Petrogenesis of the Late Oligocene Takht batholith, Southeastern Iran: Implications for the Diachronous Nature of the Arabia–Eurasia Collision

Zircon U-Pb age, mineral compositional, elemental and Sr-Nd isotopic data are documented for intermediate to felsic rocks in the Takht batholith, a Late Oligocene igneous complex in the southeastern segment of the Urumieh-Dokhtar belt, Iran, to investigate magma genesis in the context of regional tectonics. A large part of the magmatic belt formed by northward subduction of the Neotethys before the Arabia–Eurasia continental collision. Zircon U-Pb age data indicate the batholith crystallized at ∼25 Ma, an age consistent with previous results. Geochemical data indicate that the rocks share features typical of calc-alkaline magmas and I-type granitoids. The least evolved magma inferred from the data has basaltic andesite composition, consistent with either of the following origins: (i) a partial melt of the mantle followed by differentiation, or (ii) a partial melt of the lower crust. With increasing SiO2, (87Sr/86Sr)25Ma increases from 0.7053 to 0.7073, and εNd25Ma decreases from −0.3 to −2.9, consistent with increasing effects of assimilation–fractional crystallization, although processes such as magma mixing and melting of heterogeneous source rocks cannot be entirely ruled out. Aluminum-in-hornblende thermometry indicates that the batholith might have emplaced at ∼6.5 km or shallower paleo-depths. Moho depth proxies Sr/Y and (La/Yb)N yield crustal thickness of ∼20–30 km, showing no evidence for magma processing within thick crust. Overall, our results indicate that the Takht batholith is a Cordilleran-type batholith formed beneath a continental or transitional arc of normal crustal thickness. This is consistent with the notion that Arabia and Eurasia have collided in a diachronous manner, propagating from the northwest to the southeast since the Late Eocene–Early Oligocene.


INTRODUCTION
Subduction and collision are important tectonic processes that take part in orogenic cycles. These processes not only bring together continental masses but also create and shape mountain belts, playing a key role in ocean and atmospheric circulation (Yin and Harrison, 2000;Allen and Armstrong, 2008;Jagoutz et al., 2016). Also, subduction and collision lead to development of thick crust that resists recycling, contributing to continent formation (Cawood et al., 2013;Hawkesworth et al., 2017;Moyen et al., 2017). Despite the above importance, tracking subduction and collision events in orogens is harder than generally appreciated. In a given orogenic system, there could be multiple events of subduction and collision in different time and space, superimposed and variably preserved in the geologic record. Unraveling details of the record requires studies within and across sub-disciplines of earth sciences.
In magmatic orogens, igneous record gives useful constraints on magma genesis, which could yield indirect clues in tracking subduction and collision. For example, thick lithosphere generated by collision might facilitate formation of alkaline magmas either by relatively low degrees of mantle melting (Aldanmaz et al., 2000) and/or increased influence of continentderived materials in magma genesis (Wang et al., 2017). Magmas formed over thick crust in subduction and collision zones tend to have high Sr/Y, La/Yb, and Dy/Yb because of increased influence of residual garnet over plagioclase in magma genesis (Chung et al., 2003;Topuz et al., 2005;Ji et al., 2012;Goss et al., 2013;Ducea et al., 2015). For instance, Sr/Y and La/Yb of intermediate to felsic magmas have been taken as indicators of Moho depths in arcs and continental collision zones Profeta et al., 2015;Hu et al., 2017).
This study is focused on the petrology and geochemistry of the Late Oligocene Takht batholith (Fazlnia et al., 2014;Hosseini et al., 2017a), which crops out in the southeastern segment of the Urumieh-Dokhtar belt, Iran within the Arabia-Eurasia collision zone. The Late Oligocene is a critical timing in this particular collision zone because it coincides with most estimates for the age of initial collision. Here, we use zircon U-Pb ages, mineral composition, geochemical and Sr-Nd isotopic data to (i) describe the mineralogical and geochemical variations of different rocks from the batholith, (ii) constrain the petrogenesis of these rocks, and (iii) explore the potential implications of magma genesis in the context of regional tectonics. Our data indicate that the Takht batholith formed in an arc setting, showing no signs of any collision-induced crustal thickening. Because the Arabia-Eurasia collision has initiated at the Turkish-Iranian borderlands by the time the batholith formed, our results hint that the collision might have occurred in a diachronous manner.

