Chemo-stratigraphy, petrology and U-Pb geochronology of South-eastern part of the Yuntdağ volcano (Karakılıçlı – Manisa) in Western Anatolia

ABSTRACT There are a number of volcanic eruption centres that form large caldera structures, and possibly feeding from relatively large magma chambers during the Miocene period in Western Anatolia. One of them is Yuntdağ volcano, and its products have formed a thick volcanic succession covering large areas. In this study, an area where the volcanic sequence is seen well in the south-eastern part of the volcano was selected. Geochemistry and geochronology analyses were carried out along a line between Kalpakkaya and Çamlık Hills in order to better understand the petrological evolution of melting, assimilation, storage and homogenisation processes in the magma chamber (and/or chambers) feeding the volcano within a limited time period (~4 Ma). Samples acquired from the bottom, middle and upper portions of these sections display Early-Middle Miocene ages of 17.64 ± 0.20, 17.22 ± 0.15, 16.16 ± 0.17 and 16.36 ± 0.13, 15.79 ± 0.71 and 13.61 ± 0.20 Ma, respectively. Although, felsic crystal melts, magma mixing and multiple mafic inputs were the dominant processes, fractional crystallisation was less efficient. Also, geochemistry data also indicate that the studied volcanic rocks generated by the melting of the lithospheric mantle, by slab retreat and roll-back of the Hellenic – Cyprus slab, and evolved in shallow magma chambers during the Early – Middle Miocene.

There are several volcanic eruption centres (i.e. Yamanlar, Foça, Dumanlıdağ, Örpekkaya, Dikili and Yuntdağ volcanoes) forming important large caldera structures, and possibly feeding from large magma chambers for the Western Anatolian Miocene magmatism (Akay & Erdoğan, 2004;Altunkaynak et al., 2010;Karaoğlu, 2014). This study examined two successive volcano-stratigraphic levels around the Karakılıçlı volcanic field (south-eastern part of the Yuntdağ volcano, north of Manisa, Figure 2), revealing the variation of the chemical compositions of these volcanic sequences, and aimed to understand the changes in a large magma chamber between 17.64 ± 0.20 Ma and 13.61 ± 0.20 Ma (at about 4 My) periods in the Western Anatolian Volcanic Province (WAVP). The different lava flows and pyroclastic units in the volcano-stratigraphic sections (the Kalpakkaya and Çamlık Hills (Figure 3)) were sampled from bottom to top, and whole-rock geochemistry analysis were done. Moreover, it is the first time, zircon U-Pb dating for the studied volcanic successions have been conducted on lower, middle and upper portions in order to better correlation of the spatial and geochemical evolution of the volcanism in the region. The chemo-stratigraphy sections were prepared to understand the petrological evolution of melting, assimilation, storage and homogenisation (MASH) processes of the studied volcanic rocks.  (Okay & Tüysüz, 1999). b) Distribution of Oligocene -Miocene volcanic rocks in Western Anatolia (simplified from MTA 1/500.000 scaled geology map). Abbreviations: IPS: Intra-Pontide suture, IAES: 3 İzmir-Ankara-Erzincan suture, ITS: Inner Tauride suture.

Zircon U-Pb LA-ICP-MS geochronology
Six different samples were selected from the different layers of the two measured sections (three samples per each section) in the region. For each sample, 2 kg of samples were crushed in a jaw crusher, and their grain size was reduced below 250 microns. The sample size of 63 to 250 micron was selected for magnetic separator and heavy liquid (sodiumpolytungstate) to separate zircon minerals. Finally, hand-picked zircon grains mounted into the epoxy resins under a binocular microscope. After polishing, zircon grains imaged by using optical cathodoluminescent (Citl CL8200 Mk5-1) at the Istanbul Technical University, Geochemical Analysis Laboratories (ITU-JAL). By the guide of these images, the specific spot has been selected for geochronology analysis. Zircon U-Pb analyses were conducted by using Perkin Elmer NexION 2000 ICP-MS combined with the ESI NWR-213 solid phase laser established at the Geochronology and Geochemistry Laboratory of Istanbul University-Cerrahpaşa. Spot size of the zircon grains vary between 25 and 40 micron, and helium was used as carrier gas (0.5 l/sec). Zircon reference material 91,500 was selected as a primary standard, and AusZ-10 (38 Ma) and MudTank (750 Ma) as secondary reference materials. All the measurements were evaluated by the Iolite (version 2.5) data reduction software (Paton et al., 2011) package.

