Interaction of the lithospheric mantle and crustal melts for the generation of the Horoz pluton ( Ni ğ de , Turkey ) : whole-rock geochemical and Sr – Nd – Pb isotopic evidence

The Horoz pluton includes granitic and granodioritic rocks, with widespread mafic microgranular enclaves (MMEs). Petrochemically, the rocks of the pluton show calc-alkaline to shoshonitic and metaluminous to slightly peraluminous composition. The rocks also exhibit an enrichment in large ion lithophile elements, e.g. Rb, K, and depletions of high field strength elements such as Y, Lu, and Mg#, Ni, with a slightly concave-upward rare earth element pattern. Both granitic and granodioritic rocks exhibit geochemical characteristics of tonalite, trondhjemite and granodiorite assemblages, possibly developed by the partial melting of a thickened lower crust. The granitoids have high concentrations of Na2O (2.6–4.5 wt%), Sr (347–599 ppm), intermediate-high (La/Yb)N (8.2–18.1, mostly >11 ), Al2O3 (13.2–16.9 wt%, average 15.3 wt%), low MgO (0.2–1.4 wt%, average 0.84 wt%) and Co (0.7–10.3 ppm). The MMEs include higher Na2O (4.5–5.5 wt%), Sr (389–1149 ppm), Al2O3 (16.9–19.2 wt%, average 17.8 wt%), MgO (1.4–4.4 wt%, average 2.75 wt%) and Co (6.2–18.7 ppm) contents in comparison with that of their hosts. There is a lack of negative Eu anomalies, except a few samples. Both host rocks and MMEs have a low initial Sr/Sr ratio (respectively 0.7046–0.7051 and 0.7047–0.7058), low Nd value (–1.8 to –0.2 and –0.6 to 0.7 at 50 Ma) and highly radiogenic Pb/Pb ratios (39.43–39.47 and 39.39–39.54). Whole-rock chemistry and isotopic data suggest that parent magmas of both MMEs and their hosts have derived from melts of the mixing between the lithospheric mantle and crustal end members, than fractional crystallization processes in crustal levels.

Mafic microgranular enclaves provide significant information on the nature of source rocks, the genesis of granitic magma (Pin et al. 1990;Didier & Barbarin 1991; Barbarin & Didier 1992;Anderson et al. 1998;Waight et al. 2001), the coexistence of two contrasting magma types (Dorais et al. 1990;Vernon 1990), the rheology of host magmas and the tectonic environments of granitoid rocks, as well as on the interaction between the continental crust and the mantle (Didier et al. 1982). Therefore, their origin is of essential significance in interpreting the history of plutons. The Horoz pluton (HP) is a typical example of bimodal magmatism on the northern margin of the Tauride belt. We present detailed whole-rock chemical and Sr-Nd-Pb isotopic data of the MMEs and host granitoids from the HP, and use these data to constrain in granitic plutonism.

