Introduction

The Central Asian Orogenic Belt (CAOB; Fig. 1), also termed the Altaid Tectonic Collage (Sengör et al. 1993), is a complex tectonic mosaic of subduction-related complexes, island arcs, ophiolitic assemblages and slivers or terranes of Precambrian crystalline basement. The development and amalgamation of these orogenic domains are related to subduction-accretion processes within the Palaeo-Asian Ocean from early Neoproterozoic to late Palaeozoic time (Sengör et al. 1993; Buslov et al. 2001; Badarch et al. 2002; Khain et al. 2003; Kröner et al. 2007; Windley et al. 2007).

Fig. 1
figure 1

Sketch map showing the position of Mongolia within the Central Asian Orogenic Belt (modified after Jahn et al. 2000; Lu et al. 2008). The Central Asian Orogneic Belt (light grey) is predominantly composed of early Neoproterozoic to late Palaeozoic rock assemblages. The main identified Precambrian microcontinents, and their inferred extension, in Mongolia, northern China and southern Siberia are shown in dark grey (adapted from Kozakov et al. 2007). The ophiolitic complexes mentioned in this paper are shown in black. Location of Fig. 2a is also shown

Full size image

A striking feature of the CAOB is the predominantly juvenile character of the newly formed crust as revealed by Nd isotopic studies from Palaeozoic arc-related magmatic rocks (e.g., Jahn et al. 2000; Kovalenko et al. 2004; Helo et al. 2006). However, the concept of juvenile crust defined exclusively by Nd isotopic data may not always be correct as suggested by Hargrove et al. (2006) from a study of granitoids in the Arabian–Nubian Shield. These authors demonstrated that isotopically juvenile magmas assimilated significant amounts of old crustal material as revealed by large proportions of inherited zircon within these rocks. In a similar way, Precambrian ages documented by xenocrystic and detrital zircons from Palaeozoic felsic volcanic rocks and arc-related clastic sediments in northern and central Mongolia (Kröner et al. 2007) suggest a substantial involvement of old material during the generation of these rocks.

Within most parts of the CAOB, and in particular Mongolia, terrains composed of high-grade metamorphic rocks were usually assumed to be Precambrian in age, though reliable radiometric ages are frequently missing (e.g., Mitrofanov et al. 1981; Badarch et al. 2002). These crystalline rocks were either interpreted as remnants of a single Precambrian microcontinent (Tuva-Mongolian; Sengör et al. 1993; Windley et al. 2007) or as individual microcontinents (e.g., Didenko et al. 1994 and references therein), namely Tuva-Mongolia, Dzabkhan and South Gobi. An alternative interpretation of the Tuva-Mongolia microcontinent (Kozakov et al. 1999; Salnikova et al. 2001) is based on U–Pb zircon ages which demonstrate that the earliest metamorphic event occurred at c. 536 Ma, followed by a regional phase of upper amphibolite-facies metamorphism between c. 497 and 489 Ma. These authors therefore suggested that discrete crustal domains, with distinct pre-metamorphic histories (Kozakov et al. 2005), were tectonically juxtaposed prior to c. 497 Ma and then underwent a phase of high-grade metamorphism as part of the evolution of the CAOB.

The existence of undoubted Precambrian crystalline rocks in central and southern Mongolia is documented by zircon dating for a variety of gneisses from the Baidrag (central Mongolia) and the Totoshan-Ulanul (southern Mongolia) blocks which are parts of the Dzabkhan and South Gobi microcontinents, respectively (Fig. 1). Within the Baidrag Block zircon dating yielded Archaean to Palaeoproterozoic emplacement ages of c. 2.80–1.82 Ga for orthogneiss protoliths (Mitrofanov et al. 1985; Kotov et al. 1995; Kröner et al. 2001; Kozakov et al. 2007), whereas early Neoproterozoic intrusion ages of c. 0.95 Ga characterize granitic protolith of gneisses from the Totoshan-Ulanul Block (Yarmolyuk et al. 2005).

We investigated the Baga Bogd massif in southern Mongolia, which is part of the Gurvan Bogd Mountain range (Fig. 1). The occurrence of high-grade metamorphic rocks, including granitic gneisses and migmatites, led this range to be considered as Precambrian in age (Badarch et al. 2002). This study focusses on SHRIMP zircon dating of a variety of granitoid gneisses in order to identify pre-Palaeozoic magmatic events and to possibly correlate the above rocks with other Precambrian crustal fragments in southern Mongolia.

