The South Armenian Block: Gondwanan origin and Tethyan evolution in space and time

The geodynamic evolution of the South Armenian Block (SAB) within the Tethyan realm during the Palaeozoic to present-day is poorly constrained. Much of the SAB is covered by Cenozoic sediments so that the relationships between the SAB and the neighbouring terranes of Central Iran, the Pontides and Taurides are unclear. Here we present new geochronological, palaeomagnetic, and geochemical constraints to shed light on the Gondwanan and Cimmerian provenance of the SAB, timing of its rifting, and geodynamic evolution since the Permian. We report new 40 Ar/ 39 Ar and zircon U-Pb ages and compositional data on magmatic sills and dykes in the Late Devonian sedimentary cover, as well as metamorphic rocks that constitute part of the SAB basement. Zircon age distributions, ranging from (cid:1) 3.6 Ga to 100 Ma, ﬁrmly establish a Gondwanan origin for the SAB. Trondhjemite intrusions into the basement at (cid:1) 263 Ma are consistent with a SW-dipping active continental margin. Maﬁc intraplate intrusions at (cid:1) 246 Ma (OIB) and (cid:1) 234 Ma (P-MORB) in the sedimentary cover likely represent the incipient stages of breakup of the NE Gondwanan margin and opening of the Neotethys. Andesitic dykes at (cid:1) 117 Ma testify to the melting of subduction-modiﬁed lithosphere. In contrast to current interpretations, we show that the SAB should be considered separate from the Taurides, and that the Armenian ophiolite complexes formed chieﬂy in the Eurasian forearc. Based on the new constraints, we provide a geodynamic reconstruction of the SAB since the Permian, in which it started rifting from Gondwana alongside the Pontides, likely reached the Iranian margin in Early Jurassic times, and was subject to episodes of intra-plate ( (cid:1) 189 Ma) and NE-dipping subduction-related ( (cid:1) 117 Ma) magmatism.


Introduction
The present-day tectonic setting of the Arabian-Eurasian collision zone is the result of a complex Late Palaeozoic to Cenozoic geodynamic evolution that is partially preserved in large-scale tectono-stratigraphic terranes stretching from the Mediterranean to Tibet.Integral to this evolution is the Permian breakup of Gondwana and the formation of a collection of microcontinents in the Tethyan realm, termed Cimmeria (S ßengör and Yilmaz, 1981), which drifted away from the NE margin of Gondwana during the Permian-Triassic as the Neotethys Ocean opened (e.g., Stampfli and Borel, 2002;Torsvik and Cocks, 2013).These terranes, presently stretching from Turkey to southern China, have successively amalgamated to the southern Eurasian continental margin during the Mesozoic and Cenozoic, closing the Palaeotethys and Neotethys Oceans.
One of these Gondwana-derived fragments is the South Armenian Block (SAB; Knipper and Khain, 1980), a continental fragment presently separated from the former southern Eurasian margin by the Sevan-Akera suture zone in the north and east, and juxtaposed along another ophiolite-bearing suture zone against the easternmost Taurides and Iran to the south (Fig. 1; Knipper, 1975;Adamia et al., 1981).Its kinematic evolution within the Tethyan realm between the Permian and Late Cretaceous remains enigmatic, chiefly because of the limited amount of available geological evidence.Although its Gondwanan origin has long been inferred (Belov and Sokolov, 1973;Aghamalyan, 1978), its affinity with neighbouring Gondwana-derived terranes, especially Central Iran, the Pontides and the Taurides, is not well-understood.In the absence of palaeomagnetic constraints on the position of the SAB during its northward drift, it has been interpreted as a contiguous part of Iran (e.g., Stampfli et al., 1991;Brunet et al., 2003;Adamia et al., 2017), the Taurides (e.g., Okay and Tüysüz, 1999;Barrier and Vrielynck, 2008;Rolland et al., 2012;Meijers et al., 2015), and as a separate micro-continent (van Hinsbergen et al., 2020).Moreover, no unequivocal constraints have yet been placed on the timing of rifting of the SAB from the Gondwanan margin and, as a result, the inferred ages range from Late Permian ($260 Ma) to Early Jurassic ($174 Ma) (S ßengör and Yilmaz, 1981;Mart, 1987;Gealey, 1988;Kazmin, 1991;Bazhenov et al., 1996;Stampfli and Borel, 2002;Robertson et al., 2004;Moix et al., 2008).
Currently no consensus exists on the provenance and geodynamic evolution of the SAB within the Tethyan realm.Key questions yet to be answered by observational data include: (1) When did the SAB start drifting from the Gondwanan margin?; (2) What is its relation to the neighbouring terranes of present-day Turkey and Iran?; and (3) How did the SAB evolve and interact in the Mesozoic Tethyan realm?Here, we present the first U-Pb geochronological and trace-element data on zircons, coupled with geochemical compositions of their metamorphic host rocks that I.K. Nikogosian, Antoine J.J. Bracco Gartner, Paul R.D. Mason et al. Gondwana Research 121 (2023) 168-195 constitute part of the uplifted SAB basement, as well as 40 Ar/ 39 Ar ages and Sr-Nd-Pb isotope and geochemical compositions of hitherto unreported mafic to intermediate Mesozoic magmatism in the Late Devonian sedimentary cover of the SAB.These new results are used to reconstruct the geodynamic history of the SAB and interpret its evolution in the context of the Permian-Triassic breakup of the NE Gondwanan margin and the Mesozoic kinematic history of the Tethyan realm.

Geological setting
The Middle East-Caucasus region comprises a series of Gondwana-derived continental blocks that successively accreted to the southern Eurasian margin upon closure of the Palaeotethys and Neotethys Oceans in Palaeozoic to Cenozoic times (e.g., Barrier and Vrielynck, 2008).Here the focus is on the SAB, the central region between the contrasting palaeotectonic settings of Turkey and Iran (Fig. 1).

