Lithos 180–181 (2013) 58–73

Contents lists available at ScienceDirect

Lithos

j ourna l homepage: www.e lsev ie r .com/ locate / l i thos

Geochemistry and origin of ultramafic enclaves and their basanitic hostrock from Kula Volcano, Turkey

Tobias Grützner a,b,⁎, Dejan Prelević b,c, Cüneyt Akal d

a Westfählische-Wilhelms-Universität Münster, Institut für Mineralogie, Correnstr. 24, 48149 Münster, Germanyb Johannes Gutenberg-Universität Mainz, Institut für Geowissenschaften, Becherweg 21, 55099 Mainz, Germanyc University of Belgrade, Faculty of Mining and Geology, Djušina 7, 11000 Belgrade, Serbiad Dokuz Eylül Üniversitesi, Mühendislik Fakültesi, Jeoloji Mühendisliği Bölümü, TR-35160 Buca, Izmir, Turkey

⁎ Corresponding author at: Westfählische-Wilhelms-UMineralogie, Correnstr. 24, 48149 Münster, Germany.

0024-4937/$ – see front matter © 2013 Elsevier B.V. All rihttp://dx.doi.org/10.1016/j.lithos.2013.08.001

a b s t r a c t

a r t i c l e i n f o

Article history:Received 25 February 2013Accepted 1 August 2013Available online 17 August 2013

Keywords:MediterraneanContinental volcanismcumulatesRhöniteGreen-core clinopyroxeneOlivine

The Quaternary Kula Volcanic Province is located in western Anatolia, Turkey. This Na-alkaline anorogenic volca-nism includes exposures of around 80 cinder cones, lava flows, and tuffs, representing one of the youngest volcanicactivities in this region (1.9–0.026 Ma). The magmatism is related to an extensional regime and is interpreted asbeing derived predominantly from the asthenospheric mantle.The lavaflows aremostly of a basanitic composition andhost rare comagmatic enclaves. The enclaves are composedof two dominant lithologies: amphibolites and clinopyroxenites with and without olivine. Amphibole is usuallyresorbed and replaced by a rhoenite-rich breakdown corona. The mineral composition of the breakdown coronasuggests an eruption temperature slightly below 1100 °C. Pressure and temperature calculations show thatclinopyroxene crystallization began in the magma at 12–15 kbar and 1150–1200 °C. Together with amphiboleand later crystallized olivine, clinopyroxene was probably stored as cumulates at the walls of feeder dykes orsmall chambers.The compositional variation of clinopyroxene from cumulates and lavas indicates two compositionally differentclinopyroxene types. Mg-rich clinopyroxene is present in both lavas and cumulates, while the green-coreMg-poor clinopyroxene is observed exclusively in host lavas. Trace element analyses of clinopyroxene indicatethat the clinopyroxenes from cumulates and lavas are comagmatic and crystallized in equilibrium with liquidswhose compositions were the same as those of the lavas. On the other hand, the green-core clinopyroxeneswere derived from a different melt source from the lavas. They crystallized in the lithospheric mantle andwere incorporated into the basanitic melt.In contrast to the primitive composition of lava olivines, the more evolved composition of enclave olivines is aresult of fractional crystallization processes, whose cumulate products (clinopyroxenites and amphibolites) aredirectly observable. For example, a lower Fo component and Sc and V depletion in olivine from enclaves aremirrored by Sc and V enrichment in olivine from lavas.In the initial phases of Kula volcanism, mantle-derived primary melts underwent deep-pressure fractionation ofpyroxenites and amphibolites at the base of the crust. Calculations of the ascent rate suggest that the ascent of themagma from theMoho to the surface took only 4–11 days. After a period of time, a second batch ofmelt rose andincorporated the cumulates as enclaves. This newmeltmost probably originated from a slightly different mantlesource, indicated by the presence of the green-core clinopyroxenes.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

It is well known that mantle-derived melts which intrude into con-tinental lithosphere very often stagnate on their way to the surface. Thecrystallizing assemblages may be stored as high-pressure cumuliticmaterial. The fractional crystallization of these early crystal phaseswill affect the composition of the evolving lavas, both in terms of

niversität Münster, Institut für

ghts reserved.

major and trace element contents. Ne-normative silica-undersaturatedalkaline basic magmas like basanites, tephrites, and phonotephriteshave been the subject of numerous studies, which focused on melt pro-duction by partial melting of peridotite (e.g. Falloon et al., 2001; Hirose,1997; Hirose and Kushiro, 1993; Jaques and Green, 1980), or phase equi-libria in fractionating basaltic magmas at low pressure (e.g., Baker andEggler, 1987; Bowen, 1928; Grove and Bryan, 1983; Jakobsson andHolloway, 1986; Longhi, 1991; Nielsen and Dungan, 1983; Shaw,1999; Shaw et al., 1998). A few studies which investigated high pres-sure phase equilibria of Ne-normative silica-undersaturated alkalinebasic magmas (Pilet et al., 2010; Sack et al., 1987) demonstrated that

59T. Grützner et al. / Lithos 180–181 (2013) 58–73

clinopyroxene ismore stable at higher pressure,while olivine is restrict-ed to lower pressures.

The Kula Volcanic Region inwestern Anatolia, Turkey represents oneof the best preserved Quaternary volcanic provinces in the AegeanRegion. In contrast to older surrounding volcanic regions in westernAnatolia (Fig. 1) that are dominated by K-enriched volcanism and typi-cally have anorogenic geochemical signature (e.g. Innocenti et al., 2005;Lustrino and Wilson, 2007; Prelević et al., 2012; Wilson and Biancini,1999), the Kula volcanics exhibit a primitive Na-alkaline characterwhich is interpreted as being derived dominantly from the astheno-spheric mantle (Alıcı et al., 2002; Sölpüker, 2007; Tokçaer et al., 2005).

Ultramafic enclaves are common in the Kula lavas (Richardson-Bunbury, 1996). Richardson-Bunbury (1996) described these enclavesas comparable with lherzolitic mantle material. Gülen (1990) calledthese igneous enclaves “mantle enclaves”, but without presenting anydata. Holness and Bunbury (2006) presented geochemical data andinterpreted them as igneous-related, cumulitic glassy enclaves. Theyassumed that such comagmatic enclaves cannot be transported to thesurface by monogenetic feeder systems. This was in contrast withthe widely accepted opinion that the approximately 80 cinder coneswithin the Kula Volcanic Province each represent a single eruptionevent (e.g. Richardson-Bunbury, 1996). According to Holness andBunbury (2006), the feeder dykes that transported enclave-bearingmagmas to the surface had to be used at least twice: the first batchof magma stored the enclaves in the feeder system, whereas thesecond batch consumed them during its ascension.

Here,we aim to study the potential genetic relationship between en-claves, lavas, and their mantle source from the Kula Volcanic Province.In order to tackle this issue, we studied the major and trace elementcompositions of minerals from enclaves and lavas. We confine pressureand temperature limitations for several mineral assemblages. Our ulti-mate goal was to assemble a comprehensive model for the evolutionof the enclave-bearing magma from Kula Volcanic Province.

2. Geological setting

The geological history of western Anatolia is dominated by colli-sional tectonics that caused major lithospheric thickening south of

Fig. 1. Volcanic rocks and tectonic structures in Western Anatolia, Turkey. The map displays throcks from Lower to Middle Miocene represent trachitoids and ultrapotassic rocks, whereas ol

the Izmir–Ankara suture zone during Late Cretaceous to Early Tertiarytimes (Fig. 1). After Oligocene time, this changed into a dominantly ex-tensional regime, and there are differentmodels explaining the cause ofextension: themostwidely acceptedmodel is based on subduction thatplaces the Aegean region and southwestern Anatolia in a back-arc set-ting (Fytikas et al., 1984; Le Pichon and Angelier, 1979), with backwardmigration (rollback) of the North Tethyan subducting slab representinga cause of the widespread extension. Alternatively, the orogenic col-lapse is proposed to be a major driving force of the extensional regimedeveloped after the late Oligocene time (Seyitoğlu and Scott, 1996).This post-Oligocene NE–SW extension within western Anatolia initiatedthe formation of prominent NW–SE trending grabens (Fig. 1).

Widespread volcanism is contemporaneous with the post-Oligoceneextensional tectonics within southwest Anatolia. The volcanism isdominantly high-K calc-alkaline and ultrapotassic, with a strong“orogenic” signature (sensu Lustrino and Wilson, 2007), similar toother Mediterranean volcanic provinces (Conticelli et al., 2002,2009; Lustrino and Wilson, 2007; Lustrino et al., 2011; Prelevićet al., 2005, 2012). On the other hand, the Na-alkaline volcanismwith an “anorogenic” geochemical signature occurs in this region andtypically postdates the orogenic one. This transition from orogenic toanorogenic volcanism is interpreted as being due to substantial changesin the composition of the (lithospheric)mantle source. Thiswas probablyaffected by delamination during the last 20 My: Upwelling of astheno-spheric mantle through a slab tear or window (Prelević et al., 2012),which causes the interaction, is recognized by several geophysicalmodels(Biryol et al., 2011; Faccenna et al., 2004; Özacar et al., 2010; Spakmanet al., 1988, 1993).

The Kula Volcanic Province represents the youngest volcanic activityin southwestern Anatolia. Around 80 volcanic edifices are located onthe north flank of the Alaşehir Graben. Their field covers an area of300–400 km2 (Richardson-Bunbury, 1996; Tokçaer et al., 2005) ina rectangular shape. The erupted material has a volume of about2.3 km3, which is small compared to other volcanic provinces inthe Aegean region (Richardson-Bunbury, 1996). It erupted in three dif-ferent stages, which are mainly distinguished by their morphology (e.g.Ercan, 1993; Hamilton and Strickland, 1841; Richardson-Bunbury,1996; Westaway et al., 2004): The Burgaz stage (stage 1) is the oldest

e NW–SE migration of volcanism and the E–W extension of the major grabens. Light grayder rocks (dark gray) are mainly andesites and dacites.