GEOLOGICAL BACKGROUND
Mountain ranges of Iran are parts of the Alpine-Himalayan orogenic system. In particular, the Zagros orogen has received much attention because it formed by colliding continents of Arabia and Eurasia. The orogen consists of three subparallel, NW-SE-trending structural elements including (i) the Urumieh-Dokhtar belt, (ii) the Sanandaj-Sirjan zone, and (iii) the Zagros fold and thrust belt (Figure 1). A conventional wisdom regarding Zagros orogeny is that the Arabia-Eurasia collision has followed Neotethyan subduction from the Jurassic to the Tertiary (Berberian and King, 1981;Alavi, 1994;Hassanzadeh and Wernicke, 2016). Increasing evidence points to the Late Eocene to the Early Oligocene as the initial age of collision (Allen and Armstrong, 2008;Mouthereau et al., 2012;McQuarrie and van Hinsbergen, 2013;Gholami Zadeh et al., 2017;Koshnaw et al., 2018), along what is now the Bitlis-Zagros suture (or the Main Zagros Fault), the major tectonic boundary between the Sanandaj-Sirjan zone and the Zagros fold and thrust belt.
The Zagros orogen has an extensive igneous record, particularly focused in the Sanandaj-Sirjan zone and the Urumieh-Dokhtar belt. The latter is a ∼1800-2000-km-long and ∼50-80-km-wide belt composed mainly of volcanic and intrusive rocks of calc-alkaline and to a lesser extent tholeiitic or alkaline affinities Alavi, 1994;Shahabpour, 2005Shahabpour, , 2007Honarmand et al., 2013Honarmand et al., , 2016Kananian et al., 2014;Babazadeh et al., 2017;Sarjoughian et al., 2018). It is generally accepted that a large portion of the Urumieh-Dokhtar belt represents an Andean-type paleo-arc formed by northward subduction of the Neotethys beneath terranes that had accreted to the southern margin of Eurasia (Berberian and King, 1981;Alavi, 1994;Aftabi and Atapour, 2000). Neotethyan subduction might also be responsible for Eocene magmatism that affected the northwestern part of the Sanandaj-Sirjan zone (Fazlnia, 2019). In the southeastern segment of the Urumieh-Dokhtar belt where the Takht batholith crops out, published ages of igneous rocks range between ∼82 and ∼5.4 Ma (Chiu et al., 2013;Hosseini et al., 2017b), although magmatism likely peaked over the Eocene (Verdel et al., 2011).
The Takht batholith was dated at ∼25 Ma using in situ zircon U-Pb isotopic analysis on a granitic rock (Hosseini et al., 2017a). In that study, an age of ∼24.2 Ma was also reported for a granodiorite porphyry associated with high-grade Cu mineralization at the Takht-e-Gonbad mine. These ages are consistent with not only information shown on geological maps (Khan-Nazer and Emami, 1996) but also field relations indicating that the batholith intruded Eocene strata and was cut in places by dykes with inferred Miocene ages (Dimitrijevic, 1973;Hassanzadeh, 1993;Shafiei, 2008). Yet, the hosts to other porphyry Cu deposits in the southwestern segment of the Urumieh-Dokhtar belt, including Sarcheshmeh and Meiduk to the northwest of the Takht batholith, appeared to have formed slightly later at the Middle Miocene (McInnes et al., 2005;Zarasvandi et al., 2018 and references therein).