Whole-rock geochemistry
Twenty-three samples were selected for the wholerock geochemistry analyses that represent the two different stratigraphic sections. The major oxides of Figure 2. Geological map of the study area (Adapted from Dönmez et al., 2014;Türkecan, 2015), coordinate system: WGS1984_UTM_Zone35N). these samples were analysed by the X-ray fluorescence (XRF) spectrometer (Bruker S8 Tiger Wavelength Dispersive XRF) at the Geochemistry Analysis Laboratory of Istanbul Technical University (ITU-JAL). The major oxide values were analysed on samples, which turn into fused glass discs that were created by the LiBO 2 fusion in the platinum-golden crucibles. The loss on ignition values were calculated from the mass loss after the sample was heated for 2 hours at 1050°C. The trace and rare earth elements of these samples were measured by using a Perkin Elmer ELAN DRC-e at ITU-JAL, and Thermo XSeries II ICP-MS established at the Geochemistry Laboratory of General Directorate of Mineral Research and Exploration.

Description of stratigraphic sections
In order to describe in detail, the south-eastern part of the Yuntdağ volcanics, two successive stratigraphic sections measured along a line between Kalpakkaya and Çamlık Hills (in the Karakılıçlı volcanic field) was prepared where the characteristics of the volcanic sequence are best preserved (Figure 3, 4), and samples were taken from volcanic rocks that different lithological features. The Kalpakkaya Hill section has been started with reddish-grey porphyritic dacite from the bottom (Figure 3a,b,c)). The section continues with porphyritic textured and banded dacitic lavas that consist of thin tuff interlayer. The section continues monotonously and displays approximately at 250metres thicknesses. On the other hand, the Çamlık Hill section starts with porphyritic dacite with pyroclastic inliers in the bottom and overlain by porphyritic and banded dacite, contrary to Kalpakkaya Hill section, the reddish-brown and grey andesitic lavas are situated at the top of the section (Figure 3d,e,f, 4).

Petrography
Main mineral assemblages of the dacidic, and andesitic lavas observed in volcano-stratigraphy composed of plagioclase (oligoclase-andezine), hornblende (oxyhornblende and/or kaersutite), biotite, rarely clinopyroxene (in only andesitic lavas), a small amount of sanidine, and opaque minerals in glassy (rarely microlitic) groundmass ( Figure 5). They mainly display porphyritic, glomerulo porphyritic, vitrophyric and microlithic textures. Plagioclases are the main rockforming minerals in lava samples, they show euhedralsubhedral shapes, twinning, and have oscillatory zoning. Hornblende is the most common dark-coloured mineral in lavas. Hornblendes are generally reddish brown in colour, and have euhedral -subhedral shapes, and are partially or completely oxidised phenocrysts from the edges. In some places they are in the form of ghost crystals due to being resorbed by magma. Biotite, is the second most common dark coloured mineral, it is fox red -reddish brown in colour, and consists of platy crystals with on direction cleave planes. In some places, radioactive halos are observed around zircon inclusions. Clinopyroxenes have only been seen in andesitic lavas, and are usually composed of small anhedral crystals. While the secondary mineral phase is represented by chlorite, calcite, and clay minerals, accessory phase represented by mainly zircon.

