Age and evolution of late Mesozoic metamorphic core complexes in southern Siberia and northern Mongolia

Numerous Cretaceous metamorphic core complexes (MCCs) extend from Transbaikalia in Russia to northern Mongolia within the Central Asian Orogenic Belt. We investigated the Buteel and Zagan MCCs in detail. Shear sense indicators in mylonitized rocks show footwall-to-the-NW tectonic transport. Single zircon dating of footwall rocks in the Buteel MCC establishes the emplacement of granitoid orthogneiss precursors at 240–211 Ma, a felsic metavolcanic rock at 265.0 ± 1.2 Ma, a syenite at 265.5 ± 1.2 Ma and a metarhyolite of the pre-granitoid basement at 553.6 ± 2.9 Ma. A peralkaline granite intruding orthogneisses of the Zagan MCC has a new U–Pb zircon age of 151.6 ± 0.7 Ma. 40Ar/39Ar ages of 133.5 ± 1.8 Ma of hornblende from amphibolite and 122.6 ± 1.8 Ma of biotite from mylonitized gabbro–dolerite of the Buteel MCC are interpreted as cooling ages representing the time of deformation in the footwall. Geological data suggest that the MCCs in Transbaikalia and northern Mongolia formed as a result of extension in a crust that had previously been thickened by abundant calc-alkaline magmatism in an Andean-type setting on the border of the closing Mongol–Okhotsk ocean, by widespread collisional to post-collisional thrusting, and by extensive alkaline–peralkaline magmatism.

The Central Asian Orogenic Belt, one of the largest accretionary complexes on Earth, is situated between the Siberian craton in the north and the North China (or Sino-Korean) and Tarim cratons in the south ( Fig. 1) (Zonenshain et al. 1990;Kovalenko et al. 2004). Salnikova et al. (2001) and Kovalenko et al. (2004) suggested that the Central Asian Orogenic Belt is made up of old microcontinental 'composite' blocks and sublinear (mobile or fold) belts ranging in age from early and late Palaeozoic to early Mesozoic.
The Mongol-Okhotsk orogen (Fig. 1), located in northern Mongolia and Transbaikalia (Russia), is the youngest orogenic segment in the Central Asian Orogenic Belt, which developed as a result of closure of the Mongol-Okhotsk ocean (Tomurtogoo et al. 2005). The ocean closed with a scissor-like motion that began in the west in the Triassic to late Jurassic (Zonenshain et al. 1990) or at the early-mid-Jurassic boundary (Zorin 1999;Parfenov et al. 2001) and became progressively younger eastwards in the late Jurassic-early Cretaceous (Sengör & Natal'in 1996;Yakubchuk & Edwards 1999;Kravchinsky et al. 2002;Cogné et al. 2005). After closure of the Mongol-Okhotsk ocean late Mesozoic extension resulted in formation of metamorphic core complexes (MCCs) in Transbaikalia and northern Mongolia (Sklyarov et al. 1994Zorin et al. 1997;Zorin 1999).
MCCs commonly form as a result of large-scale regional extension related to crustal thickening and shortening during previous collisional or accretionary events (Coney 1980). However, the specific geodynamic setting responsible for MCC development can vary. The classical Cordilleran MCCs formed as a result of extension in the hinterland of an active continental margin (Davis & Coney 1979;Coney & Harms 1984;Lister & Davis 1989). Some MCCs developed in the Cretaceous along or near the border between Mongolia and China (Davis et al. 2001(Davis et al. , 2002Wang et al. 2004;Liu et al. 2005) as a result of postcollisional Jurassic thrusting after Permo-Triassic closure of the Palaeoasian ocean (Xiao et al. 2003). Other Cretaceous MCCs in central China between the Solonker and Dabie Shan sutures developed as a result of massive delamination of the subcontinental lithospheric mantle under eastern China . The MCCs in Transbaikalia and northern Mongolia developed in an intracontinental setting after closure of the Mongol-Okhotsk ocean. Thus, three separate belts of MCCs formed in eastern Asia following closure of two different oceans and continental collisions. The problem we wish to address is: what caused the widespread extension in the Central Asian Orogenic Belt that was responsible for the occurrence of MCCs in southern Siberia and northern Mongolia? Sklyarov et al. (1994Sklyarov et al. ( , 1997 first described more than 10 MCCs in Transbaikalia and northern Mongolia (Fig. 2). To improve our understanding of the reasons for such widespread extension we undertook detailed structural and geochronological studies of the two most representative and well-exposed complexes. Prior to 1994 the metamorphic rocks of the footwall of the MCCs were generally mapped as Precambrian basement (Yanshin 1983), their Precambrian age being postulated on the basis of their strong deformation and high-grade metamorphism, in contrast to surrounding, much less deformed, unmetamorphosed rocks. Sklyarov et al. (1994) first recognized that metamorphic fabrics in the core complexes formed during Mesozoic extensional events.
The Transbaikalian and northern Mongolian MCCs occur in a broad zone of NE-SW-trending extension that also contains a parallel belt of early Cretaceous extensional basins containing sediments and lavas (Fig. 2). The majority of MCCs are situated on the northern side of the Mongol-Okhotsk suture. Topographically they are situated in mountain ridges separated by Mesozoic basins.
We present structural and geochronological data from the Buteel and Zagan complexes and a new model for the formation of the MCCs of Transbaikalia and northernmost Mongolia. We also report new zircon ages of rocks that form the footwall of the Buteel complex, exhumed from a mid-crustal level during late Mesozoic extension. Implications concerning MCC generation in this intracontinental setting will also be discussed.

Geological setting
The Buteel complex is situated in Transbaikalia of Russia, and in northernmost Mongolia (Fig. 3). The Russian part of the complex is also named the Burgutoy complex ( Fig. 4) (Mazukabzov et al. 2006).