The Takht batholith consists of a wide spectrum of intermediate to felsic intrusive rocks (e.g., Dargahi, 2007;Shafiei, 2008), a view ascertained during our field work. In general, the rocks tend to be richer in mafic minerals at the margin of the batholith and in felsic minerals toward its center (Figure 1). According to aerial view, the intermediate and felsic units make up about 65 and 35%, respectively, of the exposed batholith. Although generally richer in mafic minerals, the intermediate  Khan-Nazer and Emami, 1996). I D = intermediate unit dominated by diorite, I GD = intermediate unit dominated by granodiorite, F = felsic unit. Inset shows the main tectonic units of the Zagros orogen (Stöcklin, 1968).
unit show complex contact relationships between dioritic and granitic rocks in places (Figure 2A). In addition, mafic to felsic enclaves of various sizes are locally observed in outcrops ( Figure 2B)

ANALYTICAL METHODS
A large set of samples was collected from outcrops of the Takht batholith. Samples that are generally fresh and representative were selected for geochronological, mineral chemical and geochemical analyses.
Zircon was extracted from the samples by conventional methods involving heavy-liquid and magnetic separation, and hand-picking under binocular microscope. Zircon grains were mounted in epoxy, polished, and analyzed for cathodoluminescence (CL) imaging and U-Pb isotopes by laser ablation-ICP-MS at the Beijing Quick-Thermo Science & Technology, Co., Ltd. Details of the analytical procedures are similar as described by Ji et al. (2020). The instrument used is an ESI New Wave NWR 193UC (TwoVol2) laser ablation system connected to an Agilent 8900 ICP-QQQ. The analytical spots are 20 µm in diameter. 91500 zircon and Plešovice zircon were used as primary and secondary reference materials, respectively. The latter yielded a mean age of 336.3 ± 2.0 Ma (MSWD = 0.8, n = 22), consistent with the recommended values within errors (Sláma et al., 2008). The zircon U-Pb age data are given in Supplementary Table 1.
Six samples were selected for electron probe microanalysis (EPMA) at the University of Kiel, Germany. Using a JEOL JXA8900 microprobe, compositions of feldspar and amphibole were determined at a probe current of 15 or 20 nA and an accelerating voltage of 15 kV. The EPMA data are given in Supplementary Tables 2, 3.
Sixteen samples were selected for whole-rock major and trace element analysis, which was performed at the ACME Analytical Laboratories, Canada. Samples were crushed into small chips and pulverized in an agate mill. Loss-on-ignition (LOI) was determined by heating 1 g of dried rock powder at 1000 • C for 2 h. Major element oxides were measured by X-ray fluorescence (XRF) spectrometry and trace elements were analyzed by inductively coupled plasma-mass spectrometry (ICP-MS) followed by lithium borate fusion. Precision and accuracy were monitored by replicate analyses of samples and in-house standards (Supplementary Table 4). The major and trace element data are given in Table 1.
Eight samples were analyzed for whole-rock Sr-Nd isotopic analysis at the ACME Analytical Laboratories, Canada. A 200-400 mg aliquot of rock powder was dissolved using mixed hydrofluoric, nitric, and hydrochloric acids. Strontium was purified using cation exchange resin and 2.5 N HCl. Sample solution was loaded onto pre-outgassed and clean rhenium filaments with phosphoric acid and tantalum oxide. Strontium isotope analysis was performed using a thermal ionization mass spectrometer with five Faraday collectors in dynamic mode and 88 Sr = 3.0 V. Strontium isotopes were normalized to 86 Sr/ 88 Sr = 0.1194. Neodymium was purified using HDHEP-coated resin and 0.25N HCl, and its isotopes were analyzed using multicollector ICP-MS with seven Faraday collectors after dissolution in 2% HNO 3 . Neodymium isotopes were normalized to 146 Nd/ 144 Nd = 0.7219. The whole-rock Sr-Nd isotopic data are given in Table 2.    (Faure and Mensing, 2005).