Geological framework

The Gurvan Bogd Mountain range is located at the eastern end of the Gobi-Altai in southwestern Mongolia and consists of three WNW–ENE striking massifs (Ih Bogd, Baga Bogd and Artz Bogd) with peaks up to 3,957 m (Fig. 2a). The present-day topography of this mountain system (and the Altai and Gobi-Altai regions in general) resulted primarily from Cenozoic transpressive deformation in response to far-field stresses associated with the Indo-Eurasian collision (Taponnier and Molnar 1979; Cunningham 2005). One of these massifs uplifted during this deformation is the Baga Bogd massif (Vassallo et al. 2007), an asymmetric elliptic massif bounded in the north by the left-lateral strike-slip Bogd fault (Bayasgalan et al. 1999). It consists predominantly of variably deformed and metamorphosed Precambrian rocks, intruded by late Neoproterozoic granitoids as inferred from the 1:200,000 geological map (Zabotkin 1988; Fig. 2b). Permian granitoids and Jurassic intracontinental sediments are locally found, whereas the northeastern part of the massif is characterized by the occurrence of a late Palaeozoic volcano-sedimentary sequence, which is mostly in tectonic contact with the metamorphic rocks.

Fig. 2
figure 2

a Grey-scaled digital topographic map of the Gurvan Bogd Mountains showing the massif of interest for this study (Baga Bogd). b Geological sketch map of the Baga Bogd massif (modified after Geological Map L47-XXX, 1:200,000; Zabotkin 1988), with location of samples dated by SHRIMP

Full size image

Three N–S valleys have been investigated along the southern slope of the Baga Bogd massif, namely the Borbituutin and Tsagaanburgas valleys and a third, unnamed valley to the east (Fig. 2b). The rocks exposed along these valleys were deformed under upper greenschist- to amphibolite-facies metamorphism, locally leading to incipient migmatization. A strong penetrative foliation and tight folding affected all metamorphic rocks, leading to parallelism of most lithological contacts. Along the Tsagaanburgas valley, the predominant lithology is represented by strongly foliated dark augen-gneisses and layered migmatites of presumed igneous origin. These rocks are cut by undeformed dykes of mafic to intermediate composition, about one to several metres in width and consistently striking NE–SW and dipping from 45° to 75° NW or SE. The predominant rock types exposed along the Borbituutin and unnamed valleys are light grey augen-gneisses (Fig. 3a), layered migmatites (Fig. 3b), biotite-quartz-feldspar gneisses, and a higher proportion of clastic metasedimentary rocks, including metaquartzite and quartz-biotite schist, as compared to the Tsagaanburgas valley. Strongly foliated dioritic and granodioritic gneisses are also widespread in the unnamed valley. These rocks were, in turn, intruded by leucogranitic veins that locally display post-emplacement folding and boudinage; some veins contain a high proportion of garnet up to 1 cm in size (Fig. 3c). Undeformed dykes of mafic to intermediate composition (Fig. 3d), striking E–W and dipping S at 60–85°, also occur along these valleys.

Fig. 3
figure 3

Field photographs of selected rock types from the Baga Bogd massif. a Strongly foliated light grey augen-gneiss (sample Mo385), Borbituutiin valley. b Banded migmatite, Borbituutiin valley. c Garnet-rich leucocratic veinlets cutting a fine-grained gneiss. d Cross-cutting relationship between basaltic/andesitic dykes and surrounding migmatitic country rock. Hammer for scale is 32 cm long

Full size image

Sample description and geochronological results

Eight samples representing different varieties of granitoid gneisses were collected for SHRIMP zircon dating on the southern slope of the Baga Bogd massif (Fig. 2b). Separation of zircon grains followed standard procedures using heavy liquid and magnetic techniques. Representative zircons were hand-picked under a binocular microscope and mounted in epoxy resin together with chips of the reference zircons Temora 1 or CZ3. The resin disk was then polished to expose the interiors of the zircons. The internal structures of the grains were documented by cathodoluminescence (CL) images acquired with a JEOL JXA-8900 RL Superprobe in the Department of Geosciences, Mainz University (operating conditions were 15 kV accelerating voltage and 12 nA beam current on carbon-coated mounts). The U–Pb isotopic analyses on single zircons were carried out using the Sensitive High Resolution Ion Microprobe (SHRIMP II) at the Centre of Isotopic Research (CIR) in St. Petersburg (Russia) and in the Beijing SHRIMP Centre of the Chinese Academy of Geological Sciences. The instrumental characteristics are outlined by Larionov et al. (2004) and De Laeter and Kennedy (1998), respectively, and the analytical procedures are described in Claoué-Long et al. (1995), Nelson (1997) and Williams (1998). The uncertainty in the ratio 206Pb*/U during analysis of all CZ3 standard zircons during this study was 0.75% for the St. Petersburg instrument and 1.18% for the Temora 1 standard for the Beijing instrument. Sensitivity was about 56 cps/ppm/nA Pb for the standard CZ3 in St. Petersburg and about 23–24 cps/ppm/nA Pb for the Temora 1 standard on the Beijing instrument. The isotopic data and ages are reported in Table 1. Errors on individual analyses are given at the 1σ level, based on counting statistics, and include the uncertainty in the standard U/Pb age (Nelson 1997). Errors for pooled analyses are given at the 95% confidence interval. The ages and errors of intercepts of the best-fit line with concordia were calculated using the Isoplot program (Ludwig 2003). These errors were not multiplied by the square root of the MSWD since the absolute value of the intercept error is strongly model dependent.