The footwall of the Buteel MCC is mainly composed of orthogneisses and foliated to mylonitized granitoids that contain layers of quartz-sillimanite gneiss, quartzite, amphibolite and rare felsic metavolcanic rocks of rhyolitic composition (Figs 3 and 4). The protolith ages of the quartz-sillimanite gneiss and amphibolite are unknown, and we determined a late Neoproterozoic age for one metarhyolite as presented below. Moreover, we have dated the orthogneisses. The orthogneisses and granitoids are intruded by bodies of slightly foliated quartz syenite and alkaline granite that have a U-Pb zircon age of 178 AE 3 Ma  Mossakovsky et al. 1993;Meng et al. 2003). MCC locations after Sklyarov et al. (1997). The Mongol-Okhotsk orogen is shown after Zorin (1999). The main direction of subduction is shown by arrowheads along the Mongol-Okhotsk suture. The location of Figure 2 is indicated.    Mazukabzov et al. 2006). (Mazukabzov et al. 2006). Still in the footwall, mylonitized Permian to Triassic volcanic and sedimentary rocks (age equivalents to unmetamorphosed Permian-Triassic rocks of the hanging wall) have tectonic contacts with the orthogneisses and are intruded by a gabbro-dolerites and granites (Fig. 4). The mylonitized rocks of the footwall grade structurally upwards towards the detachment fault from amphibolite-facies mylonites to greenschist-facies mylonites.
The hanging wall consists of unmetamorphosed, brittledeformed Permian and Triassic volcanic and sedimentary rocks, subdivided into various lithostratigraphic formations that are intruded by Permian and Triassic granites. These granites are analogous to dated granites in the hanging wall of the Zagan MCC. The age of the Ungurkuy Formation is constrained by late Permian to early Triassic flora (Kotlyar & Popeko 1974): Cladophlebis nervosa Pryn. and Cordaicarpus triangularis Radcz. The mid-to late Triassic age of the Chernoyarovsk Formation is indicated by the presence of age-sensitive flora (Skoblo et al. 2001): Cladophlebis pseudoinchiinensis Radcz. and Diplasites tamirica Vlad et al. Early Cretaceous non-marine conglomerates, sandstones, siltstones, argillites, organic limestones, brown coal and basalts (Figs 3 and 4) infill NE-SWtrending basins. The basalts have K-Ar ages of 130-143 Ma (Ivanov et al. 1995). The age of the sedimentary rocks is constrained by early Cretaceous fossil insects (Skoblo et al. 2001): Mongolianella cf. subexsortis Scob and Limnocypridea grammi Ljub. et al.
The footwall and hanging wall of the Buteel MCC are separated by a 200 m wide detachment fault zone containing blastomylonites and ultramylonites that is exposed on the southern and northern margins of the Buteel complex (Figs 3 and 4). The blastomylonites have a penetrative, prominent, thinly banded parallel fabric. The detachment fault is commonly covered by Cenozoic deposits or is cut by brittle faults.

Structural signatures
All figures illustrating structural signatures of the Buteel and Zagan MCCs are available online at http://www.geolsoc.org.uk/ SUP18291. Dominant structures. Gneissic and mylonitic foliations in the footwall and the mylonitic fabric of the detachment fault in the Mongolian Buteel MCC and in the Russian part of this complex (Burgutoy) are similar. The predominant foliation defines the domal structure of the Buteel complex. Older deformation fabrics in footwall rocks are commonly overprinted by a younger foliation that formed during exhumation.
Stretching lineations (slickenlines) and mineral lineations (amphibole, biotite, sillimanite) are ubiquitous in rocks of the footwall and the detachment fault. Lineations plunge from subhorizontal to 308 towards the SE and NW.
Folds. In the footwall the axial planes of isoclinal and tight folds dip to the SE and NW, and fold hinges plunge NE. Asymmetric flexural and isoclinal folds are common in mylonites of the footwall near the detachment fault. Locally the axial planes of these folds are subparallel to the mylonitic foliation.
Kinematic indicators. Shear sense indicators of mesoscopic and microscopic structures are well developed in foliated and mylonitized rocks of the footwall and in blastomylonites of the detachment fault. The orientation of the axial planes of asymmetric folds indicates northwestward tectonic transport of the underlying rocks.
The two main mesoscopic kinematic indicators within the footwall are as follows.
(1) Mylonitized polymictic conglomerates, which belong to a Permian to Triassic volcanic and sedimentary sequence. The pebbles are oblate and stretched, and their extensions reach 5:1 to 7:1. Clast deformation was accompanied by formation of asymmetric pressure shadows around pebbles, and S-C and C9 fabrics. The orientations of stretched pebbles in the conglomerates and shear sense fabrics indicate a footwall-to-the-NW shear sense.
(2) Synkinematically intruded amphibolite dykes, the best preserved of which contain characteristic features of basic dykes intruded during shear; that is, on the one hand the dykes are clearly discordant to intruded gneiss foliation, but on the other hand they contain an internal sigmoidal shear foliation, and their apophyses have been broken into lenses by the same shear zones into which they were emplaced, a combination of features that can be best interpreted in terms of syntectonic intrusion. The features of these dykes are consistent with them having been intruded into a gneiss foliation that was undergoing extensional shearing.
Most mylonites in the footwall and blastomylonites in the detachment fault contain only microscopic shear sense indicators, including S-C and C9 fabrics and rare ä-type porphyroclasts; these fabrics also consistently indicate footwall-to-the-NW movement. In summary, all kinematic indicators indicate simple shear deformation and footwall-to-the-NW tectonic transport.
Brittle structures. Kink bands varying in size from a few centimetres to 1.5 m are one of the latest structures in mylonite zones near and within the detachment fault. The majority of kink bands occur in the SE of the complex near the hanging wall. The kink bands strike NE-SW, dip to the NW and deform the mylonitic foliation and lineations. They formed in response to reverse faulting and indicate brittle-ductile conditions of deformation and footwall-to-the-NW tectonic transport during the final stages of MCC exhumation. Late, brittle, listric normal faults cut the mylonitic foliation. They dip to the SE and their strike is subparallel to the boundaries of Mesozoic basins.