Fig. 2. a), b) Clinopyroxene-rich enclavewith partially resorbed amphibole (kaersutite = Krs); c) amphibole-rich xenoliths. Amphibole is now totally resorbed. Clinopyroxene and apatiteoccur only at the rim of voids; d) green-core clinopyroxene in a basanitic lava sample. Abbreviations for the minerals are taken from Kretz (1983).

60 T. Grützner et al. / Lithos 180–181 (2013) 58–73

volcanic unit. Its lavas cover the Neogene sediments and formed wide-spread plateaus of basalt (Hamilton and Strickland, 1841; Tokçaer et al.,2005). The Elekçitepe stage (stage 2) represents the main phase of vol-canic activity in Kula region, indicating an increase in extensional activ-ity (Ercan, 1993, and references therein). The youngest volcanics aresubsumed under the Divlittepe stage (stage 3). Richardson-Bunbury(1996) counted 11 cones in this group. They occur together with largeand very well preserved lava flows.

Due to a topographic inversion, the oldest lavas (stage 1) often coverthe highest hills and plateaus. This inversion is the result of rapidweathering and downcutting, caused by an uplift of the Kula region:The Neogene sediments were easily eroded, whereas the regions cov-ered by lava remained as plateaus or erodedmore slowly. Subsequently,younger lava flewout on a lower level in the topography (Hamilton andStrickland, 1841; Richardson-Bunbury, 1996; Tokçaer et al., 2005).

It is generally accepted that Kula volcanism started at about 2 Ma.In several publications, Kula volcanism is described as recent orHolocene volcanism. However, the youngest verified age was 26 ±6 ka (e.g. Göksu, 1978; Richardson-Bunbury, 1996; Westawayet al., 2004).

3. Material and methods

In the course of this study, we analyzed 30 samples including maficenclaves, their host lavas, and surrounding lava flows from the samestage of volcanism, as well as a flow from the youngest stage ofvolcanism. All 30 samples were taken directly from in situ lava flows.The enclaves vary in size, shape, and color. Their size ranges fromabout 1 cm up to 20 cm in diameter. Large enclaves show a roundedshape, whereas the small samples are angular. The transition betweenenclaves and host lava is clearly visible in all samples and shows noreaction rims.

Whole-rock major elements, as well as Ni, Cr, Ga, and Pb, were deter-mined by X-ray fluorescence spectrometry using a Philips MagiX PROspectrometer on fused disks and pressed pellets at the University ofMainz. Other trace elementswere analyzedby laser ablation (LA)-ICP-MSusing an Agilent 7500ce ICP-MS system coupled with a New WaveUP-213 LA system at the University ofMainz. Rock powdersweremeltedto form glass beads without any fluxing agent on an iridium strip heaterin an argon atmosphere and analyzed by laser sampling of the glasses(Nehring et al., 2008). Major element compositions of analyzed mineralswere determined by electron microprobe (JEOL JXA 8900RL) at theUniversity of Mainz. Operating conditions during measurements ofclinopyroxene, amphibole, and rhoenite were generally 15 kV accelerat-ing voltage, 12 nA beam current, 1–5 μm beam diameter, and 15–30 scounting time on peaks. For olivinemeasurements, the operating condi-tionswere 20 kV accelerating voltage, 20 nAbeam current, 2 mmbeamdiameter, and extended counting times on peaks (100 s for Mn, Ni, Cr,Ca, andAl). Synthetic andnaturalmineralswere used for standardization.Intra-crystal variationswere explored in rim-to-rim traverses; San Carlosolivinewas analyzed as a secondary standard. Details on the accuracy andreproducibility of the analyses are reported in Prelević et al. (2013).

Trace elements in Cpx phenocrysts were measured by Laser-ICP-MSat the University ofMainz using a NewWaveUP-213 LA system coupledto an Agilent 7500ce ICP-MS system. The laser was operated at a repeti-tion rate of 10 Hz and typical energies of 0.5–1 mJ per pulse, allowingdata collection from single grains in polished thick sections (100 μm)for at least 60 s. All analyses were carried out with spot sizes of either50 or 80 μm. Helium and argon were used as the carrier gas, with flowrates of 1.2 and 0.7 L min−1, respectively. Data collection was moni-tored in time-resolved format and the data were processed on-lineusing GLITTER software (Van Achterbergh et al., 2000). Calibrationwas based on the NIST 612 trace element glass standard with referencevalues from Pearce et al. (1997). All minerals and glasses were

Fig. 3. Oxides (in wt.%) were measured in clinopyroxene and plotted against the Mg#. The data from basanitic and phonotephritic clinopyroxenes include the data of the Mg-rich rimand mantle of green-core clinopyroxenes. f) black and white data points are taken from the very inner core of green-cores (gc), their rim, the contact of their corresponding mantle(inner mantle) and the rim of the mantle.

61T. Grützner et al. / Lithos 180–181 (2013) 58–73

calibrated using 29Si, except for clinopyroxene, which was calibratedusing 44Ca. The calibration protocol involves standardization at the be-ginning and end of each analytical run to correct for instrumentaldrift. During each run, BCR2g and San Carlos olivine were analyzed assecondary standards. Details on the accuracy and reproducibility ofthe analyses are reported in Prelević et al. (2013).

4. Results

4.1. Petrology of investigated lavas and enclaves

The lavas comprise phenocrysts of olivine and clinopyroxene in afine-grained crystallized matrix (Fig. 2).

Two morphological groups of olivine crystals are recognized:i) polyhedral crystals with a diameter of several millimeters can be richin Cr-spinel inclusions; this group is dominant; and ii) another group isformed by skeletal hopper olivineswith a diameter of about 0.5 to 2 mm.

Three types of clinopyroxene were found. Clinopyroxene of the firstgroup reaches sizes of several millimeters in diameter, has a gray color

and a polyhedral shape, and bears a lot of fractures. Clinopyroxenecrystals of the second group are smaller. They are often euhedral witha size of about 1 mm in diameter and show a brownish or reddish togray color. These grains form the majority in the lava samples. Thethird group includes grains with a green to brownish core surroundedby a mantle that varies extremely in its size relative to the core. Largeand small mantles are brownish or reddish in color, similar to the grainsdescribed as the second group of clinopyroxenes. The core is roundedand does not resemble the shape of the core's rim. Similar grains ofgreen-core clinopyroxenes are widespread in alkali basalts (Duda andSchmincke, 1985). The round shape can be interpreted as a phase ofremelting and/or resorption before the mantle crystallized (Duda andSchmincke, 1985).

Twogroups of enclaves are recognized, based on theirmineralogy. Thefirst group consists mainly of amphibole. It represents about 80 vol.% ofthe rock. The amphibole in these enclaves is mostly resorbed (Fig. 2).Only a black breakdown corona can be seen, which distinguished thegrain boundaries of former amphiboles. Amphibole-rich enclaves arevesicular. Clinopyroxene (~12%) and apatite (~8%) crystallized at the

Table 1Major and trace elements of representative clinopyroxenes.

Sample Elekçitepe Divlittepe Enclaves Green-cores

11KUL05 11KUL05 11KUL07 11KUL07 10KULX01 10KULX01 09MEN43 11KUL07

SiO2 (wt.%) 44.5 46.33 44.23 44.98 41.21 44.56 48.37 44.45 43.6TiO2 3.19 2.63 2.4 3.26 5.15 5.13 1.8 2.48 3.02Al2O3 9.91 9.11 9.84 9.78 10.99 7.34 7.06 9.36 11Cr2O3 0.219 0.056 0.024 0.26 b0.01 0.035 0.282 0.125 0.01FeO 5.96 7.51 11.51 5.58 7.32 5.83 4.61 11.04 9.17MnO 0.151 0.154 0.27 0.098 0.093 0.064 0.093 0.257 0.147MgO 12.29 11.05 8.6 12.72 10.86 12.6 14.22 8.7 8.82CaO 22.2 22.69 21.64 22.2 23.41 23.49 22.52 21.26 22.98Na2O 0.675 b0.02 1.191 0.722 0.489 0.448 0.681 1.6 0.807K2O b0.007 b0.006 0.01 0.014 b0.006 b0.006 0.055 0.007 0.012Total 99.1 99.53 99.72 99.61 99.52 99.5 99.69 99.28 99.57Mg# 79 72 57 80 73 79 85 58 63En 0.39 0.35 0.28 0.4 0.34 0.38 0.43 0.29 0.29Fs 0.11 0.13 0.21 0.1 0.13 0.1 0.08 0.21 0.17Wo 0.5 0.51 0.5 0.5 0.53 0.51 0.49 0.5 0.54Rb (ppm) 0.108 – b0.41 b0.36 0.024 b0.02 – 0.77 0.115Ba b0.35 – 2.76 b0.55 b0.126 b0.118 – 2.02 0.54Th 0.324 – 0.421 0.239 0.154 0.09 – 0.66 0.418U 0.050 – 0.075 0.054 b0.022 b0.02 – 0.227 0.031Nb 3.34 – 25.19 8.04 1.89 0.85 – 6.4 5.17Ta 0.604 – 4.69 1.82 0.515 0.182 – 1.57 1.31La 14.05 – 53.9 26.25 6.77 5.41 – 18.01 9.32Ce 45.87 – 137.1 72.91 23.6 18.29 – 45.45 27.91Pr 7.09 – 18.4 10.28 3.65 2.92 – 6.98 4.31Sr 200 – 1052 363 109 118 – 169 124Nd 35.34 – 74.05 47.87 20.17 15.2 – 36.27 21.87Sm 8.66 – 14.59 10.41 5.77 4.16 – 10.24 5.87Zr 160 – 336 320 112.7 66.7 – 328 232Hf 5.11 – 7.15 8.36 4.66 2.69 – 11.98 7.56Eu 2.88 – 3.8 3.08 1.95 1.46 – 2.41 1.98Gd 7.22 – 10.97 8.67 5.74 4.63 – 8.3 5.66Tb 1.01 – 1.44 1.20 0.83 0.66 – 1.17 0.80Dy 5.98 – 8.7 6.39 5.02 3.57 – 7.68 4.87Y 25 – 38.82 29.81 21.38 16.27 – 35.45 21.03Ho 1.06 – 1.58 1.28 0.89 0.67 – 1.45 0.91Er 2.46 – 3.41 3.13 2.22 1.76 – 3.22 2.28Tm 0.30 – 0.60 0.37 0.27 0.20 – 0.33 0.30Yb 2.01 – 3.46 2.47 1.66 1.38 – 2.73 1.81Lu 0.25 – 0.47 0.34 0.20 0.17 – 0.44 0.27

62 T. Grützner et al. / Lithos 180–181 (2013) 58–73

rim of the vesicles. The clinopyroxenes have a size of 100–500 μm indiameter, a polyhedral shape, and reddish to gray color, similar to theclinopyroxenes in the second group of lava samples.

