Interactive comment on “ Metamorphic history and geodynamic significance of the Early Cretaceous Sabzevar granulites ( Sabzevar structural zone , NE Iran ) ”

-Point 3. The mentioned K-Ar and Rb-Sr ages pertain to the ophiolitic sequences of the Sabzevar structural zone hosting the granulite complex as a tectonic slice. The hosting ophiolitic sequence has a Late Cretaceous formation (Shojaat et al., 2003) and Tertiary deformation ages (Barotz et al., 1978) The granulites bodies that are the object of this study are thus an exotic fragment(s) within the Tertiary Sabzevar structural zone. We agree with the arguments stated by the Referee, and we will modify the text in order to better clarify these points . The Late Cretaceous formation age for the Sabzevar ophiolitic domain was already introduced on line 20 at pg. 3 along with the relevant references.

High-grade, granulite facies metamorphism is not commonly reported in modern (Phanerozoic) subduction zone settings, which are dominated by cold subduction of oceanic plates into the deep mantle (Hacker et al., 2003;Peacock et al., 1994;Stern, 2002;Brown, 2010).In these contexts, environments of formation of mafic high-pressure granulite suites in orogenic settings have been generally ascribed to two-end member processes: arc maturation, i.e. formation in consequence of magmatic loading at the mature arc stage (e.g.Garrido et al., 2006;Berger et al., 2009); or melting remnants of a former oceanic crust in high heat-flow oceanic subduction settings (Peacock et al., 1994;Liu et al., 1996;García Casco et al., 2007).The correct interpretation of the granulite-facies metamorphism in ophiolitic sutures is thus critical to understand both processes and history of oceanic subduction at convergent plate margins.
In this paper, we present new geochemical and petrological data on the Early Cretaceous migmatitic mafic granulites found within the Tertiary Sabzevar ophiolitic suture zone of Published by Copernicus Publications on behalf of the European Geosciences Union.showing the main tectonic domains in Iran, with the main (Neotethyan) ophiolitic belts (in white) indicated (modified after Shojaat et al., 2003;Bagheri and Stampfli, 2008).The rectangle indicates the study area.CEIM: Central East Iranian Microcontinent.
northeast Iran (Rossetti et al., 2010) (Fig. 1), and we describe their pressure-temperature (P -T ) history.These data are used to shed light on the P -T regimes associated during closure of the ephemeral oceanic basins formed in the upperplate of the Neo-Tethyan subduction during the long-lasting Alpine convergence involving Eurasia and Arabia.Implications in terms of the regional geodynamics and orogenic complex evolution are also discussed.
The NW-SE trending ophiolitic belt of the Sabzevar Range is one of the Tertiary ophiolitic suture zones that surround the Central Iranian Microcontinent (CEIM; Figs.1b and 2a).This ophiolitic domain is the remnant of one of the peri-Tethyan oceanic branches (referred as the Sabzevar Ocean; Sengör et al., 1988;McCall, 1997;Bagheri and Stampfli, 2008) that bordered the northern margin of the CEIM toward the Asian Turan plate (Fig. 1b).It constitutes the largest ophiolitic massif (150 km long and 10-30 km wide) of north-eastern Iran and consists of a dismembered ophiolitic suite, with a tectonised and extensively serpentinised mantle section (harzburgite, dunite, and lherzolite) intruded by mafic magmas and a volcano-sedimentary sequence, Late Cretaceous to Paleocene in age (Alavi-Tehrani, 1977;Lensch et al., 1977;Delaloye and Desmons, 1980;Baroz et al., 1984;Shojaat et al., 2003).The geochemical signature of the ophiolitic rocks attests to a general MORB-like signature, suggesting that the oceanic lithosphere developed at an oceanic spreading centre (Baroz et al., 1984;Shojaat et al., 2003).These ophiolitic rocks dominantly occur dispersed as centimetre-to kilometre-size blocks into a highly sheared serpentinite matrix to form a major ophiolitic tectonic mélange.The blocks consist of serpentinised peridotites and intrusive and extrusive rocks, but foliated metabasites (blueschists, greenschists, and amphibolites) are also present.Syn-to-late tectonic, which are variably-sized granitic bodies, intrude the ophiolitic mélange, postdating main fabric development in the ophiolitic mélange and inducing contact metamorphism in the host rocks (Baroz et al., 1984).Another mélange unit underlies the tectonic mélange and consists of SE-verging imbricated thrust slices of red cherts and volcanic-volcaniclastic rocks, with features characteristic of a subduction-accretion complex (e.g.von Huene and Scholl, 1991) and form the frontal part of the range.The timing of metamorphism in the Sabzevar structural zone has been constrained by K-Ar muscovite and Rb-Sr whole rock and mica geochronology.These data document that granitic magmatism occurred in the Early Eocene (at about 50 Ma), synchronous with the main greenschist-to-amphibolite retrograde metamorphism recorded in the metabasite blocks (Baroz et al., 1984).
Two exposures of km-scale mafic granulite bodies were recognised within the serpentinite mélange exposed in the NW portion of Sabzevar Range, to the north of the villages of Novdeh in the north and Zarghan in the south (Fig. 2a).Structurally, the granulite bodies occur as tectonic slivers embedded within the surrounding sheared serpentinite mélange (Fig. 2b-c).The field relationships with the mélange are usually obscured by intense fracturing and faulting related to regional, post-orogenic strike-slip tectonics.The age of the peak granulite metamorphism was established through in situ U(-Th)-Pb zircon and titanite geochronology, which point to an Early Cretaceous age (ca.105 Ma) (Rossetti et al., 2010).This metamorphic age thus contrasts with the Late Cretaceous formation age of the Sabzevar ophiolites and the Tertiary age of their structuration (Baroz et al., 1984;Shojaat et al., 2003).Consequently, the Sabzevar granulites have to be considered as allocthonous tectonic slivers preserved within the Tertiary nappe edifice, and this evidence imposes re-consideration of the current tectonic models for the Alpine convergence history of the region.