Geochemical Classification
The geochemical data obtained are useful in characterizing rock types and magma affinity. Major element data are recalculated to millications and casted on a P-Q diagram of Debon and Le Fort (1988), on which fields for rock types are used after Bonin et al. (2020). Samples of the intermediate suite plot in fields of quartz diorite, tonalite, and granodiorite ( Figure 4A); however, those classified as tonalites contain too much orthoclase to be tonalite in a strict petrographic sense. For simplicity, therefore, the former two are grouped as diorite sensu lato. Samples of the felsic suite plot above all the fields of Bonin et al. (2020) (Figure 4A). Two samples are classified as syenogranite based on relatively low molar [K -(Na + Ca)] compared with the others, which are all grouped as monzogranite. This classification, although imperfect, appears justified with respect to petrographic observations (Figure 3). Besides, all samples display calc-alkaline affinity on a binary diagram of modified alkali lime index (MALI) versus SiO 2 ( Figure 4B). In addition, samples of the intermediate suite are metaluminous, whereas those of the felsic suite are either metaluminous or slightly peraluminous ( Figure 4C). This feature, together with the occurrence of hornblende and titanite in the majority of samples, indicates an affinity to I-type granites (Chappell and White, 1974).

Zircon U-Pb Geochronology
Most zircon grains in sample I-12 (diorite) are euhedral and prismatic, showing simple oscillatory zoning typical of igneous zircon. According to Corfu (2013), igneous zircon commonly has Th/U greater than 0.2. Relatively high Th/U of zircon grains in sample I-12 (∼0.5-1.1) is thus consistent with an igneous origin (Supplementary Table 1). Nineteen spot analyses yield a mean 206 Pb * / 238 U age of 26.2 ± 0.6 Ma with MSWD of 1.3 (Figures 5A,B). While zircon grains in sample E-15 (monzogranite) are also prismatic, euhedral form is less common. Oscillatory zoning is evident from CL images for most grains, yet zoning pattern appears complex for some grains. The igneous origin of the grains is indicated by relatively high Th/U (∼0.4-1.9) (Supplementary  Figures 5C,D). In summary, the zircon U-Pb data support a crystallization age of the Takht batholith at ∼25 Ma, consistent with earlier age results reported by Hosseini et al. (2017a).

Mineral Composition
Compositions of plagioclase and orthoclase are given in Supplementary Table 2

Whole-Rock Geochemistry
The studied samples have variable SiO 2 contents from ∼55 to ∼76 wt.% in the following order: diorite sensu lato < granodiorite < monzogranite < syenogranite ( Table 1). As shown in Figure 7, the studied samples display continuous trends in the Harker diagrams. Most major elements, including TiO 2 , Al 2 O 3 , Fe 2 O 3T , MgO, CaO and P 2 O 5 , decrease with increasing SiO 2 (Figures 7A-F). The combined concentrations of Na 2 O and K 2 O show an increasing trend with SiO 2 (Figure 7G), excluding two monzogranites with low K 2 O contents (samples E-14-1 and B-2, Table 1). The above trends are consistent with fractionation of ferromagnesian silicates, plagioclase, Fe-Ti oxides (and/or titanite) and apatite in generating the major element variations. Covariations of minor or trace elements and SiO 2 provide additional clues into the probable fractionated phase(s). Decreasing trends of V and Sr with increasing SiO 2 (Figures 8A,B) are consistent with fractionation of Fe-Ti oxides and plagioclase. Yttrium increases with SiO 2 in the intermediate suite and plummets in the felsic suite (Figure 8C), suggesting Y-bearing accessory minerals such as xenotime might have started to fractionate from the magma at a late stage of differentiation. Zirconium exhibits a similar trend as Y (Figure 8D), consistent with zircon fractionation. Niobium, La and Th display increasing trends with SiO 2 (Figures 8E-G), suggesting that these elements, like Na and K, are incompatible during the entire course of magma differentiation.