Turkey
Turkey's tectonic history is markedly different from that of Iran.Two continent-derived fold-and-thrust belts-the Pontides and the Anatolide-Tauride block-are separated by the Izmir-Ankara suture zone (Fig. 1).The Pontides has been part of the southern Eurasian margin since at least Jurassic time (Dokuz et al., 2017), and is thought to have drifted away from Gondwana during the Late Triassic ($240 Ma) reaching the southern Eurasian margin during the Early Jurassic, and opening the Neotethyan Ocean in its wake ($180 Ma;S ßengör and Yilmaz, 1981;van Hinsbergen et al., 2020).Continental lithosphere of the Anatolide-Tauride block, in central and southern Turkey, separated from Gondwana during the Early Jurassic (200-190 Ma;van Hinsbergen et al., 2020).The Anatolides comprise several metamorphosed and exhumed mas-sifs, such as the Kırs ßehir block, Tavs ßanlı zone, Afyon zone, and the Menderes massif.In contrast, the Taurides are composed of mainly non-metamorphosed sedimentary rocks in the form of a thin-skinned fold and thrust belt.Both Anatolide and Tauride units are buried below Late Cretaceous ophiolites (Özgül, 1984), where the Taurides host the non-metamorphosed, foreland accreted equivalents of some of the Anatolide massifs that were buried deeper (van Hinsbergen et al., 2016).The Pontides were once separated from the Anatolide-Tauride block by one or more strands of the Neotethys Ocean, relics of which are found in the form of mélanges and in ophiolites throughout Turkey (e.g., Yilmaz and Yilmaz, 2013;Dilek and Furnes, 2019).Much of eastern Anatolia is known as the Eastern Turkish High Plateau, which has long been described as a subduction-accretion prism (S ßengör and Yilmaz, 1981;S ßengör et al., 2019a).However, below the widespread Upper Neogene volcanic rocks of that plateau are Paleozoic to Cretaceous metamorphosed and non-metamorphosed continental rocks that are overlain by Cretaceous ophiolites and intruded by Upper Cretaceous and younger volcanic arc rocks (Kuscu et al., 2010;Yilmaz et al., 2010;Topuz et al., 2017).These rocks are equivalent to the northern Taurides and show that continental crust of the Taurides continues to the Iran-Turkey border (van Hinsbergen et al., 2020).

South Armenian Block
The Lesser Caucasus and Armenian Highland form a central region between the contrasting palaeotectonic systems of Turkey and Iran.These regions were separated by a plate boundary since Permian times (Stampfli and Borel, 2002), likely in the form of a major transform fault system.Palaeogeographic reconstructions suggest that this fault system could have been reactivated since the Early Eocene (Barrier and Vrielynck, 2008).The Arax valley fault (Fig. 1c; Jackson and McKenzie, 1984) has been suggested to be a vestige of this transform system, based on substantial differences in convergence between the Lesser Caucasus and Talysh of NW Iran (van der Boon et al., 2018).
In recent years, the SAB is often assumed to be part of the Anatolide-Tauride block, based on similarities in basement, stratigraphy and especially the obduction ages for the Jurassic Sevan-Akera ophiolites in the Late Cretaceous (e.g., Barrier and Vrielynck, 2008;Rolland et al., 2012;Hässig et al., 2013a;Meijers et al., 2015;Menant et al., 2016;Hässig et al., 2017).In a Mesozoic-Cenozoic kinematic reconstruction of the Mediterranean, van Hinsbergen et al. ( 2020) provide a comprehensive interpretation of the SAB's tectonic contacts with neighbouring blocks and proposed an alternative model.They infer that a suture zone (Yilmaz et al., 2014), termed Kag ˘ızman-Khoy suture, separates the SAB from the easternmost Taurides.It consists of a belt of ophiolites extending from the Kag ˘ızman ophiolite (easternmost Turkey) to the Khoy ophiolite (NW Iran near the Turkish border).Two different subduction systems were involved: the first was intra-oceanic and emplaced Cretaceous supra-subduction zone ophiolites onto the Taurides (causing $78-83 Ma metamorphism; Topuz et al., 2017), whereas the second closed the ocean basin and juxtaposed the easternmost Taurides and the SAB.Van Hinsbergen et al. (2020) note that the latter closure likely happened after $75-80 Ma, when the SAB collided with the Transcaucasus (Rolland et al., 2012) and as a consequence the subduction system moved south.
Van Hinsbergen et al. (2020) further interpreted that the SAB detached from the Gondwanan continent at $240 Ma as part of the Pontides, and separated it from the Pontides by a ridge jump at $230 Ma.In contrast, uncoupling of the Taurides from Gondwana did not start earlier than $190-200 Ma (van Hinsbergen et al., 2020).
The Khor Virap section, located 30 km south of Yerevan (Fig. 3b), consists of Upper Devonian-Lower Carboniferous shallow marine carbonates and siliciclastics (Ginter et al., 2011) with several distinct meter-scale mafic sills exposed on the hills near the Khor Virap monastery (Fig. 3).The section forms part of the Ararat depression, which is filled mostly by Quaternary lacustrine sediments, and represents an uplifted unit situated on a horst (Milanovsky, 1968) or related to local contractional tectonics associated with thrust faulting (Avagyan et al., 2015).The three sills (samples KV-1, KV-2, KV-3) lie conformably within the Famennian sedimentary units (Fig. 3b; Ginter et al., 2011).KV-3 is sampled from the largest well-exposed sill ($4 m thick), located between limestone (upper contact) and siliciclastics (lower contact) with well-pronounced thermal contacts on both sides (Fig. 3d).The KV-1 sill as well as an adjacent quartzite (metamorphosed Upper Devonian sandstone) were studied to obtain palaeomagnetic data (details in Supplementary Data S1).

