Recognition of Late Cretaceous Hasanbag ophiolite-arc rocks in the Kurdistan Region of the Iraqi Zagros suture zone: A missing link in the paleogeography of the closing Neotethys Ocean

The Zagros orogenic belt extends NW-SE from eastern Turkey through NE Iraq and along SW Iran to the Strait of Hormuz and into northern Oman (Fig. 1). It is situated at the tectonic crossroads of the Alpine–Himalayan belts and preserves a 150 m.y. record of the protracted convergence history between Eurasia and Arabia across the Neotethys, from initial intra-oceanic subduction/obduction processes to present-day continental collision. Several recent publications have provided new insights into the tectonics of the Zagros orogenic belt (Vernant et al., 2004; Masson et al., 2005; Walpersdorf et al., 2006; Mouthereau et al., 2007; Oveisi et al., 2009; Nissen et al., 2011; Saura et al., 2011; Vergés et al., 2011; Moghadam and Stern, 2011). However, the NE Iraq portion of the Zagros orogenic belt is much less documented than its adjacent counterparts (i.e., SW Iran, eastern Turkey, and northern Oman).

The presence of a Tertiary arc has long been recognized in the Iraqi sector, with final collision between the Arabian and Iranian plates occurring in the Eocene–Oligocene (Aswad, 1999; Aswad et al., 2011). On the Arabian margin, this was superimposed onto the Cretaceous (Balambo) carbonate platform with the deep-water radiolarite Qulqula Group emplaced tectonically on top, followed by deposition of the Tanjero flysch in the Maastrichtian. This must have involved the collision of an older arc with the Arabian continental margin in Iraq (Ali, 2012), as already recognized in Oman, Turkey, and Cyprus ophiolites. This postulated arc along the Iraqi-Iranian Zagros section has previously not been recognized. In this paper, we demonstrate that an upper tectonic slice, previously thought to be the Tertiary Walash arc system, is in fact this missing Cretaceous arc along the Iraq-Iran boundary, named here the Hasanbag arc complex. This was achieved via integrated ⁴⁰Ar-³⁹Ar dating of magmatic hornblende and whole-rock petrography and geochemistry.

The study area is situated along the Iraq-Iran-Turkey border, in a rugged elevated topography (e.g., Hasarost Mountain is the highest mountain in Iraq at just over 3600 m) dissected by irregular steep valleys. The area contains unexploded landmines, restricting sampling to the margin of main roads. The study area forms part of the Tethyan arc-related volcanic assemblage in the Kurdistan Region, northeast corner of Iraq (Figs. 1 and 2), and it is an integral part of the Zagros suture zone. The Zagros Mountains developed in response to the collision of the Arabian and Iranian plates as the Neotethys basin was consumed (Alavi, 2004). The particular igneous assemblage discussed here, the Hasanbag igneous complex, outcrops as a structurally high nappe. The Hasanbag igneous complex occurs between longitudes 44°38′46.41″E and 44°38′3.36″E and latitudes 36°43′8.85″N and 36°43′33.83″N near Sidekan, 100 km northeast of Erbil city, Kurdistan Region, within the Iraqi portion of the Zagros suture zone (Fig. 2). The complex is part of the ophiolite-bearing “Upper Allochthon” terranes (designated as the Gemo-Qandil Group), which are composed of metavolcano-sedimentary rocks of Late Cretaceous age (Jassim et al., 2006). The Hasanbag igneous complex is preserved as a nearly 700-m-thick volcanic sequence dominated by andesite and basaltic andesite, crosscut by microgabbroic and dioritic dikes. No significant sedimentary beds were found between the lava flows, indicating a setting devoid of significant clastic input. The lack of sediment precludes age determination via biostratigraphy, and we rely on radiometric dating (see following). Stratigraphic correlation across such a huge volcanic pile is difficult to establish. Although, in general, the rocks are altered, most of the igneous textures are still preserved. Megascopically, Hasanbag samples have variable colors, from dusky yellowish brown to greenish gray, dark-greenish gray, and gray, reflecting variable degrees of weathering. Based on field and petrographic observations, seven dikes crosscutting the Hasanbag lavas have been identified. These are six microgabbros (samples H13, H14, H15, H25, H28, and H29) and one diorite dike (H20). All the former are medium to fine grained and normally nonfoliated.

Only the least altered basaltic rock samples were chosen for geochemical analysis. Whole-rock major- and trace-element X-ray fluorescence (XRF) analysis was carried out with a Spectro-Analytical Instrument (XEPOS) energy-dispersive spectrometer fitted with a Si-diode detector at the School of Earth and Environmental Sciences, University of Wollongong, Australia, following the methods of Norrish and Chappell (1977). Major elements were measured on samples fused with Li borate, while trace elements were analyzed from pressed pellets bonded with polyvinyl acetate (PVA). Calibration was made against a wide range of international reference materials and laboratory standards previously calibrated against synthetic standards. Loss-on-ignition was determined by heating a separate aliquot of rock powder at 1000 °C. The samples were additionally analyzed at the Australian Laboratory Services (ALS) at Brisbane, Australia, for their rare earth element (REE) and other trace-element concentrations using inductively coupled plasma–mass spectroscopy (ICP-MS).