Trace element patterns shown on primitive mantlenormalized diagrams for the intermediate suite ( Figure 9A) have the following features: (i) enrichment of large-ion-lithophile elements (LILE) in the order of 30 to 100 times primitive mantle abundances, (ii) negative Nb-Ta-(Ti) anomalies relative to K and La, (iii) positive Pb anomaly, and (iv) limited fractionation between middle to heavy rare-earth elements (REEs). A diorite sample (sample I-9) has a different pattern compared with the above in having lower abundances of Rb, Ba, and K ( Figure 9B). Generally, trace element patterns for the felsic suite share similar features as those for the intermediate suite ( Figure 9C). Within the felsic suite, the syenogranites in particular have low abundances of Sr, P, Eu, Ti and the heavy REE compared with monzogranites, a feature presumably related to fractionation of feldspars, Fe-Ti oxides (and/or titanite) and apatite ( Figure 9C). The two monzogranites showing low K 2 O contents display  (Vermeesch, 2018). Asterisk denotes correction for unradiogenic lead by isochron regression by the Ludwig (1998) algorithm. In frames (B,D), filled bars denote analyses used for mean age calculation and empty ones do not.  Leake et al., 1997). Mineral abbreviations are used after Whitney and Evans (2010), except for minerals that have no recommended abbreviations: Ab = albite, Act = actinolite, An = anorthite, And = andesine, By = bytownite, Fac = ferroactinolite, Fhb = ferrohornblende, Fts = ferrotschermakite, Lab = labradorite, Mhb = magnesiohornblende, Olg = oligoclase, Tr = tremolite, Ts = Tschermakite. relatively low abundances of Rb and Ba in trace element patterns ( Figure 9D).
The Sr-Nd isotopic data for samples chosen for analysis show limited variability. In detail, 87 Sr/ 86 Sr ranges from 0.7054 to 0.7077, and 143 Nd/ 144 Nd ranges from 0.5125 to 0.5126 ( Table 2). After age correction to 25 Ma, the isotopic data plot in the lower right quadrant of a Sr-Nd isotope space characteristic of high time-integrated Rb/Sr and low time-integrated Sm/Nd relative to Bulk Silicate Earth (BSE) (Figure 10A). Using equations of Ickert (2013), the maximum errors propagated for the calculation of initial isotopic ratios are estimated to be ± 0.0005 for ( 87 Sr/ 86 Sr) 25Ma and ± 0.30 for εNd 25Ma . Generally, the Sr and Nd isotopic ratios are negatively correlated except a syenogranite (sample B-12) having lower 87 Sr/ 86 Sr than expected. A probable reason is that the Sr isotopic ratio of this sample is over-corrected due to post-magmatic Rb-gain or Srloss, consistent with its high Rb/Sr (∼1.5). Besides, there exists a negative correlation between Nd isotopic ratios and SiO 2 concentration ( Figure 10B).

Depth of Magma Emplacement
We estimate the crustal level at which the Takht batholith was emplaced using the aluminum-in-hornblende geobarometer (Hammarstrom and Zen, 1986), a tool which was widely applied to calc-alkaline intrusive rocks (Putirka, 2016 and references therein). Since its original formulation, the geobarometer was calibrated by various researchers including Hollister et al. (1987); Johnson and Rutherford (1989), Blundy and Holland (1990), Schmidt (1992), among many others. For most hornblende compositions obtained in this study (Supplementary Table 3), these calibrations yield pressure estimates between 1.3 and 5.5 kbar, with a spread of ∼1-1.5 kbar for a given hornblende analysis using different calibrations. Generally, the median of pressure resulted from different calibrations appears to follow the order: diorite sensu lato (∼2.7-5.1 kbar), granodiorite (∼1.8 kbar),  Sun and McDonough (1989). and monzogranite (∼1.5 kbar). Given the maximum distance between the mafic and felsic units in the field (Figure 1), however, it is unlikely that the entire pressure range reflects magma solidification at different paleo-depths. Instead, it is more likely that hornblende in the diorites might have formed at mid-crustal levels and were entrained by magma(s) that ascent and eventually froze at the final emplacement depth. If that is the case, the final emplacement of the Takht batholith might occur at 2 kbar or less, equivalent to about 6.5 km or shallower paleo-depths, an estimate close to the lower bound of ∼5-10 km made by Fazlnia et al. (2014).