Whole-rock elements and isotopes
Whole-rock compositions of the studied samples were obtained by X-ray fluorescence spectrometry (XRF; for major elements) and inductively coupled plasma mass spectrometry (ICP-MS; for trace elements) at the Vrije Universiteit Amsterdam, using a Philips PW1404/10 XRF and Thermo Electron X-series-II ICP-MS, following the procedure of Klaver et al. (2017) and a modified procedure after Eggins et al. (1997), respectively.USGS reference material BHVO-2 was used as a secondary standard throughout ICP-MS analysis and indicated accuracy within 10% of the GeoReM preferred values (Jochum et al., 2016) for all reported trace elements (Supplementary Data S3).Uncertainties are typically < 2% (2 RSD) for major oxides and < 5% (2 RSD) for trace elements.
Isotope analyses for U, Pb and Sm, as well as Nd in samples ARP-3 and ARP-4, were conducted using a Thermo Fisher Neptune Multicollector (MC-)ICP-MS, following standard procedures (Font et al., 2012).Neodymium (for samples KV-1,2 and ARP-1, due to low Nd contents) and Sr isotope compositions were measured by thermal ionisation mass spectrometry (TIMS) using a Thermo Scientific Triton Plus instrument, following procedures outlined in Koornneef et al. (2013) and Klaver et al. (2015).Rubidium was measured by TIMS on a Finnigan MAT 262 RPQ-plus running in static mode.Small, 20-30 ng aliquots of synthetic reference materials (JNdi-1, NBS-987 Sr and NBS-984 Rb) were measured alongside the respective samples to check for accuracy and reproducibility ( 143 Nd/ 144 -Nd = 0.512101 ± 0.000011 2SD, n = 2; 87 Sr/ 86 Sr = 0.710254 ± 0.000021, n = 3; ( 85 Rb/ 87 Rb) raw = 2.627 ± 0.006, n = 4; respectively).Total procedural blanks (from sample dissolution to loading on filament) were 22 pg for Sr, 100 pg for Rb, 0.04 pg for Nd and 42 pg for Pb, and are insignificant in comparison to the amounts of respective elements analysed.USGS reference material BHVO-2 was included throughout to validate the chromatographic and analytical procedures.
Age corrections were applied to the isotopic values to allow evaluation of initial ratios (denoted with the subscript i).Parental/daughter ratios were calculated using atomic and isotopic constants by De Laeter et al., (2003) and age corrections were performed using conventional decay constants.Initial e Nd values [(e Nd ) i ] were calculated assuming ( 143 Nd/ 144 Nd) o CHUR = 0.512638 (Bouvier et al., 2008) and ( 147 Sm/ 144 Nd) o CHUR = 0.1967 (Jacobsen and Wasserburg, 1980).

Zircon trace elements and U-Pb isotopes
Zircon grains were separated from the crushed fractions of samples TSK-1,3,5,7 and ARP-1,3,5 by conventional gravimetric (heavy liquid) and magnetic separation techniques at the Vrije Universiteit Amsterdam.They were handpicked under a binocular micro-scope, mounted in epoxy resin and polished to expose grain centres for characterisation of internal structures by backscattered electron (BSE) and cathodoluminescence (CL) imaging at Utrecht University.
Zircon trace element and U-Pb isotope analyses were obtained by laser ablation (LA) ICP-MS, using a GeoLas 200Q Excimer laser ablation system (193-nm wavelength) coupled to a Thermo Finnigan Element 2 sector field ICP-MS instrument, at Utrecht University.
The laser was operated using a spot size of 20-40 lm, a pulse repetition rate of 10 Hz and an energy density of 5 J/cm 2 .For the traceelement analyses, offline time-integrated normalisation for instrumental drift was performed using the GLITTER software.For the trace-element analyses of zircons, the samples were calibrated to the NIST SRM 612 glass standard using an assumed fixed concentration of 32.45 wt% SiO 2 (151,684 ppm Si) in zircon (Anczkiewicz et al., 2001) measured on the interference-free isotope 29 Si.For U-Pb dating, the four Pb isotopes, 232 Th, 235 U and 238 U were measured.The 91500 zircon reference material (1,065 Ma; Wiedenbeck et al., 2004) was used as an external standard.Results for 91500 zircon ages yielded mean age of 1,065 ± 7 Ma (95% confidence level, MSDW = 0.72, probability = 0.92; Supplementary Fig. S1).Uncertainties for trace elements in zircon are typically smaller than 10% (2 RSD; Mason et al., 2008).U-Pb concordia diagrams, probability density plots and weighted averages were calculated using Isoplot 4. Cathodoluminescence images of the zircons with spot locations are in Supplementary Data S4.

40 Ar/ 39 Ar dating
Samples KV-1, KV-2, ARP-3, and ARP-4 were crushed with a rock splitter and jaw crusher and consecutively washed and thoroughly cleaned in an ultrasonic bath.The samples were sieved into different size fractions.Phenocrysts were separated from the groundmass for sample ARP-4 and $100 mg groundmass was irradiated together with a Drachenfels sanidine fluence monitor (25.

Table 1
Metamorphic basement and igneous intrusive rock sample information.Tectonic units refer to the legend shown in Fig. 2. [1]   In addition, two plagioclase samples (KV-2, ARP-3) and one amphibole sample (KV-1) were separated using standard heavy liquid and magnetic mineral separation procedures and $100 mg of each sample was irradiated in the same irradiation.All samples underwent a final handpicking step under an optical microscope before irradiation.

Metamorphic basement rocks
Descriptions and mineral contents of the SAB metamorphic basement rock samples are given in Table 1.Detailed petrographic observations of the rocks, as well as their isotope, major and trace element geochemistry, are provided in Supplementary Data S2 and S3, respectively.Two of the samples, TSK-3 and TSK-7, represent igneous intrusive bodies that penetrated the basement of the SAB.