The second group of enclaves is clinopyroxene-rich (40–60%). Allsamples consist mainly of intergrown clinopyroxene and amphibole(15-45%) and are up to several millimeters in size. Clinopyroxeneresembles the first type recognized in the lava samples. It has a graycolor and a lot of fractures and rarely shows a subhedral or euhedralshape. Amphibole has brown to yellowish color. It can form subhedralgrains as well. Additionally, the enclaves can contain olivine (up to25%), calcite (up to 6%), apatite (b1%), or plagioclase (b1%) and rarelynepheline, leucite, or ilmenite. The enclaves are only slightly vesicular.The vesicles are partially filled with glass which contains phenocrystsof apatite and plagioclase. In comparison to the amphibole-rich enclavesand the lavas, clinopyroxene-rich enclaves contain the least resorbedamphibole grains.

4.2. Mineral chemistry

4.2.1. ClinopyroxeneFig. 3 shows several oxide concentrations plotted against the Mg#

(Mg/(Mg + Fetot) ∗ 100) of clinopyroxenes. They contain high amountof Al2O3, ranging from14 wt.% to 4 wt.%. According to Morimoto et al.(1988), the nomenclature for these clinopyroxenes would be ferrianalumian diopside, formerly known as fassaite. Major and trace elementdata measured in clinopyroxenes are listed in Table 1.

The Mg# of clinopyroxenes from lavas and enclaves ranges fromabout 85 down to 70 (for the main field of data points). The range ofgreen cores lies between 70 and 45. With decreasing Mg#, Na2O,Al2O3, MnO, and TiO2 increase, whereas Cr2O3 decreases (Fig. 3e). Thecomposition of investigated clinopyroxenes generally follows twodifferent trends that intersect at Mg-rich compositions and show anegative correlation on an Al2O3 and TiO2 versus Mg# diagram(Fig. 3b, c). The trend occupied by clinopyroxenes from the enclaves ischaracterized by a very gentle slope in this diagram. On the otherhand, clinopyroxenes from enclaves show the steepest slope whereasthe lava clinopyroxenes resemble them, but their variation is morescattered (Fig. 3b, c).

Clinopyroxenes from enclaves are generally unzoned, while thosefrom the lavas are often zoned. Normal zoning is common forclinopyroxenes from lavas and enclaves, whereas the green coreclinopyroxenes show reverse zoning. The transition between theFe-rich core and the mantles is always abrupt, but the cores do notshow any compositional variation. This can also be seen in the backscattered electron (BSE) images. In Fig. 3f, only clinopyroxenes fromlavas and green cores are plotted. In the diagram every single green-core shows a uniformcompositional plateau, but considerable differencesare demonstrated by different grains plotted on a differentiation trend.

Fig. 4 displays anREE (rare earth elements) plot normalized to chon-drite (after Sun and McDonough, 1989) with representative data. Allsamples are enriched relative to chondrite. Clinopyroxenes from thelava and from the enclaves are subparallel, showing a convex upwardpattern. Clinopyroxenes from the lava are more enriched than those

Fig. 4. REE diagram of representative clinopyroxene data normalized on chondrite.Reference data are from Sun and McDonough (1989).

Table 2Major and trace elements of representative olivines.

Sample Elekçitepe Divlittepe enclave

11KUL01-2 11KUL01-2 11KUL02-4-1 11KUL07 10KULX01

SiO2 (wt.%) 40.12 39.01 40.59 39.55 39.03TiO2 0.03 0.03 0.03 0.01 b0.008Al2O3 0.04 0.02 0.04 0.04 0.03Cr2O3 0.02 b0.007 0.03 0.02 b0.008FeO 16.46 16.24 13.38 13.47 18.51MnO 0.28 0.36 0.21 0.24 0.35MgO 42.69 43.71 45.34 46.17 42.62CaO 0.26 0.31 0.22 0.24 0.18NiO 0.05 b0.01 0.1 0.13 0.05Total 99.95 99.68 99.94 99.87 100.77Mg# 82 83 86 86 80Li (ppm) 3.56 3.35 2.14 2.41 3.35Al 176.8 183.8 248.5 150.5 85.5P 269.8 271.6 204.2 199.0 125.6Ca 1685 1751 1398 1264 430Sc 6.65 6.53 8.1 6.13 2.01Ti 87.9 88.1 111.7 76.8 111.9V 4.65 4.39 7.76 5.04 2.07Co 182.5 177.4 170.5 162.8 175.8Cu 0.44 0.61 0.84 1.00 0.50Zn 108.9 107.1 75.8 100.7 155.3Ga 0.111 0.136 0.116 0.144 0.124Sr 0.015 0.02 0.029 b0.014 0.043Y 0.131 0.152 0.107 0.103 0.021Zr 0.057 0.044 0.057 0.077 0.084Nb 0.019 0.032 0.033 0.066 0.014Ba 0.065 0.119 0.125 b0.071 0.172La 0.013 0.012 0.019 b0.009 b0.006

63T. Grützner et al. / Lithos 180–181 (2013) 58–73

from the enclaves, being 13 to 21 timesmore enriched than chondrite inthe former and 6 to 11 timesmore enriched than chondrite in the latter.Green-core clinopyroxenes do not show a convex upward pattern but asteeper and almost linear trend. The enrichment of La ranges from 40 to83 times that of chondrite.

4.2.2. OlivineMajor and trace elements of representative olivines are listed in

Table 2. Fig. 5 displays different oxides plotted against the Mg#, whichranges from 88 to 78 for the main field of data. The rims of the pheno-crysts can reach down to 70. Amphibole-breakdown olivines show aconstant Mg# of 81 to 80. With decreasing Mg#, increases in CaO andMnO and a decrease in NiO can be seen for the lava olivines.Mg# versusCr2O3, Al2O3, and TiO2 does not show significant trends.

Data from olivine from enclaves show a strong vertical trend fornearly all oxides, which is remarkable (Fig. 5). This vertical trend canbe found in some lava phenocrysts, as well. Especially for NiO, but alsofor MnO and Cr2O3, the points are scattered over the whole range ofmeasured data. The grains lack any visible zoning, as shown in Fig. 6.

Fig. 5. Oxides (in wt.%) were measured

Their measuring lines range over more than 500 μm. A small changein MnO and the Mg# can be seen towards the rim, whereas inside thegrain the Mg# is constant.

4.2.3. AmphiboleAmphibole occurs in enclaves and as phenocryst in lavas. Nearly all

grains are partially or even completely replaced by the amphibolebreakdown corona. Representative data of amphibole in enclaves are

in olivine and plotted against Mg#.

Fig. 6. Olivine grains with Mg#-dependent (a) and independent oxide variations (b). Oxide data (in wt.%) are taken from the line of measuring. We performed an equal number of mea-surements per length unit for both lines. Grain a) is taken from a lava sample which shows a small variation in oxides for constant Mg# and plots in the diagonal differentiation trend inFig. 5. Grain b) shows a strong variation in oxides independent of the Mg#.

64 T. Grützner et al. / Lithos 180–181 (2013) 58–73

Table 3Major and trace elements of representative amphibole, spinel and plagioclase data.

Sample Amphibole Fe–Cr-spinel Ulvöspinel Fe-spinel Plagioclase

10KULX01 10KULX01 10KULX01 11KUL02-11 11KUL07 11KUL07

SiO2

(wt.%)39.88 40.07 0.04 0.16 0.07 50.43

TiO2 4.36 4.18 0.51 16.19 1 0.23Al2O3 14.52 13.99 39.86 9.3 42.72 29.44Cr2O3 0.04 0.41 12.78 b0.02 16.88 0.05FeO 8.82 8.25 32.17 65.35 20.72 0.67MnO 0.13 0.11 0.31 0.48 0.17 0.04MgO 13.57 14.15 13.79 5.41 17.29 0.06CaO 12.01 12.01 0.03 0.4 0.01 12.68Na2O 2.52 2.5 0.08 b0.01 b0.02 3.8K2O 1.63 1.72 b0.01 b0.02 b0.01 0.4Total 97.74 97.68 99.76 97.3 99.16 97.8Mg# 73 75 – – – –

An/(An + Ab) – – – – – 0.78

65T. Grützner et al. / Lithos 180–181 (2013) 58–73

listed in Table 3. According to Hawthorne and Oberti (2006), the aver-age measurement, NaCa2Mg3FeAl0.5Ti0.5[Si6Al2O22]O2, can be classifiedas a Fe-bearing kaersutite (kaersutite = NaCa2Mg3TiAl[Si6Al2 O22]O2).Fig. 7a presents a large grain of amphibole with a former nearlyeuhedral shape. The breakdown corona seems to grow inside thegrain along the cleavages. Fig. 7b displays the same region in higherresolution. Intergrowth of small clinopyroxenes, together with olivine,rhoenite and glass is common. The grains in Fig. 7c are much smaller.They are part of the breakdown corona which surrounds an olivinephenocryst. The corona consists of olivine, clinopyroxene, rhoenite,and plagioclase. It also shows a small bright rim of clinopyroxenebetween the corona and the olivine phenocryst. Fig. 7d presents partiallyresorbed amphibole in a lava sample. In comparisonwith the lavamatrix,the breakdown corona bears larger crystals which aremore homogenousin size.

