Zircon in amphibolites from Naxos, Aegean Sea, Greece ...people.geo.su.se/uwe/publications/Bolhar_et_al... · Zircon in amphibolites from Naxos, Aegean Sea, Greece: origin, signiﬁcance

Zircon in amphibolites from Naxos, Aegean Sea, Greece: origin,significance and tectonic setting

R. BOLHAR,1 , 2 U. RING3 AND T. R. IRELAND4

1School of Geosciences, University of the Witwatersrand, Johannesburg 2001, South Africa ([email protected])2School of Earth Sciences, University of Queensland, Brisbane, Qld 4072, Australia3Institutionen f€or Geologiska Vetenskaper, Stockholms Universitet, 10691 Stockholm, Sweden4Research School of Earth Sciences, Australian National University, Canberra, ACT 0200, Australia

ABSTRACT We report U–Pb zircon ages of c. 700–550 Ma, 262–220 Ma, 47–38 Ma and 15–14 Ma from amphi-bolites on Naxos Island in the Aegean extensional province of Greece. The zircon has complex inter-nal structures. Based on cathodoluminescence response, zoning and crosscutting relationships aminimum of four zircon growth stages are identified: inherited core, magmatic core, inner metamor-phic (?) rim and an outer metamorphic rim. Trace element compositions of the amphibolites suggestigneous differentiation and crustal assimilation. Zircon solubility as a function of saturation tempera-tures, Zr content and melt composition indicates that the zircon did not originally crystallize in themafic bodies but was inherited from felsic precursor rocks, and subsequently assimilated into themafic intrusives during emplacement. Zircon inheritance is corroborated by the complex, xenocrysticnature of the zircon in one sample. Ages of c. 700–550 Ma and 262–220 Ma are assigned to inheritedzircon. Available geochemical data suggest that the 15–14 Ma metamorphic rims grew in situ in theamphibolites, corresponding to a high-grade metamorphic event at this time. However, the geochemi-cal data cannot conclusively establish if the c. 40 Ma zircon rims also grew in situ, or whether theywere inherited along with the xenocrystic cores. Two scenarios for emplacement of the mafic intru-sives are discussed: (i) Intrusion during late-Triassic to Jurassic ocean basin development of theAegean realm, in which case the 40 Ma zircon rims would have grown in situ, and (ii) emplacementin the Miocene as a result mafic underplating during large-scale extension. In this case, only the 15–14 Ma metamorphic outer rims would have formed in situ in the amphibolitic host rocks.

Key words: amphibolite; Cyclades; inheritance; Naxos; saturation; zircon.

INTRODUCTION

Zircon is a common accessory mineral in many conti-nental, SiO2-rich magmatic, metamorphic and sedi-mentary rocks, largely due to its ability to form andrecrystallize under wide-ranging pressure–temperature(P–T) conditions, and its chemical and mechanicalresilience. Conversely, zircon has not been consideredas common in mafic volcanic rocks, although severalstudies have recognized its presence in mafic plutonicrocks formed along mid-oceanic ridges (e.g. Muiret al., 1998; Coogan & Hinton, 2006; Grimes et al.,2007). Owing to its robustness and reliability as ageochronometer, zircon has been employed to estab-lish crystallization ages of mafic intrusives, includinggabbros and dolerites (e.g. Ring et al., 2002; Denget al., 2014; Campanha et al., 2015), in cases whererelatively uniform age populations are present withinthe respective rock. Curiously, it must be consideredunlikely for zircon to crystallize from a mafic (SiO2