Microprobe analyses were carried out on polished thin sections utilizing a fully automated, Cameca SX100 electron microprobe at Macquarie University, fitted with five wavelength dispersive spectrometers (WDS) and a Princeton Gamma Tech (PGT) energy dispersive system (EDS). The operating conditions were: accelerating voltage 15 kV; beam current 20 nA; beam size focused 1–2 μm; and count times 20 s (10 s peak, 10 s background). The standards used were: Si, Ca—wollastonite; Al—kyanite, Na—albite; Ti—rutile; Cr—Cr metal; Fe—hematite; Mn—spessartine; Mg—forsterite; K—orthoclase; and Ni—Ni metal.

Hornblende separates from two samples H20 (dike) and H23 (lava) were selected for ⁴⁰Ar/³⁹Ar analyses. Mineral separates were obtained using standard crushing, de-sliming, sieving, and magnetic and heavy liquid separation methods. Microprobe analyses show that the amphiboles are unaltered and kaersutite in composition, typical of magmatic hornblende (see Table 3 in supplementary item 1¹). Prior to irradiation, all samples were rinsed with deionized water in an ultrasonic bath.

Samples were loaded into aluminum foil packets and placed in quartz tubes (UM#43) along with the flux monitor GA1550 biotite (98.8 ± 0.5 Ma; Renne et al., 1998) and irradiated in cadmium-lined cans in position 5c of the McMaster University reactor, Hamilton, Canada.

The ⁴⁰Ar/³⁹Ar step-heating analyses were conducted at the University of Melbourne, Australia, largely following procedures described by Phillips et al. (2007) and Matchan and Phillips (2011). Due to low potassium contents and limited mineral separate volumes, hornblende aliquots were step-heated using a CO₂ laser linked to a MM5400 mass spectrometer with a Daly detector.

Argon isotopic results are corrected for system blanks, mass discrimination, radioactive decay, reactor-induced interference reactions, and atmospheric argon contamination. Decay constants used are those reported by Steiger and Jäger (1977). Correction factors (±1σ) for interfering isotopes were (³⁶Ar/³⁷Ar) Ca = 0.000280% ± 3.6%, (³⁹Ar/³⁷Ar) Ca = 0.000680% ± 2.9%, (⁴⁰Ar/³⁹Ar) K = 0.000400% ± 100%, and (³⁸Ar/³⁹Ar) K = 0.0130% ± 3.9%. System blank levels were monitored between analyses and found to be essentially atmospheric (⁴⁰Ar/³⁶Ar)atm = 295.5 ± 0.5 (Nier, 1950).

Plateau ages were calculated using ISOPLOT (Ludwig, 2003) and are defined as including at least 50% of the ³⁹Ar, distributed over a minimum of three contiguous steps with ⁴⁰Ar/³⁹Ar ratios within agreement of the mean at the 95% confidence level. Inverse isochron plots were also constructed with ISOPLOT (Ludwig, 2003), using the same data points included in the calculation of the weighted mean ages. With the possible exception of sample H23, the inverse isochron regressions show that the trapped argon component (⁴⁰Ar/³⁶Ar) is of atmospheric composition within the analytical uncertainties, justifying the interpretation of weighted mean ages as crystallization ages. Calculated uncertainties associated with mean and plateau ages include uncertainties in the J-values but exclude errors associated with the age of the flux monitor and the decay constant.

Petrographic studies were undertaken on 22 thin sections of Hasanbag igneous complex rocks. Most of the studied samples are lavas. There is a wide range of igneous textures, namely, porphyritic, amygdaloidal, glomeroporphyritic, microlitic-porphyric, decussate, ophitic, and aphyric (Figs. 3A–3D). Plagioclase, clinopyroxene, and iron oxides are the main phenocrysts, but they also occur as small laths in the groundmass. Kaersutite was observed in only one volcanic sample (H23; Fig. 3E). Alkali-feldspar (anorthoclase) is found as small crystals normally comprising less than 1% of the total rock volume (e.g., H23), and Ti-Fe oxides (1%–5%) are accessory minerals. The ophitic textural relationships suggest early crystallization of plagioclase followed by clinopyroxene and finally by primary opaque minerals.