Granulite texture and petrography
The granulite bodies are dark, medium to fine-grained rocks showing granoblastic groundmass or weak foliation (Fig. 3a).Their texture is dominated by the occurrence of millimetric-to-centimetric leucocratic patches and layers (leucosomes) interlayered within the mafic host rock mineral matrix (melanosome), giving the rock a migmatitic appearance (Fig. 3a-c).A list of the analysed samples, together with their geographic coordinates and mineralogical assemblages, are presented in Table 1.This study reports data on four granulite samples, selected from a representative regional sampling on the basis of: (i) preservation of the granulite facies assemblage and lack of pervasive amphibolite retrogression (sample NG353 and NG421); and (ii) evidence of retrograde amphibolite replacement (samples SZ290, EG354D).In the following sections, mineralogical abbreviations follow Whitney and Evans (2010).

Host rock
The mafic host rock makes up the bulk (>90 %) of rocks and primarily define the main, coarse foliation in the outcrop and in the hand samples.The mineral assemblage mainly consists of amphibole + garnet + clinopyroxene with minor plagioclase ± quartz.Rutile, titanite, ilmenite, zircon, and apatite occur as accessory minerals (see Table 1 for a synoptic view).The texture is granoblastic, with the exception of quartz and plagioclase that mostly occur as interstitial, curvilinear grains.Quartz also occurs as rounded inclusions in coarser grained matrix amphibole, garnet, and clinopyroxene.In samples NG353 and NG421, matrix garnets are medium-grained (1 to 5 mm) porphyroblasts and are usually poikiloblastic, hosting clinopyroxene (Cpx 1 ), amphibole (Amp 1 ), plagioclase (Pl 1 ), quartz, rutile, ilmenite (Ilm 1 ), and titanite (Ttn 1 ) as inclusions (Figs.3d-e and 4).
In particular, Amph 1 , Pl 1 , Ilm 1 /Ttn 1 , and Qz form composite inclusions hosted at the garnet cores, whereas Cpx 1 occurs in the rim regions (Fig. 4a-h).Particularly relevant for the metamorphic evolution of the Sabzevar granulites is the sequence in the Ti mineral phases with respect to garnet crystallization and matrix assemblages.The earliest Ti mineral record, preserved in garnet, is represented by titanite 1 , often preserved shielded in ilmenite 1 or in equilibrium with ilmenite 1 (Fig. 4b, d, and g).Also mm-sized euhedral ilmenite-clinopyroxene-bearing pseudomorphs after titanite are often observed (Fig. 4d).The formation of Ttn 1 predated the major garnet growth episode.This is also the case for a few single-phase Ilm 1 inclusions in garnet and for the replacement of Ttn 1 by Ilm 1 .Small single-phase Rt inclusions are abundant in garnet cores and rims (Fig. 4c and h).Accordingly, this mineral was stable during the main growth episode of garnet.As observed in some garnet grains, Rt crystallized at the expense of Ilm 1 (Fig. 4c).In contrast, no Rt is present in the matrix.It exclusively survived as shielded inclusions in garnet or in some lager ilmenite grains.After or at the very end of the main garnet growth period, abundant Ilm 2 crystallized in the matrix, partly directly at the expense of matrix Rt (Figs. 3d and 4i).Accordingly, the formation of Ti minerals can be chronologically grouped and related to major garnet growth: Ttn 1 → Ilm 1 → Rt → Ilm 2 .Both Ttn 1 and Ilm 1 predated garnet growth, Rt was stable during the major garnet growth period, and Ilm 2 postdate garnet.A second generation of titanite (Ttn 2 ) occurs in the leucosome.In the heavily retrogressed samples, garnet is porphyroclastic, usually showing strong evidence for resorption with development of Pl + Amp symplectites and final break-down to form Chl-Ep-Amp composite association (Fig. 3f).