Fig. 7. BSE images of amphibole phenocrysts and their associated breakdown coronas in enclRhoe = rhoenite.

4.2.4. RhoeniteIn general, rhoenite can occur as phenocrysts in Na-alkaline

Ti-rich lavas together with Ti-augite, olivine, nepheline, plagioclase,Ti- magnetite, Ti-amphibole, and apatite (Grapes et al., 2003;Kunzmann, 1999). It also can crystallize as a breakdown productof Ti-rich Ca-amphibole (e.g. kaersutite) together with olivine,clinopyroxene, plagioclase, spinel, and magnetite (Grapes et al.,2003; Kunzmann, 1999). All rhoenite found in this work is assignedto the second type. The grain size is below 50 μm in diameter. Thecrystals occur in opaque breakdown coronas surrounding the amphi-boles. There is a small difference between rhoenite from basaniticlava and enclave rhoenite. Rhoenite in lavas has lower amounts of MgO,TiO2, Al2O3, and Cr2O3, but higher amounts of FeO and MnO comparedto the amounts in enclave rhoenite (Table 4). The rhoenite in lavas hasa Mg# of 50, while that in enclaves has a higher Mg# of around 60.

aves (a, b) and lavas (c, d). Abbreviations for the minerals are taken from Kretz (1983).

Table 4Major and trace elements of rhoenite.

Sample Elekçitepe Enclaves

11KUL01-2 10KULX01 10KULX01 10KULX01 10KULX01

SiO2 (wt.%) 26.07 24.92 25.52 24.94 24.79TiO2 9.74 11 10.6 10.87 10.52Al2O3 16.48 18.51 18.31 18.41 18.39Cr2O3 0.07 0.1 0.11 0.15 0.47FeO 21.3 17.24 16.66 17.37 17.85MnO 0.21 0.1 0.13 0.15 0.12MgO 12.12 14.72 14.49 14.12 14.02CaO 11.83 12.39 12.45 12.09 11.85Na2O 1.2 1.04 1.17 1.19 1.1K2O 0.06 0.02 0.06 0.12 0.01Total 99.06 100.04 99.49 99.38 99.1Mg# 50 60 61 59 58Cs (ppm) – – b0.22 0.22 –

Rb – – 3.69 1.26 –

Ba – – 28.25 9.18 –

Th – – b0.19 b0.19 –

U – – 0.203 b0.12 –

Nb – – 51.2 71.4 –

Ta – – 4.32 5.46 –

La – – 5.01 4.99 –

Ce – – 15.51 19.05 –

Pb – – 13.81 8.13 –

Pr – – 2.4 2.83 –

Sr – – 201.4 241.8 –

Nd – – 13.49 13.86 –

Sm – – 3.81 4.44 –

Zr – – 77.7 87.9 –

Hf – – 3.08 3.25 –

Eu – – 0.67 1.49 –

Gd – – 3.64 3.09 –

Tb – – 0.42 0.69 –

Dy – – 3.42 3.48 –

Y – – 13.66 12.8 –

Ho – – 0.51 0.49 –

Er – – 1.78 1.06 –

Tm – – 0.22 0.39 –

Yb – – 1.29 2.31 –

Lu – – 0.21 0.46 –

Table 5PT-calculations for lava phenocrysts and Mg-rich mantles of green-core clinopyroxeneswith the clinopyroxene–liquid thermobarometer of Putirka et al. (2003).

Cpx phenocrysts Cpx green-core rims

T in °C P in kbar T in °C P in kbar

1191.8 13.7 1181.8 12.71167.5 11.8 1195.0 14.11189.4 13.9 1207.5 15.01163.8 11.8 1163.2 12.21147.6 12.3 1162.9 12.3

1150.5 12.31153.9 12.7

66 T. Grützner et al. / Lithos 180–181 (2013) 58–73

4.2.5. Other mineralsPlagioclase can be found in every sample. It is part of the matrix in all

lavas and occurs in the breakdown coronas of amphibole (Fig. 7c). Repre-sentative data are listed in Table 3. The An# (An/(An + Ab) ranges from72 to 77 in enclave breakdown coronas, whereas all measured grains inlava breakdown coronas have a constant An# of 78.

Spinel inclusions occur in olivine and clinopyroxenes. Ti-rich spinelsare abundant inclusions of clinopyroxene whereas the Fe–Cr spinelsoccur in olivines. Representative data are listed in Table 3.

4.3. Thermobarometric limitations

Pressure–temperature estimations were made using the empiricalclinopyroxene–liquid thermobarometer of Putirka et al. (2003). Valuesfor the liquid composition were taken from whole rock analyses of thelava samples. All whole rock analyses include the phenocrysts. Yet, therelatively low amount of Cpx phenocrysts in the lavas, as well as theobservation that the phenocrysts and matrix are in equilibrium, allowsus to use the whole rock data as representatives for the liquid(cf. Putirka et al., 2003). The estimated crystallization pressure of theclinopyroxenes ranges from 11.8 up to 13.9 ± 0.9 kbar at temperaturesranging from 1190 to 1150 ± 14 °C (Table 5). Errors were set as 1standard deviation. A test for the equilibrium between clinopyroxeneand the melt can be derived from Fe/Mg ratios: Kd(Fe − Mg) betweenclinopyroxene and liquid, calculated as FeO(Cpx) × MgO(liquid) /FeO(liquid) × MgO(Cpx) (molar), should be 0.27 ± 0.03 (Draper andGreen, 1999; Grove and Bryan, 1983). Clinopyroxenes from the lavasamples are plotted with a derivation smaller than 10%. On the otherhand, clinopyroxene in the enclave shows a value of 0.33 and is there-fore not in equilibrium with the liquid anymore.

An additional test represents a comparison between the predictedand measured components in the clinopyroxene for the given pressureand temperature and liquid compositions (Putirka, 1999). Our estima-tions show that for all lava clinopyroxenes the coefficient of correlation(R2) is 0.97 for DiHd, 0.98 for EnFs, and 0.85 for CaTi (with R2 = 1 beingthe best value).

PT calculations for green-core clinopyroxenes cannot be done,because their compositional difference clearly implies that they arexenocrysts not in equilibriumwith the liquid. On the other hand, calcula-tions can be done for theirmantles. The calculated pressures and temper-atures for the clinopyroxene mantle are very similar to those for thephenocrysts: the pressure ranges from 12.2 up to 15.0 ± 0.9 kbar. Theestimated temperatures range from 1150 up to 1207 ± 18 °C. Theequilibrium between clinopyroxene mantles and liquid was verifiedwith the KD(Fe − Mg)Cpx − liq of all mantles, which deviates by lessthan 10% from 0.27. R2 = 0.97 for the DiHd component, R2 = 0.97 forthe EnFs component, and R2 = 0.94 for the CaTi component (afterPutirka, 1999).

4.4. Geochemistry

4.4.1. Major and compatible trace elements of lavas and enclavesWhole rock analysis data of 10 samples are listed in Table 6. All rocks

are SiO2-undersaturated. SiO2 ranges from 44.5 to 47 wt.% in the lavasand from 37 to 42 wt.% in the enclaves. The K2O/Na2O ratio for allrocks ranges from 0.5 to 0.7, which can be expected for the sodic,anorogenic Kula volcanism. The Mg# (100 ∗ Mg / (Mg + Fe) rangesfrom 69 to 77 for lavas and from 76 to 87 for the enclaves. In Fig. 8,whole rock data from Alıcı et al. (2002), Holness and Bunbury (2006),and Tokçaer et al. (2005) are added to the diagram. They are separatedinto the three volcanic stages: Burgaz (stage 1), Elekçitepe (stage 2), andDivlittepe (stage 3).

All lavas can be classified as basanites/tephrites, trachybasalts, orphonotephrites. To distinguish basanite from tephrite, CIPW-norm calcu-lations were done after Cox et al. (1979) and Kelsey (1965). FeO/Fe2O3

ratios and the amount of FeO were calculated after Middlemost (1989).

All stage 2 lavas are very close to the border between basanite andtephrite. Their olivine content ranges from 8.0 to 11.9%.

With decreasing MgO, increases in Al2O3, Na2O, and K2O, and adecrease in CaO can be seen in the lava samples (Fig. 9). The trendsare compatible with the fractionation of minerals like olivine,clinopyroxene, and spinel. The plotted trace elements support thistrend of olivine and clinopyroxene fractionation: Ni and Cr decreasewith decreasing MgO while Nb increases. Zr shows a weak increase.Due to the strong increase of Al2O3, it can be assumed that no largeamount of plagioclase was fractionated from the melt.

For the major elements, the enclaves seem to follow the lava trend,whereby they form the high MgO end member of the plotted samples.

67T. Grützner et al. / Lithos 180–181 (2013) 58–73

The composition of the enclaves, which consist predominantly of am-phibole and clinopyroxene, enhances the assumption that amphiboleis one of the minerals which controlled the differentiation process.