Zircon geochronology and geochemistry
A total of 130 spots on 58 zircon grains from samples TSK-1,3,5,7 was analysed for U-Pb ages by LA-ICP-MS (Supplementary Data S3 and S4).Back-scattered electron (BSE) and transmitted light images of representative zircons are shown in Fig. 6.The results for magmatic samples (TSK-3 and TSK-7) are plotted on weighted average (WA) and concordia diagrams (Fig. 7).For detrital samples (TSK-1 and TSK-5), the results are presented on probability/density plots in Fig. 8 (with all uncertainties are given at the 2r level).In the following discussion, U-Pb ( 238 U/ 206 Pb) ages are used for zircons younger than 1.0 Ga, and Pb-Pb ( 207 Pb/ 206 Pb) ages are used for older zircons.
The U-Pb zircon ages for granite gneiss sample TSK-3 spread along the concordia from $ 600 to 440 Ma (Fig. 7).Core ages range from 600 to 520 Ma (WA = 540 ± 9 Ma), whereas rim ages range from 520 to 440 Ma (WA = 461 ± 14 Ma).The probability-density diagram (not shown) shows a peak at $540 Ma for the zircon cores.This age is defined by the most concordant (98-102%) zircons, suggesting that this is the age of granite formation.
The U-Pb zircon ages for trondhjemite sample TSK-7 yield a WA age of 262.5 ± 4.3 Ma (Fig. 7c) and concordia intercept of 262.2 ± 5.0 Ma (Fig. 7d).The former is taken as the intrusion age.
Mica schist TSK-1 and metaarkose phyllite TSK-5 are host to a substantial number of detrital zircons.U-Pb and Pb-Pb zircon ages show considerable variability, ranging from 3,650 Ma to 86 Ma (Fig. 8).These zircons can be divided into 5 age groups.Remarkably, the core of one grain from TSK-1 yields Eoarchean (3,650 Ma) ages (photo in Fig. 6).A Neoarchean-Palaeoproterozoic group comprising mostly TSK-1 zircons and one TSK-5 zircon, defines an age peak at $ 2,500 Ma.Another Palaeoproterozoic group (1,655-1,850 Ma) is observed, consisting only of TSK-1 zircons, as is a Meso-Neoproterozoic group at 970-1,040 Ma.The majority of the zircons, both from TSK-1 and TSK-5, constitute a Neoproterozoic group, ranging from 527 to 850 Ma.This group defines a marked peak at $600 Ma.TSK-1 also hosts a few Cretaceous (86-120 Ma) zircons.
All zircon grains were analysed for trace elements by LA-ICP-MS (Supplementary Data S3 and S4).Virtually all zircons have Th/U ratios between 0.3 and 1 (Fig. 9a), indicative of magmatic zircons (Teipel et al., 2004 and references therein).The overall range in Y content versus U/Yb ratios plot entirely within the field of continent-derived zircons (Grimes et al., 2007) and predominantly within that of continental granitoids (Fig. 9b) (Ballard et al., 2002;Belousova et al., 2006).Rare-earth element concentrations for zircons, normalised to C1 chondrite (McDonough and Sun, 1995), are typical of growth under magmatic conditions, with TSK-3 and TSK-1,5 showing minor negative Eu anomalies (Fig. 9c,d).TSK-7 has no negative Eu anomaly.They collectively indicate HREE enrichment up to about 5,000 times C1 chondrite, with marked HREE fractionation (Yb N /Gd N > 1).
Andesites ARP-3 and ARP-5 collectively have different primitive mantle-normalised patterns (Fig. 10c) and are relatively enriched in the most incompatible trace elements.They show patterns similar to P-MORB, except for substantial negative Nb-Ta and Ti anomalies, slight positive Th and U anomalies and higher contents in several LILE (e.g., Ba, Rb, Cs), and can therefore be classified as subduction-related.
Sr-Nd-Pb isotope compositions of samples KV-1, KV-2, ARP-1, ARP-3 and ARP-4 are shown in Fig. 10e-f (and listed in Supplementary Data S3).Age-corrected values were calculated using the age data from the same samples (next section).Arpi rocks show restricted ( 87 Sr/ 86 Sr) i values, ranging from 0.70452 to 0.70511, whereas Khor Virap basalts have slightly higher values (0.70535-0.70642).The highest value corresponds to KV-2, which also shows a distinct trace elemental composition (P-MORB affinity).
Two replicate heating experiments of amphibole KV-1 (VU109-I3) were performed.The first sample (VU109-I3_1) yields a ''plateau" age of 188.5 ± 1.1 Ma (Fig. 11a) for the steps with higher radiogenic yields (>59%), but contains only 29.1% of the 39 Ar K released.The atmospheric isochron intercept overlaps with air at 2r (289.0 ± 10.3).The majority of the individual heating steps in the full age spectrum range between 169 and 189 Ma.The second experiment (VU109-I3_2) yields a ''plateau" of 187.3 ± 0.8 Ma for the middle part of the heating spectrum ( 39 Ar K = 40.1%;MSWD = 2.7).The inverse isochron age is identical at 188.4 ± 1.4 Ma with an Two replicate heating experiments of plagioclase sample KV-2 (VU118-I1) show disturbed age spectra with ages starting from $110 Ma, increasing to $245 Ma (I1a) and $250 Ma (I1b) and decreasing to $185 Ma.Note that the initial age step of $110 Ma is remarkably close to the magmatic event observable in the Arpi area (samples ARP-3 and ARP-5) and Darasham (Khanzatian, 1992).The preferred age is 233.7 ± 5.1 Ma (Fig. 11b), comprising 52.9% of the total 39 Ar K .
For plagioclase of sample ARP-3 (VU109-I4) one incremental heating experiment was performed.The sample shows a decreasing age spectrum.Six consecutive lower temperature heating steps yield a weighted mean age of 124.5 ± 0.5 Ma, comprising only 18.5% of the total 39 Ar K .Four consecutive higher temperature heating steps seem to define a ''plateau" of 112.8 ± 0.5 Ma (Fig. 11c; comprising 28.9% 39 Ar K ) with an atmospheric 40 Ar/ 36 Ar intercept (296.3 ± 9.5).This result is in good agreement with the U-Pb zircon ages of 116.7 ± 1.5 Ma for ARP-3 and 115.0 ± 1.4 Ma for ARP-5.
Two incremental heating experiments were performed on the groundmass of sample ARP-4 (VU109-I5).The first experiment did not result in a reliable plateau, but shows an increase in age followed by a decrease from $80 Ma to $230 Ma to $210 Ma.The second experiment shows similar behaviour, resulting in ages from $83 Ma to $247 Ma to $218 Ma.Note that the initial heating steps of 85 Ma are remarkably close to the SAB ophiolite obduction age (Rolland et al., 2009;Hässig et al., 2016a, 2016b, 2017, Rolland et al., 2020).The radiogenic yields are high and an isochron could not be defined due to clustering of data points.Weighted mean ages of the four oldest consecutive steps yielded 230.9 ± 1.5 Ma (MSWD = 35.9; 39Ar K = 34.0%, 40Ar* = 88.6%;K/Ca = 0.116 ± 0.007) and 245.4 ± 1.4 Ma (Fig. 11d; MSWD = 7.4; 39 Ar K = 28.5%,40 Ar* = 88.4%;K/Ca = 0.143 ± 0.011), respectively.The latter age is preferred and is in remarkable agreement with the U-Pb zircon age of 246.0 ± 3.3 Ma for ARP-1.