Geological setting
The Zagan MCC is located about 50 km NE of the Buteel-Burgutoy complex (Fig. 2).
The footwall is mainly composed of granitoid orthogneisses and weakly to strongly foliated and mylonitized granitoids that contain xenoliths of biotite or biotite-hornblende gneiss and amphibolite (Fig. 5). The zircon age of the orthogneisses and granitoids is unknown, but weakly foliated granites and granodiorites have a whole-rock 10-point Rb-Sr isochron age of 289 AE 23 Ma . These rocks are intruded by foliated granites and granosyenites that have U-Pb zircon ages of 153 AE 1 Ma and 160.7 AE 1.2 Ma  and peralkaline granites, one of which in the Mangirtui massif we have dated (Fig. 5). On the northern and southern flanks of the Zagan complex foliated and mylonitized Permo-Triassic volcanic and sedimentary rocks are intruded by granites (Fig. 5). In common with the Buteel MCC, mylonitized rocks of the footwall of the Zagan MCC grade structurally upwards towards the detachment from amphibolite-facies mylonite to greenschistfacies mylonite and ultramylonite. Syntectonic hornblende from amphibolite of the footwall of the Zagan MCC yielded a plateau age of 127 AE 2 Ma . Syntectonic biotite from the same sample has a K/Ar age of 112 Ma .
The hanging wall of the Zagan MCC largely consists of unmetamorphosed, brittle-deformed Permo-Triassic volcanic and sedimentary rocks, subdivided into various lithostratigraphic units, as well as Permian granites and late Triassic alkaline granites (Fig. 5). Volcanic rocks from the Zagan-Khuntey Formation have a whole-rock Rb-Sr isochron age of 208.5 AE 2.6 Ma (Yarmolyuk et al. 2001). Age-sensitive flora (Kozubova & Radtchenko 1961) suggest that the Alentuy Formation has a late Permian age: Grassinervia cf. pentagonata Gorel., Cordaites cf. mitinaensis (Goreb.) et al. We have already discussed the above evidence for a late Permian to early Triassic age for the Ungurkuy Formation and a mid-to late Triassic age for the equivalent Chernoyarovsk Formation. The ages of Permian granites are close to the age range of 279 AE 2 Ma and 283 AE 5 Ma of syenites and granites in the Bryansky Complex located about 100 km to the NE (U-Pb, Litvinovsky et al. 2002). The ages of Triassic granites are comparable with whole-rock Rb-Sr isochron ages of 209.0 AE 2.7 Ma to 213.6 AE 8.6 Ma of alkaline granites and volcanic rocks of the Kharitonovo association situated about 50 km to the north (Yarmolyuk et al. 2001). The above-described volcanic and sedimentary rocks are overlain by unmetamorphosed early Cretaceous basalts and sediments that are lithologically similar to early Cretaceous rocks in the hanging wall of the Buteel MCC.
The detachment fault separating the footwall and hanging wall on the southern and northern margins of the Zagan complex consists of strongly mylonitized rocks (Fig. 5). Sklyarov et al. (1997) recorded pseudotachylites within the blastomylonites within the detachment fault on the southern margin of the Zagan MCC.
Structural signatures Dominant structures. The gneissic and mylonitic foliation of the footwall and the mylonitic foliation of the detachment fault in the Zagan complex are similar to those in the Buteel complex. The low-angle dip (10-308) of the gneissic and mylonitic foliation in the footwall and the major detachment fault delineate the domal structure of the complex. In contrast to the Buteel complex, however, the oldest metamorphic fabrics have not been found in the Zagan complex.
Mineral lineations (amphibole, biotite, quartz) and stretching lineations (slickenlines) are common in rocks of the footwall, and the detachment fault dips SE or NW irrespective of the orientation of the foliation. The lineation has a low-angle plunge from subhorizontal to 308.
Folds. Folds range in size from a few centimetres to metres and less often to 35-40 m. Flexural, reclined and isoclinal large-scale folds and small-scale asymmetric folds occur in rocks of the footwall. Fold hinges plunge variably between the NE (c. 0408) and NW (c. 3258). The fold hinges are well grouped and define two maxima that, we suggest, were caused by the curvature of primary NE-SW-trending asymmetric folds into NW-SE-trending asymmetric folds during nonlinear tectonic transport within shear zones, resulting in the formation of sheath folds.
Kinematic indicators. Mesoscopic shear sense indicators are well developed only in mylonitized rocks of the footwall in the north of the complex. These include S-C fabrics, asymmetric folds, shear bands, and asymmetric quartz or ore mineral pressure shadows around porphyroclasts .
Well-developed kinematic indicators occur in Permian mylonitized stretched conglomerates of the footwall. These include Sshaped pebbles with asymmetric quartz pressure shadows and stretched, cigar-shaped pebbles. S-C structures and shear bands occur in the mylonitized conglomerate matrix and stretched pebbles. All shear sense indicators in mylonitized rocks of the footwall of this complex consistently suggest footwall-to-the-NW simple shear.
Microscopic structures observed in oriented thin sections provide shear sense indicators in mylonitized rocks of the footwall and in blastomylonites of the detachment fault in the southern Zagan complex . The most common shear-sense indicators are asymmetric microfolds, S-C oriented quartz lenses, quartz segregation fabrics, quartz c-axis fabrics, S-C fabrics, K-feldspar ó-type porphyroclasts, asymmetric quartz-feldspar pressure shadows around porphyroclasts, and asymmetric mica fish.
The structures in the mylonitized rocks and blastomylonites of the southern Zagan complex indicate formation in an isotropic stress field in a simple shear environment. All shear sense indicators imply footwall-to-the-NW tectonic transport.
Brittle structures. Small-amplitude (1 cm-5 cm) kink bands define the latest deformation of mylonites in the footwall and in the detachment fault of the southern Zagan complex. The asymmetric form of the kink bands indicates footwall-to-the-NW tectonic transport during the final phases of deformation.