The primary igneous minerals are generally partly altered to chlorite, calcite, fine-grained clay minerals, sericite, zeolites, and secondary albite. Plagioclase grains are idiomorphic to subidiomorphic and modally form 22%–66% of the samples. Plagioclase phenocrysts range from 0.13 to 1.6 mm. Some plagioclase phenocrysts (Fig. 3B) have been altered to white mica (sericite) and clay minerals. Clinopyroxenes (Fig. 3F) are 0.2–0.9 mm, generally subhedral to euhedral, modally constitute 2%–20%, and are partially altered to chlorite and secondary amphibole. Alkali feldspar (anorthoclase) is located in the groundmass as microlites. The primary groundmass constituents are predominantly plagioclase laths with variable proportions of subordinate clinopyroxene. They are commonly altered to chlorite, biotite (Fig. 3G), epidote, secondary Fe-Ti oxide (magnetite and ilmenite), titanite and quartz.

Secondary quartz (2%–8%) in the groundmass is present and is associated with chlorite in veins and vugs along with epidote (Fig. 3A). Chlorite is a secondary mineral formed by uralitization of clinopyroxene and amphibole, or less commonly replacing glassy patches in the groundmass (H6). In other samples, it appears filling the amygdules along with epidote (H11; Fig. 3A). Modally, chlorite ranges between 10% and 40%. Epidote ranges between 0.5% and 5% and is found as veins or filling amygdules along with chlorite.

The dikes display ophitic, subophitic, poikilitic, and/or intergranular textures. The primary mineral assemblage for all the microgabbro dikes is plagioclase + pyroxene + iron oxides. Moreover, the textural relationships among the primary phases suggest early crystallization of plagioclase, then clinopyroxene and, finally, opaque minerals. The diorite dike (H20) assemblage is plagioclase + kaersutite + pyroxene + iron oxides. Like the volcanic rocks, the minerals are variably altered. The microgabbro dikes contain mainly subhedral to euhedral tabular plagioclase megacrysts altered to albite (An_0.93-An_6.67; see Table 1 in supplementary item 1 [see ^{footnote 1}]) crystals occurring as both phenocrysts and in the groundmass. Clinopyroxene in all the microgabbro dikes is as 0.89–1.35 mm euhedral to subhedral grains. It shows sector zoning and contains inclusions of titanomagnetite (e.g., H15; Fig. 3H). Clinopyroxene is often well preserved, but phenocrysts and those in the groundmass can be partially replaced by actinolite and/or chlorite in both large crystals and in the matrix (Fig. 3H). All amphiboles in the microgabbro dikes appear to be secondary after igneous pyroxene, and they are mostly actinolitic. Other secondary minerals include chlorite (brunsvigite), epidote, zeolites, prehnite, and clay minerals. The diorite dike is less altered than the microgabbro dikes. Magnetite, ilmenite, apatite, titanite, and zircon are accessory minerals in the diorite dike.

In the volcanic rocks, albite forms the main composition of the subhedral laths and microlites in the matrix. However, oligoclase is preserved in samples H5 and H10 (An₁₀ and An₂₇; see Table 1 in supplementary item 1 [see ^{footnote 1}]). All clinopyroxenes are augite (Wo_44.5-En_46.5-Fs₉; see Table 2 in supplementary item 1 [see ^{footnote 1}]). Igneous amphiboles also occur in the Hasanbag volcanic rocks. Five igneous amphibole analyses from sample H23 indicate that they are kaersutites in composition, according to the classification diagram of Yavuz (1999; see Table 3 in supplementary item 1 [see ^{footnote 1}]).

In the dikes, pyroxene phenocrysts (Wo_44–49En_42–46Fs_6–13; H15, H28; Table 2 in supplementary item 1 [see ^{footnote 1}]) are magnesian-augite (Morimoto, 1988; Rock, 1990; Yavuz, 2001). Both Ti-magnetite and ilmenite occur in the microgabbro dikes. The Ti-magnetite has a high ulvöspinel component (see Table 6 in supplementary item 1 [see ^{footnote 1}]; Frost and Lindsley, 1991). However, because the analyzed iron oxides fall to the left of the magnetite-ulvöspinel join, they are more accurately termed Ti-maghemites (Banerjee, 1991), probably resulting from oxidation of Ti-magnetite at relatively low temperatures (<300–400 °C). This compositional change likely occurred during the pervasive chlorite-epidote–forming metamorphism event (Banerjee, 1991). Subhedral to euhedral tabular plagioclase crystals in the diorite dike show oligoclase to albite compositions An_29.35–An_1.84 (see Table 1 in supplementary item 1 [see ^{footnote 1}]). Textural evidence supports a primary origin for the amphibole, with mostly kaersutite/magnesio-hastingsite or tschermakite having up to 5.31 wt% TiO₂ (see Table 3 in supplementary item 1 [see ^{footnote 1}]) and a discontinuous outer zone of pale-brown lower-Ti tschermakite (3.94 wt% TiO₂). Those crystals with lower TiO₂ contents are free of Fe-Ti–oxide exsolution lamella, indicating that they were not produced by subsolidus replacement of the higher-Ti amphiboles in superimposed alteration. Generally, kaersutites occur in both alkaline and subalkaline rocks (e.g., Deer et al., 1966; Best, 1970; Wilkinson, 1974; Mitchell, 1990; Martin, 2007).