Zircon geochronology
In addition to 40 Ar/ 39 Ar dating, samples containing zircons were also U-Pb dated, considering the altered nature of the samples studied.A total of 35 zircon grains from samples ARP-1,3,4,5 and KV-1,2,3 were analysed for U-Pb ages by LA-ICP-MS (Supplementary Data S3).The results for ARP-1,3,5 are reported in concordia diagrams and weighted average (WA) plots (Fig. 12).The mean U-Pb zircon age is 246.0 ± 3.3 Ma for ARP-1; 116.7 ± 1.5 Ma for ARP-3; and 115.0 ± 1.4 Ma for ARP-5.The age for ARP-3 is in good agreement with the 40 Ar/ 39 Ar plateau age of 112.8 ± 0.5 Ma.The zircons in this sample were also analysed for trace elements by LA-ICP-MS (Supplementary Data S3).
Samples KV-1,2,3 and ARP-1,4 also contain zircons that are marked by U-Pb ages ranging from about 440 to 1,848 Ma (Supplementary Data S3), far exceeding the 40 Ar/ 39 Ar ages of their host rocks as well as the age of the sediments that host the intrusions.They show remarkable overlap with the metamorphic basement zircons on a density/probability plot (Supplementary Fig. S2) and  (Grimes et al., 2007) and continental granitoids (green) (Ballard et al., 2002;Belousova et al., 2006).Note that all zircons are continent-derived.Rare-earth element concentrations for (c) magmatic (TSK-3, TSK-7) and (d) detrital zircons (TSK-1, TSK-5) normalised to C1 chondrite (McDonough and Sun, 1995).(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)can therefore be considered as being ''xenogenic", i.e., entrained during magma ascent.
Most of these Gondwana-derived terranes contain evidence for Late Ediacaran-Early Cambrian magmatism, thought to be associated with a widespread continental arc along the northern margin of newly-formed Gondwana (Gessner et al., 2001;Ramezani and Tucker, 2003;Hassanzadeh et al., 2008).Evidence for this magmatism is found in the neighbouring terranes of Central Iran, where Neoproterozoic magmatic products have been found in the basement (Ramezani and Tucker, 2003;Hassanzadeh et al., 2008), and the Menderes massif of the Anatolide-Tauride platform, where active-margin type granites of Ediacaran-Cambrian age are exposed (Gessner et al., 2001).The magmatic zircons of mica granite-gneiss TSK-3, a metamorphosed intrusion into the SAB basement (Fig. 8), is consistent with this extensive magmatic episode in Ediacaran-Early Cambrian times.Its whole-rock composition is similar to typical volcanic arc granites and suggests formation in an active continental margin (Fig. 5).

Cimmerian continent: Rift initiation in the SAB
The Cimmerian continent was originally coined by S ßengör and Yilmaz (1981) as an arc ripped from the NE margin of Gondwana above a SW-dipping Palaeotethys subduction zone in Late Permian-Early Triassic times.Recent evidence suggests that the opening of the Neotethys might have occurred as a result of back-arc spreading (S ßengör et al., 2019b), as opposed to Atlantic-type continental margins on both sides of the Cimmerian continental ribbon.Since its original discovery, a substantial amount of Permian-Triassic rift-related magmatism has been identified along the collision belt, from Iran to China (e.g., Lapierre et al., 2004Lapierre et al., , 2007;;Chauvet et al., 2008;Shellnutt et al., 2014;Shakerardakani et al., 2018;Wang et al., 2019;Zeng et al., 2019).The north-westernmost end of this continental ribbon appears to lie east of the Taurides (NE Turkey), as there Cretaceous ophiolites are found overthrusted by Weighted mean (''plateau") ages are reported with 2r analytical uncertainty.Corresponding zircon ages from this study are indicated in (c) and (d).The ages of relevant large-scale tectonic events are also indicated.Detailed analytical data are provided in Supplementary Data S5.

Table 2
Summary of 40 Ar/ 39 Ar results.MSWD = mean square weighted deviate; N = number of steps included (excluded) in the plateau age; 39 Ar K (%) = percentage of 39 Ar K released by plateau steps.Uncertainties are given at the 2r level.Ages in bold face are the preferred ages.Detailed analytical data are provided in Supplementary Data S5.

Sample
Irradiation Jurassic ophiolites of the southern Pontide margin during the Cenozoic (Topuz et al., 2013b).Details on the movement of the SAB within the Tethyan realm after its separation from Gondwana are lacking except for a single palaeomagnetic study on volcanics in southern Nakhichevan (Bazhenov et al., 1996), which positioned the SAB (21.4°N ± 3.7°) at the African margin around the Early Jurassic, founded on results from four sites yielding positive fold and conglomerate tests, and a rock age inferred from geological mapping and stratigraphic relationships with sediment suites.Limestones and dolomites in the Julfa area, thought to be Middle-Upper Triassic (Karyakin, 1989), have later turned out to be Lower Triassic based on fossil fauna (Grigoryan, 1990).These rocks are overlain discordantly by pre-sumed Lower Jurassic (devoid of any fossil fauna) and Middle Jurassic sedimentary sequences (Azizbekov, 1962;Grachev and Karyakin, 1983).This revised stratigraphic interpretation implies that the volcanics studied by Bazhenov et al. (1996) can be any age between 247 and 174 Ma.The uncertainty questions the robustness of earlier geodynamic reconstructions, which have often relied on this single palaeomagnetic constraint.Due to the lack of local geological evidence, existing geodynamic views for the Mesozoic rifting evolution of the SAB often depend on an assumed association with a neighbouring terrane.One group considers the SAB to be a contiguous part of the Anatolide-Taurides block, which did not start rifting until the Early Jurassic (e.g., Okay and Tüysüz, 1999;Barrier and Vrielynck, 2008;Rolland et al., 2012;Rolland, 2017).Another group links it to Central Iran (e.g., Stampfli et al., 1991;Brunet et al., 2003;Adamia et al., 2017), which began to drift northward in Early Permian times as part of the Cimmerian blocks and reached Eurasia during the Late Triassic (Zanchi et al., 2009, and references therein).More recently, the SAB has been interpreted as an isolated microcontinent that, together with the Pontides, drifted away from the Taurides during the Triassic (van Hinsbergen et al., 2020).