GEOLOGICAL SETTING
Turkey is an essential east-west trending constituent of the Alpine-Himalayan orogenic system and contains several continental and oceanic fragments assembled due to the closure of different Tethyan oceanic basins during the Late Cretaceous-Early Tertiary period (Fig. 1a). One of these basins in southern Turkey, namely the Inner Tauride Ocean (Gorur et al. 1984;Dilek et al. 1999;Ozer et al. 2004), was formed between the Central Anatolian Crystalline Complex (CACC) and the Tauride carbonate platform. The CACC is the largest metamorphic block exposed in Turkey and consists of Upper Palaeozoic (Kocak 1993;Kocak & Leake 1994) interlayered metacarbonate and metapelitic rocks. The ocean was then consumed as a result of north-dipping subduction and closed during the latest Cretaceous to early Cenozoic times (Parlak et al. 2013a), as evidenced by the existence of discontinuous exposures of the Cenomanian-Turonian suprasubduction zone ophiolites (i.e. Alihoca, Aladag) and mélanges by latest Cretaceous time (Clark & Robertson 2002) along the Inner-Tauride Suture Zone (Fig. 1a, b). Though the ophiolitic exposures along the suture zone are covered by the Ulukisla Basin strata, the existence of high positive magnetic anomalies corresponding to the Inner-Tauride Suture Zone (Kaynak & Akçakaya 2006) also supports the development of the Inner Tauride Ocean and associated oceanic lithosphere through the late Mesozoic and thus the derivation of the Tauride ophiolites from this oceanic root. The collision of Tauride and CACC continental blocks during the Palaeocene led to the southward transport of the alreadyemplaced ophiolites and mélanges and flysch formation together with folding. The NE-SW trending HP is situated in the eastern part of the Bolkar Mountains as a part of the Tauride Platform (Fig. 1b). The pluton formed nearby the Inner-Tauride Suture Zone and intruded into the Bolkar Mountain units, which include variably metamorphosed, Upper Permian-Upper Triassic platform carbonates with siliciclastic intercalations. The HP contains granite and granodiorite members (Fig. 2) and has sharp and discordant contacts, with hornfels formation, which suggests a shallow-crustal emplacement depth. The HP is intruded by two types of mafic dykes with sharp contacts: (i) the blackish-coloured dyke closely associated with granitoids and dismembered in places, forming smaller angular enclaves and (ii) the greenish-coloured dyke forming relatively alteration-resistant, higher topographic levels in the northwestern part of the study area. It contains also some small felsic enclaves and has sharp contacts with their granitoid host rocks, representing possibly the youngest magmatic unit in the pluton. The HP was intruded into the Upper Cretaceous Alihoca ophiolites, which include ultrabasic rocks, volcanosedimentary rocks, volcanic rocks, diabases, spilite and glaucophane-bearing schists. In comparison with the felsic granite, the granodiorite is relatively coarse-grained, less fractured/altered and has more enclaves. Based on the existence of pebbles of the Horoz granitoid (Alan et al. 2007) in the Middle Eocene clastic rocks of the Ulukisla-Çamardı basin, the age of the Horoz granitoid is constrained as the Palaeocene-early Eocene. Geochronological studies suggest a crystallization age between 49 and 56 Ma by U-Pb zircon dating (Kadioglu & Dilek 2010;Kuscu et al. 2010;Parlak et al. 2013b) and 40 Ar-39 Ar dating (Kuscu et al. 2010). The HP was unroofed due to crustal uplift and erosion throughout the Palaeogene by 23.6 ± 1.2 Ma (Dilek et al. 1997). However, a recent study (Whitney et al. 2015) suggests that the Horoz granitoid records two main pulses of cooling: (1) an initial stage at ~38-31 Ma, possibly linked with a regional event that is recorded in other crystalline rocks in Central Anatolia and (2) a later stage that may correspond to at least ~2 km of erosionrelated exhumation associated with late Miocene uplift of the southern margin of the Central Anatolian Plateau.
The MMEs in the dioritic dykes may include a 'double enclave' structure where they partially or fully contain smaller, finer-grained, more mafic enclaves. The MMEs sometimes show core-tail structures, in which the tail displays an S-shaped bend, suggesting that they were at least partly plastic when introduced into the felsic magma.

ANALYTICAL METHODS
Whole-rock major and trace element analyses of 29 samples were performed at Acme Lab. Ltd. (Vancouver, Canada). Major oxide and trace element compositions were determined by the inductively coupled plasma emission spectrometer from pulps after 0.2 g rock powder was fused with 1.5 g LiBO 2 and then dissolved in 100 mm 3 5% HNO 3 . Rare earth elements (REEs) were analysed by inductively coupled plasma mass spectrometry from pulps after 0.25 g rock powder was dissolved with four acid digestions. Analytical uncertainties vary from 0.1% to 0.04% for major elements, from 0.1% to 0.5% for trace elements and from 0.01 to 0.5 ppm for REEs.