Table 1 SHRIMP isotopic results for single zircon grains of granitoid gneisses from the Baga Bogd massif
Full size table

Sample Mo283 is a dark and well-foliated, medium-grained dioritic orthogneiss collected in the unnamed valley (N44°45′58.9″, E101°44′35.6″). It represents the most dominant rock type in this valley and is composed of large biotite flakes, plagioclase, recrystallized quartz showing undulous extinction, and relicts of clinopyroxene. Accessory minerals include apatite, zircon and epidote. The rock shows a strong penetrative foliation, defined by biotite flakes, which conforms to the foliation developed in the surrounding fine-grained biotite gneiss and quartzo-feldspathic layered gneiss. Field relationships suggest the dioritic gneiss to be intrusive into the above gneisses, but the strong ductile deformation has largely obliterated the original contacts. This sample contains a population of zircons dominated by clear, euhedral and long prismatic grains with short and slightly rounded terminations. The colour varies from slight yellowish to pinkish, and the grain size ranges from 150 μm up to 300 μm in length. The grains display oscillatory zoning in CL images (Fig. 4a), characteristic of igneous growth. A few smaller grains <150 μm in length are also present and are characterized by a dark colour, short prismatic and subhedral habit with slightly rounded faces. Nine grains were analysed on the St. Petersburg SHRIMP II (Table 1). Eight grains yielded a tight cluster of concordant results (Fig. 4a) with a weighted mean 206Pb/238U age of 498 ± 3 Ma which is interpreted to reflect the time of crystallization of the gneiss protolith. One grain yielded a significantly older, concordant result (Fig. 4a) with a 206Pb/238U age of 656 ± 10 Ma and is interpreted as a xenocryst, probably inherited from the basement rock into which the gneiss protolith was emplaced.

Fig. 4
figure 4

Concordia diagrams showing SHRIMP analyses of single zircons for samples from the unnamed valley. Data boxes for each analysis are defined by standard errors in 207Pb/235U, 206Pb/238U and 207Pb/206Pb. Errors on pooled ages are given at 95% confidence level. Insets show CL images of zircon grains, white circles indicate SHRIMP analytical spots; spot numbers refer to Table 1

Full size image

Sample Mo286 was collected some 15 m from sample Mo283 and is a strongly foliated, medium-grained biotite gneiss, originally a granodiorite, into which the above dioritic gneiss may originally have been intrusive. The texture is lepidoblastic, and the rock contains small biotite flakes mostly altered into chlorite, plagioclase with polysynthetic twinning, K-feldspar with a high degree of alteration and quartz showing undulous extinction. It contains accessory apatite, zircon, epidote and an oxide phase. The original relationship with the surrounding rocks is not clear since they now share the same foliation, and undoubted cross-cutting relationships are not preserved. The zircons from this sample are translucent, yellowish to slightly reddish in colour and predominantly long-prismatic or needle-like in shape and between 100 and 250 μm in length. Cathodoluminescence images reveal simple oscillatory zoning indicative of a magmatic origin and do not show inherited cores or metamorphic overgrowth (Fig. 4b). Five zircon grains were analysed on the St. Petersburg SHRIMP II (Table 1), and these produced concordant or near-concordant results (Fig. 4b) with a weighted mean 207Pb/206Pb age of 499 ± 3 Ma. We interpret this to reflect the time of emplacement of the granodioritic protolith. Both samples Mo283 and Mo286 obviously reflect the same magmatic event and may represent one phase of late Cambrian granitoid plutonism in the Baga Bogd massif.

Sample Mo380 is a leucocratic, fine-grained biotite granite-gneiss cut by garnet-rich leucogranitic veins and was collected higher up in the unnamed valley (N44°46′11.7″, E101°44′41.0″), some 400 m north of samples Mo283 and Mo286. It is composed of plagioclase, quartz, microcline, and small biotite flakes. The zircon grains are translucent, subhedral, long- to short-prismatic with a dominant brownish colour and vary in length between 100 and 250 μm. Cathodoluminescence images for the long-prismatic grains display simple igneous-related oscillatory zoning, whereas short-prismatic and more rounded grains tend to have partly recrystallized rims and convolute internal zoning (Corfu et al. 2003). Six grains were analysed on the Beijing SHRIMP II (Table 1) and produced concordant or near-concordant results which are spread along the concordia from 930 to 500 Ma (Fig. 4c). At the lowest end of the spread, two grains yielded concordant results, identical within error, with a mean 206Pb/238U age of 501 ± 4 Ma (Fig. 4c). The four remaining grains have ages between 930 and 815 Ma (Fig. 4c) with two grains at the end of the spread yielding similar results with a mean 207Pb/206Pb age of 928 ± 6 Ma. In view of the igneous origin of the zircons the age of 501 ± 4 Ma is interpreted as the time of emplacement of the gneiss precursor, whereas the remaining older grains suggest inheritance and involvement of Neoproterozoic basement material in the generation of the gneiss protolith. If this interpretation is correct, the gneiss protolith belongs to the same magmatic pulse as samples Mo283 and Mo286.

