A new look at the geodynamic development of the Ediacaran–early Cambrian forearc basalts of the Tannuola-Khamsara Island Arc (Central Asia, Russia): Conclusions from geological, geochemical, and Nd-isotope data

2.2.1 Izinziul’ site

At the Izinziul’ site, the Chingin Formation is a 4,400 m thick monoclinal fold dipping to the east at 45–60°. The formation is in tectonic contact with the Kurtushiba ophiolites in the west, and unconformably overlain by the middle-upper Cambrian terrigenous Alasug Group of the early Khemchik-Systygkhem molasse basin deposits [45] to the east. The Chingin Formation is partitioned into sub formations which are simply termed the “Lower Subformation” (3,400 m) that is conformably overlain by the “Upper Subformation” (1,000 m). Within the Chingin Formation system, the lower subformation is in general considerably thicker than the upper subformation. The lower subformation chiefly consists of basaltic lavas and tuffs, as well as interlayers of tuffaceous sandstones and undifferentiated siliceous rocks. Porphyritic pyroxene basalts of the lower subformation exhibit a concentration of SiO₂ = 50–54 wt%, a high concentration of MgO = 8–13 wt%, and a low concentration of TiO₂ = 0.3–0.5 wt%. These rocks are defined as boninites [45]. Igneous rocks of an enriched mid-ocean ridge basalts (E-MORB) like composition, however, predominate the basaltic units of the lower subformation while boninites only occur in a relatively smaller amount. The upper subformation of the Chingin contains carbonaceous, siliceous, tuffaceous clay shales, to a lesser extent basalt, tuffs, metagreywackes, and gravelites, as well as black dolomites [22,45]. Sampling of the lower subformation of the Chingin Formation was carried out along the northern part of the Izinziul’ River Valley.

2.2.2 Koiard site

The Koiard site is a large, northwest dipping overturned fold, close to the metasedimentary units of the Dzhebash subzone (Figures 2 and 5). The Chingin Formation occurs in the limbs of the fold and the Kurtushiba ophiolites in the core area [46]. At this locality, the Chingin Formation is in conformable contact with strata of the Kurtushiba ophiolite. Porphyritic (KK-17/16) (Figure 5a) and aphyric (KK-18/16) pillow basalts of the upper subformation of the Chingin were sampled at Koiard. The pillow basalts are interbedded with tuffs. The pyroxenes in one basalt sample from the upper Chingin Formation at the Koiard site display a spinifex texture but the chemical composition does not correspond to that of komatiites [47]. However, komatiites have been reported from the Chingin Formation and other areas of the Kurtushiba subzone from exploration expeditions [47,48], but were not studied in detail and their origin is elusive. Possibly, the skeletal and spinifex-like textures are explained best with rapid cooling of the Chingin basalts.

2.2.3 Saryg-Tash, Tlangara, and Kopsek sites

In the southwestern part of the Kurtushiba subzone, small fragments of the Chingin Formation and Kurtushiba ophiolites are present, which are exposed as discontinuous bands along the contact zone to the Dzhebash Group and early Paleozoic Khemchik-Systygkhem molasse basin deposits (Figure 1). In this area, the Chingin Formation crops out at the Saryg-Tash, Tlangara, and Kopsek sites (Figure 1b). Saryg-Tash is a complex anticlinal structure. The limbs of the anticline are folded middle Cambrian–Silurian sediments. In the axial part, podiform blocks with a width of 0.1–0.5 km protrude the Ordovician bedrock (conglomerates and sandstones) and consist of foliated Chingin basalts, massive serpentinites, pyroxenites, gabbroids, diorites, listwänites, and listwänitized rocks. At Tlangara, a serpentinite mélange with a thickness of ∼100 m contains blocks of Chingin basalts, 40 × 60 m in size. At Kopsek, Chingin basalts are present as inclusions in a serpentinite mélange that is framing an ophiolite allochthon, as well as in the form of olistoliths among the Dzhebash metasediments.

2.3.1 Shat study area

The thickness of the Aldynbulak Formation at the Shat site (Figures 3 and 5c) is about 3,000 m. It consists mainly of the basaltic “Lower Subformation” and a largely sedimentary “Upper Subformation.” The subformations are equally thick at the study locality. The lower subformation of the Aldynbulak Formation is dominated by pillow lavas, which are associated with tuffs, tuffaceous sandstones, tuff breccias, as well as undifferentiated siliceous and carbonate rocks. The upper subformation consists of slaty siliceous rocks and metapelites with rare interlayers of carbonaceous black metashales, as well as single, smaller (∼1 m thick) limestone horizons.