Sr, Nd and Pb isotope compositions were determined using a Finnigan Mat 262 mass spectrometer at the GEOMAR research centre (Kiel, Germany). Replicate analyses of Sr-Nd-Pb isotopes on the same samples at GEOMAR were within the analytical uncertainties. Sr was measured in static mode and 87 Sr/ 86 Sr normalized withinrun to 86 Sr/ 88 Sr = 0.1194. NBS 987 gave an 87 Sr/ 86 Sr ratio of 0.710240±0.000008. The acid washed samples were boiled in 6N HCL for 1 h. The 143 Nd/ 144 Nd ratio was normalized within-run to 146 Nd/ 144 Nd = 0.7219 and measured in static mode where the Nd standard La Jolla yielded an average ratio of 143 Nd/ 144 Nd = 0.51196276. All Pb isotope analyses were corrected to NBS 981 (Todt et al. 1996)

Whole-rock geochemistry
The whole-rock chemical compositions of representative samples from HP host rocks and their MMEs are listed in Table 1.
The granitic rock samples of the HP plot mostly in the granite, monzogranite (adamellite) and granodiorite fields with minor tonalite, whereas samples of MMEs primarily plot in the fields of quartz monzodiorite/ monzogabbro, quartz diorite/gabbro ( Fig. 4a) with   (Fig. 4b), most of the samples from the HP rocks and MMEs plotted in the metaluminous field, only a few Horoz host rocks are found in the peraluminous field. All HP samples show in the high-K calc-alkaline features in the K 2 O versus SiO 2 diagram (Fig. 4c). Besides, the granite member of the HP displays a slightly more potassic character by plotting within the shoshonite field.
In comparison with the MMEs and their host rocks, the greenish dyke samples show more enrichment in MgO, Cr, Co, Pb, Sr, Ba, La and more depletion in Na 2 O, Nb, Ta, Yb and Lu. With increasing SiO 2 (Fig. 5), a negative correlation exists in CaO, MgO, FeOt, TiO 2 and P 2 O 5 (not shown). However, the samples are scattered in SiO 2 versus trace element diagrams (Fig. 5).

Sr-Nd-Pb isotopes
Host rocks have a low and variable initial 87 Sr/ 86 Sr ratio (0.7047) and negative epsilon values (Fig. 8a)  The 206 Pb/ 204 Pb ratios are between 19.333 and 19.362 in host rocks and between 19.691 and 19.829 in the enclaves. In the Pb-isotope ratios (Fig. 8b) diagram, although MME seems to plot in the mid-ocean ridge basalt (MORB) field, those samples are also at the Northern Hemisphere Reference Line (NHRL), indicating that the subduction-related component was dominated by the material contributed by aqueous fluids rather than by sediments. The MMEs are found on the area of Pacific MORB, while host rocks plot into the field of oceanic sediments and at the boundary of enriched mantled-II.

Source characteristics and origin of the MMEs and dykes
The MMEs and dykes show low Ni (4-14 ppm and 47-53 ppm, respectively) and Cr (mostly < 30 ppm and 190-200 ppm) values, which are lower than the Ni-Cr concentrations (Ni = 250-300 ppm, Cr = 500-600 ppm) expected for a primitive basaltic magma derived from a mantle peridotite source (e.g. Wilson 1989). All these features suggest that the studied MME and dyke magmas could have undergone fractional crystallization (Taylor & McLennan 1985) and/or crustal contamination. In the MME samples, SiO 2 shows a negative correlation with MgO, FeO t , CaO and TiO 2 , suggesting crystal fractionation of hornblende (±pyroxene) and Fe-Ti oxide (Fig. 5).
The MMEs and dykes have relatively low SiO 2 contents (54-63% and 55-57%) and intermediate to high molar Mg# (34-56 and 56-58), which is inconsistent with the partial melting of the mafic lower crustal rocks and requires a mantle-derived component. The Nb/Ta ratio is a good indicator of pressure, and the ratio decreases as pressure increases (Azizi et al. 2015). Accordingly, the Nb/Ta and Zr/Sm ratios of the MMEs and dyke samples can also be used to discriminate their formation under eclogite-facies or amphibolite-facies conditions (Hoffmann et al. 2011). The fairly low Nb/Ta (10-20) and Zr/Sm (11-51) ratios for the MME rocks from the HP suggest that they formed under garnet amphibolitefacies rather than under eclogite-facies conditions.