Sample Mo381 was collected in the same valley, about 600 m north of Mo380, and is a strongly foliated, fine-grained granite-gneiss composed of biotite and muscovite flakes, quartz, K-feldspar and garnet. K-feldspar is moderately to intensely sericitized, and garnet is extensively replaced by chlorite. The rock has a lepidoblastic texture defined by alternation of quartzo-feldspatic and micaceous bands. The sample contains a small population of subhedral, translucent and brownish zircons, and CL images reveal oscillatory zoning as well as partially recrystallized rims for most of the grains (Fig. 4d). Four zircons from this sample were analysed on the Beijing SHRIMP II (Table 1). One grain yielded a concordant analysis, and the three remaining grains produced discordant results (Fig. 4d). However, all analyses are aligned along a chord (MSWD = 0.06) yielding an upper intercept age of 1525 ± 26 Ma and a lower intercept at 60 ± 170 Ma reflecting recent lead-loss. All analyses have a weighted mean 207Pb/206Pb age of 1519 ± 11 Ma, identical within error to the upper intercept age, which we consider to best approximate the time of emplacement of the granitic protolith. Although sample Mo381 exhibits the same foliation as the other rocks described above, it is clearly much older and must be part of the basement into which the late Cambrian pluton was emplaced.

Sample Mo385 is a light grey augen-gneiss representing the main lithology in Borbituutiin valley and was collected at N44°46′05.3″, E101°41′29.0″. The gneiss is strongly foliated, coarse-grained, and is composed of quartz, microcline and millimetre-size biotite that is mostly altered into chlorite. K-feldspar is extensively sericitized, and the augen are up to 5 cm in size. The zircons are translucent, long- to short-prismatic, with variable aspects ratios and rounded terminations. The length ranges from 100 to 280 μm, and the colour varies from light to dark brown. Cathodoluminescence images reveal oscillatory zoning (Fig. 5a), characteristic of magmatic growth, and the presence of inherited cores in some grains. Six zircons were analysed on the Beijing SHRIMP II (Table 1) and display variable degrees of discordance (Fig. 5a). All analyses, with one virtually concordant, are well aligned along a chord (MSWD = 0.13) (Fig. 5a) and define a weighted mean 207Pb/206Pb age of 954 ± 8 Ma. The above chord defines an upper concordia intercept age of 959 ± 23 Ma and a lower intercept at 89 ± 330 Ma. The upper intercept age is identical to the mean 207Pb/206Pb age but less precise because of the large error in the lower intercept. This lower intercept is geologically meaningless and clearly reflects recent lead-loss. We therefore consider the mean 207Pb/206Pb age of 954 ± 8 Ma to most closely approximate the time of emplacement of the gneiss precursor. Again, this sample is part of a Precambrian basement.

Fig. 5
figure 5

Concordia diagrams showing SHRIMP analyses of single zircons for samples from the Borbituutiin and Tsagaanburgas valleys. Data boxes for each analysis and errors on pooled ages as in Fig. 4. Insets show CL images of zircon grains, white circles indicate SHRIMP analytical spots; spot numbers refer to Table 1

Full size image

Sample Mo387 is a medium-grained dark grey augen-gneiss collected farther north in Borbituutin valley at N44°46′12.0″, E101°41′33.5″. It is composed of strongly seriticized K-feldspar porphyroblasts, quartz, and biotite flakes with chloritized rims. It has a grano-lepidoblastic texture with a gneissic fabric underlined by biotite and quartz ribbons. The zircons are long-prismatic with a high aspect ratio, or short-prismatic and are translucent, mostly subhedral with rounded edges and light grey to brown colour. Cathodoluminescence images reveal well developed oscillatory zoning due to igneous growth (Fig. 5b) and core-rim relationships in some grains. Five zircon grains from this sample were analysed on the Beijing instrument (Table 1). Four concordant and discordant analyses (Fig. 5b) are aligned along a discordia line (MSWD = 0.02) which defines a lower intercept at 8 ± 190 Ma and an upper intercept age of 984 ± 27 Ma, identical within error to the weighted mean 207Pb/206Pb age of 983 ± 6 Ma of the four analyses. Thus, the age of 983 ± 6 Ma is considered to reflect the time of emplacement of the granitic protolith. The fifth grain yielded a much older 207Pb/206Pb age of 1701 ± 14 Ma (1σ) (Fig. 5b), which, as in previous samples, is interpreted as a xenocryst inherited from a late Palaeoproterozoic crustal source.

Sample Mo388 is a strongly deformed very fine-grained, laminated leucocratic gneiss, interlayered with the surrounding augen-gneiss and collected still farther north in Borbituutin valley at N44°47′03.5″, E101°41′14.1″. The mineral assemblage consists of K-feldspar, quartz and minor biotite flakes. The zircons are of two types. Type 1 is represented by clear, transparent, long- to short-prismatic and euhedral grains, 100–150 μm in length and displaying typical magmatic growth features under cathodoluminescence (Fig. 5c). Type 2 grains are mostly short-prismatic, subhedral with rounded terminations and dark brown colour. They range in length between 100 and 250 μm and display complex internal growth structures as revealed by CL images. Five grains of type 1 grains were analysed on the Beijing SHRIMP II (Table 1) and yielded concordant results (Fig. 5c) with a weighted mean 206Pb/238U age of 502 ± 6 Ma which we interpret as the time of emplacement of the gneiss protolith.