Middle-Late Permian trondhjemite intrusions at Tsakhkunyats
The trondhjemite (''plagiogranite") intruded into the previously-metamorphosed basement of the SAB at 262.5 ± 4.3 M a.Based on the same Middle-Late Permian trondhjemite suite, Galoyan et al. (2020) surmised the existence of a long-lived S-dipping subduction zone by linking the petrogenesis of these rocks to the Carboniferous subduction zone that generated metagranites in the Afyon zone in western Turkey (Candan et al., 2016), though it is unclear whether the disparate occurrences are related.Our palaeomagnetic data for this sample (Supplementary Data S1) point to a position at the NE margin of Gondwana at this time, next to the Pontides and the Iranian blocks (Fig. 14b).Its geochemical signature (Fig. 5) suggests magma genesis in an active continental margin.The trondhjemite is characterised by significant enrichment in Na over K (Na 2 O/ K 2 O = 7), relatively high Sr (302 ppm), low Y (1.3 ppm) and Yb (0.2 ppm) and high Sr/Y (234), similar to adakitic melts.High Na 2 O/K 2 O ratios and incompatible-element patterns of the studied sample, as well as of samples from the same trondhjemite suite reported by Galoyan et al. (2020), more closely resemble modern subduction-related adakites and are different from thickened lower continental crust-derived adakites based on (La/Yb) N vs. (Ba/Zr) N (Fig. 5e).Sr-Nd isotope systematics (Fig. 5f) further demonstrate the affinity of the Tsakhkunyats trondhjemites to compositional fields of subduction-derived adakites, rather than lower continental crust adakites.
Middle-Late Permian volcanic arc-type intrusions into the SAB imply an active SW-dipping subduction at the NE margin of Gondwana at that time, whereby the SAB was part of the overriding plate (Fig. 15).This inference is in good agreement with recent evidence of the presence of arcs recorded as Upper Permian-Lower Triassic rocks across the Cimmerian continent, which became dispersed during the Alpine evolution (S ßengör et al., 2019a).Similar Upper Permian-Lower Triassic arc-related rocks have been documented for the Pontides and Sakarya Zone (e.g., Eyuboglu et al., 2011;Karsli et al., 2016;Topuz et al., 2018), but are absent in the Anatolide-Tauride block.This observation corroborates the palaeo-position of the SAB as the SE extension of the Pontides during the Gondwanan assembly (Fig. 15).(Aysal et al., 2012;Ustaömer et al., 2012), (c) Pontides (Ustaömer et al., 2013;Okay et al., 2014), (d) Central Iran (incl., Sanandaj-Sirjan zone; Fergusson et al., 2016;Chiu et al., 2017), and (e) the SAB (this study; detrital zircons from samples TSK-1,5).Kent and Irving (2010) and Torsvik et al. (2012).Grey data shows the secondary component of TSK-7.Black arrows and green data points show restored vertical axis rotation related to the counterclockwise rotation of the SAB (see section 6.2 for further explanation).(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
It is generally assumed that sills and dykes, similar to the ones reported here, represent the plumbing systems of ascending mantle-derived magmas (e.g., Coetzee and Kisters, 2017).The studied igneous rocks (section 3.2) likely represent parts of former feeder dykes or remnants of fissure eruption conduits.It is noteworthy that outcrops of these subvolcanic bodies abound only in a relatively small area ($1,500 km 2 ) of Upper Palaeozoic sequences of the SAB in Armenia and Nakhichevan, but that they probably signal an episode of magmatism at a much wider scale, since similar dyke and sill swarms are widespread in neighbouring Gondwana-derived units (e.g., Jones et al., 2001;Gaggero et al., 2012;Xu et al., 2016;Svensen et al., 2018;Wang et al., 2019).Particularly in Iran, a wide assortment of mafic subvolcanic intrusions in Upper Devonian and Lower Carboniferous sequences has been documented: in the Azerbaijan province (NW Iran; Alavi and Bolourchi, 1973), the Nain-Kerman region (Central Iran; Wendt et al., 2002;Hairapetian and Yazdi, 2003) and the Alborz Mountains (N-NE Iran; Ghavidel-Syooki, 1994, 1995;Mahmudy Gharaie, 2002;Ghavidel-Syooki and Owens, 2007).Radiometric age dates and geochemical details are generally lacking, but many have been linked to large-scale trap volcanism (Mahmudy Gharaie  , 2004) and are thought to be considerably younger than the Upper Devonian country rocks (Wendt et al., 2002).Available data for one location confirm an intraplate (OIB-type) affinity (Mahmudy Gharaie, 2002).
The transition from OIB-to P-MORB-type magmatism within the short time interval of $10 Ma is consistent with a change from a more enriched to a more depleted asthenospheric mantle source during melting at gradually shallower levels.Such associations of MORBs variably enriched by OIB-type components are typical for many peri-Mediterranean ophiolite complexes (e.g., Saccani and Photiades, 2005, and references therein), the Kermanshah ophiolite and Sistan suture zone in Iran (Saccani et al., 2010(Saccani et al., , 2013a)), Oman ophiolites (Lapierre et al., 2004;Chauvet et al., 2011) and modern ocean basins (e.g., Le Roex et al., 1983, 1985;Haase and Devey, 1996).The generation of basaltic melts with OIB to MORB signatures is often an expression of asthenospheric upwelling and lithospheric extension that accompany initial continental rifting and subsequent (incipient) oceanic spreading (McKenzie and Bickle, 1988;Gorring et al., 2003;Saccani et al., 2013a).
The intrusions in the sedimentary cover of the SAB at Arpi, Darasham and Khor Virap (Fig. 1) between $246 and 234 Ma thus suggest that the SAB experienced an episode of extension during latest Middle Triassic times.Together with the slightly earlier emplacement of arc-related granitoids in the metamorphic basement at $262 Ma (closer to the active NE margin), this provides time constraints on rift initiation in the region.We infer that the asthenosphere-derived magmatic bodies record the incipient stage of breakup of the NE margin of Gondwana in the SAB region, and hence the opening of the Neotethys Ocean between the SAB and Africa in Middle Triassic times (Fig. 16).
The palaeomagnetic declinations of ARP-1,4 ($246 Ma) and KV-1 ($189 Ma) (Supplementary Data S1) plot between the values inferred for Africa and Eurasia (Fig. 14a).If we account for the assumed counterclockwise rotation of the SAB during its drift from the African to the Eurasian position, the declinations are more in agreement with Africa before 250 Ma, and more with Eurasia after 250 Ma (Fig. 14a).Accordingly, the results suggest that the rotation relative to Africa started before 250 Ma, and that the rotation relative to Eurasia was accomplished at 246 Ma (green and purple dots in Fig. 14).Additional support for this interpretation comes from the elongation of directional data, where declination deviations at 246 Ma might suggest rotational motion during that time and more latitudinal motions at 189 Ma led to inclination deviations (Fig. 14).Overall, the palaeomagnetic data are consistent with a rotational movement of the SAB from a position juxtaposed north of Africa and south of Eurasia (next to the Pontides and the Iranian blocks) between $263 Ma and $189 Ma, with most of the rotation having been completed before 240 Ma.