Most of the samples with no Eu anomalies accompanied by positive Sr anomalies (Fig. 6b) could reflect melting at pressures above the plagioclase stability field (>15 kbar, >~55 km) or plagioclase accumulation. The high Sr abundances (389-1149 and 892-909 ppm, respectively) in the MMEs and dykes also support this suggestion. However, the samples have high Y (23-76 and 21-22 ppm) and correspondingly low Sr/Y (9-24 and 41-42 ppm), which indicates that plagioclase was possibly in the residue. Unfractionated HREE (and Y) patterns generally suggest that the mafic magma was possibly produced outside the garnet stability field (i.e. plagioclase stable without garnet; Drummond & Defant 1990;Rapp et al. 1991;Springer & Seck 1997;Martin 1999;Pe-Piper et al. 2002). But, garnet as a residual mineral would be able to produce (Gd/Yb) N ratios > 1 (e.g. Martin 1999;Klein et al. 2000;Martin et al. 2005). Therefore garnet may exist in the source of the mafic rocks. Accordingly, the experimental melting of metabasalts under fluid-absent conditions (Rapp et al. 1991;Rapp & Watson 1995) indicates that pressures >0.8 GPa are required to stabilize garnet, and ≥1.2 GPa for garnet throughly replaces plagioclase. Alternatively, most of the mafic rocks may have been derived from sources located at depths between 30 and 44 km by assuming 1 kbar = 3.7 km for the continental crust (Tulloch & Challis 2000). The amphiboles from the enclaves yield a maximum pressure of 4.1 ± 0.6 kbar at 730 °C (Kocak et al. 2011), suggesting the crystallization of the mafic magma at least at 15 km. The existence of the greenish-coloured dyke resembling enclaves in adjacent granitoids in the HP may suggest that the magma of the MMEs can exist either independently of, or as a separate layer in, their host granitoid magma bodies. In general, the dykes exhibit a similar pattern to that of the host rocks, with a high (La/Lu) N ratio (10.4-11.3). In comparison with the scattered smaller MMEs, the dykes seem to be less differentiated (more enrichment in MgO, Cr, Co), but more enriched in Pb, Sr, Ba and LREEs. This may indicate either distinct parental magmas, or different mechanisms or degrees of interaction of the mafic magma with the partially crystallized host-granite, or both. Along with their identical occurrences with the MMEs, this precludes that the dykes correspond to several pulses of mafic magmas. The higher content of large ion lithophile elements (LILEs) and LREEs in the dykes could suggest stronger interactions with the granites, diffusion of water and alkalis, etc. A pronounced negative Nb, P, Ti anomaly ( Fig. 6b) may also support this suggestion.

