that introduction of heat by conduction, radioactive decay, upwelling asthenosphere, or a combination of these processes will raise the temperature of a slice of continental crust trapped deep in the mantle to temperatures close to the dry granite solidus. The calculations presented above suggest that these mechanisms will take tens of millions of years after collision, providing the necessary delay to explain why some PO granitoids intrude long after compressional orogeny has ceased.

The diapir model Melting reactions are endothermic, so a terrane that gradually heats to the dry granite solidus will undergo only limited melting, initially. Adiabatic decompression may still be required to induce larger-scale melting. Relatively rapid decompression can occur through the upward movement of diapirs (Whitney et al. 2004; Wallis et al. 2005). Rocks become ductile at temperatures near their melting solidus (i.e. at high homologous temperatures). Even small degrees of melting will further increase the ductility of the diapir dramatically (Brown & Solar 1998; Whitney et al. 2004). The density contrast between quartz-rich granitic crust and the overlying, static mantle wedge should allow crustal rocks to move upward in a fashion analogous to salt or gneiss domes (Fig. 2b) or perhaps as plumes (Gerya et al. 2006). Ascent in a largely solid state would limit the interaction of the diapirs with the enclosing mantle peridotite so that the bulk chemistry of the ascending mass should not change significantly. Upwelling rates are not necessarily rapid if the granitic rocks have been eclogitized

and hence are only slightly less dense than the peridotite of the mantle (i.e. Massonne et al. 2007). An essentially spherical diapir will lose heat more slowly than a tabular slab. There will be interplay between heat loss to the surrounding mantle as the diapir moves upward, new heat produced by radioactive decay, and the limited melting produced by this heat (e.g. Whitney et al. 2004, figs 4 and 5). The essential question becomes, where will the diapir melt extensively; in the mantle or in the crust?

Post-orogenic granitoid chemistry and other features The major and trace element chemical and isotopic characteristics of PO granitoids vary considerably (see, for example, Ku¨ster & Harms 1998, and other papers in Lithos, 1998, Volume 45). Some recurring features include high silica content (i.e. monzonite, tonalite, granodiorite, granite, high-silica rhyolite), high alkali content, particularly K (i.e. peralkaline, high-K calcalkaline), but low alkaline-earth contents (Ca, Ba, Sr), variable but generally low Al content (metaluminous), high halogen content, and high incompatible element content (i.e. Rb, Cs, Th, REE; Frost et al. 2001; Vander Auwera 2003; Best 2003). Some PO granitoids contain zircons with older cores of metamorphic origin (Paterson et al. 1992; Zhao et al. 2007). Most of these features are consistent with the melting of old sialic continental crust. However Sr, Nd and O isotopic ratios and trace element discriminators vary considerably, commonly displaying both ‘mantle’ and ‘crustal’ signatures (i.e. Best 2003, p. 389; Hess 1989, pp. 214 and 228; Vigneresse 2004, fig. 8; Andersen et al. 2001; Bolle et al. 2003). Major element chemistry also indicates that many granitoid suites, including the alumina-deficient ‘Atype’ granites considered here, are ‘hybrid’ rocks with both mantle and crustal components (Patinõ-Douce 1999). This interaction between crust and mantle to produce hybrid magmas is usually modelled as the upwelling of mafic magmas from the mantle into the base of the crust where the resultant high temperatures result in crustal melting (e.g. Liankun & Kuirong 1991, fig. 5). The model presented here turns this relationship on its head: sialic melts generated from slices, diapirs or plumes of continental crust within the mantle interact with the ultramafic mantle as they rise through it. This mechanism avoids several problems. With some exceptions, the lower continental crust is richer in Fe and Mg and lower in Si, K, and Na than the upper crust, making it more difficult to generate the essentially upper crust characteristics (i.e. high 87 Sr/86 , high large ion lithophile element (LILE) concentrations, etc.) of many granitoid suites by melting the bottom of the lower crust alone. The rise in temperature required to melt the lower crust requires either thickening of the continental crust or the introduction of heat by mantle magmatism and/or upwelling of the asthenosphere. The subduction model eliminates these difficulties, as it introduces the upper crust into the high-temperature realm of the upper mantle, creating an ideal mechanism for melting this crust and obviating the need to generate mantle magmas. Furthermore, these magmas can interact with the mantle to produce hybrid characteristics. Finally, the subducted crust model allows melting to occur at higher pressures than at the base of the crust, and, generally speaking, increasing pressure results in melts that are less ferromagnesian and richer in alkalis (Patinõ-Douce & McCarthy 1998; Hermann & Green 2001). If the subduction model is correct, the degree to which the chemistry and isotopic patterns of granitoid melts are hybridized by interaction with the mantle depends on whether they initially rise through the mantle as melts or solids.