Late, steeply dipping, brittle faults occur in the detachment fault and hanging wall. NW-and SE-dipping brittle faults are widespread near the boundary between the Zagan topographic axial ridge and low-lying sedimentary Tugnui and Khilok basins. These normal and strike-slip faults have small displacements with vertical slip.

Age of footwall rocks (zircon geochronology and Nd isotopic systematics)
We have undertaken zircon dating of single grains recovered from highly deformed orthogneisses and granitoids, felsic metavolcanic rock and an undeformed syenite in the footwall of the Buteel complex. The analytical work on these rocks collected on the Mongolian side was undertaken at the Max-Planck-Institut für Chemie in Mainz (zircon evaporation), Curtin University of Technology in Perth, Australia (sensitive high-resolution ion microprobe (SHRIMP) analyses of sample M02/107), and the Beijing SHRIMP Center (analysis of sample M05/287). The analytical techniques for all methods are included in the Supplementary Publication (see p. 408). All uncertainties quoted in the text are at the 2ó level.
Sample NM 1 (508159360N, 1058329150E, Fig. 3) is a mediumgrained, well-foliated and -lineated, but not sheared, orthogneiss derived from a porphyritic granite. The zircons constitute a homogeneous population of light grey, euhedral and stubby grains free of inclusions. Evaporation of six grains yielded a mean 207 Pb/ 206 Pb age of 211.4 AE 1.2 Ma (Table 1, Fig. 6a) that we consider to reflect the time of protolith emplacement. The E Nd(211 Ma) value for whole-rock sample NM 1 is 2.3, and a mean crustal residence age of 590 Ma can be calculated ( Table 2). The relatively young mean crustal residence age makes it unlikely that much, if any, ancient crustal material was involved in the generation of the granite melt, and such old crust has also not been identified in this part of the Central Asian Orogenic Belt. However, Neoproterozoic crust occurs in the Baikalian part of the Central Asian Orogenic Belt in southern Siberia (Kozakov et al. 2007), and we therefore consider it likely that the source material of sample NM1 is predominantly of this age. Kozakov et al. (2007) also found that most granitoids in the central Central Asian Orogenic Belt, regardless of emplacement age, have Neoproterozoic Nd model ages, and they related this to melting of Riphean rocks.
Sample NM 2 (508129090N, 1058389100E, Fig. 3) is a fresh, well-foliated but not mylonitic granodioritic gneiss interlayered with numerous mafic (gabbroic) dykes. As in the previous sample, the zircons constitute a homogeneous, euhedral population of light grey grains. Evaporation of five grains yielded a mean 207 Pb/ 206 Pb age of 230.7 AE 1.2 Ma (Table 1, Fig. 6b), slightly older than NM 1, which we interpret to date the emplacement of the gneiss protolith. The E Nd(230 Ma) value for whole-rock sample NM 2 is 2.7, and the mean crustal residence age is 540 Ma (Table 2), suggesting an origin for the gneiss protolith similar to that of sample NM 1.
Sample M02/107 (498549039N, 1048499199E, Fig. 3) is a slightly porphyritic, sheared granodiorite in contact with a layer of felsic metavolcanic rock. The layer and contact are parallel as a result of strong shearing and transposition, and there are anastomosing small shear zones in both rocks where not overprinted by mylonitic fabrics. The zircons define a homogeneous population of long-prismatic grains with slight rounding at their terminations. Three grains were evaporated and provided identical isotopic ratios that combine to a mean 207 Pb/ 206 Pb age of 239.6 AE 1.2 Ma (Table 1, Fig. 6c). Six spots on five grains of this sample were also analysed on the SHRIMP II ion-microprobe and yielded similar and well-grouped isotopic ratios that provide a mean 206 Pb/ 238 U age of 240 AE 2.6 Ma (Table 3, Fig. 7a). The two ages are identical and reflect the time of protolith emplacement. The E Nd(240 Ma) value for whole-rock sample M02/107 is 3.9 with a corresponding mean crustal residence age of 510 Ma (Table 2), suggesting an origin similar to that of the previous samples.
Sample M02/109 (498589110N, 1048539110E, Fig. 3) represents a very fresh, coarse-grained, biotite-rich and strongly foliated, partly mylonitic, granite. The zircons are stubby to longprismatic and vary between idiomorphic and slightly rounded at their terminations. Evaporation of six grains produced identical Pb isotope ratios with a mean 207 Pb/ 206 Pb age of 229.2 AE 1.2 Ma (Table 1, Fig. 6d) that again reflects emplacement of the gneiss protolith. The E Nd(229 Ma) value for whole-rock sample M02/109 is 3.4 with a corresponding mean crustal residence age of 500 Ma (Table 2), again suggesting a similar genesis to that of the previous samples.
Sample NM 4 is from a plug of coarse-grained unfoliated syenite collected near the border between Mongolia and Russia (508239020N, 1058219150E, Fig. 3) and possibly representing a low-strain zone in the MCC. The zircons constitute a homogeneous population of stubby to thin, long-prismatic and euhedral grains. Six grains were evaporated and provided identical isotopic ratios resulting in a mean 207 Pb/ 206 Pb age of 265.5 AE 1.2 Ma (Table 1, Fig. 6e). The E Nd(266 Ma) value for whole-rock sample NM 4 is À3.2 with a corresponding mean crustal residence age of 990 Ma (Table 2); this is distinctly different from the previous samples and suggests that the syenite was derived from either remelting of an early Neoproterozoic crustal source or a mixture of more ancient crustal material with a more juvenile component.
Sample M02/111 (508189110N, 1058059430E, Fig. 3) is from a sequence of well-foliated, north-dipping, sheared felsic and porphyritic meta-volcanic rocks, interlayered with more intermediate biotite-rich (dacitic?) meta-lava. No contact with the surrounding sheared granite-gneisses was observed. The zircons define a uniform population of clear to light yellow-brown, thin, long-prismatic and idiomorphic grains. Four zircons were evaporated, and the isotopic ratios yield a unimodal distribution with a mean 207 Pb/ 206 Pb age of 265.0 AE 1.2 Ma (Table 1, Fig. 6f) that reflects the time of volcanic activity generating the volcanic sequence. The E Nd(265 Ma) value for whole-rock sample M02/111 is 0.7 with a corresponding mean crustal residence age of 690 Ma (Table 2), again suggesting a similar genesis to that of the previous samples.