Magma mixing versus restites or autoliths
Based on their mafic composition, the MMEs could have a cumulate or autolith origin (e.g. Noyes et al. 1983;Chappell et al. 1987;Clemens & Wall 1988;Shellnutt et al. 2010;Dahlquist 2002;Donaire et al. 2005), which ignores the grain size differences between the MMEs and host granitoids. Some major and trace elements, such as Al 2 O 3 , Na 2 O, Rb, Ba, Sr and Zr (Fig. 5) exhibit nonlinear variations. Among these, SiO 2 versus Ba and Sr (Fig. 5) is of particular interest, in which Ba and Rb contents change significantly for little change in SiO 2 in the MMEs. Dispersion on the diagrams is of paramount evidence of biotite 'cumulus'. The small amount of biotite may induce an increase in Ba and Rb contents in mafic samples due to their high partition coefficients (K D ) for Ba (6.36, Philpotts & Schnetzler 1970) and for Rb (3.53, Matsui et al. 1977). Plagioclase and K-feldspar have low K D for Ba (0.36, López-Ruíz & Cebriá 1990) and for Rb (0.07-0.76, Icenhower & London 1996), respectively, therefore they are unlikely to cause this enrichment. However, the MMEs are fine-grained, 10 to 20 times smaller in comparison with the same phases in the host granitoid and have low Ni and Cr, which suggests that the enclaves as a whole cannot be a cumulate of the pluton itself. It is also remarkable that all the enclaves and host rocks have similar total REE concentrations and sub-parallel REE patterns, which are inconsistent with the autolith model. The presence of MMEs in the granitoids could also be indicative of the evolution of the HP through the restite unmixing mechanism (Chappell et al. 1987;Chen et al. 1989;Chappell & White 1992;Chappell 1996;White et al. 1999). The field characteristics of the MMEs, and lack of linear trends for K 2 O (Fig. 4c), Ba, La, Zr with SiO 2 (Fig. 5) exclude restite origin. Therefore, the MMEs in the studied HP could have developed mostly by mingling/mixing between near-contemporaneous mafic and felsic magmas (e.g. Vernon 1984;Wiebe et al. 1997;Barbarin 2005;Hawkesworth & Kemp 2006;Kocak 2006;Feeley et al. 2008;Chen et al. 2009;Kocak et al. 2011;Liu et al. 2013).
The magma mixing process could take place before or after the injection of the enclave-forming magma into the felsic host magma. Since the MME samples are mostly non-porphyritic, fine-grained and enclosed in another enclave, the mixing process probably happened prior to the injection of the enclave-forming magma. Furthermore, the mafic enclaves are characterized by relatively low Mg, Ni and Co, suggesting that they were much evolved before their injection into the host felsic magmas. This implies that significant fractionation of hornblende (± pyroxene) had occurred before and during the process of crustal contamination/magma mixing at depth. The MMEs are characteristically enriched in P, Ti, Y, Nb, and HREEs, possibly due to selective interdiffusion of these elements into the less polymerized magmas. These elements were consequently concentrated in apatite, titanite and hornblendes due to their high K D for these elements (López-Ruíz & Cebriá 1990;Klein et al. 1997), keeping their low activity in the melt. Such low activity in the mafic melt gives rise to the continuity of 'Uphill' diffusion, as described for K due to crystallization of biotite by Johnston & Wyllie (1988). Selective diffusion of these elements may be attributed to the crystallization of biotite as cumulate. The Ba depletion in most enclaves could be connected to the complicated feldspar transfer processes at the contact with host magmas and somewhat to the dilution effect induced by the inward migration of Si and alkalis as suggested by some researchers (Bussy 1991;Debon 1991;Orsini et al. 1991).
In general, MME samples have higher  Nd than their hosts (Fig. 8a, Table 2), suggesting isotopic equilibration between mafic and felsic magmas, which is usually more easily achieved than chemical equilibration since isotopic exchanges proceed more quickly than chemical exchanges (Lesher 1990). The MMEs have generally 87 Sr/ 86 Sr values close to or in the range of that for the respective host granite, suggesting a granite-enclave interaction (Fourcade & Javoy 1991). The MMEs and host rocks have distinct 206 Pb/ 204 Pb, but similar 208 Pb/ 204 Pb values (Fig. 8b). Small variations in the amount and composition of Pb contributed to the mafic melt by the relatively elevated 206 Pb/ 204 Pb of zircon compared to 208 Pb/ 204 Pb would result in large relative changes in 206 Pb/ 204 Pb with only small changes in 208 Pb/ 204 Pb.