Mantle interactions Granitoid melts should react extensively with the mantle. They will not cause mantle melting and mixing as a result of the much higher melting temperatures of peridotite relative to granite. However, magma bodies will undergo exchange reactions at their margins with the host peridotite to form pyroxenite and orthopyroxene-rich peridotite (Kelemen et al. 1998). Pyroxene crystallization requires the transfer Mg, Cr and Fe(?) from the peridotite into the magma and Si, Al and Ca from the magma into the mantle. Extensive melt–mantle interactions probably produced the Mg-rich hybrid magmas (melagranites, quartz monzonites, Gerdes et al. 2000) that occur within some PO suites. K and LILE such as Rb and Cs are not taken up by pyroxenes or other anhydrous mantle phases, so these incompatible elements will remain in the granite melt, provided that H2 O is absent. If water is dissolved in the magma, K-rich phlogopite or amphibole will crystallize and change fractionation trends significantly by depleting the magma in K and the LILE. Therefore the amount of H2 O dissolved in the magma is a significant variable. Isotopic patterns will be modified if the mantle wedge contains significant clinopyroxene. Exchange reactions between mantle clinopyroxenes and magmas would lower the 87 Sr/86 Sr and raise the 143 Nd/144 Nd ratios of the magma so that the enriched isotopic signature of melted continental crust will become progressively more mantle-like or depleted. A significant geochemical signature of PO granitoids is the relatively low FeO/(FeO + MgO) of post-orogenic granites (‘magnesian’ granitoids; Frost et al. 2001) compared with the significantly higher ratios of ‘A-type’ or anorogenic granites (‘ferroan’ granitoids). Other differences are that anorogenic granitoids form much later than PO granitoids, tend to melt at higher temperatures (.950 8C), have higher concentrations of high-field strength elements (HFSE) such as Zr and Nb, and tend to lack hydrous minerals, relative to post-orogenic granites. These chemical differences suggest that some hydrous phases existed in the source rock for the post-orogenic granites whereas the source rock for anorogenic granites lacked hydrous minerals. The amount of hydrous minerals in the source terrane may also influence the degree of hybridization with the mantle. A ‘somewhat wet’ scenario (some hydrous minerals) will feature relatively low-temperature dehydration melting (relative to the dry solidus) early in the heating history of the trapped slab. For example, phengite dehydrates at a temperature of as low as 850 8C at 100 km depth in the presence of trace amounts of H2 O (Fig. 1; Hermann & Green 2001; Auzanneau et al. 2006). The small melt volumes generated should initiate early ductile behaviour and the early formation of ascending diapirs (Fig. 2). The resultant decompression melting is likely to occur while the diapir is still within the mantle (Fig. 2b), increasing the likelihood that interaction of the melt with the mantle will form relatively early ‘magnesian’ granitoids and other granitoids with ‘mantle’-like signatures, as discussed above. The higher temperatures required to melt a (‘dry’) terrane that is free of hydrous phases requires that extended time must pass before conductive and/or radioactive heating can initiate ductile upwelling. A solid diapir will not react extensively with the surrounding mantle as it rises. Decompression melting will also be delayed and may not occur until the diapir has reached relatively shallow levels (arrow ‘C’ in Fig. 1), which can be close to the Moho, depending on the thickness of the crust (Figs 1 and 2b). Melts generated at shallow levels will not react as extensively with the mantle as those generated at deeper levels, although some interaction presumably must occur before they

757

leave the mantle. Afterwards, they will rise more or less isochemically through the crust to form large-scale post-orogenic intrusion complexes retaining strong ‘crustal’ geochemical signatures. Melting near the Moho will occur within the plagioclase stability field. The geochemical signatures of anorogenic granites are low Al2 O3, CaO, Sr and Eu concentrations (Patinõ-Douce 1999), requiring melting within the plagioclase stability field (Patinõ-Douce 1997). A possible problem with the subducted crust melting model is that extensive interaction of granitoid melts with the mantle could result in the complete crystallization of the melt in the mantle, preventing the magmas from ever rising into the crust. However, a pyroxenite carapace is likely to form between the ascending melt and the enclosing mantle, which would armour the melt from further contact with the mantle. This shield may remain in place rather than disintegrating, because the mantle would be essentially static as a result of the termination of subduction. Dismembered fragments of this carapace might be produced if there is residual flow in the mantle. These fragments could be the source of some of the mafic xenoliths or enclaves that characterize many PO granitoids (Didier & Barbarin 1991); for example, the pyrope-bearing peridotite xenoliths within perpotassic granites in the Bohemian Massif (Vrańa 1989). Mantle enclaves can react extensively with the host melt and will not necessarily be distinguishable from mafic enclaves of crustal origin, as both hydration and dehydration processes accompany disintegration and mixing (Beard et al. 2005). For example, gabbronorite and hornblendite xenoliths from the ‘Big Jim Complex’ in the Cascade Range of Washington State would be difficult to distinguish from lower crustal enclaves even though they formed through the interaction of a large peridotite body with a diorite magma (Kelemen & Ghiorso 1986). Similarly, mafic and intermediate granulite xenoliths within the Hannuoba basalts of North China could be mislabelled as fragments of the lower crust event though they formed through silicic melt– peridotite reactions in the mantle (Liu et al. 2005).

Orogenic cycles within the Fennoscandian Shield The evolution of the Fennoscandian (or Baltic) Shield and the Caledonian system that occurs along the shield’s western margin in Scandinavia provides several possible examples of subducted continental crust that melted during the Proterozoic. The connection between orogeny and the intrusion of ‘anorogenic’ rapakivi granites and rhyolites in this shield is not a new one and has been proposed previously for both the Svecofennian (Windley ˚ ha¨ll et al. 2000) orogenies. New discov1993) and Gothian (A eries related to the Sveconorwegian orogeny have strengthened evidence for this connection. The Svecofennian orogeny occurred between 1.91 and 1.87 Ga, overlapping the intrusion of anatectic granites (the Central Finland Granitoid Complex) at 1.89–1.88 Ga, followed by the intrusion of anatectic potassium granites at 1.84–1.82 Ga, then by the Transscandinavian Igneous Belt at 1.81–1.65 Ga (Nironen 1997). The subsequent emplacement of Rapakivi Granite Suites at 1.65–1.50 Ga may also be related to Svecofennian subduction (Windley 1993); however, recent evidence suggests that rapakivi emplacement maybe related to the enigmatic Gothian orogeny ˚ ha¨ll et al. 2000). The tectonic scenarios for Fennoscandian (A evolution are complex but generally involve the accretion of arcs and microcontinents to the southwestern margin of an Archaean nucleus (present coordinates) along a NW–SE-trending suture (the Lulea´ –Jakkmokk and Ladoga–Bothnian Bay Zones, Nironen 1997, fig. 1) during the early phases of this evolution