Metamorphic evolution and reaction history
The petrographic and textural features described above show that five main metamorphic stages (M 1 to M 5 ) can be recognised in the Sabzevar ganulites (Table 2).This metamorphic evolution is primarily derived from the inclusion assemblages hosted in the garnet cores and clinopyroxene grains in samples NG353 and NG421, and by the reaction textures overprinting peak assemblages in samples SZ290 and EG354D.
The M 1 stage corresponds to the early prograde metamorphic history that is typified by the inclusion assemblage found in the garnet, i.e. it predates the main episode of garnet growth.The assemblage comprises Amp 1 + Pl 1 + Qz + Ttn 1 + Ep.Probably formed in a later stage, Ilm 1 is also enclosed in Grt, and Ttn 1 , completely pseudomorphed to Ilm 1 + Cpx + Qtz (Fig. 4d).Replacement of Ttn 1 could have been due to the prograde reaction (Frost et al., 2000): The M 2a stage corresponds to the main phase of garnet growth close to the metamorphic peak and to the crystallization of the matrix assemblage Amp 2 +Pl 2 + Cpx 2 ± Qtz.   -------▬▬▬▬▬▬▬▬▬----------------- ------▬▬▬▬▬▬▬▬▬--------Pl   -------▬▬▬▬▬▬▬▬▬----------- In particular, presence of rutile in spiral trails of inclusions in the garnet rims reflects the prograde history during garnet growth, likely after destabilisation of matrix ttn 1 and ilm 1 .This assemblage can be then tentatively ascribed to a reaction such as (Manning and Bohlen, 1991;Tropper et al., 2002;Page et al., 2003): After the main Grt growth episode, Ilm 2 crystallized at the expense of rutile in the matrix in association with Cpx 2 , Amph 2 , Grt, and Plag 2 (stage M 2b ).This assemblage typifies the main equilibrium assemblage observed in fresh and unaltered granulite.Crystallization of the granulitc garnet-clinopyroxeneamphibole matrix is associated with leucosome segregations.Leucosome textures such as those described in this study show striking similarities with those reported in Hartel and Pattison (1996), who interpreted the quartz + plagioclase leucosomes as melt textures produced by partial melting and hence product of migmatisation.In particular, the skeletal nature of the quartz and the plagioclase + quartz films argue for an internal origin of such melts (see Rossetti et al., 2010).Hence, the M 2 stage involves amphibole dehydratation melting during prograde granulite facies migmatisation of a basic protolith according to the following generalised reaction (Hartel and Pattison, 1996): Accordingly, the metamorphic stage M 2 is related to granulite formation associated with minor melt segregation.M 2 was followed by development of M 3 symplectitic amphibole, plagioclase, quartz ± ilmenite forming coronas around garnet and clinopyroxene associations.The observed reaction texture indicate garnet-consuming reaction such as (Mengel and Rivers, 1991;Willard and Adams, 1994;Zhao et al.;2001) Grt Progress of this reaction needs fluid supply probably in response to external fluid infiltration during the uplift and related retrogressive evolution (Shervais et al., 2003;Engvik et al., 2007).Development of amphibole + plagioclase kelyphites around garnet have been observed in many granulite terranes and is generally related to a predominantly isothermal decompression after metamorphic peak (e.g.Harly, 1989;Carswell, 1990;Thost et al., 1991;O'Brien, 1997).
In addition, growth of matrix ilmenite at the expense of rutile indicates retrogression of the HP assemblage according the bulk reaction (Zhao et al., 2001): This reaction is mainly pressure-dependent and proceeds to the right with decompression (Liu et al., 1996;Liou et al., 1998).
Locally, a M 4 metamorphic assemblage of Amp 4 + Ep ± Chl statically grew.This assemblage formed at the expense of Grt, Cpx, and Amph 1-2-3 , indicating the following bulk reaction (Abbotto and Green Wood, 2001): The latest stage of retrograde metamorphism is represented by the occurrence of M 5 cross-cutting prehnite-bearing veins.

Whole rock geochemistry
Nine samples were geochemically analysed to constrain the paleo-tectonic setting on which their precursor rocks formed (see Table 1 for location and constituent mineralogy).
Major element analysis shows a restricted range of composition and a sub-alkaline, tholeiitic affinity.They have low silica (41-48 wt % SiO 2 ) and potassium (0.2-0.9 wt % K 2 O) contents with Al 2 O 3 between 12.5 and 15.0 wt %, for a #Mg (molar [MgO/MgO+Fe Tot ]) of 38-65 (sample SZ290).In the TAS diagram, the studied granulite samples correspond to subalkaline, picro-basaltic to basaltic compositions (Fig. 5a).A determination of the protolith is considered to be tentative because it is difficult to assess the influence of interaction with metamorphic fluids as well as the amount of melt being extracted and removed from the rocks.A discrimination according to Irvine and Baragar (1971) and Winchester and Floyd (1976), classify the basaltic protolith as tholeiitic in composition (Fig. 5b-c).The rare earth element patterns are basically flat with a slight convex-upward distribution   ([Gd/Yb] N 0.9-1.5)and no significant Eu anomaly.The (La/Sm) N values are low (0.7-1.0) and their REE concentrations range between 10 and 50 times chondritic values (Fig. 6a).On a MORB-normalized diagram, the fluid-mobile LILE scatter widely, and element ratios based on them show little or no systematic behaviour indication some disturbance by interaction with metamorphic fluids.On the other hand, the fluid immobile trace elements show flat patterns compatible with a MORB signature (Fig. 6b).This interpretation is supported by Nb/La values ranging between between 0.5 and 1.3.In a plot of (La/Sm) N vs. Nb/La (after John et al., 2003; not shown), the samples are far from the arc magmatism fields and fall in-between the back-arc and MORB ones.Values of Th/Yb range between 0.07 and 0.16, Nb/Yb between 0.96 and 1.60; values strongly resembling those of average N-MORB (see Pearce, 2008).In a Sr/Y vs. (La/Sm) N diagram (after Berger et al., 2009), the granulites have a Sr/Y ratio significantly lower than in intra-oceanic arc lavas (exemplified by Caribbean Cretaceous lava); they plot close to the fields of MORB-type lavas (Fig. 7a).In geotectonic discrimination diagrams based on "immobile" elements for basaltic rocks, majority of the samples plot in the MORB/back-arc fields (Fig. 7b-c).
Although significantly disturbed, the evaluation of the observed signatures generally suggests MORB as protolith (cf.Arevalo Jr. and McDonough, 2010).The enrichment of incompatible trace elements could be attributed to their metamorphic evolution in consequence of interactions with enriched metamorphic fluids, melt segregation and extraction.In particular, a garnet-amphibole rich residual mineralogy during partial melting may produce the relatively high level of heavy REE (HREE) in the granulite restite (e.g.Rapp et al., 1991;Sen and Dunn, 1994;Martin et al., 2005).It is beyond the scope of the present paper to address the quantification of such processes on the geochemistry of the protolith.Basically, despite possible modification of the original major and trace element patterns caused by the metamorphic evolution,, the overall whole rock data of the Sabzevar granulites suggests a MORB-type/backarc oceanic protolith.