Lastly, sample Mo389 is a coarse-grained, light grey granite-gneiss collected in Tsagaanburgas valley at N44°47′29.8″, E101°37′16.9″. This gneiss is composed of large K-feldspar, quartz, minor plagioclase, millimetre-size biotite and large idiomorphic garnet crystals altered into an assemblage of chlorite + quartz. K-feldspar is moderately sericitized and biotite is mostly transformed into chlorite. Quartz is mostly recrystallized. The zircons are mainly short-prismatic with rare long-prismatic and transparent grains. Most grains are euhedral to subhedral with slightly rounded terminations, vary in length from 80 to 250 μm, and are light yellowish to dark brownish in colour. Cathodoluminescence images reveal well-developed oscillatory zoning (Fig. 5d), characteristic of igneous crystallization, and do not show cores or overgrowth. Seven zircons were analysed on the Beijing instrument and yielded well-aligned results one of which is concordant, one reversely discordant and five exhibit variable degrees of discordance (Table 1; Fig. 5d). All analyses define a weighted mean 207Pb/206Pb age of 956 ± 3 Ma, whereas linear regression (MSWD = 0.02) resulted in an upper concordia intercept age of 955 ± 19 Ma and a lower intercept at −13 ± 240 Ma which clearly reflects recent Pb-loss. As in the case of sample Mo385, both ages are identical, and we consider the more precise mean 207Pb/206Pb age as most closely reflecting the time of protolith emplacement.

Whole-rock major and trace elements chemistry

Major and trace element data for the dated samples were obtained by X-ray fluorescence spectrometry, using a Philips MagiXPro spectrometer in the Department of Geosciences, Mainz University. Loss on ignition (LOI) was determined by weight loss after heating at 1050°C for 4 h. The results are listed in Table 2.

Table 2 Major and trace element analyses of granitoid gneisses from the Baga Bogd massif
Full size table

Zircon geochronology revealed three groups of ages, one at around 500 Ma (samples Mo283, Mo286, Mo380 and Mo388), another with early Neoproterozoic ages ranging from 983 to 954 Ma (samples Mo385, Mo387 and Mo389) and one with a much older age of 1.52 Ga (sample Mo381). The distinction between the three groups was already suspected from field relationships although intrusive features displayed by the youngest group were partly obliterated during post-intrusion deformation. In the following, these samples will be treated as two separate groups: Precambrian (Meso- and early Neoproterozoic) gneisses on one side and the late Cambrian gneisses on the other side. The chemical indices of alteration (CIA; Table 2) of 46–55 for all granitoid gneisses are comparable to those of unaltered granitic and mafic rocks (CIA for average fresh granite: 45–55; Nesbitt and Young 1982) except for the granite gneiss Mo381 which has a CIA index of 69 and the highest LOI value of 1.75 (Table 2). Nevertheless, the relatively high degree of alteration of K-feldspar and some biotite observed in thin section indicates that some elements and the alkaline elements in particular, may not reflect primary compositions.

Major element chemistry

The early Neoproterozoic gneisses are granitic and calc-alkaline in composition (Fig. 6; Table 2) and show limited variation in chemical composition with restricted SiO2 contents of c. 71–79 wt% (Table 2). They are characterized by homogeneous and low TiO2, Fe2O3t, MgO and CaO contents and relatively high K2O contents of 5.6–6.4 wt% (Table 2), whereas significant scatter is observed for Al2O3, Na2O and P2O5 (Table 2). These samples are all slightly peraluminous with A/CNK (Aluminium Saturation Index; molecular Al2O3/(CaO + Na2O + K2O) and A/NK values of 1.1 and 1.2–1.3, respectively (Fig. 7a; Table 2). Compared to these samples, the Mesoproterozoic gneiss has lower SiO2, CaO, Na2O, K2O, P2O5 contents and higher Al2O3, Fe2O3t, MgO and TiO2 contents (Table 2). This sample is clearly peraluminous (Fig. 7a; Table 2) with A/CNK and A/NK values of 2.2 and 2.6, respectively, in agreement with its mineralogy. All samples with Precambrian ages are moreover characterized by low Na2O/K2O ratios (0.3–0.5) and an enrichment in iron as suggested by FeOt/(FeOt + MgO) ratios that are systematically higher than 0.8 at a given SiO2 content (Fig. 7b).