758


(Windley 1993; Nironen 1997). Structural, sedimentological and geochemical criteria suggest that the Archaean Shield was subducted to the SW beneath the arcs. The above-mentioned plutonic suites occur to the SW of the suture in southern Finland and central and southern Sweden. The SW edge of the Archaean Shield (i.e. the Karelian Province) underwent high-temperature but low-pressure metamorphism, locally into the low-P granulite facies (Nironen 1997), suggesting that only the shallowest portion of the subducted Archaean slab was exhumed and that the more deeply subducted portions were trapped in the mantle and ultimately melted to generate the numerous PO granitoid suites, most of which have geochemical and mineralogical characteristics of magmas generated from older continental crust and mixed with a mantle component (Nironen 1997; Elliot 2003). The 1.89–1.82 Ga ages of the two earlier episodes of anatectic melting, as well as the older c. 1.81 ages from the Transscandinavian Igneous Belt overlap with or are up to 80 Ma younger than the Svecofennian orogeny, consistent with the tens of millions of years required for subducted crust to heat through a combination of radioactive decay and conductive heating. The 1.65 Ga Transscandinavian Igneous Belt is more than 200 Ma younger than Svecofennian collision, which may be too long to relate them to the collision as proposed by Windley (1993). It is more likely that these younger plutons are related to the ˚ ha¨ll et al. (2000) directly related subsequent Gothian orogeny. A three distinct convergent events during the Gothian orogenic cycle at 1.69–1.65, 1.62–1.58 and 1.56–1.55 Ga with subsequent phases of rapakivi magmatism (locally associated with anorthosites) at 1.65–1.62, 1.58–1.56 and 1.55–1.50 Ga, respectively. Both the collisional belts and the resultant rapakivi belts trend north–south, and both belts migrated from east to west as three terranes docked successively outboard of the southwestern margin of Fennoscandia. A problem with this model is that the rapakivi belts occur significantly further east (300–600 km) than the collisional belts. Possibly the subducted Gothian crust was carried eastward within the asthenosphere after collision. More significantly, the two initial convergent events involved the collision of island arc systems with Fennoscandia, which implies that subduction was outboard, beneath the arcs, instead of inboard, beneath Fennoscandia. Outboard subduction could not have emplaced continental crust beneath Fennoscandia as proposed in the model presented here. The third event, however, involved the development of a continental arc on the Fennoscandian margin followed by collision with a microcontinent (‘proto˚ ha¨ll et al. 2000). Eastward subduction of southwest Norway’; A oceanic lithosphere beneath the western margin of Fennoscandia during the evolution of the arc would have been followed by the subduction of the leading edge of the microcontinent into the mantle, which would be entirely consistent with the model presented here.

The Sveconorwegian orogeny Subsequently, in southern Norway and southwestern Sweden, the Sveconorwegian (Grenville) orogeny occurred at 1.13–0.96 Ga (Bingen et al. 2008c) followed, with some overlap, by the intrusion of post-orogenic granites at 1.0–0.93 Ga and the famous anorthosite–mangerite–charnockite Rogaland Complex (also known as the Egersund Igneous Complex) of southern Norway at c. 0.93 Ga (Andersen et al. 2001; Bingen et al. 2008a–c). The Sveconorwegian orogeny involved the successive accretion of terranes with metamorphic ages that ‘young’ from west to east (So¨derlund et al. 2008; Bingen et al. 2008b,c); namely, the 1.14–1.08 Ga Bamble–Kongsberg terranes, the

1.05–1.025 Ga Idefjorden terrane and the 0.98–0.96 Ga ‘Eastern Segment’ (Fig. 3). A younger metamorphism between 0.97 and 0.90 Ga in the westernmost Telemarkia terrane (the ‘Dalan phase’ of Bingen et al. 2008b) appears to violate this general pattern, but the metamorphism not necessarily date an accretionary event but rather may date metamorphism related to the intrusion of PO granitoids discussed below. The tectonic evolution of the easternmost Eastern Segment (Andersson et al. 2008; the ‘Falkenber phase of Bingen et al. 2008b) fits very well with the proposed crustal subduction– melting model. It is a north–south-trending (Fig. 3) belt of eclogite-bearing high-pressure granulite-facies rocks in southwestern Sweden (Mo¨ller 1998; Andersson et al. 2008). It is separated to the east from relatively unmetamorphosed largely magmatic rocks of the Transcandinavian Igneous Belt (discussed above) by a large-scale tectonic boundary, the Sveconorwegian Frontal Deformation Zone, which extends northward all the way to the west coast of the Western Gneiss Region. It is separated to the west by a major mylonite zone (Andersson et al. 2002, 2008; Mo¨ller et al. 2007) from the Idefjorden terrane, a c. 1.6–1.53 Ga Gothian volcanic arc sequence (Brewer et al. 1998, 2004) that was metamorphosed at high pressures 50 Ma before the Eastern Segment (So¨derlund et al. 2008; the Agder phase of Bingen et al. 2008b). All of the voluminous Sveconorwegian PO granitoids occur west of this mylonite zone, within the Idefjorden and Bamble–Kongsberg terranes and, further to the west, within the Mesoproterozoic Telemarkian terrane (Fig. 3). The eastern and western boundaries of the Eastern Segment terrane can be interpreted (Fig. 3) as the lower thrust and upper low-angle normal detachment, respectively, of an exhumed HP slab (Hegardt et al. 2003). East–west stretching lineations (Andersson et al. 2008) with a top-to-the-west movement along the mylonite zone (Andersson et al. 2002) provide evidence that the mylonite zone was a low-angle normal fault during exhumation. If subduction was in the opposite direction (i.e. slab subduction to the west) a reasonable tectonic model would have the western edge of Fennoscandia subducted beneath arcs (i.e. the Idenfjordan and Bamble–Kongsberg terranes), microcontinents (i.e. ‘Telemarkia’, Bingen et al. 2008c) and perhaps even beneath the eastern edge of a major exotic craton (Amazonia?, Bingen et al. 2008b). The subducted western edge of Fennoscandia at the time was probably crust formed during the earlier Gothian orogeny (i.e. the Transcandinavian Igneous Belt), an inference supported by orthogneisses within the Eastern Segment that give U–Pb zircon crystallization ages of 1.7–1.68 Ga (So¨derlund et al. 2002). Eclogite-facies metamorphism of the ‘Eastern Segment’ occurred between 0.98 and 0.96 Ga (Johansson et al. 2001; Hegardt et al. 2005; Mo¨ller et al. 2007). Here it is proposed (see Fig. 4) that the subducted continental crust separated (boudined?) into two slabs: a shallower one that was exhumed essentially intact to expose the HP eclogites of the eastern segment, and a deeper segment that remained trapped within the mantle (Fig. 4). Evidence for a granulite-facies overprint accompanied by migmatization (Mo¨ller et al. 2007) suggests that exhumation of the HP eastern segment was slow enough to cause heating and partial melting of the slab during exhumation (Hegardt et al. 2005). It seems likely that a stranded more deeply subducted terrane further to the west also melted and intruded the overlying complex of arcs and microcontinents to form the 0.97–0.92 Ga PO granitoids of southern Norway. These intrusions are associated with the Dalan phase of Sveconorwegian tectonism (Bingen et al. 2008b,c), which was characterized by the develop-