Fig. 8. Total alkali-silica plot after Le Maitre et al. (1989): The ellipses represent glassyenclaves from Holness and Bunbury (2006) and lava samples from Tokçaer et al. (2005).

4.4.2. Incompatible trace elements of lavas and enclavesAll lavas are strongly enriched in incompatible trace elements in

comparison with the primitive mantle (Fig. 10). The enrichment ofvery incompatible elements is less strong in enclaves than in the lavas.All investigated samples show a trough in Pb, U, and Th.

The lavas show no orogenic influence, which would be typical forMiocene volcanism within western Anatolia. They were generated bypartial melting of an isotopically depleted spinel peridotite mantlesource, free of any subduction-related influence (cf. Alıcı et al., 2002;Lustrino and Wilson, 2007; Sölpüker, 2007; Wilson and Biancini,1999). Fig. 11 shows REE variation. The lavas are highly enriched in allREEs, relative to Ocean island basalts (OIB) The enrichment of LREE(light REEs) is much higher than the enrichment of HREEs (heavy rareearth elements). Basanites are slightly more enriched than thephonotephrite in LREEs and converge with them at Dy. Compared toOIB, the value of enrichment in LREEs and HREEs is much steeper forthe lavas of this work. The pattern flattens for the HREEs and becomesnearly parallel.

Fig. 11 shows no Eu anomaly. The values of HREEs (Dy–Lu) alsoshow no depletion. They are nearly parallel to chondritic compositionand to mid ocean ridge basalts (MORB).

Table 6Major and trace elements for volcanic lavas and enclaves from Kula Region.

Sample Elekçitepe Divlittepe enclaves

09MEN43 09MEN43-1 11KUL08 11KUL02-4 11KUL05 11KUL01-2 11KUL07 10KulX01 11KUL02-1 11KUL02-6

SiO2 (wt.%) 44.85 44.47 44.91 45.4 45.43 45.08 47.09 41.96 42.19 37.13TiO2 2.21 2.24 2.19 2.19 2.2 2.16 1.86 2.73 3.79 4.26Al2O3 16.56 16.48 18.27 16.84 16.94 17.39 18.05 10.04 12.47 14.55Cr2O3 – – b0.002 0.014 0.014 0.004 0.01 – 0.15 0.005Fe2O3 9.33 9.38 9.64 9.34 9.22 9.87 8.16 11.02 8.53 11.85MnO 0.16 0.16 0.16 0.17 0.16 0.16 0.15 0.16 0.1 0.13MgO 7.76 7.69 4.72 7.1 7.21 5.58 5.69 18.54 12.73 9.66CaO 9.64 9.58 9.33 9.38 9.2 9.75 8.23 12.71 15.91 14.28Na2O 4.74 4.65 5.57 4.98 4.94 5 5.34 1.55 1.59 2.34K2O 2.96 2.95 2.78 2.94 3.13 2.88 3.54 0.76 1.11 1.89P2O5 0.88 0.91 1.23 1.02 0.93 1.27 0.77 0.1 0.07 2.69LOI – – 0.42 0.25 0.42 0.83 0.45 – 0.32 0.19Sum 99.09 98.51 99.23 99.61 99.78 99.97 99.34 99.57 98.96 98.97Mg# 77 76 66 75 76 69 73 87 86 76Sr (ppm) – – 1453 1058 1067 1260 962 – 507 861Ba – – 1128 918 957 974 946 – 407 734Rb – – 59.8 67.0 69.8 53.7 80.5 – 13.2 37.7Ni 99 94 13 78 83 29 62 339 113 18Cr 114 118 7 82 82 24 64 457 1067 b5.3Sc – – 14.9 20.5 21.7 18.4 19.0 – 58.5 25.5Ga 20 18 20 20 19 20 19 12 14 18La – – 79.8 59.5 62.7 71.2 52.8 – 10.0 49.2Ce – – 144 113 115 132 88 – 27.8 112Pr – – 15.3 11.8 12.2 13.7 9.5 – 4.2 13.8Nd – – 58.5 44.3 47.4 51.8 37.5 – 22.2 60.3Sm – – 9.9 7.7 8.6 8.8 6.8 – 6.0 12.3Eu – – 2.85 2.30 2.49 2.64 2.10 – 1.98 3.73Gd – – 7.4 5.8 6.6 6.8 5.9 – 5.5 9.8Dy – – 5.59 4.59 5.21 5.05 4.96 – 4.60 7.42Er – – 2.69 2.20 2.51 2.41 2.51 – 2.01 3.09Yb – – 2.36 1.95 2.24 2.05 2.24 – 1.52 2.33Lu – – 0.334 0.280 0.304 0.305 0.326 – 0.194 0.306Zr - - 234 199 227 195 237 – 85 134Hf – – 4.35 3.81 4.50 3.80 4.64 – 2.80 3.42Cs – – 1.02 0.89 1.01 0.96 1.26 – 0.08 0.33Pb 5 6 5.77 2.91 5.01 4.16 0.79 b2.7 0.36 1.23Nb – – 124 101 107 96 100 – 29.4 76Ta – – 5.71 4.90 5.52 4.58 5.93 – 1.82 3.80Th – – 9.8 7.0 7.8 8.7 8.4 – 0.54 3.13U - - 2.69 2.22 2.34 2.60 2.08 – 0.16 0.95

Fig. 9. Variation diagrams showing several oxides and trace elements versusMgO (oxides in wt.%, trace elements in ppm). White data points are taken from Alıcı et al. (2002). Black datapoints are processed from this work. The positions of the vectors are not absolute. They show relative directions for lava differentiation in case of crystal fractionation.

68 T. Grützner et al. / Lithos 180–181 (2013) 58–73

Previously it has been suggested that the Kulamagmaswere generat-ed from a mantle source in the stability field of spinel. Alıcı et al. (2002)suggest a mixture of two sources: a slightly depleted asthenosphericmantle and a lithospheric source. Sölpüker (2007) calculated severalmodels and his best fitting model assumes a source of spinel peridotiteat a pressure of 20 to 26 kbar with 7–9% degrees of melting.

5. Discussion

In order to constrain an evolution model for the enclave-bearingmagma from Kula Volcanic Province, we first aim to test whether thelavas (including green-core clinopyroxenes) and enclaves are derivedfrom the same parentalmelts, principally bymineral chemistry includingtrace element data. Thermobarometric estimation of clinopyroxene

Fig. 10. Mantle-normalized diagram for incompatible trace elements of representativeKula lavas and enclaves. Values for normalization and reference rocks (OIB andN-MORB) are taken from Sun and McDonough (1989). The gray shaded field representsorogenic lavas from south-western Anatolia (data from Prelević et al., 2012).

crystallization can help to constrain themaximum depth of magma stor-age. By calculating the ascent rate, the model acquires a temporaldimension.

5.1. Origin of the enclaves

A line of textural and compositional evidence suggests that theenclaves do not represent mantle composition (Richardson-Bunbury,1996) but are cumuliticmaterial formed from theKulamagmas by segre-gation of early formed crystals (Holness andBunbury, 2006). They showatypical cumulate texture comprised of a framework of coarse-grainedeuhedral to subhedral crystals. Quite often, the spaces in between themare filled by apatite and plagioclase acicular grains that have crystallizedin situ from intercumulus melt. The mineral composition also suggests

Fig. 11. Chondrite-normalized REE diagram for incompatible trace elements of represen-tative Kula lavas and enclaves. Values for normalization and reference rocks (OIB andN-MORB) are taken from Sun and McDonough (1989).

69T. Grützner et al. / Lithos 180–181 (2013) 58–73

that enclaves do not represent ultramafic mantle material but are segre-gated by fractionation of severalminerals from themelt. Themost impor-tant compositional differences relative to the typical composition ofmantle minerals are demonstrated by amphibole and olivine composi-tions. They have considerably lower Mg# than typical mantle amphibole(Vaselli et al., 1995) and olivine, suggesting that their Mg# was notbuffered by the peridotite but, which is more likely, by the magma fromwhich they crystallized. The compositional resemblance of clinopyroxenefrom lavas and enclaves further supports their cumulitic origin. Theclinopyroxene oxide plots in Fig. 3 show that primitive clinopyroxenesfrom both lavas and enclaves are nearly identical in composition,suggesting a comagmatic origin. This is further supported by our REEdata (Figs. 4 and 12). Clinopyroxenes from both enclaves and lavasshow similar chondrite-normalized REE patterns and very similarLREE/HREE ratios.

Our observations and measurements of enclaves imply thatclinopyroxene togetherwith amphibole crystallized first at the liquidus,followed by olivine. Phase equilibrium studies of Ne-normative silica-undersaturated alkaline basic melts have demonstrated that in the

Fig. 12. Incompatible trace elements from olivines of this work plotted versus Li (in ppm). Thegray field displays the compositional range of mantle olivines (Foley et al., 2013 and the refereMediterranean phenocryst data are taken from Prelević et al. (2013).

systemOl–Di–Ne, increasing pressure moves the cotectic curve towardsthe olivine apex (Pilet et al., 2010; Sack et al., 1987). At pressures of up to30 kbar, the olivine stability field shrinks and therefore clinopyroxenemay crystallize first on the liquidus. Our PT estimations suggest thatclinopyroxene crystallization occurs at mantle pressures of around14 kbar, thus limiting the maximum depth of the cumulates. On theother hand, the stability of Ti-rich amphiboles like kaersutite is tempera-ture dependent: it has been experimentally confirmed that the amphi-bole can be stable in the Ne-normative basaltic liquid at a pressure of15 kbar, but only at the temperatures lower than 1130 °C (Pilet et al.,2010). This, together with the absence of olivine at liquidus, indicatesthat enclaves most probably crystallized at a maximum pressure of14 kbar.