At Shat, the tectonic contact of the Aldynbulak Formation with the Khemchik ophiolites (Shatskii massif) is marked by narrow bands of olistostromes and serpentinite mélanges (Figure 3). The foliated sand-silt matrix of the olistostromes contains meter- to tens of meter-sized blocks of cherts and basalts, as well as altered ultrabasites, gabbroids, and limestones [51]. The serpentinite mélange zone is tens to hundreds of meters thick, with inclusions of basalts, quartzites, massive serpentinites, and limestones. The sampling points are shown in Figure 3.

2.3.2 Buura site

At the Buura study area, lower Cambrian strata comprise reworked fragments of the Aldynbulak Formation in an olistholithic setting. The sedimentary units are siltstones, sandstones, gravelites, and conglomerates as well as blocks of limestone, chert, basalt, sometimes gabbro, and ultramafic rocks. The sampling of Aldynbulak basalts was carried out on the largest olistolith block with pillow basalts, measuring about 500 m across.

2.3.3 Uttug-Khaia formation and Uttug-Khaia mountain

In previous mapping projects considering the southeastern edge of the Western Sayan mountains (Figure 1), most Ediacaran–lowermost Cambrian sedimentary and basalt-bearing units have either been attributed to the Aldynbulak or Chingin Formations or synonymous formations of these [14,35,36,37,42,49]. Geochemical analytics of different localities of the Aldynbulak Formation basalts (this study) have, however, shown that these basalts do group into two different geochemical variants. We thus preliminarily subdivided the Aldynbulak Formation into the Aldynbulak and Uttug-Khaia Formation (with lower and upper subformations), based on the different chemical basalt compositions and the implication that these formed in different geodynamic environments.

The Uttug-Khaia Formation basalts (Figure 5d) are best exposed at the Uttug-Khaia Mountain locality within the Khemchik-Tapsa zone (Figure 4). At Uttug-Khaia Mountain, fault-bound tectonic sections of basalt-bearing volcano-sedimentary and sedimentary rocks are found alongside the olistolithes. The basalt-bearing volcano-sedimentary units are composed of pillow lavas, foliated basaltic tuffs, interlayers of cherts, and layers of siliceous volcanogenic rocks and siliceous carbonates. The pillow lavas are cut by single dikes of plagiophyric basalts. Other sedimentary rocks are largely undifferentiated cherts. The matrix of the olistostromes is a sedimentary breccia and conglobreccia with fragments of basalts, less often cherts, carbonate rocks, and (meta-)shale (Figure 5e and f). Clastic blocks indicating episodes of considerable mass-transport are up to several tens of meters in size, composed mainly of cherts, as well as limestones, dolomites, and less often gabbroids and serpentinite. Some siliceous olistoliths are monolithic breccias composed of flints in a siliceous cement (Figure 5f). The Ediacaran–lower Cambrian Uttug-Khaia Formation is overlain by siliciclastic and reef carbonates of the Akdurug Formation from which archaeocyathids of the Sanashtykgol biohorizon (Siberian stage Botoman) have been described [40]. The lower units of the Akdurug Formation show signs of reworking of the underlying olistoliths [42]. The sampling localities are shown in Figure 4.

Figure 4

Geological map and section of Uttug-Khaia site (compiled and adapted from [42] with kind permission from the Journal Geologiya i Geofizika). Photographs of field sites are indicated. (a) Geological map showing outcrops of the Uttug-Khaia Formation (formerly assigned to the Aldynbulak Formation) and sampling localities; (b) Section A–B through the Uttug-Khaia Mountain fold system displaying the stratigraphic relations of the different complex faulted units. E – Ediacaran, Ꞓ₁ – lower Cambrian (approx. Terreneuvian – Cambrian Series 2), O – Ordovician (Khemchik-Systygkhem molasse sediments), Q – Quaternary.

Figure 5

Field photographs of the study sites. (a) Chingin pillow basalts at Koiard; (b) Dzhebash metasediments (slate) at the Koiard site; (c) Aldynbulak pillow basalts at Shat; (d) Uttug-Khaia pillow basalts at the southern foot of Uttug-Khaia Mountain; (e) Uttug-Khaia section: olistostromes in the foreground and the overlying lower Cambrian conglomeratic and archaeocyathid bearing Akdurug Formation in the background; (f) Olistolith breccia with siliceous clasts in a siliceous matrix, Uttug-Khaia section. ((a) and (b) in Figure 2; (c) in Figure 3; (d)–(f) in Figure 4).

2.3.4 Tapsa River site

The study locality in the Tapsa River area is located further east than the previously described sites. Here lower Cambrian olistostromes are in direct contact with the island arc complexes of the Ondum subzone (Figure 1b, locality 9). The volcano-sedimentary units are mainly conglomerates, gravelites, sandstones, and siltstones. The mass-transport related deposits include blocks of 1–30 m in size, composed of island arc basalts and plagiorhyolites, limestones, as well as Aldynbulak basalts (sample Tp-3/2).