759

Fig. 3. A simplified tectonic map of the Sveconorwegian province of southern Scandinavia showing the distribution of 1.13–0.96 Ga igneous intrusions relative to the Eastern Segment, here interpreted as an exhumed HP terrane bounded by faults (modified from Bingen et al. 2008c, fig. 1). Eclogite localities are from Hegardt et al. (2005).

ment of large-scale gneiss domes similar to those described by Whitney et al. (2004), suggesting that the Dalan phase was dominated by thermally driven vertical tectonics. If the end of eclogite-facies metamorphism is taken to be between 0.98 and 0.96 Ga, then the intrusion of the youngest plutons dated at 0.93 and 0.92 Ga (Andersen et al. 2001; Bingen et al. 2008a,b) suggests that 30–60 Ma were required to heat the subducted portion of the Eastern Segment to melting temperatures. Isotope and trace element studies of these plutons show an increasing component of ‘mantle derived material in the granite sources’ (Andersen et al. 2001) from east to west, as would be predicted in a tapered mantle wedge that increases in thickness from east to west (i.e. Fig. 4). Finally, the Caledonian orogenic cycle involved the closure of Iapetus and the collision of the west side of composite Fennoscandia with Laurentia during the c. 400 Ma Scandian orogeny. The HP–UHP terrane known as the Western Gneiss Region was

subducted and exhumed without significant amounts of melting (Brueckner & Van Roermund 2004). The change in behaviour from very high degree of melting of subducted continental crust during the Svecofennian and perhaps Gothian orogenies to exhumation plus melting during the Sveconorwegian orogeny to exhumation without melting during the Caledonian orogeny may signal a change in the behaviour in subducted continental crust as a result of the secular cooling of the mantle through time.

Sveconorwegian anorthosites, mangerites, charnockites and ferro-potassic granitoids The anorthosite–mangerite–charnockite Rogaland Complex in southernmost Norway (Fig. 3) is characteristically composed of anhydrous minerals. East of the Rogaland Complex is a 300 km long north–south belt intruded by ferro-potassic granitoids (also known as hornblende–biotite granitoids). The two associations

760

H . K . B RU E C K N E R Fig. 4. A crustal subduction model for the generation of Sveconorwegian eclogites, post-orogenic granites, the anorthosite– mangerite–charnockite (AMC) Rogaland Complex and ferro-potassic granites (hornblende–biotite granitoids; HBG). The subducted western edge of the Fennoscandian Shield is modelled as a sialic upper crust (diagonal shading) and a mafic lower crust (no pattern). The sialic upper crust detaches along the Conrad Discontinuity and returns towards the surface as both a coherent HP terrane and as post-orogenic granitoids (light-coloured plutons) of southern Norway. The stranded mafic crust forms eclogite at deep levels and less dense mafic granulite and amphibolite at successively shallower levels. The eclogitized crust and underlying lithosphere detach and sink into the mantle (dark arrows), exposing the mafic granulite and amphibolite to upwelling asthenosphere (white arrows) and resulting in the production of anorthosite–mangerite– charnockite (dark-coloured plutons) and hornblende–biotite granitoid (mediumcoloured plutons) magmas.

intruded contemporaneously at 0.93–0.92 Ga (Scha¨rer et al. 1996; Vander Auwera et al. 2003; Bingen et al. 2006, 2008a), roughly 30–60 Ma after the eclogite-facies metamorphism of the Eastern Segment. The apparent lack of relationship of the anorthosite–mangerite–charnockite and hornblende–biotite granitoid complexes with orogeny earned them the sobriquet of ‘anorogenic’ or A-type granitoids (Scha¨rer et al. 1996; Vander Auwera et al. 2003), but here it is argued that the source for these intrusions was put in place during the Sveconorwegian subduction of the Eastern Segment. Both rock suites required very high melting temperatures; 1200–1300 8C at depths of c. 40–50 km for the anorthosites (Duchesne et al. 1999; Longhi et al. 1999) and 1000–1050 8C for the biotite–hornblende granitoids (Vander Auwera et al. 2003). Both suites were generated by melting rocks of broadly basaltic composition, but the anorthosite–mangerite–charnockite suite requires an anhydrous source (gabbronorite or mafic granulite) whereas the hornblende–biotite granitoid suite requires a hydrous parent (amphibolite). Duchesne et al. (1999) proposed that the source rocks for the Rogaland anorthosite–mangerite– charnockite suite were mafic ‘tongues’ that projected into the mantle from the Moho. The subducted crust model presented here provides the perfect mechanism for generating these ‘tongues’ (Fig. 4), assuming the upper part of subducted crust is primarily sialic whereas the lower crust is mafic. The upper quartz-rich crust could be removed either by diapiric upwelling or delamination and exhumation along the boundary between lower and upper crust (i.e. the ‘Conrad’ discontinuity). Both processes apparently operated during the evolution of the Sveconorwegian orogenic belt; that is, melting resulted in the intrusion of the 0.96–0.92 PO granitoids of southern Norway and delamination and eduction resulted in the exhumation of the HP ‘Eastern Segment’ of southwestern Sweden. The denser and more refractory lower crust left behind would have been exposed to the introduction of heat not only from