Themineral chemistry also supports the idea that olivine in enclavescrystallized after clinopyroxene. This is demonstrated by considerablylower Fo in olivine from enclaves (Mg# 80–83) relative to olivinefrom the lavas (up to 88; Fig. 5). Trace element contents provide furtherevidence: in comparison to lava olivine, the enclave olivine showsdepletion in Sc and V (Fig. 12). These elements are highly compatible

light gray fields show the compositional range from volcanic olivines worldwide. The darknces within).

70 T. Grützner et al. / Lithos 180–181 (2013) 58–73

in clinopyroxene (Adam and Green, 2006), and they are obviouslydepleted in the melt by clinopyroxene crystallization before olivine.Olivine crystals that grow simultaneously with clinopyroxene (or evenlater) can still incorporate Zn, Ti, and Li, but will be depleted in Sc andV compared to the olivine that crystallized alone on the liquidus. Thisis taken as further evidence for the fact that olivines in enclaves crystal-lized after clinopyroxene and more primitive olivine from lavas.

The strong vertical trend of compatibleminor elements for relativelyconstantMg# in single olivine grains (Fig. 5) is still not understood. Thisobservation may be potentially explained by the processes limited tothe boundary layer of olivine crystals within the crystallizing magmabatch at the site of its stagnation. The liquid layer close to the crystalboundary differs in composition from the bulk liquid. Incompatibleelements will be enriched in the layer, and the compatible ones willbe depleted, depending on the partition coefficient of the differentelements (e.g. Pletchov and Trousovt, 2000; Smith et al., 1955). Thecompatible elements are easily incorporated into the crystal lattice.Assuming a slow growing grain surrounded by a highly mobile liquid,the olivine could consume compatible elements like Ni out of the melt(KD 5–47 in basaltic systems; Beattie, 1994; Foley et al., 2013; Hartand Davis, 1978; Kinzler et al., 1990) until it is too depleted. The nextbatch of melt carries new Ni into the boundary layer. The grain keepson growing with a stable Mg# that is independent of the Ni contentinside the boundary layer. The result is an oscillatory zoning for compat-ible minor and trace elements, especially Ni. Still, the variation in Cr istoo intensive for boundary layer effects. The highest peaks could beexplained by micro-inclusions of spinel, which can be present at veryhigh concentrations in Cr. Micro-spinels can influence the variation inAl, which correlates well with some Cr peaks (cf. Fig. 6).

Fig. 13. Clinopyroxene REE ratios vs. La/Yb. The data suggest a comagmatic character ofenclaves and Kula lavas. Green-core clinopyroxenes are derived from a different source.With increasing Dy/Yb the importance of garnet in the source increases, as well (blackdashed arrow). Different ratios in La/Yb and Nd/La can be explained by metasomatismor different degrees of partial melting in the source (gray arrows).

5.2. Origin of green-core clinopyroxenes

Fig. 13 shows variation in REE ratios between green cores andclinopyroxene from their mantles, phenocrysts, and enclaves. Whilethe clinopyroxene from enclaves and lavas is plotted in a narrowrange, the green cores show greater variation and usually a higherextent of REE fractionation, as illustrated by LREE/HREE ratios. Togetherwith systematic variations of the clinopyroxene composition (Mg#,Al2O3, and TiO2), two clearly different compositional arrays are defined(e.g. Fig. 3c). This implies that these two clinopyroxene types crystal-lized from different primary melts, which evolved as discrete volcaniclineages unrelated by fractional crystallization of a common parentalmagma.

Due to their low Mg#, green cores represent a highly differentiatedmelt. A low Mg# could be derived by small amounts of partial meltescaping from the mantle source, or by melting of the metasomatizedmantle (Duda and Schmincke, 1985). With respect to the suggestionof Duda and Schmincke (1985), we assume that small amounts ofescaped melt became stuck and started crystallizing: the homogeneouscomposition of the green cores indicates fast crystallization, and differ-ences between the grains are in accordance with the notion that theymay represent a kind of “pegmatitic” material (Duda and Schmincke,1985). The mantles around the green cores crystallized simultaneouslywith the phenocrysts, and our PT calculations suggest that thishappened in the (lithospheric) mantle depths. This indicates that thegreen cores were incorporated prior to the Mg-rich clinopyroxenecrystallization, implying that basanitic magma entrained and consumedthose “green pegmatites” while ascending. Probably the larger crystalsremained as rounded cores, protected in a mantle of clinopyroxenewhich crystallized from the basanitic melt and had an identical compo-sition to normal phenocrysts (cf. Duda and Schmincke, 1985).

Fe-rich green-core clinopyroxenes have a lower Nd/La ratiocompared to the majority of clinopyroxenes in lavas and enclaves. Weinterpret the trend in Fig. 13a for Nd/La versus La/Yb as a mixing trendbetween clinopyroxenes from lavas and enclaves and the green cores.It is viable that the large variation of La/Yb seen in clinopyroxene fromthe lavas reflects the variable extent of resorption, incorporation, andconsumption of Fe-rich clinopyroxene.

5.3. Constraints on the magma ascent

In this section we use estimates of minimum ascent rates and thesolidus–liquidus relationship of amphibole breakdown and crystalliza-tion of rhoenite to tightly constrain the ascent of the Kula magmas.

Minimum ascent rates were calculated for selected samples afterSpera (1980). The density of the enclaves ranges from about 3200to 3700 kg/m3. The basanitic host lava has a density of about2800 kg/m3. Assuming a viscosity of 100 Pa ∗ s and a yield strength24 Pa (cf. Alletti et al., 2005) in a Bingham liquid, the minimum ascentrate ranges from 0.03 to 0.09 m/s. For a depth of 30 kmwith this ascentrate, the magma ascended to the surface in 4–11 days at most. This is aminimum ascent rate, because this estimation does not consider thepresence of volatiles that will reduce the viscosity and density of themelt. The presence of the hydrous component is recognized by the am-phibole crystallization and suppression of plagioclase stability at a latestage. A fast ascent and a little interaction imply a rapid filling of themagma chambers (Alletti et al., 2005). According to Saunders et al.(1998), no largemagma chamber was detected below the Kula volcanicregion. Hence, the enclaves most probably crystallized within smallchambers or feeder dykes (cf. Alletti et al., 2005; Duda and Schmincke,1985; Holness and Bunbury, 2006). In order to accomplish the entrain-ment of enclaves, the feeder dykes have to be filled with magma.Enclaves crystallize from the first magma batches. The next or anylater batch of magma incorporates the enclaves while ascending to thesurface (cf. Holness and Bunbury, 2006). Including the number ofcones, the total volume of the erupted lava, and the volume of single

Fig. 14. Solidus–liquidus relations for Ti-rich amphiboles at low pressures (Kunzmann,1989). The stability field of rhoenite is shaded in light gray. The dark gray field displaysthe mineral composition in the breakdown corona of amphiboles of this work.

71T. Grützner et al. / Lithos 180–181 (2013) 58–73

lava flows, Holness and Bunbury (2006) calculated an eruption rate ofabout 4000 years. On average, the cones are monogenetic, formed bya single eruption event. Enclave-bearing magmas erupted in series ofmuch shorter time periods and used old feeder dykes before the com-plete solidification of the former magma (cf. Holness and Bunbury,2006).

According to Kunzmann (1989, 1999), the breakdown of kaersutiteoccurs at pressures below 500 bar and temperatures above 1060 °C.Fig. 14 presents a PT diagram showing the solidus–liquidus relationsfor Ti-amphibole at low pressures (up to 1 kbar) and the stability ofrhoenite (light gray; after Kunzmann, 1989). It should be noticedthat rhoenite is stable over a wide range of oxygen fugacity (fO2)(Kunzmann, 1989, 1999), and the pressure is a parameter that actuallyplays the most important role in this amphibole breakdown reaction.The absence of spinel in samples of this work limits the PT conditionsof the breakdown of Ti-rich amphibole, depending on the presence of

Fig. 15.Model for themagma ascent below the sampling area in theKula Volcanic Province incluPilet et al. (2010), Saunders et al. (1998) and Sölpüker (2007).

clinopyroxene (1120 °C and 100 bar down to 1090 °C and 450 bar)and olivine (1160 °C and 50 bar down to 1120 °C and 350 bar). Inspite of a very detailed examination of the breakdown corona, wehave been able to observe only clinopyroxene and olivine togetherwith plagioclase and rhoenite, whereas no spinel was found. Thisapplies to both lava and enclaves.

The coexistence of clinopyroxene and olivine together with theabsence of spinel in the breakdown corona leads us to the conclusionthat the PT conditions during the breakdown of amphibole were set atabout 1090–1120 °C and 400–100 bar. This implies that relativelyhigh temperatures still existed during the breakdown of amphiboleclose to the surface. These high breakdown temperatures also suggestthat there was only a little cooling during the ascent of the magmafrom the Moho to the surface, pointing to a fast ascent. Moreover, wecan assume that the temperature of the amphibole breakdown wasvery close to the eruption temperature.

6. Evolutionary model of the lavas from Kula Volcanic Province

The comparison of geochemical data from enclaves, their host rocks,and other lava flows gives a detailed insight into the processes ofmagma ascent below the Kula Volcanic Province. In Fig. 15 we assem-bled all our data and observations in order to constrain a comprehensivemodel for the evolution of the Kula volcanism.