Sample M05/287 (508109000N, 1058349180E, Fig. 3) is from a thinly laminated and mylonitic metarhyolite that occurs in a layer, 0.8-1.0 m thick, between strongly sheared amphibolite in the centre of the Buteel complex. This bimodal metavolcanic sequence is apparently intruded by the nearby granitoid gneisses, but the contact is not well exposed. The zircons are clear, thin and long-prismatic, and SHRIMP analysis of six grains yielded one concordant and five variably discordant results (Table 3, Fig.  7b), which are well aligned along a chord through the origin and suggest recent Pb loss. The mean 207 Pb/ 206 Pb age of these grains is 553.6 AE 2.9 Ma (Fig. 7b), which we consider to reflect the time of bimodal volcanism. This age also identifies the metavolcanic sequence as part of a late Neoproterozoic basement into which the precursors of the granitoid gneisses apparently intruded.
In summary, the emplacement ages of the granitoid gneisses making up most of the footwall rocks in the Buteel MCC are between 240 and 211 Ma (Triassic), and reflect a period of granitoid activity as also found elsewhere in the Central Asian Orogenic Belt (Kovalenko et al. 2004) and as discussed in more detail below. These granitoids were emplaced into a sequence of mafic to felsic volcanic rocks with an age of 265 Ma and represented by our sample M02/111. Locally, a sequence of basement rocks is exposed as represented by our sample M05/ 287 and dated at c. 554 Ma. The original intrusive contact was overprinted by extensive ductile deformation during core complex formation. The Nd isotopic systematics suggests derivation of the granitoid gneiss precursors from late Neoproterozoic mantle-derived sources, probably related to subduction and arc accretion within the evolving Palaeo-Asian ocean (Kröner et al. 2007). Our metarhyolite sample M05/287 may represent this late Neoproterozoic arc material. In contrast, the undeformed 266 Ma, late Permian syenite seems to have a crustal source and may reflect remelting within a Precambrian continental fragment already accreted to the Central Asian Orogenic Belt by Permian times.

Age of granites intrusive into footwall orthogneisses (zircon geochronology)
We have undertaken multigrain zircon dating of a foliated peralkaline granite (sample B611) collected within the Mangirtui massif that intruded orthogneisses of the footwall of the Zagan MCC (Fig. 5). The analytical work was undertaken at the Natural Environment Research Council Isotope Geosciences Laboratory (NIGL) in the UK.
Sample B611 yielded clear euhedral prismatic to subeuhedral stubby zircon crystals. The majority of zircons are inclusion-free and vary in length from 100 to 180 ìm. Analysed multigrain aliquots Z1 and Z2 are concordant with Z3 being reversely discordant (Table 4, Fig. 8). The variation in 207 Pb/ 206 Pb, especially for Z3, could be the result of inappropriate Stacey & Kramers (1975) model ratios for the common Pb correction. The younger age of Z2 (c. 147 Ma) combined with a high U concentration of this aliquot implies Pb loss. The most concordant aliquot (Z1) yields a concordia age of 151.6 AE 0.7 Ma (Table 4, Fig. 8), which we interpret to probably date the emplacement of the peralkaline granite. However, it is also possible that the age of the peralkaline granite could be slightly older (c. 153 Ma) on the basis of aliquot Z3. Therefore, the age of intrusion cannot be derived with certainty from these three aliquots but may lie within the range of 206 Pb/ 238 Pb ages of Z1 and Z3. Given this uncertainty we prefer the weighted mean 206 Pb/ 238 Pb age of 152.9 AE 0.9 Ma of these two aliquots.

Development of the metamorphic core complexes (Ar-Ar geochronology)
To estimate the time of metamorphic core complexes development, 40 Ar-39 Ar isotopic analyses of hornblendes and biotites from foliated and mylonitized rocks of the Buteel MCC were carried out at the United Institute of Geology, Geophysics and Mineralogy (Novosibirsk, Russia). Sample locations are shown in Figure 4. The analytical data are presented in Table 5. Detailed description of samples, chemical compositions of hornblende and thermobarometry are included in the Supplementary Publication (see p. 408).
Sample 1116 was collected from an exposure situated in the centre of the Burgutoy ridge (the footwall of the Buteel complex). It is an amphibolite from an outcrop of biotiteamphibole gneiss and amphibolite, intruded by foliated granites. The mineral lineation (biotite, amphibole) in the amphibolite  plunges subhorizontally to the NW, similar to all mineral lineations within the Buteel complex. A hornblende from this amphibolite yielded a plateau age of 133.5 AE 1.8 Ma for 76% of the gas released and a total age of 133.6 AE 3.8 Ma (Fig. 9a). The temperature of metamorphism estimated for this sample is higher than the maximum closure temperature for hornblende (580 8C at a 40-80 ìm diffusion radius; Harrison 1981). We therefore interpret the hornblende age to reflect a cooling age through the closure temperature for hornblende.
The age spectrum for biotite from sample 1116 is disruptive (Fig. 9b); minimum and maximum ages are 115 Ma and 135 Ma, respectively (except for the lowest and highest temperature steps), and the total age is 122.0 AE 3.2 Ma (Fig. 9b). An initial 40 Ar/ 36 Ar value of 296.7 AE 2.0 does not suggest the presence of excess argon in the biotite. The high 37 Ar/ 39 Ar values (.0.02) in the biotite (Table 5) are consistent with alteration of biotite by chlorite (Baksi 2007); insignificant replacement of biotite by chlorite is indicated by petrographic data. Therefore the disruptive age spectrum of this biotite is probably caused by alteration to chlorite, and accordingly by geological redistribution of 40 Ar and reactor-induced redistribution of 39 Ar. It is possible that the youngest age (115 Ma) is close to the maximum age of hydrothermal activity responsible for the alteration of biotite by chlorite.  Pb. § Corrected for blank Pb and U, and common Pb (Stacey & Kramers (1975) model isotopic composition at 153 Ma).