below, but also from above. The eventual establishment of a late Proterozoic sub-continental geothermal gradient and the heat generated by radioactive decay may not have been sufficient to heat the mafic slab to solidus temperatures (.1000 8C). Upwelling asthenosphere may still have been required to raise temperatures to the melting solidus (Ashwal 1993). The subducted slab model offers a simple mechanism for initiating upwelling by converting the deepest levels of subducted mafic crust to dense eclogite, which would cause it to break off and sink to deeper levels of the mantle, the same fate that occurs to subducted oceanic lithosphere. This ‘slab breakoff’ would create a gap allowing upwelling of the asthenosphere (Fig. 4). The shallower (,40–50 km) mafic crust would recrystallize to mafic garnet granulites, or, at still shallower levels, and provided hydrous fluids were available, amphibolites. The lower density of these plagioclase-bearing assemblages makes it unlikely that they would sink. Instead, they would remain in place to form the ‘tongues’ of Duchesne et al. (1999). The upwelling of the hot asthenosphere through the breakoff gap (Fig. 4) would engulf these mafic slices, causing them to melt and to generate, respectively, the anhydrous magmas of the anorthosite– mangerite–charnockite suite and the hydrous magmas of the hornblende–biotite granitoid suite. This model is consistent with lower temperatures and shallower melting levels that lead to the hornblende–biotite granitoid suite to the east and higher temperatures and deeper melting levels that led to the anorthosite– mangerite–charnockite to the west as would be expected for a slab that dipped westward during Sveconorwegian subduction.

Conclusions Slices of subducted continental crust may melt and return towards the surface as magmas or recrystallize and return towards the surface as relatively coherent HP–UHP terranes or, as elegantly demonstrated by the evolution of the Sveconorwe-


gian orogenic belt, exhibit some combination of these two processes. The evolution of the Fennoscandian Shield suggests that there was a shift from the former towards the latter as the Earth cooled and the nature of metamorphism changed. HP metamorphism is not unknown in the Precambrian, but most UHP eclogite-facies belts formed during the Phanerozoic whereas large volumes of PO granitoids were generated largely during the Proterozoic. Solid exhumation returns continental crust to the outer layer of the solid Earth as more or less intact and chemically unmodified terranes, whereas melting returns to the surface igneous rocks that are more silica, aluminum and alkali rich than the rocks that were originally subducted. Thus crustal subduction and melting may be an important distillation mechanism for creating ever more sialic continental crust. Simultaneously, deep-level crustal melting would leave behind residues of more mafic rocks (Patinõ-Douce & McCarthy 1998; Patinõ-Douce 2005), which would remain in the mantle as a result of their high density. In addition, deeply generated granitoid magmas will enrich the overlying peridotite as they pass through and interact with the mantle wedge. Both mechanisms can generate enriched geochemical reservoirs that would play an important role in subsequent mantle melting processes (Hermann & Green 2001; Schmidt et al. 2004; Patinõ-Douce 2005). Many mountain systems that appear to lack evidence of HP– UHP metamorphism, but are instead characterized by voluminous syntectonic and PO granitoids, may nevertheless have evolved through the subduction of continental crust into the mantle and the formation of eclogite-facies assemblages. Subtle clues for this metamorphism, such as coesite in zircons within granitoid magmas, should be sought to either prove or disprove the subducted crust model. Discussions with B. Hacker, P. Kelemen and J. Longhi, and reviews of earlier versions of the paper by A. E. Patinõ-Douce, M. Brown, B. Bingen, D. Cornell and two anonymous reviewers are gratefully acknowledged. B. Bingen kindly provided the template for Figure 3. This work was supported by NSF grant EAR-0000937 and grants from the Research Foundation of the City University of New York. LamontDoherty Earth Observatory Contribution Number 7266

References ˚ ha¨ll, K.-I., Connelly, J.N. & Brewer, T.S. 2000. Episodic rapakivi magmaA tism due to distal orogenesis?: correlation of 1.69–1.50 Ga orogenic and inboard ‘anorogenic’ events in the Baltic Shield. Geology, 28, 823–826. Andersen, T., Andresen, A. & Sylvester, B. 2001. Nature and distribution of deep crustal reservoirs in the southwestern part of the Baltic Shield: Evidence from Nd, Sr and Pb isotope data on late Sveconorwegian granites. Journal of the Geological Society, London, 158, 253–267. Andersson, J., Mo¨ller, C. & Johansson, L. 2002. Zircon geochronology of migmatite gneisses along the Mylonite Zone (S Sweden): a major Sveconorwegian terrane boundary in the Baltic Shield. Precambrian Research, 114, 121–147. Andersson, J., Bingen, B., Cornell, D., Johansson, L., Mo¨ller, C. & Søderlund, U. 2008. The Sveconorwegian orogen of southern Scandinavia: setting, petrology and geochronology of polymetamorphic high-grade terranes. 33rd International Geological Congress, Oslo, Norway, Field Excursion Guide. Ashwal, L.D. 1993. Anorthosites. Springer, Berlin. Auzanneau, E., Vielzeuf, D. & Schmidt, M.W. 2006. Experimental evidence of decompression melting during exhumation of subducted continental crust. Contributions to Mineralogy and Petrology, 152, 125–148. Beard, J.S., Ragland, P.C. & Crawford, M.L. 2005. Reactive bulk assimilation: a model for crust–mantle mixing in silicic magmas. Geology, 33, 681–684. Best, M.G. 2003. Igneous and Metamorphic Petrology, 2nd edn. Blackwell Science, Oxford. Bingen, B., Stein, J.J., Bogaerts, M., Bolle, O. & Mansfeld, J. 2006. Molybdenite Re–Os dating constrains gravitational collapse of the Sveconorwegian orogen, SW Scandinavia. Lithos, 87, 328–346.