Mesozoic Tethyan realm: Evolution of the SAB
Lack of unequivocal geological evidence from the Armenian territory, owing to the extensive Cenozoic (volcano-)sedimentary cover, has hampered the reconstruction of the Mesozoic northward drift of the SAB in the Tethyan realm.In recent years, a significant effort has been made to reconstruct the tectonic evolution of the SAB from the Permian Gondwanan breakup to the Jurassic accre- tion onto the Eurasian margin (Rolland et al., 2012) and the Miocene closure of the Neotethys by the Arabia-Eurasia collision (Okay et al., 2010;Cavazza et al., 2018).Most work has focused on the ophiolite complexes (Sevan-Akera, Vedi and Zangezur), thought to represent suture zones delimitating continental micro-blocks, and the metamorphic events preserved therein.The Jurassic-Cretaceous mafic intrusions in the Late Devonian sedimentary cover of the SAB at Khor Virap and Arpi (Fig. 1) provide more reliable, in-situ derived constraints on its Mesozoic evolution than the surrounding ophiolites, whose palaeo-positions with respect to the block are ambiguous.

Early Jurassic intrusions at Khor Virap
The Late Devonian sediments at Khor Virap also host two igneous sills consisting of intraplate OIB-type basalts dated at 188.5 ± 1.1 Ma.The only known, possibly contemporaneous, intracontinental igneous rocks in the SAB are Lower Jurassic basaltic rocks in the Negram-Julfa area (south Nahkichevan) and near the village of Aznaberd (Çalxanqala) in central Nakhichevan (Karyakin, 1989;Bazhenov et al., 1996), the locations of which are shown in Fig. 3a and 4. Their age was not radiometrically determined, but the work of Karyakin (1989) and geological mapping place the rocks unconformably between Middle-Upper Triassic and Middle Jurassic sediments.Although detailed geochemical data are lacking, these alkali basalts have an OIB-type ''continental rifting" signature (Karyakin, 1989), similar to the Khor Virap OIB intrusions.All of these occurrences, from Khor Virap to Aznaberd to Negram, are positioned along a NW-SE-striking alignment (Fig. 1).
Based on their geochemical similarities, spatial association, and potential synchronicity, it is tempting to associate this Early Jurassic magmatism to lithospheric thinning and/or asthenospheric upwelling on the scale of the entire SAB, but this seems difficult to reconcile with existing geodynamic interpretations.Van Hinsbergen et al. (2020) used the palaeolatitude constraints of Bazhenov et al. (1996) and Meijers et al. (2015) to envisage the SAB as an isolated microcontinent during this period, left behind after an apparent Late Triassic ridge jump from south to north If we disregard the palaeolatitude of Bazhenov et al. (1996), in absence of a reliable age constraint, and consider the analytical uncertainty of our new position at 188.5 ± 1.1 Ma (Fig. 14; Supplementary Data S1), a scenario is conceivable in which the SAB continued its northward drift along with the Pontides (van Hinsbergen et al., 2020), and was already close to the Eurasian margin during the Early Jurassic.This is in line with the absence of evidence for any 'stranding' of the SAB in the Neotethys behind the eastern Pontides, although tighter palaeomagnetic testing is obviously needed.In this scenario the SAB met the Iranian block at about 190 Ma (Fig. 17), remarkably coinciding with the age of the studied intraplate magmatism.We propose this scenario as an update of the 'isolated island' interpretation of van Hinsbergen et al. (2020), as it provides a plausible explanation for the magmatic event.The rise of intraplate basalts may be facilitated by the presence of a plate boundary, as in the case of the Anatolian-African-Arabian plate junction in southeastern Turkey (Nikogosian et al., 2018).It is therefore conceivable that a similar setting along the triple junction between the SAB, Pontides-Transcaucasus and Iran triggered mantle melting and emplacement of the intraplate basalts of Khor Virap, Aznaberd, and Negram (Fig. 17).

Generation of the Armenian ophiolites
Current interpretations of the SAB are based chiefly on ophiolitic remnants in Armenia and the assumed Mesozoic palaeoposition of Bazhenov et al. (1996).Many authors have adopted the view that the drift history of the SAB was identical to that of the Taurides (Okay and Tüysüz, 1999;Barrier and Vrielynck, 2008;Rolland et al., 2012;Meijers et al., 2015).An alternative reconstruction (van Hinsbergen et al., 2020) proposes that the I.K. Nikogosian, Antoine J.J. Bracco Gartner, Paul R.D. Mason et al. Gondwana Research 121 (2023) 168-195 Taurides drifted away from Gondwana much later (<200 Ma) than the SAB ($245 Ma), and that this $50 Myr 'lag' persisted throughout the Mesozoic until the collision with Eurasia.According to this scenario all of the Armenian ophiolites formed in a forearc setting close to the Eurasian margin (van Hinsbergen et al., 2020), which is consistent with our suggested Middle Jurassic position of the SAB against the western boundary of the Iranian continent, a few hundred kilometres south of the Transcaucausus (Fig. 18).It explains the occurrence of MORB, OIB and E-MORB-type rocks (e.g., Rolland et al., 2020), as well as rare boninites in the SAB (Magakyan et al., 1993).Extension of the forearc region and eventual slab rollback (Fig. 18b) could have halted the SAB drift and initiate NE-directed subduction on the southern margin of the SAB.