2.3.5 Rear part of the Sayan-Tuvan forearc zone – the Systygkhem subzone

The crustal rocks of the Sayan-Tuvan forearc are largely covered by the Khemchik-Systygkhem molasse basin deposits in the Systygkhem subzone and not further considered in this study (Figure 1b–d).

The Chingin, Aldynbulak, and Uttug-Khaia Formations, together with the Tannuola-Khamsara island arc and back-arc complexes are overlain by sedimentary and volcano-sedimentary deposits of late Early Cambrian age (roughly Cambrian Age 2–4, ca. 521–509 Ma; [52] mainly the – Tereshkin, Bayankol, Akdurug, Ilchir, Syynak, Irbitei, Terektig, and similar formations [35,36,40,42]. It has been inferred from stratigraphic reports [35,36] (and references therein), tectonic configurations, and provenance studies [50] that these sediments were formed by extremely proximal sources and accumulated in an active margin forearc-like basin setting [43,53]. Sedimentary, tectonic, and volcanic processes were presumably controlled by unstable subduction, possibly due to subduction slowdown and slab separation (slab window formation) [54].

2.4.1 Ophiolites in the Kurtushiba area

The Kurtushiba ophiolites (Figures 1 and 2) (associated with the Chingin basalts) were likely formed during subduction initiation and the onset of primitive ensimatic island arc formation. They contain younger boninites and older MOR-type basalts [56,57]. The co-occurrence of sheeted dike complexes that exhibit oceanic as well as island arc geochemical signatures [22,57] implies different formation settings of the ophiolitic basalts. Indirect evidence for a multi-generation development of the Kurtushiba ophiolites is given by magmatic breccias that comprise fragments of serpentinites in clinopyroxenites, fragments of pyroxenites and gabbro in plagiogranites, and gabbro and diabase set in a diabasic matrix [57], as well as ultramafic xenoliths in gabbro. In some cases, dunites were intruded by gabbroid and troctolitic rocks which led to the alteration of these dunites to rocks with wehrlitic and clinopyroxenitic composition [44,58].

2.4.2 Khemchik area ophiolites

The formation of the Khemchik ophiolites (Figures 1, 3 and 4) (linked with the Aldynbulak and Uttug-Khaia basalts) also took place during the development of the primitive island arc [37], but there is evidence for an inter-arc or back-arc origin [31,57]. The Khemchik ophiolites differ from the ones occurring in the Kurtushiba zone (Figure 1) by an increased amount of andesite, microdiorite, quartz microdiorite, and plagiogranite dikes [37]. In contrast, only single plagiogranite veins were found in the Kurtushiba ophiolites. The Khemchik ophiolites are characterized by a more pronounced Nb-Ta negative anomaly in comparison to the Kurtushiba ophiolite units [59]. Ar–Ar dating of amphibole from the Shatskii massif gabbro (Figure 3) revealed an age of 578.1 ± 5.6 Ma for the Khemchik ophiolites [37]. It is noteworthy that the sheeted dike complex in the Shatskii massif is not uniform, similar to those of the Kurtushiba ophiolite. The dikes exhibit different orientations and petrologic characteristics, indicating that this ophiolite evolved substantially during different episodes of growth [57].

The distribution of trace elements in the gabbro and dike complexes of the Khemchik ophiolite is similar to that of N-MORB, but the concentration of elements is lower than in N-MORBs, and they show a slight negative Nb anomaly [37,59]. A negative Nb–Ta (or Nb) anomaly is probably one of the most important geochemical signatures of magmas produced in subduction zones [60,61] and is seen, for example, in the forearc magmatic rocks of the Bonin-Mariana arc [62].

2.4.3 Origin of the Kurtushiba and Khemchik ophiolites

So far, it is unclear how the Kurtushiba and Khemchik ophiolites relate to the Tannuola-Khamsara island arc system. Berzin and Kungurtsev [14] and Berzin [43] understand the Kurtushiba and Khemchik ophiolites as part of the accretional prism area that possibly represent “accretion type” ophiolites (sensu [11]). This approach implies that these ophiolites in the Sayan-Tuvan forearc zone may not necessarily origin from the subduction processes adjacent to the Tannuola-Khamsara island arc, but instead represent random oceanic crustal parts of the Paleo-Asian Ocean stacked up during later tectonics. However, there are various indicators which suggest that the Kurtushiba and Khemchik ophiolites are not allochthonous fragments and more likely formed in association with the Tannuola-Khamsara island arc and subduction systems. The geochemical composition of plagiogranites, andesites, and diabases of these ophiolites are very similar or identical with equal rocks found in the Ondum subzone of the Tannuola-Khamsara island arc. The age range of the oldest granites in the Ondum subzone (572–562 Ma) is comparable with that of the Shatskii massif (578 Ma) and the late-early to middle Cambrian stratigraphic overlap assemblages throughout this region contain recycled material of the underlying (ophiolitic) strata [35,36,37,38,39,59] These connections and similarities were best explained, given the tectonostratigraphic arrangement of the ASFB terranes in that area, if the Kurtushiba and Khemchik ophiolites were created in the incipient and expanding proto-arc–forearc region of the Tannuola-Khamsara island arc. Samples of both ophiolite complexes are of forearc plateau basalt affinity [59].