Sample 1610 is a mylonitized gabbro-dolerite intruding Permo-Triassic mylonitized volcano-sedimentary rocks of the footwall of the Buteel complex (Burgutoy) not far from the detachment fault on the southern flank of the complex (Fig. 4). The sample exhibits a primary magmatic texture and a secondary mylonitic fabric. A mineral lineation (amphibole, biotite) plunges at a low angle to the SE.
The age spectrum for amphibole of the sample has a staircase pattern (Fig. 9c); a minimum age (the lowest temperature step) is 138.5 AE 4.6 Ma and a maximum age (the highest temperature step) is 247.3 AE 20.4 Ma. Petrographic study of sample 1610 shows that magmatic minerals are replaced by syntectonic magnesiohornblende. The dated hornblende is probably an inhomogeneous mineral and a mixture of a primary magmatic mineral (pyroxene and possible post-magmatic amphibole) and syntectonic magnesiohornblende. In this case, we can only interpret the highest temperature step of the 40 Ar/ 39 Ar age spectrum. The oldest age (247.3 AE 20.4 Ma) should be regarded as the maximum age of gabbro-dolerite. Interpretation of the younger age is ambiguous.
A biotite from sample 1610 yielded a plateau age of 122.6 AE 1.8 Ma for c. 65% of the gas released (Fig. 9d). The P-T conditions of metamorphism calculated for this sample suggest that the biotite crystallized above its closure temperature (c. 320 8C, Harrison et al. 1985). Therefore we propose that the plateau age is a cooling age after the closure temperature of the biotite.
Only two of the four dated minerals from the Buteel MCC provide information on the development of the metamorphic core complex. These are the hornblende from sample 1116 and the biotite of sample 1610. The other two dated minerals (the biotite of sample 1116 and the hornblende of sample 1610) show disruptive and staircase patterns of age spectra; the ages obtained from these minerals do not reflect the metamorphic core complex development. We interpret the 40 Ar-39 Ar plateau ages of the hornblende of sample 1116 and the biotite of sample 1610 as cooling ages representing the time of deformation in the footwall.

Discussion
From our Ar-Ar data from foliated and mylonitized rocks of the Buteel MCC and from Ar-Ar and K-Ar data for the Zagan MCC , we consider that extension responsible for development of the MCCs in Transbaikalia and northern Mongolia occurred in the early Cretaceous. We interpret the hornblende 40 Ar-39 Ar age of 133 Ma for the amphibolite and the biotite 40 Ar-39 Ar age of 123 Ma for the mylonitized gabbro-dolerite of the Buteel MCC as cooling ages. The plateau age of 127 AE 2 Ma for hornblende from the amphibolite in the footwall of the Zagan MCC can also be interpreted as a cooling age . Some differences between the hornblende 40 Ar-39 Ar cooling ages for amphibolites of the Buteel MCC (133 Ma) and the Zagan MCC (127 Ma) may reflect the different cooling histories of the two MCCs, possibly related to rates of extension or to the influence of meteoric fluids.
The Buteel and Zagan complexes contain all the elements of classic Cordilleran metamorphic core complexes (Davis & Coney 1979;Coney 1980): ductilely deformed rocks in the footwall (basement), brittle deformed rocks in the hanging wall (cover), and a low-angle detachment fault. Also, the structures in the Buteel and Zagan MCCs are similar to those in classic Cordilleran MCCs; strongly oriented mineral and stretching lineations are ubiquitous in both complexes irrespective of the orientation of the foliation.
MCCs worldwide tend to form in continental crust that has been through a process of thickening that caused or allowed it to undergo extension and gravitational collapse with consequent formation of MCCs and rift basins. Mesozoic-Cenozoic MCCs in continental orogens can be grouped according to their tectonic environment, as follows.
(1) The North American Cordillera. The MCCs formed in the upper plate of an active continental margin that had been thickened by the accretion of suspect terranes and by subductiongenerated magmatic processes (e.g. Davis & Coney 1979;Coney 1980;Coney & Harms 1984;Lister & Davis 1989).
(2) The Himalayas. This is a collisional orogenic belt that has been thickened by post-collisional thrusting. MCCs occur in the thrust-thickened, lower Indian plate south of the Indus-Tsangpo suture that has not been thickened by subduction-generated magmatism (Vannay & Hodges 1996).
(3) MCCs in central China between the Solonker and Dabie Shan sutures. These formed in the Cretaceous during a period of major extension following the delamination of c. 100 km of Archaean subcontinental lithospheric mantle in the Jurassic ).
(4) Northern China north of the Solonker suture and in southernmost Mongolia. Here, there are many MCCs for which there are diverse ideas on their generation; for example, the Yagan-Onch Hayrhan (Webb et al. 1999) is considered to have formed as a result of post-collisional thrusting related to the end-Permian Solonker suture (Xiao et al. 2003;Wang et al. 2004), extension related to westward subduction of the Kula-Pacific oceanic plate (Davis et al. 1998;Wu et al. 2005), or contraction and extension related to formation of the Mongol-Okhotsk suture and orogen (Davis et al. 1998;Meng 2003).
The tectonic setting of the MCCs in Transbaikalia and northern Mongolia was different from any of the above environments. We consider that this region underwent two major stages of crustal thickening prior to formation of the MCCs. First, subduction, mostly to the NNW of the Mongol-Okhotsk ocean, led to formation of magmatically thickened crust in an Andeantype setting. Direction of subduction to the north (where the Buteel-Zagan complexes and other MCCs of Transbaikalia are located) was established by seismic tomography (Van der Voo et al. 1999) and teleseismic receiver function data (Zorin et al. 2002). Second, after formation of the Mongol-Okhotsk suture,  Table 4. Errors for data ellipses are 2ó.  post-collisional thrusting led to crustal thickening and extensive alkaline to peralkaline magmatism.