Petrogenetic Constraints
It is important to evaluate the effect of alteration on the geochemical data obtained in this study, for some of these data are used to derive constraints related to magma genesis at high temperature. Petrographic observations indicate that most studied samples are essentially fresh with little or no signs of postmagmatic alteration (Figure 3). Almost all samples have LOI lower than 2 wt.% (Table 1), except in a monzogranite (sample E-14-1) where higher LOI (∼3.8 wt.%) might be due to partial alteration of plagioclase to sericite as observed in thin section. Further, the trace element patterns for most studied samples are coherent (Figures 9A,C), suggesting mobilization of particular trace elements, as would be expected for extensive post-magmatic alteration, did not occur.
Sample I-9, a diorite as mentioned above, requires a special note because of its unusual major and trace element features. It has low Fe 2 O 3 (∼1.4 wt.%), high CaO (∼12.1 wt.%) compared with other diorites (Figures 7C,E). Also, its trace element pattern is different from other diorites in showing low abundances of Rb, Ba, and K (Figure 9B). After a detailed petrographic check, we note several unique features: (i) a high content of plagioclase compared with the other diorites, (ii) plagioclase occurs as a framework of coarse grained crystals commonly observed in cumulates (Figure 3C), and (iii) the rare presence of hornblende and absent of biotite. Based on these observations, we suggest the unusual geochemical features of sample I-9 might be related to its probable origin as a plagioclaserich cumulate.
negative correlation between Nd isotopes and SiO 2 ( Figure 10B). The trend appears to require increasing contribution of isotopically enriched materials to the magma as it became more differentiated. To model the process, we calculate assimilationfractional crystallization (AFC) trajectories using data from the ChahJam-Biarjmand metamorphic complex, which was exhumed from and likely representative of the lower to middle crust of Iran (Shafaii Moghadam et al., 2015). Three major rock types in that study are used as end-member assimilants, including a metagranite, a paragneiss and a granitic gneiss. As shown in Figure 10A, the least evolved diorite would have undergone ∼10-15% AFC to explain the observed Sr-Nd isotopic variations. We then explore whether the intermediate and felsic units of the Takht batholith satisfy mass balance by undertaking energy-constrained (EC)-AFC simulations Spera and Bohrson, 2001) with similar compositional parameters for the initial magma and the assimilants as the classical AFC models presented above (Supplementary Table 5).
The results indicate that, for all three assimilants, the initial dioritic magma would have undergone ∼42% differentiation (mainly by fractional crystallization) before being able to form rocks with Sr-Nd isotopes observed in the felsic unit due to significant addition of anatectic melt from the wallrocks. Also, the mass ratios of the intermediate unit to the felsic unit predicted by these models range from 6:1 to 9:1 (Supplementary Table 6).
Thus, it appears that the current exposure of these units (65 to 35%, ∼2:1) requires a considerable amount of dioritic rocks be hidden at a deeper level of the batholith. Another possibility is that mechanisms apart from AFC might have played a role in the formation of the felsic suite, such as crustal melting, which is discussed below.