2.4.4 Ophiolites in the back-arc subzone of the Tannuola-Khamsara island arc

The back-arc–forearc subtype of suprasubduction ophiolites (for definition of this type see [31]) is located in the Agardag back-arc subzone (Figure 1b). The age of the Agardag ophiolites has been constrained by dating a plagiogranite dike associated with the gabbro of the ophiolite massif (569.6 ± 1.7 Ma, Pb–Pb dating of zircons) and Sm-Nd age for the gabbro of the Karashat ophiolite massif (546 ± 18 Ma) [63]. The straight nature of the dated plagiogranite dike may also indicate that the host ophiolite is older than 570 Ma [63]. Detrital zircon age spectra from the Terektig Formation of the Agardag subzone do suggest that the Agardag ophiolite is slightly older than 570 Ma [63] (580–574 Ma [64]), which is similar to the age of the Khemchik ophiolites (Shatskii massif, 2.4.2). It has been proposed that basalt dikes and lavas exposed on the northern side of the Teskhem River (Figure 1b) are part of ophiolites [63,65,66]. On the other hand, the lower Cambrian volcanic Karakhol and the archaeocyathid-bearing Terektig formations were mapped (Gibsher et al. 1987 unpublished, map published in [67]) at exactly the position used for the subsequent studies [63,65,66]. The age of the Karakhol and Terektig formations is approximately ∼525–509 Ma [67] or ∼530–520 Ma [64] and according to available geochemical datasets, the volcanic rocks of the Karakhol Formation were formed in an active continental margin setting [67]. Yang et al. [32], however, interpret the Agardag zone as suprasubduction-related terrane. The likely broader involvement of subduction-processes has also been acknowledged by Pfänder and Kröner [63]. According to Dobretsov et al. [16], the Agardag ophiolites formed during rifting of continental lithosphere are comparable to the rifting system of the modern Red Sea. This shows the need for a reevaluation of the Agardag mélange-suture zone and creation of a detailed geological framework in order to understand its complex geological history and relation to adjacent terranes.

The Ediacaran–lower Cambrian Kuskunnug Formation is also part of the Agardag back-arc subzone. There, ocean island basalts (OIB) and E-MORB-like rocks have been identified in the eastern part of the Teskhem River site, as well in blocks within the mélanges of the Agardag site [16,64,66,67]. Considering the age, geographical distribution of these geodynamic units, and their relative location to the Dzhebash accretional zone and Tannuola-Khamsara island arc, it implies that the Ediacaran–early Cambrian Chingin, Aldynbulak (and Uttug-Khaia), and Kuskunnug volcanogenic formations (which all contain enriched basalts) may have formed in a related process, but at different distances in relation to the subduction zone.

Ophiolites of the Kaakhem back-arc subzone (Figure 1b–d) have been found to carry similar geochemical signatures like the enriched back-arc basalts of the Woodlark Basin in the southwestern Pacific Ocean, that was formed during spreading processes involving the subcontinental lithosphere (in the case of the Kaakhem subzone, the Tuva-Mongolian microcontinent) [16]. The lower Cambrian carbonate-terrigenous Tapsa Formation is associated with the Kaakhem subzone [68], which suggests that the age of the Kaakhem ophiolites is comparable to that of the Tapsa Formation.

4.2.1 Major and trace element contents of the basalts

Aldynbulak basalts. The Aldynbulak basalts are alkaline to subalkaline rocks and plotted in the fields of trachybasalt and basalt on a Na₂O + K₂O vs SiO₂ diagram (Figure 7a and b) (after [69]). A large diversity and spread of major elements (low to high aluminum Al₂O₃ = 11.6–21.5 wt%, moderate to ultra-titanium TiO₂ = 1.6–4.5 wt%, low to ultra-potassium K₂O = 0.20–2.63 wt%) mainly reflects the varying geochemical compositions of the different Aldynbulak sampling sites. On a Nb/Th vs Zr/Nb diagram (after [70]), the Aldynbulak basalts occupy a boundary position between the fields of the oceanic island basalts and oceanic plateau basalts (Figure 9a) and in a ternary graph Nb*2−Zr/4−Y (after [71]), the Aldynbulak samples are plotted in the fields of intraplate alkaline basalts and intraplate tholeiites (Figure 9b). The elemental ratios La _n /Yb _n = 3.0–12.8 and Th _n /Yb _n = 2.6–9.4 (Table 1) suggest an OIB and E-MORB-like geochemical composition (OIB: La _n /Yb _n = 11.7 and Th _n /Yb _n = 10.1 [72]; E-MORB: La _n /Yb _n = 2.9 and Th _n /Yb _n = 2.2 [73]). These results are supported by a Tb*3–Th–Ta*2 ternary diagram (Figure 9c) (after [74]) and the spider diagrams (Figure 8a and b) in which the Aldynbulak samples mainly overlap with OIB and E-MORB-like basalts.