Active continental margin
During closure of the Mongol-Okhotsk ocean (Zonenshain et al. 1990;Zorin 1999;Kravchinsky et al. 2002;Yarmolyuk et al. 2002;Cogné et al. 2005) subduction gave rise to abundant calcalkaline to peralkaline batholiths, plutons and lavas in active continental margins to the NW and SE (present coordinates) (Zorin 1999;Yarmolyuk et al. 2002;Tomurtogoo et al. 2005). Thus, the suture zone is directly bordered on both sides by Triassic-early Jurassic belts dominated by calc-alkaline basalts, granodiorites and granites. Farther from the suture zone to the NW and SE these belts are succeeded by belts that contain predominantly subalkaline basalts, syenites and alkaline granites (Kovalenko et al. 1995). The increase in alkalinity away from the suture zone is comparable with that across modern subductiongenerated active continental margins (Zorin et al. 1995). Following formation of the suture many calc-alkaline and alkaline magmatic rocks continued to form, particularly farther east in the Cretaceous, as a consequence of the eastward younging of the suture closure time. This huge magmatic development that lasted from the late Palaeozoic to the late Mesozoic gave rise to more than 900 granitic intrusions in a belt 1000 km wide and 3000 km long along the Mongol-Okhotsk orogen (Koval et al. 1999;Kovalenko et al. 2004).
On the northern side of the suture and closest to it is the Khangai-Khentei-Daurian belt (Fig. 10), which contains largely poorly dated calc-alkaline granodiorites and granites, some of which intrude Carboniferous rocks and are overlain by Lower Triassic sediments (Parfenov et al. 2001). Conventional multigrain U-Pb zircon formation ages are 252 AE 3 and 253 AE 2 Ma (Budnikov et al. 1999;Jahn et al. 2004). Takahashi et al. (2001) pointed out that magnetite-series granitic rocks are comparable with those on the Asian continent side of the Japanese arc. Detailed petrochemistry of the Khangai batholith has been summarized by Dergunov (2001).
The most prominent and well-studied intrusion of the Selenga belt is the Erdenet granitic pluton in NE Mongolia, which contains a major porphyry copper-molybdenum deposit (Gerel & Munkhtsengel 2005). The age of the mineralization is provided by a Re-Os date of 241 AE 0.8 Ma on molybdenite (Watanabe & Stein 2000). According to Gerel & Munkhtsengel (2005), rifting of a continental margin was accompanied by emplacement of subaerial Permian trachyandesites and pyroclastic rocks. The shallow-level Erdenet plutons intrude these Permian volcanic rocks and are overlain by Triassic trachyte, trachyandesite and basaltic trachyandesite flows. Gerel &  Munkhtsengel concluded that the Erdenet rocks were emplaced into high-level continental crust in an active continental margin.
On the northern side of the Selenga belt is the .150 000 km 2 Vitim batholith, reputed to be the world's largest batholith (Yarmolyuk et al. 1997). Ninety per cent of the main constituent Barguzin Complex consists of calc-alkaline biotite-hornblende tonalite, granodiorite and granite; their U-Pb zircon ages vary from 292 AE 1 Ma to 286.0 AE 1.1 Ma. The very minor Zaza Complex contains granite, quartz syenite, biotite granite and leucogranite; their U-Pb zircon ages vary from 303.0 AE 7.3 Ma to 286.0 AE 1.1 Ma. Because of the presence of minor younger alkaline granitic rocks Yarmolyuk et al. (1997) concluded that the whole batholith was derived from a mantle plume and crustal anatexis. In contrast, we suggest that this predominantly calcalkaline granitic batholith was generated by subduction processes and emplaced in the northern active continental margin of the orogen.
Calc-alkaline volcano-plutonic belts of late Jurassic to early Cretaceous age defining the northern active continental margin continue along strike of the Selenga belt east of 1248E in the Stanovoi and Uda belts (Parfenov et al. 2001), but these are east of the metamorphic core complexes.
At 498N, 1098E the 279 AE 2 Ma and 283 AE 5 Ma Bryansky Complex comprises peralkaline syenites and granites (Litvinovsky et al. 2002). Geochemical data and the presence of high-temperature melt inclusions suggest that the magma of the Bryansky Complex formed at the base of a 60-70 km thick continental crust (Litvinovsky et al. 2002), which Parfenov et al. (2001) and Tomurtogoo et al. (2005) considered had been magmatically thickened in an Andean-type environment.

Post-collisional thrusting and magmatism
Final closure of the Mongol-Okhotsk ocean and final collision of Siberia and Southern Mongolia-North China, giving rise to the Mongol-Okhotsk suture, occurred in the early or mid-Jurassic in eastern Mongolia and Transbaikalia (Zorin 1999;Tomurtogoo et al. 2005) and in the late Jurassic and Cretaceous farther east (Yakubchuk & Edwards 1999;Parfenov et al. 2001;Kravchinsky et al. 2002). The kinematics during the final stages of ocean closure and during collision were probably responsible for the formation of post-collisional thrusts on the side of the Mongol-Okhotsk suture.
The Devonian-Carboniferous Onon island arc representing a fault slice series is situated within the Mongol-Okhotsk suture zone (Figs 1, 2 and 10). During the collisional and postcollisional deformation the arc rocks were thrust southwards over late Permian-early Jurassic passive margin sediments and their basement belonging to the Southern Mongolia-North China continent; horizontal displacements on the thrusts reached 200 km (Zorin 1999). Zorin et al. (1995) estimated that the resultant nappe that includes the entire displaced island arc is 8-12 km thick. East and SE boundaries of the Onon island arc (Figs 1, 2 and 10) were considered by Zorin (1999) as the Onon branch of the north-to NW-dipping Mongol-Okhotsk suture.
North-vergent thrusts are prominent to the north of the suture where the MCCs are located. The late Jurassic north-vergent Angara thrust near Irkutsk has carried Precambrian rocks northwards for several hundred metres over mid-Jurassic coal-bearing sediments. Although this thrust is situated 600 km north of the Mongol-Okhotsk suture, Zorin (1999) considered that it was caused by the collision tectonics of the Mongol-Okhotsk suture.