A primarymelt that is parental tomost of the Kula lavas is generatedby partial melting of asthenospheric mantle. This takes place within thestability field of spinel peridotite at pressures between 20 and 26 kbar(Sölpüker, 2007). The ascendingmantle-derived basaltic melts stagnateat pressures of 12–15 kbar, and a magma storage region, probablyincluding small chambers or feeder dykes, has developed. At tempera-tures of 1150–1200 °C the most primitive clinopyroxenes start crystal-lizing. The REE ratios imply that they have a comagmatic characterwith clinopyroxenes from the lavas. Clinopyroxenes are accompaniedby amphibole and Sc-depleted olivine. Cumulates consisting mainly ofamphibole and clinopyroxene are stored in the feeder system. A verticaltrend demonstrated by variation of compatible elements in olivine

ding information fromAlıcı et al. (2002), Holness and Bunbury (2006), Kunzmann (1989),

72 T. Grützner et al. / Lithos 180–181 (2013) 58–73

indicates low viscosity and little circulation in themagma at the bound-ary layer of the grains within this magma storage region.

After a period of time, a second batch of melt rises and incorporatesthe cumulates as enclaves. This time period has to be short enough forthe enclave host magma to use the old feeder dyke system before itcrystallizes (Holness and Bunbury, 2006). This new melt originatesfrom the same parental magma as the first batch, but is most probablyinfluenced by a slightly different mantle source: Fe-rich green-coreclinopyroxenes with Mg# down to 50 were found only in the lavasamples. Their REE ratios suggest the presence of heterogeneities in thelithospheric mantle. Due to the fact that their Mg-rich mantles crystal-lized under the same PT conditions as themost primitive clinopyroxenes,these xenocrysts had to be incorporated into themagma at the same time(or even before) the primitive clinopyroxenes started crystallizing. Therounded shape of the green cores suggests a resorption of the formerphenocrysts as (or before) they were incorporated into the basaniticmelt.

Close to the surface, at pressures below0.6 kbar and temperatures ofabout 1100 °C, amphibole is resorbed and forms a breakdown coronawhich contains rhoenite. The calculated minimum ascension ratesuggests an ascension time of about 4–11 days from the Moho to thesurface. Therefore, the lava must have erupted out at a temperatureslightly below 1100 °C.

Acknowledgments

We are indebted to Stephen Foley, Dorrit Jacob and Stefan Buhrewho enabled measurements at the ICP-MS and at the Microprobe inMainz. We thank the staff of Mainz, especially Nora Grohschopf andAntje Huttenlocher for the XRF analysis, as well as Adalbert Becker forhis help with the sample preparation. We would like to thank reviewerMichel Grégoire and guest editor Ioan Seghedi for their constructivecomments which improved the paper significantly. We have to thankMarina Alletti for the favorable review. Finally, our special thanks goto Sebastian Fischer, Stephan Klemme, S.M. and Valeria Handke forexcellent suggestions and fruitful discussions.

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.lithos.2013.08.001.

References

Adam, J., Green, T., 2006. Trace element partitioning between mica- and amphibole-bearing garnet lherzolite and hydrous basanitic melt: 1. Experimental results andthe investigation of controls on partitioning behavior. Contributions to Mineralogyand Petrology 152, 1–17.

Alıcı, P., Temel, A., Gourgaud, A., 2002. Pb–Nd–Sr isotope and trace element geochemistryof Quaternary extension-related alkaline volcanism. a case study of Kula region(western Anatolia, Turkey). Journal of Volcanology and Geothermal Research 115,487–510.

Alletti, M., Pompilio, M., Rotolo, S.G., 2005. Mafic and ultramafic enclaves in Ustica Islandlavas: inferences on composition of lower crust and deepmagmatic processes. Lithos84 (3–4), 151–167.

Baker, D.R., Eggler, D.H., 1987. Compositions of Anhydrous and Hydrous Melts Coexistingwith Plagioclase, Augite, and Olivine or Low-Ca Pyroxene from 1 atm to 8 kbar -Application to the Aleutian Volcanic Center of Atka. AmericanMineralogist 72, 12–28.

Beattie, P., 1994. Systematics and energetics of trace element partitioning between olivineand silicate melts; implications for the nature of mineral/melt partitioning. FifthInternational Symposium on Experimental Mineralogy, Petrology and Geochemistry.Blackwell Scientific Publications, Oxford, International, p. 3.

Biryol, C.B., Beck, S.L., Zandt, G., Özacar, A.A., 2011. Segmented African lithosphere beneaththe Anatolian region inferred from teleseismic P-wave tomography. GeophysicalJournal International 184, 1037–1057.

Bowen, N.L., 1928. The evolution of the igneous rocks. Princeton University Press,Princeton, New Jersey.

Conticelli, S., D'Antonio, M., Pinarelli, L., Civetta, L., 2002. Source contamination andmantle heterogeneity in the genesis of Italian potassic and ultrapotassic volcanicrocks: Sr–Nd–Pb isotope data fromRoman Province and Southern Tuscany. Mineralogyand Petrology 74, 189–222.

Conticelli, S., Guarnieri, L., Farinelli, A., Mattei, M., Avanzinelli, R., Bianchini, G., Boari, E.,Tommasini, S., Tiepolo, M., Prelević, D., Venturelli, G., 2009. Trace elements andSr–Nd–Pb isotopes of K-rich, shoshonitic, and calc-alkalinemagmatism of theWesternMediterraneanRegion: genesis of ultrapotassic to calc-alkalinemagmatic associations ina post-collisional geodynamic setting. Lithos 107, 68–92.

Cox, K.G., Bell, J.D., Pankhurst, R.J., 1979. The Interpretation of Igneous Rocks. G. Allen &Unwin, London, Boston.

Draper, D.S., Green, T.H., 1999. P-T phase relations of silicic, alkaline, aluminous liquids:new results and applications to mantle melting and metasomatism. Earth and Plane-tary Science Letters 170, 255–268.

Duda, A., Schmincke, H.-U., 1985. Polybaric differentiation of alkali basaltic magmas.evidence from green-core clinopyroxenes (Eifel, FRG). Contributions to Mineralogyand Petrology 91, 340–353.

Ercan, T., 1993. Interpretation of geochemical, radiometric and isotopic data on Kulavolcanics (Manisa — W. Anatolia). Geological Bulletin of Turkey 36, 113–129.

Ercan, T., Günay, E., Dinçel, A., Türkecan, A., Küçükayman, A., 1980. In: Rapor, M.T.A. (Ed.),Kula-Selendi yörelerinin jeolosjisi ve volkanitlerin petrolosjisi (IV, 01.0.01.07.02).

Faccenna, C., Piromallo, C., Crespo-Blanc, A., Jolivet, L., Rossetti, F., 2004. Lateral slab deforma-tion and the origin of the western Mediterranean arcs. Tectonics 23, TC1012.

Falloon, T.J., Danyushevsky, L.V., Green, D.H., 2001. Peridotite Melting at 1 GPa: ReversalExperiments on Partial Melt Compositions Produced by Peridotite-Basalt SandwichExperiments. Journal of Petrology 42, 2363–2390.

Foley, S.F., Prelević, D., Rehfeldt, T., Jacob, D.E., 2013. Minor and trace elements in olivinesas probes into early igneous and mantle melting processes. Earth and Planetary Sci-ence Letters 363, 181–191.

Fytikas, M., Innocenti, F., Manetti, P., Mazzuoli, R., Peccerillo, A., Villari, L., 1984. Tertiary toQuaternary evolution of volcanism in the Aegean region. In: Dixon, J.E., Robertson,A.H.F. (Eds.), The Geological Evolution of the Eastern Mediterranean, 17. GeologicalSociety, Special Publication, London, pp. 687–699.

Göksu, Y., 1978. The TL age determination of fossil human footprints. Archaeo-Physica 10,445–462.

Grapes, R.H., Wysoczanski, R.J., Hoskin, P.W.O., 2003. Rhönite paragenesis in pyroxenitexenoliths, Mount Sidley volcano, Marie Byrd Land, West Antarctica. MineralogicalMagazine 67 (4), 639–651.

Grove, T.L., Bryan, W.B., 1983. Fractionation of pyroxene-phyric MORB at low pressure:An experimental study. Contributions to Mineralogy and Petrology 84, 293–309.

Gülen, L., 1990. Isotopic characterization of Aegean magmatism and geodynamic evolutionof the Aegean subduction. In: Savaşçin, M.Y., Eronat, A.H. (Eds.), International EarthScience Congress on Aegean Regions, vol. 2. Izmir — Turkey, pp. 143–167.

Hamilton, W.J., Strickland, H.E., 1841. I.—on the geology of the western part of Asia Minor.Transactions of the Geological Society of London Series 2 6 (1), 1–39.

Hart, S.R., Davis, K.E., 1978. Nickel partitioning between olivine and silicate melt. Earthand Planetary Science Letters 40, 203–219.

Hawthorne, F.C., Oberti, R., 2006. On the classification of amphiboles. The Canadian Min-eralogist 44, 1–21.

Hirose, K., 1997. Melting experiments on lherzolite KLB-1 under hydrous conditions andgeneration of high-magnesian andesitic melts. Geology (Boulder) 25, 42–44.

Hirose, K., Kushiro, I., 1993. Partial melting of dry peridotites at high pressures; determina-tion of compositions of melts segregated from peridotite using aggregates of diamond.Earth and Planetary Science Letters 114, 477–489.

Holness, M.B., Bunbury, J.M., 2006. Insights into continental rift-relatedmagma chambers.Cognate nodules from the Kula Volcanic Province, Western Turkey. Journal of Volca-nology and Geothermal Research 153, 241–261.

Innocenti, F., Agostini, S., Di Vincenzo, G., Doglioni, C., Manetti, P., Savaşçin,M., Tonarini, S.,2005. Neogene and Quaternary volcanism in Western Anatolia: magma sources andgeodynamic evolution. Marine Geology 221 (1–4), 397–421.

Jaques, A.L., Green, D.H., 1980. Anhydrousmelting of peridotite at 0–15 kbar pressure andthe genesis of tholeiitic basalts. Contributions to Mineralogy and Petrology 73,287–310.