Figure 7

Petrochemical diagrams of major elements for the studied forearc basalts. (a) Na₂O + K₂O vs SiO₂ [69]: A – andesite, B – basalt, BA – basaltic andesite, BSN – basanite, BTA – basaltic trachyandesite, PB – picrobasalt, TB – trachybasalt; (b) Al₂O₃ vs MgO; (c) K₂O vs SiO₂ [104]; and (d) TiO₂ vs MgO. Here and further: “Late Chingin” refers to the “Upper Subformation.” Data from Table 1 and supplementary Table S1 were used for the charts in (a)–(d). For additional information the reader is referred to Table S1.

Figure 8

Spider diagrams of chondrite- and primitive mantle-normalized [72] trace element patterns. (a) and (b) Aldynbulak basalts; (c) and (d) Uttug-Khaia basalts and dike KhU-69/12; and (e) and (f) – Chingin basalts and late Chingin basalts.

Uttug-Khaia basalts. The basalts and basaltic andesites (Figure 7) of the Uttug-Khaia Formation are marked by low Al₂O₃ = 13.1–15.2 values. TiO₂ = 1.44–2.44 and K₂O = 0.13–1.13 wt% is on average lower compared to the Aldynbulak basalts (Figure 7b–d). On the discrimination diagram in the study by Meschede [71], they fall in the field on N-MORB and take an intermediate position between N-MORB and volcanic arc basalts on the diagram in the study by Condie [70] (Figure 9a and b). Most samples are plotted as N-MORB in Tb*3–Th–Ta*2 ternary diagram (after [74]); one sample, however, falls in the field of forearc/back-arc basalts (Figure 9c). On spider diagrams (Figure 8), the samples broadly follow the trend of N- and E-MORBs. A negative Nb anomaly is present, but not well defined (Figure 8d). Elemental ratios of La _n /Yb _n = 0.6–0.9 and Th _n /Yb _n = 0.3–0.6 (Table 1) are similar to that of N-MORB: La _n /Yb _n = 0.6 and Th _n /Yb _n = 0.2 [72], but the Uttug-Khaia samples show in general an increased and unusual trace element concentration compared to average N-MORBs (Figure 8c and d).

Figure 9

Discriminant diagrams for the basalt and dike samples. (a) Nb*2 – Zr/4 – Y after [71]; (b) Zr/Nb vs Nb/Th after [70]; and (c) Tb*3–Th–Ta*2 after [74]. Compositional fields of all diagrams: AB – alkaline basalts, Arc – volcanic-arc basalts, CAB – calc-alkaline basalts, CT – continental tholeiites, FBB – forearc and back-arc basalts, IAT – island arc tholeiites, N- and E-MORB – normal and enriched mid-oceanic ridge basalts, OPB – oceanic plateau basalts, OIB – oceanic island basalts, WPAB – intraplate alkaline basalts, WPT – intraplate tholeiites.

Chingin basalts. The petrochemical composition (after [69]) of the Chingin basalts corresponds mainly to basalt (28 samples), less often basaltic andesite (4 samples), and in one case to trachybasalt (Figure 7a). A large scatter of the magnesium content data points are recognized as MgO = 3.2–10.2 wt% (Mg# = 0.35–0.64) which is also in overall higher compared to the Aldynbulak (MgO = 2.0–6.8 wt% and Mg# = 0.19–0.55) and Uttug-Khaia (MgO = 5.8–6.9 wt% and Mg# = 0.44–0.55) samples (Figure 7b and c). The TiO₂ = 0.6–3.1 and K₂O = 0.02–0.99 wt% values are noticeably lower than in the Aldynbulak basalts (Figure 7). The basalts sampled from the upper and lower subformations of the Chingin Formation exhibit an almost identical composition (Figures 8 and 9), even though a high content of TiO₂ = 3.05 wt% and the highest concentration of trace elements was found in the upper subformation (sample KK-18/16*) (Table 1, Figures (6e and f) and (7d)). On the Nb*2−Zr/4−Y (after [71]) and Nb/Th vs Zr/Nb (after [70]) diagrams, the Chingin samples occupy the compositional fields of N-MORB and intraplate tholeiites and ocean plateau basalts, respectively (Figure 9a and b). However, on the Tb*3−Th−Ta*2 ternary diagram (after [74]), the Chingin samples are plotted mainly in the field of E-MORB and ocean island basalts. Three samples are plotted in the field of forearc/back-arc basalts and two samples are even of continental tholeiitic composition (Figure 9c). The E-MORB-like composition also reflected in the trace element ratios La _n /Yb _n = 1.3–4.1 and Th _n /Yb _n = 0.6–2.8 of the Chingin basalts are similar to E-MORBs: La _n /Yb _n = 2.9 and Th _n /Yb _n = 2.2 and T-MORBs (transitional mid-ocean ridge basalts): La _n /Yb _n = 1.1 and Th _n /Yb _n = 0.7 [73] (Table 1) and in the spider diagram (Figure 8e and f).