761

Bingen, B., Davis, W.J., Hamilton, M.A., Engvik, A.K., Stein, H.J., Øyvind, S. & Øystein, N. 2008a. Geochronology of high-grade metamorphism in the Sveconorwegian belt, S. Norway: U–Pb, Th–Pb and Re–Os data. Norwegian Journal of Geology, 88, 13–42. Bingen, B., Nordgulen, Ø & Giulio, V. 2008b. A four-phase model for the Sveconorwegian orogeny, SW Scandinavia. Norwegian Journal of Geology, 88, 43–72. Bingen, B., Andersson, J., So¨derlund, U. & Mo¨ller, C. 2008c. The Mesoproterozoic in the Nordic countries. Episodes, 31, 29–34. Bohlen, S.R. & Boettcher, A.L. 1982. The quartz–coesite transformation: A pressure determination and effects of other components. Journal of Geophysical Research, 87, 7073–7078. Bolle, O., Damaiffe, D. & Duchesne, J.-C. 2003. Petrogenesis of jotunitic and acidic members of an AMC suite (Rogaland anorthosite province, SW Norway): a Sr and Nd isotopic assessment. Precambrian Research, 124, 185– 214. ˚ ha¨ll, K.-I. 1998. Contrasting magmatic arcs in the Brewer, T.S., Daly, J.S. & A Palaeoproterozoic of the south-western Baltic Shield. Precambrian Research, 92, 297–315. ˚ ha¨ll, K.-I., Menuge, J.F., Storey, C.D. & Parrish, R.R. 2004. Brewer, T.S., A Mesoproterozoic bimodal volcanism in SW Norway, evidence for recurring pre-Sveconorwegian continental margin tectonism. Precambrian Research, 134, 249–273. Brown, M. & Solar, G.S. 1998. Shear zone systems and melts: Feedback relations and self-organization in orogenic belts. Journal of Structural Geology, 20, 211–227. Brueckner, H.K. 2006. Dunk, dunkless and re-dunk tectonics: a model for metamorphism, lack of metamorphism, and repeated metamorphism of HP/ UHP terranes. International Geologic Review, 48, 978–995. Brueckner, H.K. & Medaris, L.G., Jr 2000. A general model for the intrusion and evolution of ‘mantle’ garnet peridotites in high-pressure and ultra-high pressure metamorphic terranes. Journal of Metamorphic Geology, 8, 123– 134. Brueckner, H.K. & Van Roermund, H.L.M. 2004. Dunk tectonics: A multiple subduction/eduction model for the evolution of the Scandinavian Caledonides. Tectonics, 23, 1–20. Chamberlain, C.P. & Sonder, L.J. 1990. Heat-producing elements and the thermal and baric patterns of metamorphic belts. Science, 250, 763–769. Chemenda, A.I., Mattauer, M. & Bokun, A.N. 1996. Continental subduction and a mechanism for exhumation of high-pressure metamorphic rocks: New modeling and field data from Oman. Earth and Planetary Science Letters, 143, 173–182. Chopin, C. 2003. Ultrahigh-pressure metamorphism: tracing continental crust into the mantle. Earth and Planetary Science Letters, 212, 1–14. Didier, J. & Barbarin, B. 1991. Enclaves and Granite Petrology. Elsevier, Amsterdam. Duchesne, L.C., Lie´geois, J.P., Vander Auwera, J. & Longhi, J. 1999. The crustal tongue melting model and the origin of massive anorthosites. Terra Nova, 11, 100–105. Elliot, B.A. 2003. Petrogenesis of the post-kinematic magmatism of the Central Finland Granitoid Complex II; sources and magmatic evolution. Journal of Petrology, 44, 1681–1701. England, P. & Thompson, A.B. 1984. Pressure–temperature–time paths of regional metamorphism. Part I: Heat transfer during the evolution of regions of thickened continental crust. Journal of Petrology, 25, 864–928. Ernst, W.G. 2005. Alpine and Pacific styles of Phanerozoic mountain building: subduction-zone petrogenesis of continental crust. Terra Nova, 17, 165–188, doi:10.1111/j.1365-3121.2005.00604.x. Ernst, W.G., Maruyama, S. & Wallis, S. 1997. Buoyancy-driven, rapid exhumation of ultrahigh-pressure metamorphosed continental crust. Proceedings of the National Academy of Sciences, USA, 94, 9532–9537. Frost, B.R., Barnes, W.J.C., Collins, W.J., Arculus, R.J., Ellis, D.J. & Frost, C.D. 2001. A geochemical classification for granitic rocks. Journal of Petrology, 42, 2033–2048. Gerdes, A., Wo¨rner, G. & Henk, I. 2000. Post-collisional granite generation and HT–LP metamorphism by radiogenic heating: the Variscan South Bohemian Batholith. Journal of the Geological Society, London, 157, 577–587. Gerya, T., Connolly, J.A.D., Yuen, D.A., Gorczyk, W. & Capel, A.M. 2006. Seismic implications of mantle wedge plumes. Physics of the Earth and Planetary Interiors, 156, 59–74. Goffe´, B., Bousquet, R., Henry, P. & le Pichon, X. 2003. Effect of the chemical composition of the crust on the metamorphic evolution of orogenic wedges. Journal of Metamorphic Geology, 21, 123–141. Hacker, B.R. & Peacock, S.M. 1995. Creation, preservation, and exhumation of UHPM rocks. In: Coleman, R.G. & Wang, X. (eds) Ultrahigh Pressure Metamorphism. Cambridge University Press, Cambridge, 159–181. Hacker, B., Luffi, P., Lutkov, V., et al. 2005. Near-ultrahigh pressure processing of continental crust: Miocene crustal xenoliths from the Pamir.