The igneous mineralogy shows variable but generally extensive alteration under greenschist-facies conditions, and the development of chlorite, tremolite/actinolite, and epidote with the albitization of igneous plagioclase. The detailed mineralogy of the alteration products is presented in Ali (2012).

Ti concentrations in the igneous kaersutites can provide constraints on the crystallization history of magmas. This is because Ti occupancy in amphibole depends strongly on temperature (T) and oxygen fugacity (log ƒO₂) (e.g., Gilbert et al., 1982; Deer et al., 1997). Under specific log ƒO₂–T conditions, the hydrous subalkaline liquids can also be kaersutite-saturated at conditions close to amphibole liquid temperatures (∼1050 °C), and the ƒ(O₂) can be controlled by the NNO (Ni–NiO) buffer (e.g., Holloway and Burnham, 1972; Helz, 1979). Anderson and Smith (1995) suggested that variation in temperature causes the aluminum of granitoid amphiboles to vary by two titanium substitutions (i.e., Ti + R²⁺ = 2^[6]Al and Ti + ^[4]Al = ^[6]Al + Si), whereas changes in oxygen fugacity would promote the Fe³⁺ = ^[6]Al exchange. A possible substitution, ^[6]Al³⁺ + OH⁻ = ^[6]Ti⁴⁺ + O²⁻, may be operative for kaersutite. Another kaersutite substitution mechanism, however, might be ^[6]R²⁺ + 2OH⁻ = ^[6]Ti⁴⁺ + 2O²⁻, where ^[6]R²⁺ = Fe²⁺ + Mg + Mn. Ernst and Liu (1998) compiled a pressure-temperature (P-T) scheme based on the Al₂O₃ and TiO₂ contents in amphiboles (inset of Fig. 4A). Using this scheme, the Hasanbag kaersutites can be classified as a high-temperature phase (i.e., Ti vs. Al; Fig. 4). Based on the empirical thermobarometric formulations of Ridolfi et al. (2010) (Fig. 5), the P-T, logƒO₂-T, and T-H₂O melt relationships (Ali, 2012) for the Hasanbag kaersutites show that they are water saturated and most likely crystallized from a magma that contained 4–6 wt% H₂O.

The age spectra of analyses H20-1 to H20-5 reveal that the older ages occur in the lower-temperature fractions (step 1), and the minimum age occurs in the middle-temperature steps (step 1; see Table 1 in supplementary item 2 [see ^{footnote 1}]).These anomalously old apparent ages from the initial steps are attributed to the rocks being strongly affected by excess ⁴⁰Ar, which would make them appear significantly older than their actual intrusion ages. Also, there is still some uncertainty concerning the role of excess ⁴⁰Ar, which might need further investigation. Excluding the low-temperature fractions, regression of ³⁶Ar/⁴⁰Ar versus ³⁹Ar/⁴⁰Ar yields an age of 91.9 Ma (95% confidence, mean square of weighted deviates [MSWD] = 0.8; see Fig. 6A; Table 1 in supplementary item 2 [see ^{footnote 1}]). Taking just the grain H20-5 step-heating spectra, we find a plateau age of 92.4 ± 1.4 Ma (MSWD = 0.5; see Fig. 6B), which is interpreted as the age of crystallization.

The textural history of sample H23 suggests that the kaersutites formed at moderate pressure (i.e., 300–400 MPa) at 950–1000 °C. They were probably brought up as early phenocrysts with ascending magma. Therefore, the studied kaersutite from sample H23 yielded scattered data in the step-heating analysis, with an imprecise ⁴⁰Ar/³⁶Ar intercept value of 313.5 ± 8.6, a high MSWD = 4.8, and a poor precision of 8.9 m.y. on the regressed age of 95.8 Ma (see Fig. 6C; Tables 3 and 4 in supplementary item 2 [see ^{footnote 1}]). Figure 6C shows that the sample produced isochron ages that are simple mixtures of radiogenic and atmospheric components, though some exhibit excess argon. This is reflected in the intercepts below 0.003384 (the isotope ratio of atmospheric argon 1/295.5) on the ³⁶Ar/⁴⁰Ar axis (i.e., 0.00318 = 1/313.5). The benefit of using the inverse isochron approach in argon age determination of the studied sample is to avoid an assumption that the (³⁶Ar/⁴⁰Ar) value of trapped Ar equals that of the air (i.e., ³⁶Ar/⁴⁰Ar ratio of modern atmosphere [= 1/295.5]). By excluding low- and high-temperature gas fractions of the plateau age spectra, an age of 106 ± 12 Ma was obtained, with a rather high MSWD = 6.6 (see Fig. 6D; Tables 3 and 4 in supplementary item 2 [see ^{footnote 1}]). The discrepancy between the isochron age and the plateau age might be due to large amounts of argon dissolved in ocean water that is recycled into Earth’s interior at subduction zones (e.g., Sumino et al., 2005). Therefore, a rock of arc affinity may show “excess” ³⁶Ar. It is possible that sample H23 incorporated material from the subducted slab and so may show argon isotopic fractionation effects (i.e., y intercept ⁴⁰Ar/³⁶Ar ratio > 295.5, the atmospheric argon value), but the atmospheric ⁴⁰Ar contamination problem is not under consideration in the Hasanbag samples because it is most troublesome for very young Quaternary samples that have very little ⁴⁰Ar*.