We then consider additional magmatic processes such as mingling or mixing of magmas, which is hinted by field observations (Figures 2A,B) and complex zoning seen from plagioclase in diorites (Supplementary Figures S1-S3), even though available data point to a coherent age of ∼25 Ma for the batholith (Figure 5 and Hosseini et al., 2017a). As magma mixing generates linear trends shown by covariation of elements and isotopes, one could argue that, for example, granodiorites and monzogranites are solidified products of mixing between dioritic and syenogranitic magmas (Figures 7, 8). However, mixing alone is hard to explain the kinked trends of Y or Zr versus SiO 2 (Figures 8C,D), which appear to record fractionation of accessory minerals such as xenotime and zircon (Förster, 1998;Lee and Bachmann, 2014). Another possibility is that differences in rock types are results of melting heterogeneous source rocks in the crust with limited mixing of the resultant magmas. That has to imply that not only the kinked trends note above but also the εNd 25Ma -SiO 2 trend ( Figure 10A) are fortuitous. If heterogeneous source rocks were to be melted in the crust, however, scattered data in terms of Sr-Nd isotopes and SiO 2 seem to be more likely. Yet, melting heterogeneous source rocks might be responsible for second-order geochemical variations in the dataset. For example, differences in major and trace element geochemistry among the monzogranites might be related to source rocks with variable contents of K-bearing phases such as FIGURE 11 | Estimation of crustal thickness using Moho depth proxies (A) Sr/Y and (B) (La/Yb) N where subscript N denotes normalization against CI chondrite (Sun and McDonough, 1989). Dotted line denotes empirical fits of data for a given volcanic arc shown by gray symbols . The yellow band and orange line denote the range and median, respectively, of data for the intermediate suite, which generally satisfy the filters proposed by Profeta et al. (2015) including ranges of SiO 2 (55-68 wt.%) and MgO (1-6 wt.%). If data for monzogranites are included, the median for Sr/Y remains unchanged (∼17) while that for (La/Yb) N is slightly higher (∼6.9).
biotite, muscovite and orthoclase (Figures 9C,D). Also, the Sr-Nd isotopic composition of the syenogranite hints derivation from source rocks that are probably different from the monzogranites ( Figure 10A). In summary, we maintain that AFC exerts the dominant control of the geochemical variations observed from the samples, while processes such as mingling and mixing of magmas, and melting of heterogeneous source rocks, cannot be completed excluded.
Next, we examine the origin of the diorite sensu lato, the arguably least evolved rock type in the Takht batholith. As a group it has the highest MgO (∼3.5-4.9 wt.%), the lowest SiO 2 (∼55.3-56.7 wt.%) and, relative to BSE, the least unradiogenic Nd isotopes and the least radiogenic Sr isotopes ( Figure 10A). Hosseini et al. (2017a) analyzed a range of igneous rocks related to the Takht-e-Gonbad deposit and their dataset includes a sample with basaltic composition (i.e., sample TAM12 with ∼6 wt.% MgO and ∼49 wt.% SiO 2 ). However, that sample belongs to Eocene volcanic/volcaniclastic rocks, which were intruded by Oligocene and Miocene igneous rocks. Thus, there is no direct evidence for magmas that are more primitive than the studied diorites in the formation of the ∼25 Ma Takht batholith. If our inference for an intermediate parental magma is correct, then two possible origins apply. First, it might be a residual liquid of a mantle-derived magma that had undergone differentiation prior to emplacement. This explains the general lack of mafic-ultramafic rocks at the current exposure. Second, the intermediate parental magma might have a lower crustal origin. Blatter et al. (2013) demonstrated by melting experiments that intermediate magmas with similar MgO and SiO 2 as the studied diorites could form by partial melting of hydrous arc basalts over a wide range of oxygen fugacity at lower crustal pressure. Importantly, hybrid mechanisms involving a combination of the two end-member scenarios are also permissive; for instance, mixing of lower crustal melts and variably fractionated mantle-derived magmas in zones of melting, assimilation, storage and homogenization (MASH) close to the Moho (e.g., Hildreth and Moorbath, 1988), or lower crustal hot zones (Annen et al., 2006).