Mineral chemistry
The chemical compositions of minerals were determined using a CAMECA SX100 electron microprobe at the Institute für Mineralogie of Universität Stuttgart.Electron microprobe analyses (EMPA) were carried out in focused beam mode at 15 kV and 15 nA, using natural minerals and synthetic phases as standards.Concentration map for major elements (Ca, Fe, Mg, Mn, Al, Na) were also produced by stepwise movement of the thin section under the electron beam.Mineral recalculation and estimation of Fe 3+ content were made using the AX2000 program in the THER-MOCALCv.3.26 package (Powell and Holland, 2008; ftp: //www.esc.cam.ac.uk/pub/minp/thermocalc/). Selected mineral analyses and compositions are shown in Table 4.

Garnet
Representative garnet analyses are presented in Table 4a.Garnet composition is rather heterogeneous.It is essentially almandine (38-52 %), but it shows a large variation in grossular (26-34 %) and pyrope (12-30 %) components; spessartine (1-7 %) is present in minor amounts only (Fig. 8a).Representative compositional profiles and X-ray compositional maps are shown in Fig. 9. Three main types of chemical zoning are recognised (cf.Spear, 1993): (i) prograde zoning documented by a more or less pronounced (core-to-rim) bell shaped spessartine profile and pyrope increase (Fig. 9a); (ii) flat chemical profiles with retrograde zoning (Mg decrease) in the rim regions (Fig. 9b); and (iii) composite patterns (Fig. 9c).In type (ii), a rimward increase of the grossular component is observed, largely compensated by a decrease of the pyrope content.In garnet with outlines pointing to major resoprtion, the rimward chemical zoning is also typified by an increase in XFe values in conjunction with spessartine increase.Flat chemical profiles are typified by the lower spessartine and higher pyrope contents.Mg-rich garnet is generally related to temperature close to peak metamorphic conditions (e.g.Spear and Selverstone, 1983), being in line with the flat chemical profile (e.g.Spear, 1993;Ganguly, 2002).A similarly high Mg content occurs at the rim of the progradely zoned garnets that we thus attribute to the metamorphic peak.The sharp increase of the Ca and decrease of Mg contents with flat Mn and Fe contents can be explained as due to garnet growth in presence of a Ca-rich melt phase, i.e. during cooling (e.g.Spear and Kohn, 1996).The general picture described above is also complicated by ion diffusion next to both inclusions and external matrix minerals.For example, in garnets from sample SZ290, a 100-150 µm wide zone with a sharp increase in Mg-and Fe-and depletion in Ca-and Mn-components is observed close to the clinopyroxene inclusions.Finally, various degrees of diffusion/resorption processes may have influenced the garnet zoning.

Clinopyroxene
Representative microprobe analyses and structural formula of clinopyroxene are reported in Table 4b.Their composition is dominated by the diopside component (49-60 %), with minor hedenbergite (10-30 %) and Ca-Tschermaks (2-18 %); Jadeite component is negligible (<0.05) (Fig. 8b).Clinopyroxene occurring both in matrix and as inclusions in garnet shows similar composition.In all of the samples, a significant chemical zoning is observed from core to rim, which is typified by an increase of the Ca and Mg contents concomitantly with a decrease of the Al content (Fig. 9).A large variation in the Ca-Tschermaks component is also evident in the NG353 sample, in which clinopyroxene inclusions hosted in garnet contain lower Al and Na, and higher X Fe than the core of clinopyroxene dispersed in the matrix (Table 4b).

Plagioclase
Representative analyses are presented in Table 4c.Composition of plagioclase is distinctly different in the preserved and retrogressed rock types (Fig. 8c).In the first case (samples NG353 and NG421), all analysed plagioclase grains, regardless of their textural context (matrix, inclusions, leucosome), have andesine compositions (An 43−51 Ab 50−56 Or 0−1 ).In the second case, most of the plagioclase grains have been pseudomorphosed into an Solid Earth, 2, 219-243, 2011 albite-prehnite symplectite (sample SZ290).Albite also occurs as constituent of symplectite or in corona structures around garnet (sample EG354D).In several cases, matrix plagioclase shows alteration into prehnite.

Amphibole
Representative analyses of amphibole are given in Table 4d.In all samples, the composition of amphibole corresponds to tschermakite (Leake et al., 2004).In particular, amphibole from NG353 sample occurring in the matrix has the same composition as in the kelyphite structures around garnet and replacing clinopyroxene [X Mg = Mg/(Mg + Fe 2+ )].On the other hand, coarse amphibole from sample SZ290 dispersed in the matrix shows higher X Mg content (up to 0.88).Amphibole occurring as inclusion in garnet has been found in samples EG354D and NG353.It has a similar composition with respect to the ones found in matrix (just smaller X Mg values).Late-stage, post-kinematic amphibole is actinolitic in composition (Fig. 8d).

Epidote
Epidote was found both as inclusions (sample NG353) in garnet and as constituent of the late retrogressive matrix assemblage.
The P -T -X (forward) modelling of the Sabzevar granulites during the M 1 -to-M 2 composite metamorphic evolution was computed using the Perple X 07 software (Connolly, 2005; http://www.perplex.ethz.ch,v. 6.6.5, downloaded 7 January 2011).Because of the unknown amount of melt extracted from the rock, the calculation is considered as tentative and, accordingly, results on P -T conditions as semi-quantitative.