The low- and high-temperature gas fractions of the studied specimen display clear fluctuations in apparent age (Fig. 6D). The fluctuations coincide with an increase of the Ca/K values from 34.47 through 38.72–49.22 (step 1 to step 3 in Fig. 6D; Tables 3 and 4 in supplementary item 2 [see ^{footnote 1}]). The very high apparent age shown by sample H23 within the first few percent of gas released could be due to secondary mineral constituents formed during postmagmatic alteration. The high-temperature gas fractions display high fluctuations in apparent Ca/K ratios and high apparent age, suggesting that the release of ³⁹Ar occurred from compositionally nonuniform sites. The latter might be revealed due to recoil of ³⁹Ar, which has the effect of increasing the apparent ages of this part of the spectrum up to 260.2 Ma (step 3), controlled by the primary mineral phase (i.e., kaersutites) that has lost ³⁹Ar to the high-Ar-retentive secondary minerals. Alternatively, the tightly correlated apparent age and Ca/K ratios of the high-temperature plateau could be interpreted as CaCl₂-rich fluid inclusions containing excess ⁴⁰Ar.

In comparison with the volcanic rocks, the diorite dikes at Hasanbag contained very significant quantities of excess ⁴⁰Ar* (Fig. 6E; i.e., ⁴⁰Ar is not exclusively produced by ⁴⁰K decay). Therefore, the anomalously old apparent ages from some of the initial steps in some cases are probably due to excess ⁴⁰Ar. Figure 6E shows the plot of isochrons determined from the argon data from the two rock types, constrained by the isochron line needing to pass through the ³⁶Ar/⁴⁰Ar ratio of modern atmosphere (= 1/295.5); it shows clearly that the volcanic rocks have an atmospheric component caused by more extensive alteration of the sample. Thus, more altered volcanic samples yielded scattered data, often with high atmospheric Ar contents, a high MSWD, and high errors on the final age. However, the poorer-precision 106 ± 12 Ma age for volcanic sample H23 is in accord with the marginally younger age of 92.4 ± 1.4 Ma for dike H20. Thus, overall, the ⁴⁰Ar-³⁹Ar kaersutite dating indicates a Cenomanian (Late Cretaceous) age for the Hasanbag igneous rocks.

The Hasanbag rocks are largely mafic to intermediate in composition. The high loss on ignition (LOI) values of 1.41%–4.2% reflect secondary alteration of the rocks plus the presence of veining and vugs filled by calcite, chlorite, and epidote. They show small variation in SiO₂ (45.86–53.42 wt%). Variations (wt%) for other oxides are Al₂O₃ 10.45–17.22; CaO 5.21–10.3; MgO 3.24–8.71; Fe₂O₃t (as total iron) 9.7–14.83; TiO₂ 0.78–1.26; P₂O₅ 0.18–0.34; and MnO 0.15–0.20 (Table 1). Na₂O and K₂O values range from 2.7 to 5.89 wt% and <0.01–0.73 wt%, respectively (averages of 4.40 and 0.31).

Most major oxides display a clear negative or positive correlation with increasing MgO content, reflecting the vital role of fractional crystallization processes during the evolution of the Hasanbag rocks. The trends for CaO, SiO₂, and TiO₂ versus MgO are suggestive of fractionation of clinopyroxene, plagioclase, amphibole, and Ti-Fe oxides (Fig. 1 in supplementary item 3 [see ^{footnote 1}]). In the AFM (A [Na₂O + K₂O], F [FeOt], M [MgO]) triangle, all the Hasanbag samples (except for two subvolcanic samples—H15 and H29) fall in the calc-alkaline field of Irvine and Baragar (1971) (Fig. 7A), and hence most samples can be classified as calc-alkaline andesites and basaltic andesites. The Hasanbag samples plot in the calc-alkaline field on the TiO₂-MnO-P₂O₅ diagram of Mullen (1983) (Ali, 2012). Most samples have low K contents (except two samples that fall in a medium-K range—H17 and H25; cf. Le Maitre, 2002).