Fig. 6
figure 6

AFM diagram for gneisses from the Baga Bogd massif showing. FeOt, total iron as FeO; dividing line between tholeiitic and calc-alkaline series is from Irvine and Baragar (1971)

Full size image
Fig. 7
figure 7

a A/NK vs A/CNK diagram, modified after Shand (1951) and b FeOt/(FeOt + MgO) versus SiO2 diagram after Frost et al. (2001). FeOt, total iron as FeO. For legend see Fig. 6

Full size image

The late Cambrian gneisses cover a wide compositional range with SiO2 contents of c. 55–68 wt% (Table 2) and show dioritic to granodioritic compositions with calk-alkaline affinities (Fig. 6). They display crude correlations of decreasing TiO2, Fe2O3t, MgO and CaO with increasing SiO2 (Table 2). Most gneisses are slightly peraluminous with A/CNK and A/NK values of 1.1–1.2 and 1.5–1.7, respectively (Fig. 7a; Table 2), whereas sample Mo283 with SiO2 < 56 wt% is classified as metaluminous with A/CNK < 1 and A/NK > 1 (Fig. 7a; Table 2). Moreover, these samples have higher Na2O/K2O ratios (0.6–1.2) and FeOt/(FeOt + MgO) ratios systematically lower than 0.8 (Fig. 7b) and are referred to as magnesian Cordilleran-type granitoids according to the classification of Frost et al. (2001).

Trace element chemistry

The Neoproterozoic granites gneisses have relatively low concentrations of Nb (4–9 ppm), Zr (85–151 ppm), Sr (87–96 ppm) and Ba (301–557 ppm). They show slight enrichments in large ion lithophile elements (LILE; Ba, Th, U) compared to high field strength elements (HFSE; Nb, Zr, Ti) and display Nb, Sr and Ti troughs on a bulk continental crust-normalized trace element patterns (Fig. 8a). The Mesoproterozoic gneiss Mo381 displays a slightly different picture with higher concentrations of Ba, Nb, Zr, Ni and Cr (Table 2). Trace elements normalized to bulk continental crust show an overall good agreement with typical crustal values except for lower concentrations in Sr and P (Fig. 8a). Using the Nb–Y trace-element discrimination diagram of Pearce et al. (1984), all Precambrian samples plot in the field of syn-collisional (syn-COLG)/volcanic arc-related granitoids (VAG, Fig. 9).

Fig. 8
figure 8

Multi-element diagrams for a Meso- and early Neoproterozoic gneisses and b late Cambrian gneisses from the Baga Bogd massif. All elements were analysed by XRF; primitive mantle and bulk continental crust normalizing values are from Sun and McDonough (1989) and Rudnick and Fountain (1995), respectively

Full size image

Compared to the Precambrian samples, the late Cambrian rocks are more enriched in Nb, Zr, Sr and Ba (Table 2).These samples are characterized by enrichment in LILE compared to HFSE when plotted in a primitive mantle-normalized spidergram (Fig. 8b). They display well pronounced negative Nb anomalies and slightly negative Ti anomalies. These gneisses display crude trends of decreasing contents of Nb, Sr, Ba and increasing Zr with increasing content of SiO2 (Table 2). In the Nb–Y discrimination diagram (Fig. 9) (Pearce et al. 1984) these samples overlap the fields of syn-COLG/VAG and within-plate granitoids (WPG, Fig. 9), which suggests formation in an active continental margin environment or in a mature island arc.

Discussion

Our SHRIMP zircon ages for granitoid protoliths of various orthogneisses from the Baga Bogd massif in southern Mongolia confirm the Neoproterozoic and Cambrian age assessments inferred on the 1:200,000 geological map (Fig. 2b). We suspect that the Neoproterozoic gneisses experienced some structural and possibly metamorphic history prior to their incorporation into the early Palaeozoic arc terrane of southern Mongolia. However, strong ductile deformation during the Palaeozoic accretion event led to parallelism of structures in all metamorphosed rocks so that no distinction could be made in the field between Palaeozoic and older structures. No angular relationships between the Precambrian and early Palaeozoic orthogneisses were observed. This is unlike some of the low-strain domains of the large Baidrag Block farther NE where Precambrian structures and a polymetamorphic Precambrian history are well preserved (Kozakov et al. 2007).

Fig. 9
figure 9

Nb versus Y discrimination diagram for granites after Pearce et al. (1984). For legend see Fig. 6

Full size image

Precambrian thermal events and possible link to identified crustal basement

The oldest dated event in the geological evolution of the study area was the intrusion of the granitic gneiss precursor of sample Mo381 at 1519 ± 11 Ma. An even older age of c. 1700 Ma is documented by a xenocrystic zircon grain from grey augen-gneiss sample Mo387. These ages document the presence of late Palaeoproterozoic to early Mesoproterozoic crust in the Baga Bogd massif.

The existence of Precambrian crust of Archaean to Palaeoproterozoic age is well documented for the Baidrag Block (Dzabkhan microcontinent; Kozakov et al. 1997) which is located to the northwest of the study area (Fig. 1). The Baidrag basement shows a complex and polyphase tectono-metamorphic evolution from late Archaean to late Palaeoproterozoic time (e.g., Kozakov et al. 1997, 2007 and references therein). The oldest rocks correspond to tonalitic and two pyroxene gneisses with protolith emplacement ages documented by conventional multigrain and SHRIMP single zircon ages of c. 2.80–2.65 Ma (Mitrofanov et al. 1985; Kozakov et al. 2007). The last major thermal event experienced by these rocks reflects the peak of granulite-facies metamorphism with a minimum age of 1839 ± 1 Ma (Pb–Pb zircon evaporation age, Kröner et al. 2001), for metamorphic zircons from a metapelitic rocks. A post-tectonic granite with a U–Pb zircon age of 1825 ± 5 Ma (conventional method; Kotov et al. 1995) marks the final consolidation of this Precambrian crystalline block.