M 1 stage
For this early metamorphic stage, characterized by the assemblage Amp, Ep, Grt Ttn, lower amphibolite facies conditions can be assumed.Mineral formula and ferric iron recalculation obtained through the software AX2000, enclosed in the THERMOCALC Package.bdl, below detection limit; c, core; i, inclusion; m, matrix

M 2b stage
The progressive transition to the M 2b stage is related to growth of garnet in association with rutile.Peak conditions were calculated using the composition of (i) Cpx + Pl + Rt ± Ilm (+Qz) inclusions within the rim domain of the garnet and (ii) Grt-Cpx-Pl-Amp matrix assemblages showing textural equilibrium.The average P -T calculations, as obtained with THERMOCALC, cluster around 750-850 • C and 1.0-1.2GPa.A significantly lower temperature but similar pressure is calculated using the outer, Ca-rich garnet domains (Fig. 10).The Grt-Cpx thermometry of Krogh Ravna (2000a) and Ellis and Green (1979) (calculated for an average P = 1.0 GPa for all samples) gave similar results, in the order of 715-765 • C. Using the composition of clinopyroxene with highest Al contents (in the rims) for the assemblage indicated above, calculated conditions are about 50-100 • C higher, suggesting that peak T was higher than obtained by the applied thermometers due to postpeak re-equilibration of clinopyroxene.The GADS barometry of Perkins and Newton (1981) and Eckert et al. (1991) gave pressure estimates of 1.1-1.2GPa for samples NG353, EG354, and NG421, and 1.3 GPa for sample SZ290 (at an average T = 750 • C for all samples), respectively.The Zr-in rutile thermometry from rutile grains hosted in the garnet rim domains provides 710-800 • C (Fig. 10).

M 3 stage
The   C for sample EG354.For all samples, the highest values were obtained when amphibole with higher X Mg is chosen The pressure conditions were calculated by the application of the Kohn and Spear (1990) calibration on the amphibole and plagioclase occurring in symplectite and corona structures in textural equilibrium with garnet rims.Results, calculated in the range from 500 to 700 • C, span from 0.35 to 0.62 GPa.

M 4 -to-M 5 stage
The late retrogressive P -T conditions have been only qualitatively inferred by using the stability fields for the considered mineral assemblage.The presence of late-formed actinolite, chlorite, epidote, albite and prehnite-zeolite assemblages collectively indicate that final equilibration occurred under greenschist-to-sub-greenschist facies P -T conditions.
In particular, the late prehnite-bearing veins indicate the final M 5 stage occurred at temeprature between 200 and 280 • C for pressure less than 0.3 GPa.Furthermore, presence of wairakite-free zeolitic assemblages (heulandite) poses an upper limit for the temperature and pressure conditions during M 5 (T < 250 • C and P < 0.2 GPa; Frey et al., 1991).

Forward modelling
Pseudosections in the NCFMASHTMnO (Na 2 O-CaO-FeO-MgO-Al 2 O3-SiO 2 -H 2 O-TiO 2 -MnO-O 2 ) chemical system were calculated for the composition of sample NG421 (Table 1), containing garnet with only minor indication for corrosion.The following solid solution mixing models offered by Perple X were considered in the calculations (details in solut09.dat;PERPLEX 07; database: hp04ver.dat):Amph(DPW) for amphibole, Grt(HP) for garnet, Cpx(HP) for clinopyroxene, feldspar for feldspar, melt(HP) for the melt phase, and H 2 O for the fluid phase; rutile and ilmenite were considered as pure phases.The small amount of whole rock K 2 O has been neglected because the current amphibole mixing models do not consider K-bearing end-members.For the calculations, water and O 2 content were fixed to 1 (close to the measured value) and 0.01 wt %, respectively.A representative pseudosection is shown in Fig. 11a.
The most important parameters to reconstruct the P -T history of the Sabzevar granulites are the garnet compositions, the inclusion assemblages in garnet, and the distribution of Ti-bearing phases, with titanite, ilmenite and rutile growth tracing the different stages in the metamorphic evolution.
The M 1 episode, predating major garnet growth, is characterized by the presence of relatively large titanite crystals.Mineral formula and ferric iron recalculation obtained through the software AX2000, enclosed in the THERMOCALC Package.m, matrix i, inclusion A tentative Perple X calculation considering H 2 O-satured conditions results (not shown here) in a (fluorine-free) titanite stability field at low T conditions (<450 • C).The presence of titanite at such conditions is consistent with only minor modal amounts of garnet (<3 %), predating the major garnet growth episodes.This early formed M 1 garnet incorporated most of the Mn available in the rock, and can still be recognized as garnet core (Fig. 9).
To trace the prograde path towards the M 2b peak, the boundaries between the Ti-minerals and the pressure dependence of garnet modal content (Fig. 11b) are particularly significant.The calculations show that a relatively narrow reaction field of coexisting ilmenite and rutile exists, bounding a low-pressure, Ilm-bearing field from a high-pressure, Rt-bearing one (Fig. 11a).The textural evidence documenting rutile being late with respect to ilmenite (Ilm 1 ) indicates pressure increase.The recorded M 2a step of the metamorphic history is assumed to be preserved close to garnet cores that record the prograde chemical zoning (Fig. 9a-c).These are characterized by a Mg content of ca.0.45 c.f.u., Ca of ca.0.7 c.f.u., and inclusions of ilmenite 1 .Because of the presence of ilmenite, pressure conditions of about 0.7 GPa at 650-680 • C are estimated for the M 2a stage (Fig. 11a).These estimates are in line with those obtained from inverse thermobarometry (Fig. 10).Transition from M 2a to M 2b is typified by crystallisation of garnet rim domains with dominant rutile instead of ilmenite.An inner rim of garnet, representing M 2b , contains inclusions of Rt, high Mg content (up to 0.6 c.f.u.) and a nearly constant Ca content.These data indicate an increase in P accompanied by minor heating along a path nearly parallel to the Ca isopleths.The estimated peak conditions of this M 2 step attained at ca. 780 • C and 1.1-1.2GPa, limited by the upper pressure stability of plagioclase.Also in this case, results are in good accordance with the Zr-in-rutile thermometry and the THERMOCALC calculations (Fig. 11a).Achievement of the peak P -T conditions during M 2b was accompanied with production of a small amount melt that is now visible as leucosome.Subsequent cooling resulted in resorption of garnet that is also indicated by Mn-rich seems (Fig. 9c), at first within the rutile Table 5. Synoptic results of the conventional and multi-equilibrium thermobarometry applied to the Sabzevar granulites.stability field and then within the ilmenite field.An important constraint on the P -T path can be derived from garnet with zoning pattern such as Ng353 grt1 (Fig. 9b).This garnet type is in contact with a leucosome rim and shows little resorption.Its outer rim zoning is characterized by a significant increase in Ca (up to 0.9 c.f.u.), compensated by a decrease in Mg.Obviously, in this case the garnet outer rim adjacent to melt grew during cooling instead of being resorbed.Considering the calculated modal abundance of garnet, this feature requires some pressure increase during cooling, compatible with an anti-clockwise P -T path (Fig. 11a).In the rock matrix, ilmenite almost completely replaced rutile.It only survived as shielded (mainly mono-phase) inclusion in garnet or in larger ilmenite aggregates.The cooling path is also imposed by the preservation of plagioclase in the matrix assemblages (Fig. 11a).Crystallisation of Ttn 2 in leucosomes likely occurred close to the solidus during cooling.In contrast to early formed, relatively low-T Ttn 1 , the Ttn 2 stability during this episode probably extended to significantly higher temperature because of the presence of significant amounts of fluorine in the residual melt (Tropper et al., 2002;Troitzsch and Ellis, 2002).Given that the water content is fixed to 1 wt % during the calculation of the pseudosection, resulting sub-solidus assemblages are mostly water-deficient, "dry" assemblages which cannot account for retrograde reactions in retrogressed samples (cf.Cpx stability at low pressure) containing minerals such as epidote, chlorite, actinolite, prehnite and zeolite appear.A Perple X calculation of subsolidus phase relation with water-saturated conditions (not shown here) shows that during cooling epidote appears in the range of 550 • C during cooling.This temperature range fixes conditions for the composite M 3 -M 4 retrogression.
We would emphasise here that various sources of uncertainty affect the results described above: (i) the amount of melt extracted during in situ melting, (ii) the incomplete mineral equilibrium (involving shielding of chemical components in garnet cores and in other minerals) and, more important, (iii) the uncertainties in the thermodynamic data set.The latter aspect particularly concerns the influence of minor elements such as Ti, F, Fe 3+ on melt, amphibole and titanite compositions, as well as "extending" the application of melt thermodynamic parameters beyond the compositional range of their specified validity.A further source of uncertainty pertains to the estimation of the water content and its variation during the changing P -T conditions.Accordingly, forward modelling of the Sabzevar granulites can be considered as tentative only and the results as semi-quantitative.Nonetheless, trial calculations with variations of H 2 O and O 2 contents, as well as addition of various amounts of tonalitic melt to the effective reactive composition did not result in significant modification of the general aspects of the derived P -T path.Particularly, the P -T path of the Sabzevar granulites is significantly constrained by (i) the stability of Ti minerals and (ii) the modal content of garnet reflect-ing episodes of growth/resorption of this mineral.In addition, the general validity of the obtained pseudosection to constrain the P -T path of the Sabzevar granulites is confirmed by the overall compatibility with the results obtained from inverse (conventional and multi-equilibrium) thermobarometry.
Fig. 10.Results from conventional thermobarometry, compared with those obtained from the THERMOCALC software in the average P -T calculation mode.The latter results are shown as best fit ellipses, with error quoted at 1σ level.The grey areas show the P -T fields calculated with the THERMOCALC software that pertain to the single reactions (numbered in the inset) that involve the Ti-bearing phases relevant for the main rock forming mineral assemblages (mineral abbreviations follow Whitney and Evans, 2010).See text for further details.