Source characteristics and genesis of the host granitic rocks
The host granitic rocks are characterized by pronounced negative Nb, Ba, P and Ti anomalies but are enriched in Rb, Th and K. These features are in accordance with those of typical crustal melts, e.g., Himalayan granites (Harris et al. 1986) and granitoids of the Lachlan Fold belt (Chappell & White 1992), and the subduction component. However, the host rocks have relatively low Sr (t) (0.7067) and high ε Nd(t) values (-0.2 to 1.8) ( Table 2, Fig. 8a), suggesting mantle material involved in the generation of the HP. Similarly, the MMEs in general, have an initial 87 Sr/ 86 Sr ratio and Nd isotope ratios similar to those of their hosts, suggesting significant input of a lithospheric mantle-derived component during magma generation. The granitoids show high-K-shoshonitic and I-type characteristics with a wide range of silica content (SiO 2 = 64-74 wt%), relatively low-intermediate Mg# (22-41) and low Ni content (1.4-10.0 ppm), all of which may indicate that they could be developed from the mixing of the lower crust and mantle-derived magmas (e.g. Barbarin 1999). Hence average Nb/Ta ratios are 17.5 for mantle-derived and 11-12 for crustal-derived magmas (Green 1995), Nb/Ta ratios for the felsic samples vary between 10.05 and 18.12, suggesting crustal-and mantle-derived magmas in the generation of the HP.
Both MMEs and felsic samples show colinear variation in the Harker diagrams, suggesting that the host granites and MMs/dykes are possibly variably differentiated products of the same parent magma which derived from mixing melts of lithospheric mantle and crustal components. SiO 2 increases with decreasing MgO, FeO t , CaO and TiO 2 and P 2 O 5 , suggesting fractionation of hornblende (± pyroxene), Fe-Ti oxide and apatite. Amphibole has a high K D for heavy REEs, but even higher for the medium and heavy REEs (such as Dy); therefore, amphibole fractionation can be traced by decreasing Dy/Yb ratios with differentiation (Davidson et al. 2007a(Davidson et al. , 2007b(Davidson et al. , 2008. Accordingly, the amphibole fractionation is indicated by concave-upward REE patterns without significant Eu anomalies (see Tepper et al. 1993, Fig. 7a, b). In the host rocks, Zr and P 2 O 5 show negative correlation with SiO 2 , suggesting zircon and apatite fractionation. Aplitic suites on the HP most probably represent such comagmatic highly differentiated late-stage melts.
The existence of MMEs with mode of occurrence and mineralogical (Kocak et al. 2011) and geochemical characteristics suggest mafic-felsic interaction and mingling (Barbarin & Didier 1992;Barbarin 1999;Ferré & Leake 2001;Kocak 2006) by the injection of hot mafic magma into felsic magma (source mixing). Langmuir et al. (1978) showed that in the ratio-ratio and ratio-element plots, data consistent with mixing lie along a hyperbolic curve for both isotopic and elemental ratios, while a linear array forms when the ratios of the concentrations of the two denominators are the same for all data points. In the samples, these characteristic hyperbolic mixing arrays are observed in plots of Al 2 O 3 /CaO versus Na 2 O/K 2 O, Ti/Ba versus Ti, and a linear trend is observed in a plot of Al 2 O 3 /CaO versus Na 2 O/CaO (Fig. 9a-c). Accordingly, mafic (lithospheric mantle) and felsic (crustal component) magma mixing may alter both the elemental and isotopic compositions of magmas prior to the assimilation/fractionation processes.  It has also been suggested that I-type granites most likely form by the mixing of crustal materials and mantle-derived magmas rather than by the remelting of ancient meta-igneous crustal rocks (Kemp et al. 2007;Li et al. 2009;Zhu et al. 2009;He et al. 2010). Besides, relative heterogeneity of the initial Sr ratios (0.7046-0.7051) in the host rocks could be attributed to a difference in the degree of contamination of magmas with upper crustal materials. Kadioglu & Dilek (2010) suggested that the HP shows chemical characteristics of high-Al adakitic compositions, which could have formed by the partial melting of the hydrated lithospheric mantle and the amphibolitic mafic lower crust that was triggered by delaminationinduced asthenospheric upwelling. However, all samples from the HP usually have lower Mg# [(molar 100  MgO/ (MgO + FeOt)) < 0.41], Ni (~4 ppm) and Cr (~9 ppm) (Fig. 10a), and higher K (~ 3.6 wt%), Ba (~712 ppm) and Rb (~86 ppm) contents than that of adakites. The samples have also high Sr contents and are found on the TTG area, rather than on the arc one in Fig. 10b.