Uttug-Khaia dikes. The dikes that cut through the Uttug-Khaia Formation basalts (samples KhU-69/12 and KhU-70/12) are trachybasalts and basaltic trachyandesites (Table 1, Figure 7a). These samples have high alumina (Al₂O₃ = 17.1 and 18.8 wt%), moderate and low-magnesium (MgO = 5.0 and 2.1 wt%, Mg# = 0.48 and 0.26), high titanium (TiO₂ = 1.94 and 2.92 wt%), and high potassium (K₂O = 1.51 and 1.57 wt%) contents (Figure 7). Only one sample was available for further trace element analytics (KhU-69/12). On the diagrams after Condie [70] and Meschede [71], this sample is plotted in the field of volcanic-arc basalts or intraplate tholeiites (Figure 9a and b). On the Tb*3−Th−Ta*2 ternary diagram (Figure 9c) (after [74]), this sample, however, is plotted in the field of continental tholeiites. On the spidergrams (Figure 8c and d) and according to La _n /Yb _n = 2.4 and Th _n /Yb _n = 2.7 ratios (Table 1), the dike sample KhU-69/12 is close to the composition of E-MORB (La _n /Yb _n = 2.9 and Th _n /Yb _n = 2.2 [73]). It was noted that the composition of the dike shows a similar trace element pattern and composition to the Chingin basalt samples (Figure 8d and f).

4.2.2 Sm–Nd isotopic composition and assessment of basalt magma sources

To estimate the phase composition of mantle protoliths and the degrees of their partial melting, the element ratios of Lu/Hf vs La/Sm (Figure 10a) (after [75]), (La/Sm) _n vs Zr/Nb (after [76,77,78] (Figure 10b), and Y/Nb vs Zr/Nb (Figure 10c) (after [77,78]) were plotted.

Figure 10

Diagrams reflecting the degree of partial melting and source. (a) Lu/Hf vs La/Sm after [75], (b) (La/Sm) _n vs Zr/Nb after [76,77,78], and (c) Y/Nb vs Zr/Nb after [77,78].

The Aldynbulak basalts plot predominantly in the intermediate region of the two peridotite phases, both showing low degrees of partial melting (Figure 10a). These basalts are further characterized by relatively low positive values of initial ε _Nd (T) = +3.7 to +5.7 (Table 2), which is probably due to the presence of a recycled primitive mantle component in their source [79]. Considering the OIB + E-MORB-like composition of the Aldynbulak basalts, it is likely that an enriched deep mantle source was of primary importance in their petrogenesis. This is also seen on the (La/Sm) _n vs Zr/Nb and Y/Nb vs Zr/Nb diagrams (Figure 10b and c) on which the samples were plotted close to the field of E-MORBs. Note that the Tp-3/2 sample with the highest Lu/Hf ratio = 0.13 is linked with the highest ε _Nd(T) = +5.7 value among the Aldynbulak basalts (Table 2, Figure 10a).

The Chingin basalts are characterized by relatively high positive ε _Nd(T) = +6.7 to +8.3 values, indicating a larger contribution of juvenile mantle components in the source [80]. The ε _Nd values of the Chingin basalts are also close to those of the depleted mantle of the respective age ε _Nd(0.57) = +8.8 [81]. Possibly, the Chingin samples represent a mixture of melts created at high degrees of partial melting of garnet and spinel peridotite (Figure 10a). The (La/Sm) _n vs Zr/Nb and Y/Nb vs Zr/Nb diagrams (Figure 10b and c) place the Chingin basalts between E- and T-MORBs, indicating multiple, including deeply rooted, mantle sources. Also note that the sample KKp-8-12 shows one of the lowest Lu/Hf ratios = 0.10 and the lowest ε _Nd(T) = +6.7 value of the Chingin basalt samples (Table 2).