Kelsey, C.H., 1965. Calculation of the CIPW norm. Mineralogical Magazine 34, 276–282.Kinzler, R.J., Grove, T.L., Recca, S.I., 1990. An experimental study on the effect of temperature

and melt composition on the partitioning of nickel between olivine and silicate melt.Geochimica et Cosmochimica Acta 54, 1255–1265.

Kretz, R., 1983. Symbols for rock-forming minerals. American Mineralogist 68, 277–279.Kunzmann, T., 1999. The aenigmatite–rhonite mineral group. European Journal of

Mineralogy 11, 743–756.Kunzmann, T., 1989. Rhönit: Mineralchmie, Paragenese und Stabilität in alkalibasaltischen

Vulkaniten (Ein Beitrag zur Mineralogenese der Rhönit-Änigmatit Mischkristallgruppe).Inaugural-Dissertation zur Erlangung des Doktorgrades der Fakultät fürGeowissenschaften der Ludwig-Maximilian-Universität. München, Germany.

Le Maitre, R.W., Bateman, P., Dudek, A., Keller, J., Lameyre, J., Bas, M.J., et al., 1989. A Classifi-cation of Igneous Rocks and Glossary of Terms. Blackwell Scientific Publication, Oxford.

Le Pichon, X., Angelier, J., 1979. Hellenic Arc and Trench System - Key to the NeotectonicEvolution of the Eastern Mediterranean Area. Tectonophysics 60, 1–42.

Longhi, J., 1991. Comparative liquidus equilibria of hypersthene normativebasalts at low-pressure. American Mineralogist 76, 785–800.

Lustrino, M., Wilson, M., 2007. The circum-Mediterranean anorogenic Cenozoic igneousprovince. Earth-Science Reviews 81, 1–65.

Lustrino, M., Duggen, S., Rosenberg, C.L., 2011. The central-western Mediterranean: anoma-lous igneous activity in an anomalous collisional tectonic setting. Earth-Science Reviews104, 1–40.

Middlemost, E.A.K., 1989. iron oxidation ratios, norms and the classification of volcanicrocks. Chemical Geology 77, 19–26.

Morimoto, N., Fabries, J., Ferguson, A.K., Ginzburg, I.V., Ross, M., Seifert, F.A., Zussman, J.,Aoki, K., Gottardi, G., 1988. Nomenclature of pyroxenes. Mineralogical Magazine 52,535–550.

73T. Grützner et al. / Lithos 180–181 (2013) 58–73

Nehring, F., Jacob, D.E., Barth, M.G., Foley, S.F., 2008. Laser-ablation ICP-MS analysis ofsiliceous rock glasses fused on an iridium strip heater using MgO dilution.Microchimica Acta 160, 153–163.

Nielsen, R.L., Dungan, M.A., 1983. Low-Pressure Mineral-Melt Equilibria in NaturalAnhydrous Mafic Systems. Contributions to Mineralogy and Petrology 84,310–326.

Özacar, A.A., Biryol, C.B., Beck, S., Zandt, G., Kaymakci, N., 2010. Crust and UpperMantle Dynamics of Turkey Inferred from Passive Seismology: Implications ofSegmented Slab Geometry. In: Dilek, Y., Bozkurt, E. (Eds.), Proceedings of theConference Tectonic Crossroads. Evolving Orogens of Eurasia-Africa-Arabia,Ankara, Turkey, p. 46.

Pearce, N.J.G., Perkins, W.T., Westgate, J.A., Gorton, M.P., Jackson, S.E., Neal, C.R., Chenery,S.P., 1997. A compilation of new and publishedmajor and trace element data for NISTSRM 610 and NIST SRM 612 glass reference materials. Geostandards Newsletters 21,115–144.

Pilet, S., Ulmer, P., Villiger, S., 2010. Liquid line of descent of a basanitic liquid at 1.5 Gpa:constraints on the formation of metasomatic veins. Contributions to Mineralogy andPetrology 159, 621–643.

Pletchov, P.Y., Trousovt, S.V., 2000. In: DGGGMSRAS, Herald (Ed.), Influence of the BoundaryLayer Effects to the Melt Inclusions Compositions. Moscow State University, Geologicaldepartment, Moscow (5, 15).

Prelević, D., Foley, S.F., Romer, R.L., Cvetkovic, V., Downes, H., 2005. Tertiary ultrapotassicvolcanism in Serbia: constraints on petrogenesis and mantle source characteristics.Journal of Petrology 46, 1443–1487.

Prelević, D., Akal, C., Foley, S.F., Romer, R.L., Stracke, A., Van Den Bogaard, P., 2012.Ultrapotassic mafic rocks as geochemical proxies for post-collisional dynamics oforogenic lithospheric mantle: the case of Southwestern Anatolia, Turkey. Journal ofPetrology 53, 1019–1055.

Prelević, D., Jacob, Dorrit E., Foley, S.F., 2013. Recycling plus: a new recipe for the formationof Alpine–Himalayan orogenic mantle lithosphere. Earth and Planetary Science Letters362, 187–197.

Putirka, K.D., 1999. Clinopyroxene + liquid equilibria to 100 kbar and 2450 K. Contribu-tions to Mineralogy and Petrology 135, 151–163.

Putirka, K.D., Mikaelian, H., Ryerson, F., Shaw, H., 2003. New clinopyroxene–liquidthermobarometers for mafic, evolved, and volatile-bearing lava compositions, withapplications to lavas from Tibet and the Snake River Plain, Idaho. AmericanMineralogist88, 1542–1554.

Richardson-Bunbury, J.M., 1996. The Kula Volcanic Field, western Turkey: the developmentof a Holocene alkali basalt province and th adjacent normal-faulting graben. GeologicalMagazine 133 (3), 275–283.

Sack, R., Walker, D., Carmichael, I.E., 1987. Experimental petrology of alkalic lavas:constraints on cotectics of multiple saturation in natural basic liquids. Contributionsto Mineralogy and Petrology 96, 1–23.

Saunders, P., Priestley, K., Taymaz, T., 1998. Variations in the crustal structure beneathwestern Turkey. Geophysical Journal International 134, 373–389.

Seyitoğlu, G., Scott, B.C., 1996. The cause of N-S extensional tectonics in western Turkey:Tectonic escape vs back-arc spreading vs orogenic collapse. Journal of Geodynamics22, 145–153.

Shaw, C.S.J., 1999. Dissolution of orthopyroxene in basanitic magma between 0.4 and2 GPa: further implications for the origin of Si-rich alkaline glass inclusions in mantlexenoliths. Contributions to Mineralogy and Petrology 135, 114–132.

Shaw, C.S.J., Thibault, Y., Edgar, A.D., Lloyd, F.E., 1998. Mechanisms of orthopyroxene dis-solution in silica-undersaturated melts at 1 atmosphere and implications for the ori-gin of silica-rich glass inmantle xenoliths. Contributions to Mineralogy and Petrology132, 354–370.

Smith, V.G., Tiller, W.A., Rutter, J.W., 1955. A mathematical analysis of solute redistributionduring solidifaction. Canadian Journal of Physics 33 (12), 723–745.

Sölpüker, U., 2007. Petrology of Kula Volcanic Province, Western Turkey. dissertationUniversity of Cincinnati (Available online at http://etd.ohiolink.edu/view.cgi?acc_num=ucin1187013478).

Spakman,W.,Wortel, M.J.R., Vlaar, N.J., 1988. The Hellenic Subduction Zone— a tomographicimage and its geodynamic implications. Geophysical Research Letters 15, 60–63.

Spakman,W., Vanderlee, S., Vanderhilst, R., 1993. Travel-time tomography of the European–Mediterranean mantle down to 1400 km. Physics of the Earth and Planetary Interiors79, 3–74.

Spera, F., 1980. Aspects of magma transport. In: Hargraves, Robert B. (Ed.), Physics ofMagmatic Processes. Princeton University Press, Princeton, N.J., pp. 265–324.

Sun, S.-s, McDonough, W., 1989. Chemical and isotopic systematics of oceanic basalts.implication for mantle composition and processes. In: Saunders, A.D., Norry, M.J.(Eds.), Magmatism in the Oceanic Basins. Geological Society, Special Publication,London, pp. 313–345.

Tokçaer, M., Agostini, S., Savaşçın, M.Y., 2005. Geotectonic setting and origin of the youngestKula volcanics (Western Anatolia), with a new emplacement model. Turkish Journal ofEarth Sciences 14, 145–166.

Van Achterbergh, E., Ryan, C.G., Jackson, S.E., Griffin, W.L., 2000. Data reduction software forLA-ICPMS: appendix. In: Sylvester, P.J. (Ed.), Laser Ablation-ICP-Mass Spectrometry inthe Earth Sciences: Principles and Applications. Mineralogical Association of CanadaShort Course Series, pp. 239–243.

Vaselli, O., Downes, H., Thirlwall, M., Dobosi, G., Coradossi, N., Seghedi, I., Szakacs, A.,Vannucci, R., 1995. Ultramafic xenoliths in Plio-Pleistocene alkali basalts from theeastern Transylvanian Basin: depleted mantle enriched by vein metasomatism. Jour-nal of Petrology 36, 23–53.

Westaway, R., Pringle, M., Yurtmen, S., Demir, T., Bridgland, D., Rowbotham, G.,Maddy, D., 2004. Pliocene and Quaternary regional uplift in western Turkey:the Gediz River terrace staircase and the volcanism at Kula. Tectonophysics391 (1–4), 121–169.

Wilson, M., Biancini, G., 1999. Tertiary-Quatenary magmatism within the Mediterraneanand surrounding regions. In: Durand, B., Jolivet, L., Horvath, F., Seranne, M. (Eds.), TheMediterranean Basins: Tertiary Extension within the Alpine Orogen. Geological Soci-ety, Special Publications, London, pp. 141–168.