Adakitic versus TTG
It is widely accepted that TTG magmas were created by the partial melting of hydrous metabasaltic rocks transformed into garnet-bearing amphibolite or eclogite, under a variety of fluid conditions (Sen & Dunn 1994;Zamora 2000). Experimental studies show that the partial melting of the mafic lower crust could produce metaluminous granitic magmas regardless of the degree of melting (Sen & Dunn 1994;Wolf & Wyllie 1994;Rapp & Watson 1995). High abundances of Al 2 O 3 (≥19 wt%) in an amphibolite-derived liquid are the result of high H 2 O (water-saturated) and/or high anorthite contents in the mafic protolith source (e.g. fig. 13 in Beard & Lofgren 1991; fig. 9 in Wolf & Wyllie 1994). Nevertheless, host rocks have lower Al 2 O 3 contents (13.17-16.4 wt%) than the liquids developed during H 2 O-saturated amphibolite partial melting experiments. Accordingly, granitoids from the HP were possibly formed under fluid-absent/vapourabsent conditions (with the only H 2 O derived from the breakdown of hydrous minerals) and/or low anorthite contents in the mafic source. Figure 11 shows that the granitoids, particularly granites, could have formed by low-pressure (100-700 MPa), 20-50% dehydration melting of a basaltic/amphibolitic source. Both the mafic rocks and dykes contain high Al 2 O 3 and plot (Fig. 11) between the fields of 'low water basalt melting' and '1000 Mpa, no-water melting of a basaltic/amphibolitic source', suggesting relatively higher-pressure conditions in comparison with their host rocks during the partial melting event.
The TTGs require two main mechanisms to account for their petrogenesis: (1) the partial melting of the subducted oceanic crust (i.e. slab melts) in a convergent margin with usually higher Mg# values and Cr and Ni concentrations (e.g. Martin 1986Martin , 1999Drummond & Defant 1990;Foley et al. 2002;Kamber et al. 2002;Smithies et al. 2003) due to the interaction of the slabderived melt with the overlying mantle wedge during ascent (Rapp et al. 1999) or (2) the melting of the thickened mafic crust or underplated basalt with low Mg# values and low Cr and Ni concentrations (e.g. Atherton & Petford 1993;Petford & Atherton 1996;Rapp et al. 1999;Smithies 2000;Condie 2005;Smithies et al. 2009). In Fig. 12a, b, samples fall mostly in the field of adakites derived from the partial melting of the thick lower crust and metabasaltic and eclogite fields, rather than in that of adakite rocks derived from the partial melting of the delaminated lower crust. The data for adakites worldwide exhibit that typical slab melts have low Rb/Sr ratios (0.01-0.05); this is in contrast with the wide range of Rb/Sr ratios (0.01-0.4) for the adakitic rocks that developed from the thickened continental lower crust (Huang et al. 2009). Hence, the relatively higher Rb/Sr ratios (0.09-0.3 in host rocks, 0.04-0.16 in MMEs) of the samples from the HP are in accordance with their derivation from the thickened lower continental crust.  Rapp et al. (1999) and Smithies & Champion (2000). (b) SiO 2 vs Ni. Boundaries are from Wang et al. (2006).

Tectonic setting
I-type post-collisional granitoids with mantle-crust signature develop in many tectonic settings, such as intracontinental rifting (Vorontsov et al. 2004;Li et al. 2005;Shu et al. 2005), back-arc basins (Hochstaedter et al. 1990;West et al. 2004), island arc (Geist et al. 1995;Qian & Wang 1999), active continental margins (Donnelly & Rogers 1980) and the rifting of the passive margin (Oberc-Dziedzic et al. 2005). All MMEs and greenish dyke have lower Nb/U (average 5.9 and 5.1) than an average continental crust (Nb/U = 8.4; Rudnick & Fountain 1995). Both MMEs and granitic rocks are characterized by pronounced negative Nb anomalies, positive Pb anomalies (Fig. 6) and enrichment in LILEs and LREEs. The negative anomalies in Nb, Ti and P are characteristic of subduction-related magmas, usually thought to have resulted from the relative enrichment of the mantle source by influx of LILEs through slab dehydration (e.g. McCulloch & Gamble 1991). Similarly, the host granite displays spikes in Cs, Rb, K and troughs in Nb and Ti (Fig. 6), which may represent the continental crust developed by the chemical differentiation of arcderived magmas (Taylor & McLennan 1995). Besides, low La/Th (1.2-5.1) and medium-high Ba/Nb (24-145) are also typical for the rocks formed in relation with the subduction zone (Sun 1980).