The Uttug-Khaia basalts are typified by an extraordinary composition. According to the Lu/Hf vs La/Sm diagram, they were formed at high degrees of partial melting of spinel peridotite (Figure 10a), which is consistent with the N-MORB-like distribution of trace elements (Figures 8c and d, 9, and 10b and c). However, the ε _Nd (T) = +6.3 value is lower and the concentrations of Ti and K are noticeably higher than in the E + T-MORB-like Chingin basalts (Table 2, Figure 7). This may suggest that the Aldynbulak basalts also influenced the largely N-MORB-like melt of the Uttug-Khaia basalts.

The E-MORB-like Uttug-Khaia dike (Figure 8c), which is derived from a deep mantle source in the garnet stability zone (Figure 10a), is consistent with relatively high concentrations of titanium, potassium, and alkaline elements (Table 1, Figure 5). The high positive ε _Nd(T) = +7.8 value (Table 2), usually characteristic for the depleted mantle [80], hints a predominantly juvenile magma source for the Uttug-Khaia dike. The combination of the trace element content and Sm-Nd isotopic composition presumably results from the chemical and isotopic heterogeneity of the deep mantle reservoir that is associated with the subduction and recycling of oceanic crust and sediments [79].

The Sm-Nd isotopic composition of a fine-grained volcano-sedimentary rock sample from the Dzhebash subzone was also studied: ε _Nd(T) = +1.2 and T _Nd(DM) = 1,043 Ma (Table 2). A low positive value of initial ε _Nd in this sample may indicate mixing of late Mesoproterozoic–early Neoproterozoic and early Paleozoic isotopic provinces [81].

5.3.1 First stage (580–578 Ma ago)

During the first stage (Figure 11b), about 580–578 Ma ago, the initiation of subduction causes decompression melting of the mantle from garnet peridotite at low degrees of partial melting, which led to the generation of basalts with OIB + E–MORB-like composition (Aldynbulak and Kuskunnug basalts, see Section 2.4), that likely arose as an intra oceanic plateau basalt structure at some distance from the newly-forming trench (Figure 11b), assumed from the stratigraphic and structural position. Yang et al. [32] indicated a “suprasubduction and plume”-type setting for the Kurtushiba and Shatskii massifs; however, no unambiguous evidence for plume magmatism was found in this study.

5.3.2 Second stage (578–572 Ma ago)

About 578–572 Ma ago (Figure 11c), a distinct slab rollback was accompanied by several suprasubduction spreading centers (Figure 11c), where mainly decompression (+fluid flux) melting of the depleted mantle took place. Paleospreading complexes of the Khemchik (∼578 Ma [37]) and Kurtushiba ophiolites were formed during these intervals and possibly, the formation of Uttug-Khaia basalts began. There was likely an additional spreading center close to the Ondum subzone, which evolved into island arc magmatism in the following stages (Figure 11d and e). This is indicated by the earliest plagiogranites of the Ondum island arc subzone with an age of 572–562 Ma and an N-MORB-like distribution of trace element values [38,39]. Data on the age of the Agardag ophiolites (Section 2.4.4) indicate that they also began to form during this period. The suprasubduction crust probably consisted not only of newly formed oceanic crust, but also included remnants of older MOR-type oceanic rocks, which were gradually replaced by suprasubduction melts (primary MOR-type crust was found in the parts of Kurtushiba ophiolites that were presumably located closest to the trench, see Section 2.4).

5.3.3 Third stage (572–562 Ma ago)

The third stage (Figure 11d), at about 572–562 Ma, marks the slab rollback slowdown and transition to normal subduction. During this stage, the formation of forearc complexes continues, and the first island arc and back-arc complexes are established. The basalt lavas that now form the dominant variants are the mixing product of different enriched and depleted melts: N-MORB-like Uttug-Khaia basalts, E-MORB-like Uttug-Khaia dikes, and the E + T-MORB-like Chingin basalts. Apparently, the introduction of boninites also took place at the same time (Figure 11d) in the Kurtushiba zone Section 5.3.5.

The third stage probably also records the transition from predominantly decompression melting to predominantly fluid-flux melting of the newly formed suprasubduction crust. Possibly, the Uttug-Khaia and Chingin basalts were formed during the final stages of a fading forearc spreading zone, while the first island arc complexes, in particular, Ondum plagiogranites, were formed closely to the suprasubduction complexes. The growth of the Agardag ophiolites with an age of ∼570–546 Ma [63] may also be associated with decompression and fluid-flux melting of the depleted mantle in the back-arc spreading center [31,65]. The Agardag zone, as possible suprasubduction ophiolite [32], implies that the basalts and ophiolites evolved in multiple episodes during that time (Kuskunnug basalts and Agardag spreading complexes, see Section 2.4.4, Figure 11b and d)).