Early Cretaceous (Albian) intrusions at Arpi and Darasham
In addition to the Middle Triassic alkaline intrusions, the Late Devonian sedimentary cover in the Arpi area also hosts an andesitic neck dated at 116.7 ± 1.5 Ma.They also have a clear parallel in the Darasham section in south Nakhichevan (Fig. 4), where an andesitic dyke (104.0 ± 2.2 Ma) and amphibole-bearing basaltic dyke (126.0 ± 2.1 Ma) have been found (Khanzatian, 1992).Their trace-element contents showing a subduction-related imprint are common geochemical signatures.Moreover, the Sr-Nd isotope compositions (Fig. 10e) of the Arpi andesite suggest source contamination by subducted crustal components.When the occurrence of slightly older (140-155 Ma) arc-type granodiorite intrusions in the SAB (Hässig et al., 2015;Galoyan et al., 2018Galoyan et al., , 2020) ) are also considered, this coexistence of subductionassociated magmatism in Armenia and Nakhichevan likely points to an active subduction system, at least during this (late) Early Cretaceous period (Fig. 19).
The arc-type granodiorite intrusions into the SAB basement at 140-155 Ma that have previously been explained by SW-dipping subduction (Hässig et al., 2015;Galoyan et al., 2018Galoyan et al., , 2020) ) in our interpretation are best explained by a NE-directed subduction zone at the southern margin of the SAB in Late Jurassic to Early Cretaceous times (Fig. 19).This subduction possibly initiated as a result of forearc spreading at the Eurasian margin, halting the SAB and accommodating continued Africa-Eurasia convergence to its south.The Early Cretaceous andesitic and basaltic sills in the Darasham section in south Nakhichevan (Khanzatian, 1992), and the Arpi andesitic intrusion of $117 Ma possibly represent a continuation of this subduction-related igneous activity.This NE-directed subduction system obviates the need for one or more separate, coexisting active margins to explain the presence of the intrusive rocks (e.g., Hässig et al., 2015;Rolland, 2017;Galoyan et al., 2020), which seems unlikely given the convergence rate required to sustain multiple subduction zones.

Conclusions
We report new geochronological, geochemical, and palaeomagnetic data on magmatic intrusions into the Late Devonian sedimentary cover, and metamorphic rocks that constitute part of the basement of the South Armenian Block.These data are used to place new constraints on the origin and geodynamic history of the SAB in the context of Permian-Triassic breakup of the NE Gondwanan margin, the opening of the Neotethys Ocean, and the Mesozoic kinematic history of the SAB.Our conclusions can be summarised as follows: 1.The characteristic Neoproterozoic-Palaeozoic U-Pb age peaks of detrital zircons derived from the Armenian metamorphic basement firmly establish a Gondwanan origin of the SAB.
2. The geochemistry of trondhjemite (''plagiogranite") intrusions into the Armenian metamorphic basement at $263 Ma demonstrates their adakite/TTG affinity and reflects magma genesis in an active continental margin, consistent with a SW-dipping subduction zone active at the NE Gondwanan margin (Pontides and SAB) during the Middle-Late Permian.3. Mafic alkaline OIB-like sills in the Late Devonian sedimentary cover in the Arpi (south central Armenia) and Darasham (south Nakhichevan) areas, dated at $246 Ma, are products of asthenospheric melting beneath the SAB.A mafic P-MORB-like intrusion at Khor Virap, dated at $234 Ma, reflects melt derivation from a more depleted, shallower mantle source.This set of intrusions is typical of initial continental rifting and early-stage oceanic spreading and suggests a phase of extensional tectonics in the SAB during the Middle Triassic.We infer that this activity marks the incipient breakup of the NE Gondwanan margin and subsequent opening of the Neotethys Ocean in the area.4. Mafic alkaline OIB-type sills within the Late Devonian sedimentary cover at Khor Virap (south central Armenia), dated at $189 Ma, testify to another episode of magma production in the shallow asthenospheric mantle beneath the SAB.In our interpretation the SAB continued its northwards drift alongside the eastern Pontides and reached the Iranian block at about 190 Ma.The intraplate magmatism is likely associated with the triple junction between the SAB, Pontides-Transcaucasus and Iran. 5. Andesitic dykes in the Late Devonian sedimentary cover in the Arpi ($117 Ma) and Darasham (104-126 Ma) areas exhibit a ''subduction-related" geochemical signature, consistent with melt derivation from subduction-modified lithospheric mantle, which also applies to other sporadic occurrences of Late Jurassic to Early Cretaceous igneous products in the SAB.This subduction-related magmatism can be explained by a NEdirected subduction system at the southern margin of the SAB, driven by forearc spreading in the Eurasian margin, which led to cessation of the SAB drift and accommodated compression south of the SAB in Late Jurassic-Early Cretaceous times.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
The five argon isotopes were measured simultaneously with40 Ar on the H2-Faraday position with a 10 13 O resistor amplifier, 39 Ar on the H1-Faraday with a 10 13 O resistor amplifier, 38 Ar on the AX-CDD, 37 Ar on the L1-CDD and 36 Ar on the L2-CDD (CDD = Compact Discrete Dynode).Gain calibration for the CDDs and Faraday cups is done by peak jumping a CO 2 reference beam on all detectors in dynamic mode.In a few cases calibration of Faraday cups was checked by peak jumping the 40 Ar beam between H2 and H1.All intensities were corrected relative to the L2 detector.Air pipettes were run every ten hours and were used for mass discrimination corrections.The atmospheric air value 40 Ar/ 36 Ar = 298.56fromLee et al. (2006) was used in age calculations.Detailed analytical procedures for the Helix MC are described inMonster (2016).The correction factors for neutron interference

Fig. 14 .
Fig. 14.Plots of age (in Ma) versus (a) mean declination and (b) palaeolatitude for the investigated SAB sites (Supplementary Data S1).Shaded areas show the respective values based on the apparent polar wander paths of Kent and Irving (2010) and Torsvik et al. (2012).Grey data shows the secondary component of TSK-7.Black arrows and green data points show restored vertical axis rotation related to the counterclockwise rotation of the SAB (see section 6.2 for further explanation).(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 15 .
Fig. 15.Tectonic reconstruction of the SAB in the Tethyan realm during the Permian (260 Ma; map modified after van Hinsbergen et al., 2020).

Fig. 18 .
Fig. 18.Tectonic reconstruction of the SAB in the Tethyan realm during the Middle Jurassic (170 Ma; map modified after van Hinsbergen et al., 2020).