762


Journal of Petrology, 46, 1661–1687. Hegardt, E.A., Claesson, L., Cornell, D. & Simakov, S.K. 2003. Eclogites in the central part of the Sveconorwegian parautochthonous eastern segment of the Baltic Shield: support for a subduction–extrusion model. Geophysical Research Abstracts, 5, 02482. Hegardt, E.A., Cornell, D., Claesson, L., Simakov, S., Stein, H. & Hannah, J. 2005. Eclogites in the central part of the Sveconorwegian Eastern Segment of the Baltic Shield: Support for an extensive eclogite terrane. GFF, 127, 221–232. Hermann, J. 2002. Experimental constraints on phase relations in subducted continental crust. Contributions to Mineralogy and Petrology, 143, 219–235. Hermann, J. & Green, D.H. 2001. Experimental constraints on high-pressure melting in subducted crust. Earth and Planetary Science Letters, 188, 149– 168. Hess, P.C. 1989. Origins of Igneous Rocks. Harvard University Press, Cambridge, MA. Huang, W.L. & Wyllie, P.J. 1973. Melting relations of muscovite-granite to 35 kbar as a model for fusion of metamorphosed subducted oceanic sediments. Contributions to Mineralogy and Petrology, 42, 1–14. Johansson, L., Mo¨ller, C. & So¨derlund, U. 2001. Geochronology of eclogite facies metamorphism in the Sveconorwegian Province of SW Sweden. Precambrian Research, 106, 261–275. Kelemen, P.B. & Ghiorso, M.S. 1986. Assimilation of peridotite in zoned calcalkaline plutonic complexes: evidence from the Big Jim complex, Washington Cascades. Contributions to Mineralogy and Petrology, 94, 12–28. Kelemen, P.B., Hart, S.R. & Bernstein, S. 1998. Silica enrichment in the continental upper mantle via melt/rock reaction. Earth and Planetary Science Letters, 164, 387–406. Ku¨ster, D. & Harms, U. 1998. Post-collisional potassic granitoids from the southern and northwestern parts of the Late Neoproterozoic East African Orogen: a review. Lithos, 45, 177–195. Liankun, S. & Kuirong, Y. 1991. A two-stage crust–mantle interaction model for mafic microgranular enclaves in the Daning granodiorite pluton, Guangxi, China. In: Didier, J. & Barbarin, B. (eds) Enclaves and Granite Petrology. Elsevier, Amsterdam, 63–82. Liu, Y.S., Gao, S., Lee, C.-T.A., Hu, S.G., Liu, X.M. & Yan, H.G. 2005. Melt– peridotite interactions: Links between garnet pyroxenite and high-Mg# signature of continental crust. Earth and Planetary Science Letters, 234, 39–57. Longhi, J., Vander Auwera, J., Fram, M. & Duchesne, J.C. 1999. Some phase equilibrium constraints on the origin of Proterozoic (massif) anorthosites and related rocks. Journal of Petrology, 40, 339–362. Massonne, H.-J., Willner, A.P. & Gerya, T. 2007. Densities of metapelitic rocks at high to ultrahigh pressure conditions: What are the geodynamic consequences? Earth and Planetary Science Letters, 256, 12–27. Mo¨ller, C. 1998. Decompressed eclogites in the Sveconorwegian (–Grenvillian) orogen of SW Sweden: petrology and tectonic implications. Journal of Metamorphic Geology, 16, 641–656. Mo¨ller, C., Andersson, J., Lundqvist, I. & Hellstro¨m, F. 2007. Linking deformation, migmatite formation and zircon U–Pb geochronology in polymetamorphic orthogneisses, Sveconorwegian Province, Sweden. Journal of Metamorphic Geology, 25, 727–750. Montel, J.-M. & Vielzeuf, D. 1997. Partial melting of metagreywackes, Part II. Compositions of minerals and melts. Contributions to Mineralogy and Petrology, 128, 176–196. Nironen, M. 1997. The Svecofennian Orogen: a tectonic model. Precambrian Research, 86, 21–44. Parkinson, C.D. & Kohn, M.J. 2002. Continental subduction to depths of 200 km: implications for intra-continental ultrapotassic magmatism. In: Parkinson, C.D., Katayama, I., Liou, J.G. & Maruyama, S. (eds) The DiamondBearing Kokchetav Massif, Kazakhstan. Frontiers in Science Series, 38, 463– 476. Paterson, B.A., Stephens, W.E., Rogers, G., Williams, I.S., Hinton, R.W. & Herd, D.A. 1992. The nature of zircon inheritance in two granite plutons. Transactions of the Royal Society of Edinburgh: Earth Sciences, 83, 459– 471. Patinõ-Douce, A.E. 1997. Generation of metaluminous A-type granites by lowpressure melting of calc-alkaline granitoids. Geology, 25, 743–746. Patinõ-Douce, A.E. 1999. What do experiments tell us about the relative contributions of crust and mantle to the origin of granitic magmas? In: Castro, A., Fernandez, C. & Vigneresse, J.L. (eds) Understanding Granites. Integrating New and Classical Techniques. Geological Society, London, Special Publications, 158, 55–75.