It is notable to mention that the duration of these three stages is roughly comparable to the timescales and duration of subduction initiation and subsequent evolution of the IBM system. There, the first basaltic magmatism at subduction initiation was produced by decompression melting of the mantle 52–51 Ma before present. The change in flux melting and boninitic volcanism took about 2–4 million years and the change from fluid-flux melting in counterflowing mantle to “normal” arc magmatism is assumed to have taken 7–8 million years [97].

5.3.4 Fourth stage (562–538 Ma ago)

In the last stage, at about 562–538 Ma (Figure 11e) before present, a normal subduction regime is established. At about the same time, plagiogranite-plagiorhyolite island arc magmatism of the Ondum subzone is accompanied by basalt-andesite-rhyolite and gabbroid island arc magmatism in the Tannuola subzone [54]. The growth of the back-arc ophiolites and basalts continues in the Agardag subzone (gabbro 546 ± 18 Ma [63]) but is also reported from the late Ediacaran Anakhem complex in the Ulugo back-arc subzone [98,99]. Sedimentary deposits began to develop largely across the suprasubduction “basement” in the forearc and back-arc regions.

Ediacaran to early Cambrian sediments of the forearc basin contain single interlayers of the late Chingin basalts. For the Duushkunnug hypabyssal gabbro massif located near the Saryg-Tash site (Figure 1), an age of 537.5 ± 4.9 Ma (Ar–Ar dating on amphibole) was determined, which shows an E-MORB-like composition with weakly pronounced island arc characteristics [100]. In addition, bodies of the Izinziul’ microgabbro-diorite-plagiogranite subvolcanic complex, including E-MORB-like diabase-microgabbros, dated at 538 ± 4 Ma (unpublished data, SHRIMP-II analysis of zircons from plagiogranite on the right bank of the Izinziul’ River), are associated with the Chingin basalts (see Table S1). The age of the Duushkunnug gabbro and our unpublished preliminary data imply a magmatic forearc impulse at about 538 Ma, which is likely linked to the emplacements of the late Chingin basalts. The source(s) of these later E-MORB-like forearc magmatites (compare mantle wedge in Figure 11e) remains uncertain. If their genesis was explained using the “accretionary model,” the gabbro of the Duushkunnug massif, the Izinziul’ complex, and the late Chingin basalts were magmatic bodies that intruded the accretionary prism during subduction (accretion and collision of the oceanic plateau and the Tannuola-Khamsara island arc began at the Ediacaran – Cambrian boundary, although subduction continued until the middle of the early Cambrian [54]). However, the stratigraphic intact arrangement of the late Chingin basalts and the predominant localization of subvolcanic bodies of the Izinziul’ complex on the boundary of the lower and upper Chingin subformations make it unlikely that those magmatics intruded the accretionary prism itself as, e.g., unrelated irregular dikes.

Data by Borisova et al. infer that the chemical evolution of oceanic basaltic magmas depends on (1) the depth of their interaction with the overlying oceanic lithospheric mantle (serpentinized by seawater-derived fluids) and (2) the rate of the basaltic melt transport from their upper mantle source, i.e., the time the oceanic melts interacted with the serpentinized lithospheric mantle [101]. These data are not completely consistent with our proposed model of forearc basalt genesis in the Sayan-Tuvan forearc zone, but they emphasize the importance of oceanic crust thickness for composition of mantle-derived basaltic melts.

5.3.5 Possible causes of mantle melting and the influence of oceanic crust thickness on basalt formation

Recent observations of the Hawaiian plume systems [75] indicate the degree of partial melting and trace element contents may be strongly depended on the thickness of the oceanic crust (but see also Section 5.3.4). In the case of the Sayan-Tuvan forearc zone, this mechanism may have influenced the formation of the enriched early Aldynbulak basalts over thicker crust followed by the depleted Uttug-Khaia basalts and the Chingin basalts (and boninites) and also Uttug-Khaia dikes which formed over the relative thinner parts of the crust caused by spreading. This process has also been suggested to explain the co-occurrence of geochemically different basalts in a similar geodynamic setting in the Paleo-Asian Ocean, the Gorny Altai accretionary prism (Kuznetsk-Altai, Figure 1a) [29]. General decompression melting most likely took place in response to the subduction initiation, slab rollback, and overall thinning of the crust. We argue that the mantle melting was not caused by a mantle plume but by another process of supraregional tectonic nature and scale, which led to the initiation of subduction – followed by decompression melting of the mantle. The Aldynbulak basalts differ from the subsequent generations of Uttug-Khaia and Chingin basalts by lower degrees of partial melting of the mantle and larger proportions of enriched sources. Possibly, the changing basalt geochemistry was caused by a thinning of the oceanic crust over the area of mantle magma generation as a result of suprasubduction spreading.