Zircon U-Pb crystallisation ages
The selected zircon grains for dating mainly show transparent, pale brown colours and display prismatic shapes. CL images mostly display magmatic oscillatory zoning and mainly homogeneous ( Figure 6). Considering the CL images, both the core and edge zones of zircon grains were measured. In some zircon grains, measurements were made in different zones. No inherited zircon was found, and it was observed that measurements made in different zones were generally close to each other. The results of the zircon crystallisation ages for the Kalpakkaya and Çamlık Hill sections (three samples from lower, middle and upper levels per each section) are outlined below.  (Figure 7c).

Section of Çamlık Hill
Çamlık Hill section displays quite similar age population like the Kalpakkaya Hill section.
The Concordia age calculated as 15.79 ± 0.71 and the mean age deduced as 15.9 ± 1.1 Ma (Figure 7e).

Whole rock geochemistry
The results of the major oxide and trace element analysis of twenty-three samples from Kalpakkaya and Çamlık Hill are given in Table 2. These volcanic rocks show narrow SiO 2 contents ranging from 61.70 to 66.53 wt.%, TiO 2 from 0.68 to 0.85 wt.%, Al 2 O 3 from 13.88 to 16.77 wt.%, Fe 2 O 3 from 4.48 to 6.19 wt.%, MgO from 1.10 to 2.28 wt.%, CaO from 2.86 to 5.80 wt.%, Na 2 O from 2.39 to 3.36 wt.% and K 2 O from 2.94 to 3.80 wt. % (Table 2). Na 2 O/K 2 O ratios of the Kalpakkaya Hill volcanic units are vary from 1.02 to 1.10, and they have slightly sodic character. In contrast, Çamlık Hill volcanic units display lower Na 2 O/K 2 O ratios and have more potassic character. Magnesium number (Mg#) of the samples are between 30 and 49 (Mg# = 100 x molar Mg/(Mg+Fe)). The results of the analysis are plotted in total alkali (Na 2 O+K 2 O) versus SiO 2 discrimination diagram (Le Bas et al., 1986), the studied volcanic rocks mainly plot within the dacite, and rarely andesite fields ( Figure 8a). On the K 2 O versus SiO 2 diagram (Peccerillo & Taylor, 1976), all of the samples fall into the high -K calc-alkaline series field ( Figure 8b).
On N-type MORB normalised spider diagrams, all of the volcanic samples display apparent enrichment in LILEs (Large Ion Lithophile Elements; Sr, K, Rb, Ba, Th) and much lower but still prominent enrichment in LREEs (Light Rare Earth Elements; La, Ce and Nd). On the other hand, HFS (High Field Strength) elements (Tb, Ti, Y and Yb) show apparent depletion ( Figure  9a). There is clear depletion through Th, Ta, and Nb and a similar depletion also be found through La to Nd  Figure 9a). In addition, Hf, Zr, and Sm elements display nearly flat patterns (Figure 9a). Even the Nb and Ta display very apparent depletion relative to the neighbouring elements, they still enriched relative to the N-MORB values (Figure 9a). The apparent Nb and Ta anomaly mainly attributed to the subduction-related arc settings while it may have been generated by the relict metasomatized portions generated in the collision-related settings.
Chondrite normalised rare-earth element (REE) variation diagrams of the volcanic units display similar patterns for all series. In the selected units of the both volcanic sections, light rare earth elements (LREE) of the samples display relative enrichment regarding to the other elements and the majority of the heavy light rare earth elements (HREE) display similar values to the chondritic values ( Figure 9b). Furthermore, in all series there is an apparent negative   Pearce et al. (1984) that using Nb vs Y values, our samples plotted along the volcanic arc granites (VAG) and syncollisional granite field and one samples plotted along the within-plate field (Figure 10a). At the tectonic discrimination diagram of Harris et al. (1986) that utilising Rb/30 -Hf-3xTa; our samples dispersed along the volcanic arc and late and post-collisional fields (Figure 10b). The Th/Yb versus Ta/Yb (Schandl and Gorton 2002) reveal that the volcanic units were generated along the active continental margin ( Figure   10c). These diagrams imply that the magma source area of the studied volcanic samples were affected by the subduction related modifications and possibly effected by crystal-related modifications.