The ranges for Cr, Ni, and Co in the Hasanbag samples are 10–140, 2–70, and 9–67 ppm, respectively (see Table 1). Vanadium ranges from 157 to 619 ppm. Vanadium shows a significant positive correlation (predominantly for the subvolcanic samples) with Fe₂O₃. Ga shows a significant positive correlation with Zr, indicating it was concentrated in the remaining melt phase. All compatible elements (Cr, Co, Ni, and V) show significant positive correlation with MgO, indicative of an igneous fractionation trend (Figs. 2A–2D in supplementary item 3 [see ^{footnote 1}]). However, most incompatible elements are poorly negatively correlated with MgO, whereas incompatible high field strength elements (HFSEs) such as Zr, Y, Th, Nb, Ta, and Ce show clearer negative correlation trends with MgO (Figs. 2G–2H in supplementary item 3 [see ^{footnote 1}]). Moreover, K₂O displays a significant positive correlation with Rb and Ba, equated with secondary alteration. Based on the Th versus Co (Hastie et al., 2007) diagram, all Hasanbag rocks are calc-alkaline basalt, andesite/basalt, or andesite, except one subvolcanic sample (H20) that falls in the calc-alkaline dacites field (Fig. 7B).

The chondrite-normalized REE patterns for all the Hasanbag samples are similar, with enrichment in light REEs (LREEs) (normalizing values from Nakamura, 1974), with (La/Yb)_N and (La/Sm) ranges of 2.9–7 and 1.8–2.93, respectively. REE abundances vary between 10 and 100 times the chondrite values (Fig. 7C). The regular parallel pattern in all the Hasanbag samples indicates a very low secondary mobility. Although plagioclase is a significant fractionating phase, no marked Eu anomaly is observed in any of the Hasanbag rocks (Eu/Eu* = 0.92–1.11; Eu/Eu* = Eu_NÖ[{Sm_N} × {Gd_N}]; see Fig. 7C). It is also clear that the Hasanbag samples show the same trend as the mean of Hassanabad arc rocks in the Iranian Neyriz ophiolite (Babaie et al., 2001), but with somewhat higher concentration (see Fig. 7C).

N-MORB–normalized trace-element patterns for the Hasanbag samples show enrichment in the large ion lithophile elements (LILEs; e.g., Th, Ba, Rb, and Cs) and depletion in the high field strength elements (HFSEs; e.g., Nb, Zr, and Ti; Fig. 7D; Sun and McDonough, 1989), confirming the subduction-related character of the Hasanbag samples (Pearce, 2008).

The calc-alkaline arc affinities are corroborated in two diagrams: Zr-Ti and Zr-Ti/100-Y×3 (Pearce and Cann, 1973). In addition, in four tectono-magmatic discrimination diagrams proposed by Pearce (2008), Shervais (1982), Meschede (1986) (Figs. 8A and 8B), and Vermeesch (2006) to distinguish within-plate alkaline basalts (WPB), mid-oceanic-ridge basalt (MORB), volcanic-arc basalt (VAB), and ocean-island basalt (OIB), the Hasanbag samples all fall in the VAB field, with exception of two subvolcanic samples (Fig. 8D). Moreover, in the tectono-magmatic discrimination ternary diagram SiO₂/100-TiO₂-Na₂O (Beccaluva et al., 1989; see figure inAli, 2012), the Hasanbag clinopyroxenes plotted mostly in the island-arc field, and a few in the MORB field. These observations all indicate that the Hasanbag samples display a suprasubduction-zone geochemical signature.

This paper shows for the first time the presence of Late Cretaceous ophiolite-arc complexes in the Iraqi sector of the Zagros collision zone. Thus, the Hasanbag rocks, although broadly lithologically similar, are clearly older than the Eocene–Oligocene Walash-Naopurdan arc-backarc complex. This is a major new insight into the Zagros collision zone, because it provides evidence that a Cretaceous intra-oceanic subduction-related complex might have been continuous from Oman in the south to Cyprus in the north.

Intra-oceanic subduction-related crust can be preserved as fragments trapped in continental crust due to collisional orogenic events (e.g., Shervais, 2001; Vaughan and Scarrow, 2003). Along collisional orogenic belts, the nature of the preserved subduction-related crust can vary, largely due to the serendipity of the collisional geometry and events, such as the geometry of out-of-sequence thrusting, oblique convergence, or removal of part of the assemblage by erosion. Thus, ophiolites sensu stricto, fragments of arc plutonic crust, and/or volcanic carapace, and sometimes just klippe of the upper mantle, can be preserved.