Tectonic Implications
As illustrated above, the Takht batholith possesses typical subduction-related geochemical signatures (Figure 9). A straightforward explanation for these is that the Takht batholith formed in a subduction zone, an interpretation mostly adapted in earlier studies (Shahabpour, 2005(Shahabpour, , 2007Shafiei et al., 2009;Mohajjel and Fergusson, 2014;Shafaii Moghadam et al., 2015;Rasouli et al., 2016). Another possibility is that the batholith formed in a setting without active subduction, for example, where subduction-modified domains of the mantle lithosphere and/or the lower crust were perturbed by processes unrelated directly to subduction (e.g., upwelling of the asthenosphere). If that is the case, variable magma types and chemistry (e.g., alkalinity, trace elements, and radiogenic isotopes) is expected because of the variety of source rocks that might be involved and the range of depths where partial melting might occur. In this study, however, we do not observe such variations and the geochemical data illustrated above are consistent in most aspects with arc magmas. Therefore, it is appropriate to consider the Takht batholith as a Cordilleran-type batholith, forming as part of a continental arc in the Late Oligocene at ∼25 Ma.
Our inference that the Takht batholith is of Cordilleran-type sheds indirect lights on how Arabia and Eurasia might have collided. Available data in northwestern Iran and southern Lesser Caucasus indicate that Late Oligocene magmatic rocks show geochemical features characteristic of formation over relatively thick crust, such as relatively high Sr/Y and La/Yb Rezeau et al., 2016Rezeau et al., , 2017. In view of growing evidence suggesting that the Arabia-Eurasia collision initiated between the Late Eocene and the Early Oligocene (Allen and Armstrong, 2008;Mouthereau et al., 2012;McQuarrie and van Hinsbergen, 2013;Gholami Zadeh et al., 2017;Koshnaw et al., 2018), these features have been interpreted in the context of collision-induced crustal thickening (e.g., Moritz et al., 2016). We note that none of the rocks in this study display such geochemical features. For instance, the ranges of Sr/Y and (La/Yb) N of these rocks could be used to indicate modest crustal thickness, probably between 20 and 30 km (Figure 11). Here, an intriguing question is, at a given time, how the Takht batholith with an inferred arc origin can be reconciled in a tectonic setting where continental collision appeared to have started. One possibility is that collision did initiate in northwestern Iran and southern Lesser Caucasus, but not so to the southeast of the collision zone, where the Neotethys was still subducting beneath central Iran, as depicted by Chiu et al. (2013). This is consistent with paleogeographic reconstruction by McQuarrie and van Hinsbergen (2013) (Figure 12), and the view that post-collisional magmatism propagated from northwest to southeast along the Arabia-Eurasia collision zone (Pang et al., 2015;Lin et al., 2020).
While we remain open to other possible tectonic scenarios that could explain the aforementioned spatial variations in magma composition, the simplest explanation hitherto appears to be diachronous collision.

CONCLUSION
A field, petrologic and geochemical study of the Late Oligocene Takht batholith, Iran yields the following conclusions: • The batholith was solidified at ∼25 Ma and the depth of magma emplacement was likely 6.5 km or less. It consists mainly of intermediate to felsic rocks, showing subalkaline, dominantly metaluminous compositions and I-type affinity. • The rocks display geochemical and Sr-Nd isotopic variations consistent with concurrent fractional crystallization and assimilation of lower to middle crustal rocks.
• The least evolved magma indicated by the data has intermediate composition, and might represent a differentiated liquid of a mantle-derived magma, or a partial melt of the lower crust, or any combination between them as predicted by models involving MASH or lower crustal hot zones. • The batholith is most likely of Cordilleran-type related to northward subduction of the Neotethys. In the context of regional tectonics, the paleogeographic position of the batholith can be reconciled if Arabia and Eurasia have been colliding in a diachronous manner since the Late Eocene-Early Oligocene.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

AUTHOR CONTRIBUTIONS
K-NP analyzed the data and wrote the manuscript. AF designed the study, participated in field work, and analyzed the data. W-QJ obtained zircon U-Pb isotopic data and analyzed the data. SJ and AJ participated in field work. All authors contributed to the article and approved the submitted version.