Chemo-stratigraphy
The volcanic rocks from the Kalpakkaya, and Çamlık Hill sections have mainly intermediate (mainly dacitic and rarely andesitic) composition. The Harker variation diagrams can be used to assess various magmatic evolutionary trends such as fractional crystallisation, mixing, assimilation and replenishment in magma chamber. SiO 2 versus TiO 2 , Al 2 O 3 and Fe 2 O 3 binary diagrams imply mainly fractional crystallisation trends with the increasing of SiO 2 content despite a bit scattering. On the other hand, SiO 2 versus MgO, CaO, Na 2 O, K 2 O, Sc, Co and Zr binary diagrams (Figure 11) show    Figure 8. A) Total alkali (Na 2 O + K 2 O) versus SiO 2 (.Le Bas et al., 1986), and b) K 2 O versus SiO 2 discrimination diagrams (Peccerillo & Taylor, 1976) for the classification of volcanic rocks. Western Anatolian Miocene volcanics data is taken from Ercan et al (1985;Ercan et al., 1996), Innocenti et al. (2005), Ersoy et al. (2008), Ersoy et al. (2012a), Helvacı et al. (2009), Seghedi andHelvacı (2016), Hasözbek (2017). disseminated patterns. Also, the evolutionary trends for the Ba, Rb, Sr and Y elements generally show flat trends ( Figure 11). However, chemo-stratigraphy diagrams provide valuable information about the evolutionary trends of magmatism, and it is always more easy to interpret and monitor the temporal changes of magma chamber processes. An appropriate way to monitor the temporal changes of magma chamber processes is to sampling, dating and compare compositional changes of different lava levels in volcanic fields. In order to better constrain the temporal and geochemical variations of the volcanic rocks of Kalpakkaya and Çamlık hill, the lava samples collected from different levels of these sections (lower-middleupper) were analysed, and the obtained results plotted against the height in the stratigraphic column of Kalpakkaya and Çamlık Hill to reveal the chemical stratigraphy of the volcanism throughout the time ( Figure  12, 13). When both measured stratigraphic sections are evaluated together (Figure 8, 9), they present local differences due to the mineralogical composition of some major oxides ( Figure 12). However, major oxide chemo-stratigraphic diagrams are mainly displayed in a straight line, except for these small differences.
On the chemo-stratigraphy diagrams of the Kalpakkaya and Çamlıktepe Hill, it is observed that the trace elements (Sc, V, Co, Ba, Rb, Sr, Nb and Zr) follow a linear enrichment trend towards the middle levels of the section (Figure 13). Sc, V, and Co elements are mainly used for the crystallisation of ferromagnesian minerals such as pyroxene and amphibole. Therefore, they are used to estimate the evolution of fractional crystallisation processes, and they gradually decrease during evolutionary processes in the transition from mafic to felsic magma in the magma chamber. Ba, Rb and Sr are mainly used to track the behaviour of the feldspar fractionation and they tend to get higher with the increasing SiO 2 content. Rb together with Nb are also used to evaluate the behaviour of the mica phases, and they increase with increasing SiO 2 . Likewise, Zr tends to increase with increasing SiO 2 content and is used to estimate the behaviour of the accessory phases ( Figure 13). Variations in trace element geochemistry, such as Figure 10. Tectonic discrimination diagrams of a) Nb versus Y (Pearce et al., 1984), b) Rb/30-Hf-3xTa (Harris et al., 1986), and Th/Yb versus Ta/Yb (Schandl and Gorton 2002). Abbreviations: MORB: mid-ocean ridge basalts; VAG: volcanic arc granites; ORG: ocean ridge granites; WPG: within plate granites; syn-COLG: syn-collisional granites, WPE, within-plate enrichment, WPVZ: within plate volcanic zone, ACM: Active Continental margin, SZE, subduction zone enrichment; CC, crustal contamination; FC, fractional crystallisation; UC, average upper crust.
changes in the behaviour of major oxides, also indicate that the chemical composition of lavas did not significantly change over the period of its temporal evolution in the study area.