Other discrete Precambrian crustal domains are recognized in western Mongolia in the Dariv Metamorphic Complex and the Khantaishir Complex (Fig. 1), also assumed to be part of the Dzabkhan microcontinent (Badarch et al. 2002; Kozakov et al. 2002), but display slightly different age patterns in comparison to the Baidrag Block. Within the Dariv Metamorphic Complex there are unpublished zircon ages between 840 and 600 Ma with rare xenocrysts up to Archaean in age (Kozakov I.K., pers. comm., 2007). The minimum age of emplacement for the protolith of a dioritic gneiss is 1426 ± 1 Ma (Pb–Pb zircon evaporation age; Kröner et al. 2001). The Khantaishir Complex around Altai City contains a porphyritic granite-gneiss with a crystallization age of 1127 ± 1 Ma and a 1715 Ma old xenocrystic zircon (Pb–Pb zircon evaporation ages; Kröner et al. 2001). Near Altai City, a migmatite yielded a SHRIMP zircon age of 840 ± 9 Ma and also contains inherited zircons ranging from 2,445 to 1,440 Ma in age (Zhao et al. 2006).

Compared to the crustal fragments discussed above, a distinctive feature of the Baga Bogd massif are the early Neoproterozoic granitic intrusions documented by three samples (Mo387, Mo389, and Mo385) and yielding emplacement ages of 983 ± 6, 956 ± 3 and 954 ± 8 Ma, respectively. Granitic protholiths with early Neoproterozoic ages have so far only been identified in one locality in southern Mongolia. A granite-gneiss exposed in the Totoshan-Ulanul Block of south-eastern Mongolia (Fig. 1) provided a protolith intrusion age of 952 ± 8 Ma (conventional U–Pb single and multigrain zircon age; Yarmolyuk et al. 2005). A similar event was documented to the W of the Totoshan-Ulanul Block within the Tsagan-Khairkhan Block of northern China (Fig. 1) where emplacement of a granite-gneiss protolith was dated at 916 ± 16 Ma (U–Pb zircon age; Wang et al. 2001). These two blocks are usually considered as part of the South Gobi microcontinent.

The origin of these microcontinental crustal fragments is uncertain in most cases. They are interpreted either as fragments derived from the East Gondwana margin (Mossakovsky et al. 1993; Didenko et al. 1994; Filippova et al. 2001; Dobretsov et al. 2003; Dobretsov and Buslov 2007; Kheraskova et al. 2003) or from the Siberian craton (Berzin et al. 1994; Kuzmichev et al. 2001; Yarmolyuk et al. 2006) or from both (Buslov et al. 2001). However, late Mesoproterozoic to early Neoproterozoic magmatic events have not been documented so far from the Siberian craton margin (Pisarevsky and Natapov 2003; Gladkochub et al. 2006; Smelov and Timofeev 2007), and the distribution of sedimentary successions mainly record a passive margin environment (Pisarevsky et al. 2008). For example, the Teya granite pluton, part of the Yenisey Ridge and located at the south-western margin of the Siberian Craton (Fig. 1), has long been considered to reflect a Grenvillian-age collisional event. However, Vernikovsky et al. (2007) have now shown that the oldest phase of magmatism in this area occurred between 875 and 864 Ma, making a possible link with a Grenvillian-age event unlikely. On the other hand, discrete early Neoproterozoic magmatism is documented in northwestern China. Southeast of the Tarim Basin, the Qilian and Qaidam blocks (Fig. 1) host granitoids, which are intrusive into shallow marine strata and ultramafic rocks and yielded emplacement ages of c. 930–920 Ma (conventional U–Pb zircon ages; Gehrels et al. 2003), whereas a zircon age of c. 969 Ma was reported for a granitoid gneiss in the Altyn Tagh range (Fig. 1)(ion microprobe analysis; Cowgill et al. 2003). Lu et al. (2008) listed numerous zircon ages between 1,050 and 920 Ma for granitoids from the northern and eastern margin of the Tarim craton. These authors interpret these early Neoproterozoic magmatic events as the result of what they term “Tarimian Orogeny”, leading to the final cratonization of the Tarim Craton.

The aforementioned magmatic ages compare well with our new zircon ages and suggest that the Baga Bogd massif is unlikely to be related to the Baidrag basement block but shares greater similarities with crustal fragments in southern Mongolia and northwestern China. It is therefore unlikely that the Baga Bogd block was derived from the Siberian craton. It may either be considered as a fragment of the South Gobi microcontinent, rifted off the northeastern Gondwana margin (Buslov et al. 2001) or as a fragment of the Tarim Craton. However, considering the poor correlation and the paucity of reliable ages, further data are required to draw definitive conclusions.