The most mafic compositions in the granite (lowest in SiO 2 , and highest in MgO and Co) have the highest K 2 O and Na 2 O contents as well as anomalously high LREE, P, Zr and Th contents and slight negative Eu anomalies, which are characteristic of A-type granites. However, they differ from A-type granites in their unelevated Rb/Sr contents or intermediate-high Ca and Sr contents (Kemp & Hawkesworth 2003) as well as unelevated Zr + Nb + Ce + Y contents (mostly <350 ppm).

GEODYNAMIC IMPLICATION
Horoz granitoids have lower radiogenic ( 87 Sr/ 86 Sr = 0.7045-0.7051) and higher  Nd values (-0.085 to -1.75) than granitoids from the CACC ( 87 Sr/ 86 Sr = 0.7080-0.7096;  Nd = -4.8, -6.7, Ilbeyli et al. 2004), probably owing to a combination of upper crustal contamination and heterogeneity of the magma source. Large differences in isotopic data of granitoids from the HP and Karamadaz pluton, and granitoids from the CACC may imply that two groups of magma developed in relation with the closure of the Inner Tauride Ocean and the Izmir-Ankara-Erzincan Ocean, respectively (Kocak 2008). The Inner Tauride Ocean started to develop as early as the Jurassic between the CACC to the east and the Taurides to the west, and consumed by an intra-oceanic subduction northwards (Gorur et al. 1998) along the Inner-Tauride Suture Zone during the latest Cretaceous to early Cenozoic times. Parlak et al (2013b) suggest that the HP could have been formed as a result of hard collision (continent-continent collision) after soft collision (collision of the passive margin with the subduction trench and subsequent slab break-off). However, we suggest that the HP was emplaced after the last stage of oceanic subduction, or at a hiatus of the oceanic subduction at ~50 Ma. The pluton then possibly underwent cooling in two main pulses, ~38-31 Ma and late Miocene (Whitney et al. 2015).

CONCLUSIONS
From combined field, geochemical and isotopical studies it has been concluded that the mantle-derived mafic magmas from which the MMEs crystallized were likely mostly formed by mafic-felsic interaction and mingling, or prior to the mixing crystal fractionation of hornblende (± pyroxene) and Fe-Ti oxide. The MMEs usually underwent geochemical and Nd-Sr isotopic equilibration with their host granitoids, with resultant K, P, Ti, Y, Nb and HREE enrichments. The greenish dykes are distinct from the MMEs and display stronger interactions with the granites.
The granitoids have both crustal (distinct negative Nb, Ba, P and Ti anomalies but enriched in Rb, Th and K) and lithospheric mantle [low Sr (t) (0.7067) and high ε Nd(t) values (-0.2 to 1.8)] geochemical and isotopic signatures. They exhibit geochemical characteristics of TTGs, which were possibly created by the dehydration melting of a basaltic/amphibolitic source in a thickened lower crust. The parental granitic magma underwent the mixing of mantle-derived mafic magma and crustal felsic magma, coupled with fractional crystallization during magma ascent before emplacement. Relative heterogeneity of the initial Sr ratios (0.7046-0.7051) in the granitoids could also indicate contamination of magmas with upper crustal materials.
The HP granitoids differ from granitoids of the CACC in their lower radiogenic ( 87 Sr/ 86 Sr = 0.7045-0.7051) and higher  Nd values (-0.085 to -1.75), in relation with the combination of upper crustal contamination and/or heterogeneity of the magma source.