P -T path: proposal of an anticlockwise trajectory
The metamorphic evolution reconstructed from the Sabzevar granulites attests to an early prograde stage starting from amphibolite-facies conditions (M 1 stage) with later equilibration under granulite-facies peak conditions (M 2 stage).The peak mineral assemblage described in this study (Grt + Cpx + Pl + Rt ± Amp ± Qz) is indicative of the Opx-free HP granulite facies (Pattison, 2003;O'Brien and Rötzler, 2003).According to thermobarometric calculations, peak metamorphism occurred at ca. 780 • C and 1.1-1.2GPa, as indicated by assemblage M 2a .This is compatible with (i) occurrence of rutile and garnet in MORBlike rocks, which indicate minimum pressure of about 0.1-0.12GPa (Ernest and Liu, 1998); and (ii) presence of Cpx instead of Opx (e.g.O'Brien and Rötzler, 2003).Calculated peak P -T conditions are also compatible with partial melting of amphibolite as they are located above the H 2 O-saturated basaltic solidus (Vielzieuf and Schmidt, 2001).
Textural evidence for in situ amphibolite dehydratation melting (see also Rossetti et al., 2010) attests to migmatisation occurring during this prograde history and during attainment of the granulite climax.In addition, according to the calculated % grt isopleths (Fig. 11b), garnet growth close to peak temperature indicates significant pressure in-crease.Even during the initial cooling episode, further pressure increase is suggested by Ca-rich/Mg-poor garnet rims (Fig. 9b), formed during melt crystallization.The pressure increase close to peak temperature, the late-M 2 cooling with some pressure increase and the late stage formation of ilmenite-bearing matrix assemblage collectively suggest a anticlockwise P -T loop (pressure axis upward; Fig. 12), departing from a prograde Barrovian-type metamorphic gradient (in the order of 25 • C km −1 ).The metamorphic peak was followed by a polyphase metamorphic retrogression (M 3 -to-M 5 stages), tracing the progressive cooling and exhumation from upper-greenschist/amphibolite facies (assemblage M 3 ) down to the stability field of prehnite and zeolite (assemblage M 5 ) (Fig. 12).