Petrogenesis of volcanic rocks
We use the FC-Modeller software package of Keskin (2002) to modelling the fractional crystallisation processes by using the Sr/Rb versus Sr elements ( Figure  14a). For modelling parameters, Kd values compiled from GERM database and literature data, which are embedded in the FC-Modeller. In order to determine the crystal fractionation trends of plagioclase, sanidine, hornblende, biotite and zircon minerals, Rayleigh fractional crystallisation vectors were created by selecting intermediate and felsic magma compositions. The calculated crystallisation vectors show that the plagioclase, sanidine and biotite crystallisation is the dominant process while the amphibole and zircon crystallisation is less prominent. To explain the magma mixing and replenishment processes, creating a binary diagram that consists of a compatible element versus an incompatible element can give valuable insights about the magma evolution processes, and we created a Co versus Th diagram to track the possible mixing effects. If there is common fractional crystallisation related processes occurred, our samples would align along the FC-1 and FC-2 curves in Figure 14b. On the other hand, in the presence of an open system magma chamber where magma mixing and replenishment processes are effective together, the plotted samples are lined up along straight lines. When we check the plotted samples, some portion of our volcanic units follow the curved lines of FC-1 and FC-2, while a significant amount of the samples plots along the straight lines, which supports the possible existence of the magma mixing processes (mixing of mafic and felsic end members) through the evolution of the volcanism (Figure 14b). Even though the fractional crystallisation and magma mixing processes affected the magma composition in our samples, crustal assimilation can also affect the magmatic processes since our volcanic units display high silica contents. To reveal the possible assimilation processes, we used the assimilation-fractional crystallisation algorithms of DePaolo (1981) that implement to the excel spreadsheet AFC-Modeller software by Keskin (2013). We specifically select Ta/Zr and Zr values. The Ta/Zr ratios are used as an assimilation index while Zr values are interpreted as fractionation index (Figure 14c). For modelling scheme, we use the upper-crustal values of Taylor and McLennan (1995) for an assimilant, and a calcalkaline basalt from the Soma region (Ersoy et al., 2012b) as a basic end member (BEM) for the primitive lava composition (unpublished data; Zr: 58.1 ppm and Th: 0.2 ppm). The calculated vectors show that volcanic units plot along a linear trend from the BEM to assimilant, and Zr concentrations display quite similar patterns, which can be interpreted as the assimilation is much more prominent regarding the fractionation processes (r (Ma/Mc) = 0.4-0.7).
Magma source chemistry, volcano plumbing system and petrogenetic -geochronologic evolution of the study area and the surrounding areas have been investigated by different studies in the literature (Akay & Erdoğan, 2004;Altunkaynak et al., 2010;Borsi et al., 1972;Dönmez et al., 2017;Ercan et al., 1985;Fytikas et al., 1976;Hasözbek, 2017;Karaoğlu, 2014;Seghedi & Helvacı, 2016). The Miocene volcanic activity (Yuntdağ, Dumanlıdağ, Yamanlar, Foça, and Dikili volcanoes) in the Izmir Bay, west of the Menderes massif, are located within the Izmir -Balıkesir transfer zone (İBTZ). Yundağ volcanics are represented by calc-alkali volcanic rocks developed in this zone of weakness with NE -SW trending and they transition to lacustrine sediments in lateral and vertical directions. Although Akay and Erdoğan (2004) suggested that the volcanic activity, which spreads over large areas in the region and forms large calderas, is related to stress tectonics, Karaoğlu (2014) demonstrated the existence of a more complex mechanism that controls this volcanic activity. Karaoğlu (2014) suggested that the western part of the Menderes Metamorphic Core Complex (MMCC) was affected by different deformation stages instead of İBTZ, and that early Miocene transtensional tectonics, middle Miocene extensional tectonics associated with slab-retreat and late Miocene tectonic block rotation controlled magma uplift at different stages.
The studies conducted in the recent years imply that the Cenozoic magmatism in the Western Anatolia was mainly controlled by the slab retreat and the roll-back. Different studies imply that the Early-Middle Miocene magmatism in the region was generated by the roll-back and rotation of the Hellenic - Cyprus slab, thinning of the continental crust, rupturing of the slab and transfer faults and block rotations in the upper crust control the pathways of the magma intrusive systems (Akal, 2019;Hasözbek, 2017;Karaoğlu, 2014;Seghedi & Helvacı, 2016;Uzel et al., 2020). Ersoy and Palmer (2013) suggested that magmatism between the Eocene -Miocene periods was probably due to break-off of the subducted slab in western Anatolia. They argued that the Oligocene -Miocene magmatism was controlled by rotational extension in response to rotational roll-back of the Hellenic subduction system. Also, they underlined that this magmatism was closely associated with core complex formation and asthenosphere-related thermal input along a ~ N-S-trending slab tear beneath the Menderes Core Complex (Ersoy & Palmer, 2013). In a similar manner, different studies suggest that the roll-back of the Hellenic -Cyprus slab to the south/ southeast through time also control the spatialtemporal and geochemical character of the magmatism in the region (Aysal, 2015;Borsi et al., 1972;Erkül et al., 2005;Ersoy et al., 2017;Fytikas et al., 1984;Innocenti et al., 2005;Karaoğlu, 2014;Pe-Piper & Piper, 1989).
The major portion of these studies imply that the volcanic and plutonic rocks in that region were generated by the fractionation, assimilation and magma mixing processes of the shallow seated magma batches that generated from the lithospheric mantle and mantle domains (Ersoy et al., 2010(Ersoy et al., , 2012(Ersoy et al., , 2017. Chemo-stratigraphic data from the volcanic units in the two sections of the south-eastern part of the Yuntdağ volcanics agrees with these previous studies. Our data imply that volcanism was fed by shallow seated magma chamber/mush that marked by fractional crystallisation, magma mixing and assimilationrelated processes. Even though the complex shallow seated processes completely masked the original source chemistry, we tentatively postulated that our units possibly derived from melting of the mantle and/ or subcontinental lithospheric mantle regions that triggered by the roll-back and rotation of the Hellenic -Cyprus slab during the Miocene period.