Hence, the Cretaceous assemblages that were emplaced out of Neotethys onto the Arabian foreland follow this pattern. In some sections such as Oman, the obduction of the 96–94 Ma Semail ophiolite represents a spreading center above a NE-dipping subduction zone with the island-arc tholeiitic lavas at upper levels, indicating the initiation of the immature Lasail arc (Searle and Cox, 1999). Along strike to the north-northeast in Iraq, the Hasanbag igneous complex (this paper) is dominated by 106–92 Ma arc-related basalts-andesites and their hypabyssal equivalents, without any deep crustal rocks or exhumed mantle. In Cyprus, the 90–94 Ma Troodos ophiolite is from a suprasubduction setting with a section down into the upper mantle (Mukasa and Ludden, 1987).

Overall, these vestiges of intra-oceanic assemblages present in the upper crust along the eastern fringe of Arabia are a component of similar ones of Cretaceous age trapped/preserved as scattered remnants along the entire length of the Alpine-Himalayan orogenic system (Moghadam and Stern, 2011).

Overall, the Hasanbag geochemical data indicate magmatic evolution of a low-K calc-alkaline suite. Chondrite-normalized trace-element patterns show enrichment in LILEs (i.e., Sr, Rb, Ba, and Ce) and negative anomalies in Nb and Ti. These features, jointly with the high Ba/Nb (3.38–17.47) and low Nb/Y (0.3–0.46) ratios, indicate a mantle source region enriched by subduction processes (cf. Pearce et al., 1994; Pearce, 2008). The Zr/Y versus Zr modeling of Pearce and Norry (1979) suggests 15% and 25% enrichment of the mantle source by slab components (see Figs. 9A–9C). According to Ishizuka (2008), Ishizuka et al. (2006), and Reagan et al. (2010), this can involve both convecting asthenosphere and/or extremely depleted mantle with more extensive hydrous fluid input to generate island-arc mafic magmas. Furthermore, high concentrations of incompatible elements and high Nb/Yb for most Hasanbag samples imply that they have enriched (E) MORB affinity rather than flat N-MORB REE patterns. This may be a result of asthenospheric upwelling associated with subduction initiation and the subsequent formation of a suprasubduction-zone ophiolite-arc complex (Pearce et al., 1984; Shervais, 2001).

An arc setting for Hasanbag rocks is consistently indicated by several tectono-magmatic discrimination diagrams proposed by Pearce (1982), Shervais (1982), Meschede (1986), Le Roux (1986), and Vermeesch (2006). Moreover, Pearce (2008) used Nb and Th to discriminate between tectonic settings for basaltic rocks. In the Pearce model, Th is only added from slab-derived fluids/melts (Pearce et al., 1995; Hawkesworth et al., 1997), while Nb and Yb remain nearly constant. The subduction signature of volcanic and subvolcanic Hasanbag rocks is confirmed by the Nb/Yb-Th/Yb diagram (Fig. 8C), which shows that they all fall in the compositional field of arc-related rocks, above the MORB-OIB mantle array. The muted depletion of Ta and Nb (Fig. 7D) is a characteristic noted in suprasubduction-zone environments (Leat et al., 2004), and, combined with lower Ba/La ratios, this might suggest that the Hasanbag rocks formed slightly more distant from the subduction zone, possibly in the side of a backarc basin proximal to the arc, or it may just be normal variation in a heterogeneous mantle.

The chondrite-normalized REE and multi-element patterns also confirm the subduction-related character of the Hasanbag samples. Most obvious is the enrichment of the LILEs (e.g., Th, Ba, Rb, and Cs) and depletion in HFSEs (e.g., Nb, P, Ce, Zr, and Ti). Positive anomalies for Th and negative anomalies for Nb relative to other incompatible elements (Fig. 7D) are considered to represent a subduction-zone component (Gill, 1981; Pearce, 1983; Wood, 1990; Hawkesworth et al., 1991). However, K, Sr, and Pb generally display negative anomalies and show a greater variability in concentration than other LILE mobile elements. The negative troughs of K and Sr may be related to retention of amphibole and/or biotite in the source during melting (e.g., Wilson and Downes, 1991), or possibly reflect K and Sr mobility during burial metamorphism and/or modern weathering. It is also clear that the N-MORB–normalizing diagram for Hasanbag samples shows a similar trend to the Hassanabad volcanic arc rocks from Neyriz in southwestern Iran (see Fig. 7D).