A possible geodynamic scenario
The whole rock geochemical data presented above document that the Early Cretaceous (Albian) Sabzevar granulites likely derive from a MORB-type precursor and therefore represent remnants of a former oceanic crust (hereafter referred as proto-Sabzevar Ocean to distinguish it from the later formed Sabzevar Ocean) affected by subduction zone metamorphism.The thermobarometric estimates of peak metamorphism demonstrate incipient migmatisation of the oceanic protolith already at relatively shallow depths (ca.40 km) and hence suggest for a high heat flow subduction channel (cf.Peacock and Wang, 1999).
Modelling of the thermal and rheological structure of subduction zones as well as fossil and active examples of oceanic subduction zones (e.g.Peacock et al., 1994Peacock et al., , 2005;;Peacock and Wang, 1999;Hacker et al., 2003;Conder, 2005;Bebout et al., 2007;García-Casco et al., 2007;Guilmette et al., 2008) have documented that cool geothermal conditions can not be generalized to any oceanic subduction and that "warmer" geothermal environments such as those documented in this study do not necessarily require an atypical tectonic environment in oceanic subduction settings.The most favourable conditions for a hot subduction setting pertain to slab edges and windows (Kincaid and Griffiths, 2004;Thorkelson and Breitsprecher, 2005), sub-horizontal or flat subduction (Gutcher et al., 2000), onset of subduction (e.g.Gerya et al., 2002), or subduction of a young slab or a ridge (e.g.Peacock et al., 1994;Peacock, 1996;Okudaira and Yoshitake, 2004).High heat-flow subduction settings also provide appropriate conditions to cause slab melting and generate the tonalite-trondhjemite-granodiorite magmatic suites (Martin et al., 2005, and references therein).Peacock et al. (1994) documented that young (0-5 Ma) crust that is slowly subducting (1 cm yr −1 ) into a hot mantle at constant shear may produce a subduction zone metamorphism with high T /P ratios and peak metamorphic conditions compatible with those of the Sabzevar HP granulites.Such metamorphic conditions are typical of most dynamothermal soles in ophiolites (and particularly of the early stages  , 1994 andMartin, 1999) show the dry and wet solidus of tholeiite and the inferred P -T trajectories and geothermal gradients in the subduction channel as a function of the age and/or amount of the subducted oceanic lithosphere.Trajectory 1 and 2 refers to a nascent and mature subduction, respectively.Slab melting is predicted only when young lithosphere subducts or at the early stages of subduction.The dashed grey arrow schematically depicts the P -T evolution of the Sabzevar granulites.Key to labels: G, and P, line delimiting garnet and plagioclase stability, respectively; H, hornblende-out.
of subduction zone metamorphism), which also show anticlockwise P -T paths (Spray, 1984;Jamieson, 1986;Peacock, 1987;Dilek and Whitney, 1997;Wakabayashi and Dilek, 2000;Guilmette et al., 2008).Peacock et al. (1994) also showed that aging of a nascent subduction zone and/or increase of the subducted lithosphere rapidly results in cooling of the subducted crust and the subduction channel, shifting the metamorphic gradient towards higher P /T ratios.We thus propose that the anticlockwise P -T trajectory followed by the Sabzevar granulites is the record of a progressively colder thermal regime (and hence cold exhumation) within the subduction channel in consequence of the continuous underthrusting of the oceanic material at the subduction front during the transition from infant to mature subduction of the proto-Sabzevar Ocean (Fig. 13).Rossetti et al. (2010) tentatively linked the proto-Sabzevar Ocean (of probable Early Cretaceous age) to the Sistan Ocean to the East of the Central East Iranian Microontinent (Fig. 1;cf. Tirrull et al., 1983;McCall, 1997), where Early Cretaceous (ca.125 Ma in age) subduction zone metamorphism was reported based on Ar-Ar geochronology (Footohi Rad et al., 2009).New Rb-Sr results (internal mineral isochrons from the eclogitic rocks of the Sistan area indicate a regionally consistent Late Cretaceous (about 85 Ma) age clustering, with considerably younger ages than those indicated by Ar-Ar geochronology (Bröcker et al., 2011).These new chronological data thus impose a critical re-evaluation of the former interpretation.The proto-Sabzevar Ocean can be alternatively interpreted as a small marginal basin formed in the upper-plate of the Neothetyan subduction, separated from the Sistan Ocean and with an independent evolution.This oceanic basin formed in a inner position with respect to the Tertiary Sabzevar structural zone and can be tentatively correlated with the south-central Iran ophiolitic domains.Nevertheless, formation ages compatible with the Albian age of the Sabzevar granulites are only reported from the Eastern www.solid-earth.net/2/219/2011/Solid Earth, 2, 219-243, 2011 ophiolites (Sistan suture; Fig. 1b), where an Early Cretaceous (pre-Aptian) timing for opening of the Sistan Ocean has been documented (Babazadeh and De Wever, 2004).Independent of which interpretation is adopted, by integrating our reconstruction with the tectono-metamorphic evolution of the Neotethyan subduction (Zagros Blueschists) along the Sanandaj-Sirjan Zone (Fig. 1) as reconstructed in the work of Agard et al. (2006), consumption of the oceanic lithosphere along the Iranian tract of the Himalayan-Alpine convergence zone during the Early Cretaceous testifies to different thermal structures in the active subduction systems.In particular, being the thermal structure in subduction channels and orogenic convergence zones modulated by the relative importance of heat advection to diffusion (e.g.Peackock et al., 1994;Peackock and Wang, 1999;Faccenna et al., 2001;Sandiford, 2002), conduction-dominated (the nascent and warm Proto-Sabzevar subduction, thermal gradient of ca. 25 • C km −1 ) and advection-dominated (the mature and cold Neotheyan subduction, thermal gradient of ca. 10 • C km −1 ) subduction channels can be recognised in the Iranian region.This evidence suggests a heterogeneous space-time distribution in the mode and style of oceanic subduction in the upper-plate of the subducting Neo-Thetyan ocean, a likely consequence of the different types of oceanic (mature vs. young) lithosphere that were being subducting.