Conclusions
Thi study conducted in the south-eastern part of the Yuntdağ volcano that mainly focused on Kalpakkaya and Çamlık Hill sections show that the volcanic units in the region are mainly represented by dacitic, and andesitic lavas and tuff, which are intercalated with the sporadic Neogene sediments. Geochemistry of the volcanic rocks mainly shows high-K calc-alkaline character and displays volcanic arc and post-collisional tectonic settings. When all the geochemical data and models considered, it can be implying that the assimilation and magma mixing related processes are much more prominent compared to the fractional crystallisation related modifications, and/or possible continuous infiltration of the mafic or much primitive magma to the more felsic shallow seated magma chamber/mush to maintain the similar magmatic character. The chemo-stratigraphy sections suggest that 4 million years of variations implies no significant differences in magma composition and magma evolution processes, and this can be explained in two ways: (i) continuous feeding by a more mafic magma to balance the fractional crystallisation effect; (ii) homogenisation of mafic and intermediate magma during the evolution of volcanism and/or formation within an open system magma chamber.
Zircon U-Pb LA-ICP-MS ages from the Kalpakkaya Hill section show 17.64 ± 0.20, 17.22 ± 0.15 and 16.16 ± 0.17, Ma ages from bottom to top and Çamlık Hill section display 16.36 ± 0.13, 15.79 ± 0.71 and 13.61 ± 0.20 Ma (Lower -Middle Miocene) ages, which are compatible with the existing age data acquired around the region.
The volcanism in the region was generated and evolved under the influence of the roll-back/rotation of the Hellenic -Cyprus slab, and the pathways generated by the subsequent tectonic structures (İBTZ, core complex formation of Menderes Massif).