A Cretaceous ophiolite-arc complex is incompletely preserved along the Zagros suture zone due to its low preservation potential, having been emplaced onto the Arabian margin during arc-continent collision and then subsequently eroded during uplift associated with the three collision events in (1) the Paleocene (ophiolite-arc obduction); (2) Miocene Walash-Naopurdan arc-continent collision (Ali, 2012); and (3) the ongoing Arabian-Eurasian continent-continent collision. The final continent-continent collision resulted in extensive nappe formation, which may have thrust younger terranes over the Cretaceous ophiolite-arc complex by out-of-sequence thrusting. Thus, most of the evidence for Late Cretaceous arc-continent collision was completely eroded or covered by allochthonous units in the later continent-continent collision along most of the Iraqi-Iranian border. Subsequent continent-continent collision resulted in further dismemberment of the Cretaceous complex by a younger generation of thrusts that re-emplaced it as out-of-sequence thrust sheets on top of the Tertiary Walash-Naopurdan magmatic arc (Fig. 10E). Thrust sheets of metamorphosed rocks from the hinterland were also transported as younger nappes that are situated above the earlier accreted thrusts and nappes in Zagros suture zone (Ali, 2012). Cretaceous ophiolite remnants are known to the south in Oman (95 Ma; Hacker et al., 1996), Neyriz (93 Ma; Babaie et al., 2006), and Cyprus (94 Ma; Mukasa and Ludden, 1987). They are found here for the first time in the northeast Iraqi Zagros suture zone and are named the Hasanbag arc complex. Therefore, it is concluded that the Hasanbag arc complex is a remnant of once laterally continuous Cretaceous ophiolite-arc rocks that developed within the Neotethys Ocean along the entire Arabian margin and that were subsequently accreted to the Arabian plate during the Late Cretaceous to Paleocene.

The following tectonic model is suggested here for the development of the Hasanbag arc complex:

• (1) Rifting along the present Zagros fold-and-thrust belt took place in Permian to Triassic time, resulting in the opening of the Neotethys Ocean (Berberian and King, 1981; Saidi et al., 1997; Sepehr and Cosgrove, 2004; Ali, 2012; see Fig. 10A).

• (2) During the Early Cretaceous, subduction of part of the Neotethyan oceanic crust toward the east and northeast developed an ophiolite-arc complex (basaltic andesites and andesites) represented by the Hasanbag arc complex (106–92 Ma; see Fig. 10B), Oman (95 Ma), and Neyriz (93 Ma). This was distal from Eurasia, which lay on the northeastern margin of Neotethys.

• (3) Collision of the Hasanbag ophiolite-arc complex with the Arabian passive margin occurred during the Late Cretaceous (see Fig. 10C). This coincided with obduction of other ophiolites, including the Kermanshah-Penjween ophiolites, over the northeastern margin of the Arabian plate (Ricou et al., 1977; Mohajjel et al., 2003; Jassim and Goff, 2006; Ali, 2012).

• (4) Another island-arc complex developed in the Neotethys to eventually form the Eocene–Oligocene Walash-Naopurdan Group when it was subsequently accreted to the Arabian margin.

• (5) Final continent-continent collision between the Iranian-Turkish and Arabian continents started during the middle? Miocene (see Fig. 10D).

Detailed field, geochronological, petrographical, geochemical, and petrogenetic studies of the Late Cretaceous Hasanbag igneous complex within the Iraqi segment of the Zagros suture zone have led to the following conclusions:

• (1) The ⁴⁰Ar-³⁹Ar results of magmatic kaersutite from Hasanbag volcanic rocks indicate an Albian–Cenomanian eruption age (106–92 Ma).

• (2) For the first time, this study has identified the Hasanbag igneous complex as a remnant of a Cretaceous ophiolite arc in the Iraqi sector of the Zagros collision zone. Hasanbag is thus unequivocally distinct from the lithologically similar but younger Eocene–Oligocene Walash-Naopurdan arc-backarc complex.

• (3) Petrographic observations reveal that Hasanbag rocks were subjected to low-grade metamorphism, but most of the igneous textures are still preserved. The Hasanbag volcanic rocks are mainly basaltic andesites to andesites, while the Hasanbag dikes are mostly microgabbro, with one diorite body recognized. Presence of the hornblende kaersutite is consistent with a suprasubduction-zone fluid-fluxing mechanism for magma generation.

• (4) The geochemical data indicate magmatic evolution of a basaltic andesite, andesite calc-alkaline suite with low-K characteristics.

• (5) Trace-element and REE chemistry confirm a suprasubduction-zone environment, with magma being derived from fluxing of the mantle wedge by fluids from a subduction slab.

• (6) Based on all the above, the Hasanbag arc complex is a remnant of a Cretaceous ophiolite-arc assemblage that developed within the Neotethys Ocean and was subsequently accreted to the Arabian plate during the Late Cretaceous to Paleocene.

This paper is a part of the first author’s Ph.D. thesis undertaken at the School of Earth and Environmental Sciences, University of Wollongong, and was supported financially by the Ministry of Higher Education and Scientific Research, Iraqi government, and the University of Wollongong’s GeoQuest research center. We also appreciated help from Saman Asi in drawing the maps and figures.