Conclusions
The geochemical and petrological study of the Sabzevar HP granulites provide new insights on the tectono-metamorphic evolution experienced by the remnants of peri-Tethyan oceanic branches that surround the Central East Iranian Microcontinent (Fig. 1).In particular, evidence for a nascent oceanic subduction zone in central Iran (Sabzevar structural zone) during the Early Cretaceous is documented.
Based on the regional scenario, our results suggest that punctuated events of subduction of short-lived marginal oceanic domains accompanied the long-lasting (Mesozoicto-Tertiary) history of Neotethyan subduction along the Sanandaj-Sirjan zone.Consequently, diachronic and independent tectonic evolutions of the different ophiolitic domains surrounding the CEIM have to be taken into account for future investigations.

Fig. 2 .
Fig. 2. (a) Satellite image of the Sabzevar Range and location of the granulite exposures (inset).(b) and (c) field exposures and characteristics of the granulite/serpentinite contacts.The granulites occur as massive bodies enclosed within the serpentinite mass.

Fig. 3 .
Fig. 3. (A)-(C) Granulite texture at the meso-scale.Note the presence of millimetric-to-centimetric leucocratic layers and patches that impart a stromatic texture to the rock.(D)-(E) Matrix assemblages at the thin section scale showing poikiloblastic garnet and clinopyroxene.Amphibole-plagioclase forms the main inclusion assemblage.Quartz-Plagioclase films surround garnet.(F) Porpyroclastic garnet surrounded by coronitic amphibole and late chlorite overgrowth.(G) Prehnite-calcite vein cutting across matrix amphibole.(H) Plagioclase alteration in leucosome.

Fig. 4 .
Fig. 4. Back-scattered electron images showing mineral and textural characteristics of the Sabzevar granulites (samples NG353 and NG421).(A)-(C) Poikiloblastic garnet with details on the core-rim distribution of the inclusion assemblages.The Ti-phases inclusions consist of ilmenite-titanite (Ilm 1 /Ttn 1 ) at the garnet core (B) and of rutile that substitute ilmenite at the garnet rim (C).(D) Garnet core hosting ilmenite pseudomorphs after titanite.The inclusion assemblages also consists of clinopyroxene and plagioclase.(E) Large porphyroblastic garnet with details on the core-rim inclusion assemblages: amphibole occurs at the garnet core (F), whereas clinopyroxene at the garnet rim (G).Note that composite ilmenite-titanite (Ilm 1 -Ttn 1 ) assemblages occurs at the garnet core.(H) Rutile-bearing spiral trails of mineral inclusion in porphyroblastic garnet.Ilmenite occurs in the matrix assemblage (Ilm 2 ).(I) Poikyloblastic garnet and granular texture of the leucocratic quartz-plagioclase segregation.Titanite (Ttn 2 ) and ilmenite (Ilm 2 ) both occurs dispersed in the matrix assemblage.

Fig. 9 .
Fig. 9. Left: representative qualitative compositional cation map and quantitative cation profiles showing the variation in cation (Mg, Fe, Ca, Mn) distribution in garnets from the Sabzevar granulites.Noteworthy features are the rimward increase in Ca (grossular) and Mn (spessartine, Sps) and decrease in Mg (pyrope), which are indicative of retrograde diffusion and resorption.Alm, almandine; Grs, grossular; Py, pyrope; Sps, spessartine.Right: garnet compositions as obtained through EMPA chemical profiles across the same garnet grains (dashed white line).

Fig. 11 .
Fig. 11.(a) Representative P -T pseudosection calculated for the Sabzevar granulites in the system NCFMASHMnTO with a bulk composition (wt %) Na 2 O 1.35, CaO 10.87, FeO 15.53, MgO 6.51, Al 2 O 3 12.92, SIO 2 45.99, MnO, 0.25, H 2 O 1.0, Ti O 2 3.04, O 2 0.01 by using the Perple X07 software.The dashed grey circles enclose the P -T spaces in which the M 1 and M 2 (metamorphic climax) assemblages are inferred to be located.The dashed white arrow indicates the P -T path followed by the Sabzevar granulites during their prograde and retrograde history.(b) Predicted modal distribution of garnet in the modelled P -T field.Mineral abbreviations follow Whitney and Evans (2010) except for L, melt and W, H 2 O.

Fig. 13 .
Fig. 13.Conceptual two-stage geodynamic model to accomplish for the anticlockwise P -T evolution recorded by the Sabzevar granulites.This P -T path is interpreted as consequence of transition from an infant to a mature stage of subduction of the Proto-Sabzevar Ocean below the Turan plate.The top panel (modified and re-adapted after Peackock et al., 1994 andMartin, 1999) show the dry and wet solidus of tholeiite and the inferred P -T trajectories and geothermal gradients in the subduction channel as a function of the age and/or amount of the subducted oceanic lithosphere.Trajectory 1 and 2 refers to a nascent and mature subduction, respectively.Slab melting is predicted only when young lithosphere subducts or at the early stages of subduction.The dashed grey arrow schematically depicts the P -T evolution of the Sabzevar granulites.Key to labels: G, and P, line delimiting garnet and plagioclase stability, respectively; H, hornblende-out.

Table 1 .
Location, texture, and mineral assemblages of the studied samples.

Table 2 .
Relations between metamorphic stage and mineral growth in the Sabzevar granulites.

Table 3 .
Major and trace element composition of the Sabzevar granulite samples.

Table 4a .
Representative EMPA analyses of garnet.