Patinõ-Douce, A.E. 2005. Vapor-absent melting of tonalite at 15–32 kbar. Journal of Petrology, 46, 275–290. Patinõ-Douce, A.E. & McCarthy, T.C. 1998. Melting of crustal rocks during continental collision and subduction. In: Hacker, B.R. & Liou, J.G. (eds) When Continents Collide. Kluwer, Dordrecht, 27–55. Peacock, S.M. 1990. Fluid processes in subduction zones. Science, 248, 329–337. Peacock, S.M. & Wang, K. 1999. Seismic consequences of warm versus cool subduction zone metamorphism: Examples from northeast and southwest Japan. Science, 286, 937–939. Pollack, H.H. & Chapman, D.S. 1977. On the regional variation of heat flow, geotherms and the lithospheric thickness. Tectonophysics, 38, 279–296. Scha¨rer, U., Wilmart, E. & Duchesne, J.C. 1996. The short duration and anorogenic character of anorthosite magmatism: U–Pb dating of the Rogaland complex, Norway. Earth and Planetary Science Letters, 139, 335–350. Schmidt, M.W., Vielzeuf, D. & Auzanneau, E. 2004. Melting and dissolution of subducting crust at high pressures: the key role of white mica. Earth and Planetary Science Letters, 228, 65–84. Schneider, D.A., Zahniser, S.J., Glascock, J.M., Gordon, S.M. & Manecki, M. 2007. Thermochronology of the West Sudetes (Bohemian Massif): Rapid and repeated eduction in the eastern Variscides, Poland and Czech Republic. American Journal of Science, 306, 846–873. So¨derlund, U., Mo¨ller, C., Andersson, J., Johansson, L. & Whitehouse, M. 2002. Zircon geochronology in polymetamorphic gneisses in the Sveconorwegian orogen, SW Sweden: ion microprobe evidence for 1.46–1.42 and 0.98– 0.96 Ga reworking. Precambrian Research, 113, 193–225. So¨derlund, U., Hallestro¨m, F.A. & Kamo, S.L. 2008. Geochronology of highpressure mafic granulite dykes in SW Sweden: tracing the P–T–t path of metamorphism using Hf isotopes in zircon and baddeleyite. Journal of Metamorphic Geology, 26, 539–560. Sylvester, P.J. 1998. Post-collisional strongly peraluminous granites. Lithos, 45, 29–44. Thompson, A.B. 1982. Dehydration melting of pelitic rocks and the generation of H2 O-undersaturated granitic liquids. American Journal of Science, 282, 1567–1595. Thompson, A.B. & Algor, J.R. 1977. Model systems for anatexis of pelitic rocks, 1. Theory of melting reactions in the system KAlO2 –NaAlO2 –Al2 O3 –SiO2 – H2 O. Contributions to Mineralogy and Petrology, 63, 247–269. Turcotte, D.L. & Schubert, G. 1982. Geodynamics: Applications of Continuum Physics to Geological Problems, 1st edn. Wiley, New York. Vander Auwera, J. 2003. Preface. Origin and evolution of Proterozoic anorogenic magmas. Precambrian Research, 124, 105–106. Vander Auwera, J., Bogaerts, M., Lie´geois, J.-P., Demaiffe, D., Wilmart, E., Bolle, O & Duchesne, J.C. 2003. Derivation of the 1.0–0.9 Ga ferropotassic A-type granitoids of southern Norway by extreme differentiation from basic magmas. Precambrian Research, 124, 107–148. Vielzeuf, D. & Holloway, J.M. 1988. Experimental determination of the fluidabsent melting relations in the pelitic system. Consequence for crustal differentiation. Contributions to Mineralogy and Petrology, 98, 257–276. Vigneresse, J.L. 2004. A new paradigm for granite generation. Transactions of the Royal Society of Edinburgh: Earth Sciences, 95, 11–22. Vrańa, S. 1989. Perpotassic granulites from southern Bohemia—a new rock-type derived from partial melting of crustal rocks under upper mantle conditions. Contributions to Mineralogy and Petrology, 103, 510–522. Wallis, S.R., Tsuboi, M., Suzuki, K., Fanning, M., Jiang, L., Tanaka, T. & Tsuyoshi, T. 2005. Role of partial melting in the evolution of the Sulu (eastern China) ultrahigh-pressure terrane. Geology, 33, 129–132. Walsh, E.O. & Hacker, B.R. 2004. The fate of subducted continental margins: Two-stage exhumation of the high-pressure to ultrahigh-pressure Western Gneiss Region, Norway. Journal of Metamorphic Geology, 22, 671–687. Whitney, D.L. Teyssier, C. & Fayon, A.K. 2004. Isothermal decompression, partial melting, and the exhumation of deep continental crust. In: Grocott, J., Tikoff, B., McCaffrey, K.J.W. & Taylor, G. (eds) Vertical Coupling and Decoupling in the Lithosphere. Geological Society, London, Special Publications, 227, 313–326. Windley, B.F. 1993. Proterozoic anorogenic magmatism and its orogenic connections. Journal of the Geological Society, London, 150, 39–50. Yardley, B.W.D. & Valley, J.W. 1997. The petrologic case for a dry lower crust. Journal of Geophysical Research, 102, 12173–12185. Zhao, Z.-F., Zheng, Y.-F., Wei, C.-S. & Wu, Y.-B. 2007. Post-collisional granitoids from the Dabie orogen in China: Zircon U–Pb age, element and O isotope evidence for recycling of subducted continental crust. Lithos, 93, 248–272.

Received 5 March 2008; revised typescript accepted 3 February 2009. Scientific editing by Simon Cuthbert and David Peate.