Late Cambrian plutonism and geodynamic implications

Finally, our results document a phase of late Cambrian magmatism with the emplacement of granitic, granodioritic and dioritic intrusions at 502 ± 6, 501 ± 4, 499 ± 3 and 498 ± 3 Ma. Moreover, two of the late Cambrian gneisses contain xenocrystic zircons with ages of c. 928, 893, 815 and 656 Ma. These inherited grains provide further evidence for Neoproterozoic crustal material and suggest several Neoproterozoic zircon-forming events within the study area as also supported by early Neoproterozoic zircon ages recorded in samples Mo385, Mo387 and Mo389.

Evidence for an oceanic domain to the north of the Baidrag Block is indicated by the large (300 km long) Bayankhongor ophiolite mélange recording ocean-crust formation at c. 665 Ma ago (SHRIMP zircon age; Kovach et al. 2005). A minimum age for the accretion of this oceanic remnant with the Baidrag block (Buchan et al. 2001) is bracket by syn-collisional granitoids intruding the ophiolite and its boundary fault between c. 547 and 539 Ma (U–Pb and Pb–Pb zircon evaporation ages; Buchan et al. 2002; Kozakov et al. 2006). Within this interval, the southwestern margin of the Siberian craton faced the open Palaeo-Asian Ocean as indicated by a broad belt of c. 570 Ma old ophiolitic remnants. Slivers of oceanic crust are well exposed in the Khantaishir ophiolite south of Altai City (U–Pb zircon age of 568 ± 4 Ma for plagiogranite; Khain et al. 2003), in the Dariv range farther W (Bayannor ophiolite, U–Pb zircon age of 571 ± 4 Ma for plagiogranite; Khain et al. 2003) and more to the north in the Tuva area of southern Siberia with the Agadagh Tes-Chem Massif (Pb–Pb zircon evaporation age of 570 ± 1 Ma; Pfänder and Kröner 2004). The accretion of these portions of oceanic lithosphere onto Precambrian crystalline massifs likely occurred in late Cambrian time and was accompanied by widespread magmatism and regional metamorphism. A minimum time for the accretion of the Bayannor ophiolite within the Dariv Metamorphic Complex is given by the intrusion of an undeformed quartz diorite at 515 ± 8 Ma (SHRIMP zircon age; Dijkstra et al. 2006). A phase of high-grade metamorphism in this complex is moreover recorded at 499 ± 3 Ma by a U–Pb zircon age for a hypersthene-garnet granulite (Kozakov et al. 2002) and may reflect accretion of the ophiolitic sequence onto the crystalline basement. The post-tectonic intrusion of a “stitching granite” at 497 ± 1 Ma (Pb–Pb zircon evaporation age; Kröner et al. 2001) indicates that this phase of accretion was accomplished in the latest Cambrian.

Within this framework, and according to the chemistry of the dated samples, we consider that the late Cambrian phase of magmatism recorded in the Baga Bogd massif occurred in an active continental margin environment and was related to the progressive closure of the Palaeo-Asian Ocean. Moreover, these data suggest that the Baga Bogd Precambrian crustal fragment was either docked against the southward (present-day coordinates) growing margin of the CAOB or was a large enough crustal entity to develop a magmatic arc along its margin.

The field relationships, combined with our new zircon ages, indicate that the different generations of granitoid gneisses underwent a common metamorphic and deformational event. This produced the regional penetrating fabric, leading to tectonic interlayering of Precambrian and late Cambrian gneisses, and most probably culminated in migmatization of the most fertile lithologies. This event occurred after the Cambrian, but our data are insufficient to constrain the time of post-Cambrian metamorphism. We speculate that this event may signify accretion of the Baga Bogd or South Gobi basement terrane onto the southward (present-day coordinates) growing margin of the CAOB or accretion of a Japan-style basement-core magmatic arc in the course of continuous closure of oceanic basins in the Palaeo-Asian Ocean.

Conclusions

Our new SHRIMP zircon ages for granitoid gneisses in the Baga Bogd massif of southern Mongolia indicate the presence of Precambrian crystalline rocks and a late Cambrian phase of granitoid magmatism. The protolith emplacement of a granite-gneiss (Mo381) is dated at 1519 ± 11 Ma and represents, up to now, the oldest rocks in this massif. A phase of magmatism is then documented by granite-gneisses whose precursors were emplaced at 983 ± 6, 956 ± 3 and 954 ± 8 Ma. This is the second record in southern Mongolia of early Neoproterozoic granitoid magmatism. Finally, a late Cambrian subduction-related phase of magmatism is identified by several granitoid gneisses with homogeneous protolith intrusion ages at c. 500 Ma. Our results document the presence of an additional fragment of Precambrian crystalline basement in southern Mongolia and reinforce the idea that slivers of Precambrian crystalline domains are widespread in central and southern Mongolia as suggested by recent geological and isotopic studies (Yarmolyuk et al. 2005; Kozakov et al. 2007).