In the Andean-type active continental margin, where the Buteel-Zagan complexes and other MCCs are located, mid-late Jurassic north-vergent thrusts were commonly inverted in the early Cretaceous to listric, extensional normal faults (Zorin 1999). Many such faults controlled the development of Cretaceous clastic basins that are common throughout the belt of MCCs (Zorin et al. 1997) (Fig. 2).
Coeval with thrusting in mid-to late Jurassic time, A-type granites and granosyenites, and bimodal basalt-rhyolite rocks formed in Transbaikalia (Gordienko & Klimuk 1995;Gordienko et al. 1997;Sklyarov et al. 1997;Mazukabzov et al. 2006;this work). The quartz syenites and alkaline granites with an age of 178 Ma occur in the footwall of the Buteel MCC. The granites and granosyenites (154-161 Ma) and peralkaline granites (152 Ma) are distributed in the footwall of the Zagan MCC. The chemical characteristics of 178 Ma syenites and alkaline granites in the footwall of the Buteel complex (e.g. Y/Nb ratio varies from 0.7 to 1.5) indicate derivation from a mantle source; these are A 1 -type granites according to Eby (1992). Gordienko & Klimuk (1995) and Gordienko et al. (1997) also suggested mantle-derived sources for 158 Ma within-plate basalts and rhyolites. Bimodal basalt-rhyolite lavas in Mesozoic basins overlie Permian-Triassic volcano-sedimentary rocks on the northern side of the hanging wall of the Buteel and Zagan complexes. Gordienko & Klimuk (1995) suggested that these magmatic rocks were emplaced within intracontinental rifts. Gans et al. (1989) noted that in the Basin and Range province of the North American Cordillera the beginning of extension was connected with the introduction of mantle-derived basalts and rhyolites that resemble calc-alkaline andesite to rhyolite series in the Andes, and that trace element and isotopic data of the Cordilleran lavas suggest extensive contamination of mantlederived basalt by crustal melts in the deep continental crust. Partly following the model of Gans et al. (1989), we interpret the relations in Transbaikalia and northern Mongolia as follows: after extensive thrusting and magmatism gravitational collapse of thickened crust led to large-scale regional extension and thinning of the continental crust, followed by bimodal magmatism caused by decompression melting of still-hot asthenosphere. The extension was responsible for the formation of the MCCs and assisted in the emplacement of magmatic rocks in the rifts.
Based on the hornblende cooling age from amphibolite of the Buteel MCC, we assume that the main stages of extension and of ductile deformation within this complex began earlier than 134 Ma. It is possible that ductile deformation of the Zagan MCC started at the same time, because the difference in the hornblende cooling ages from amphibolites of the Buteel and Zagan MCCs can be connected with the different cooling histories of these complexes (see above). Coeval with formation of the MCCs, 130-143 Ma fissure basalts (K-Ar, Ivanov et al. 1995) were intruded within the hanging wall of the Buteel and the Zagan MCCs (Figs 3-5). It is probable that the early Cretaceous non-marine sedimentary rocks in the hanging wall of the MCCs were deposited contemporaneously with deformation within the footwall of the MCCs.
The final structures that developed within the mylonite zones and detachment faults of the footwall of the MCCs were kink bands and listric faults. These brittle structures cut mylonite foliations and mineral lineations. We assume that formation of these structures occurred in the final stage of extension that was continuous with isostatic compensation. The domal structure of the complexes formed at the same time. Figure 11 is a schematic cross-section reflecting the proposed spatial geometry of the Zagan MCC, which was one of the typical MCCs of Transbaikalia.
Thus, Transbaikalia-northernmost Mongolia contains an im-portant belt of classic early Cretaceous MCCs that are little known in the international literature. Our data from the Buteel and Zagan complexes together with published information demonstrate that the complexes formed in a crust that had been thickened, first by abundant magmatism in an active continental margin bordering the closing Mongol-Okhotsk ocean, and then by collisional to post-collisional thrusts related to formation of the Mongol-Okhotsk suture. In post-collisional times extensive basaltic fissure eruptions, bimodal alkaline to peralkaline withinplate lavas, and A-type granites developed in rifts and early Cretaceous basins bordering on or in the hanging wall of the MCCs, completing formation of one of the largest magmatic belts within a collisional orogen. All the metamorphic core complexes in various parts of eastern Asia from Transbaikalia to central China formed in the Cretaceous (Sklyarov et al. 1994;Davis et al. 1996;Webb et al. 1999;Wang & Zheng 2005;Zhai et al. 2007; this work). However, their tectonic settings and their time of formation after collision were very different. We consider that several interconnected reasons explain the synchronous widespread extension within eastern Asia and the formation of the Cretaceous MCCs. The Solonker suture and the Dabie Shan formed in the Permo-Triassic and both orogenic zones underwent massive post-collisional thrusting and crustal thickening in the Jurassic. Likewise, the western end of the Mongol-Okhotsk suture formed in the Triassic to early Jurassic and underwent major post-collisional thrusting and crustal thickening later in the Jurassic. The final closure of the Mongol-Okhotsk ocean at its eastern end was in the late Jurassic-early Cretaceous (Yakubchuk & Edwards 1999;Kravchinsky et al. 2002;Cogné et al. 2005); this means that by the early Cretaceous eastern Asia had completed its assembly into a uniform large single craton, allowing the widespread formation of extensional sedimentary basins. The result of this crustal thickening, as a result of post-collisional thrusting, was major extension and the formation of many metamorphic complexes. An additional factor that triggered extension and formation of MCCs in eastern China between the Solonker and Dabie Shan sutures in the Cretaceous was the fact that about 100 km (thickness) of Archaean mantle was delaminated under this region in the Cretaceous, the final collapse perhaps being triggered by post-collisional crustal thickening, and this led to major extension in the Cretaceous with consequent formation of more metamorphic core complexes ).