<52 wt%) melt under temperature conditions preva-lent in the crust at the time of emplacement, since zir-con requires the precipitating melt to attain

unrealistically high Zr contents, in excess of5000 ppm, for zircon saturation (Boehnke et al.,2013). It is possible that some zircon may form inhighly evolved, late-stage melt pockets (e.g. Robertset al., 2013), but it is generally more likely that zir-con, if present in a mafic host rock, was derived fromextraneous zircon-bearing sources, during ascent and/or emplacement of mafic melts in the crust. Zircon-bearing amphibolites from Naxos form the subject ofthis study and provide an opportunity to evaluateboth of the above scenarios, as well as to examinethe response of zircon in mafic host rocks to meta-morphic overprinting.The island of Naxos is located in the Aegean Sea

region, which is a world-class example of large-scalecontinental extension behind a retreating subductionzone. After the Variscan orogeny in the Carbonifer-ous, parts of the Aegean experienced Triassic mag-matism at c. 240–230 Ma (see below). Subsequently,late-Triassic to Jurassic plate divergence formed theVardar-_Izmir-Ankara ocean and the Pindos basin,and resulted in the palaeogeographic configurationand depositional geometry of the Hellenides (D€urr

© 2016 John Wiley & Sons Ltd 1

J. metamorphic Geol., 2016 doi:10.1111/jmg.12238

et al., 1978; Robertson et al., 1991). In the Cycladesin the central Aegean Sea (Fig. 1), the Hellenide Oro-geny commenced in the early Cenozoic causing sub-duction and sustained high-P metamorphism of theCycladic Blueschist Unit (CBU) between c. 53 Maand c. 30 Ma (Wijbrans et al., 1990; Tomascheket al., 2003; Ring et al., 2007). At c. 23 Ma, large-scale continental extension commenced in theCyclades in the forearc region of the southwardretreating Hellenic subduction zone (Ring et al.,2010). Extensional deformation mainly progressedunder greenschist facies conditions, but in the centralCyclades (Naxos, Paros, Ios and Ikaria islands) athermal anomaly formed as a result of extension.

On the island of Naxos a spectacular migmatitedome is exposed, some ~20 by 8 km in size (Jansen,1973; Jansen & Schuiling, 1976; Vanderhaeghe, 2004)(Fig. 2). The migmatites formed between c. 20 and14 Ma as a result of extensional deformation (Wij-brans & McDougall, 1986; Buick & Holland, 1989;Keay et al., 2001). A major question remains as towhere the heat for the thermal anomaly in the centralCyclades was sourced. Ring et al. (2010) showed thatthe central Cyclades are the most highly extendedregion of the central Aegean and as a result wereprobably subjected to large-scale magmatic under-plating, which then may have caused the thermalanomaly. Numerous amphibolite layers and lensesoccur in and around the Naxos migmatite dome(Fig. 3). The lenses are usually a few centimetres upto a few metres in thickness and laterally pinch out

after some 10 to ~100 m. The amphibolite bodies arestrongly deformed and metamorphosed during exten-sional shearing and may represent basaltic sills thatintruded during and after underplating. Alternatively,the mafic bodies could have been intruded in theMesozoic, with transformation into amphibolites dur-ing the Tertiary Hellenic Orogeny.Here, we present whole-rock and zircon geochemi-

cal data measured by laser ablation quadrupoleinductively coupled plasma mass spectrometry (LA-Q-ICP-MS) and secondary ionization mass spectrom-etry (SIMS; mid section, depth profiling) for fouramphibolites from the periphery of the Naxos mig-matite dome. Zircon data were obtained from care-fully characterized growth zones defined by CLimagery. The aims are twofold: (i) to examine theorigin and nature of mafic to intermediate rock-hosted zircon, specifically timing and conditions ofzircon growth, and assessment of in situ crystalliza-tion of zircon from mafic melts, and (ii) to test themagmatic underplating hypothesis through examina-tion of the tectono-thermal history involving maficmagmatism.

TECTONIC SETTING

The Hellenides in the eastern Mediterranean form anarcuate orogen to the north of the present-day activemargin (Fig. 1), which marks the site of northeast-ward subduction of the African plate beneath Eura-sia. The Hellenides can be subdivided from top

Fig. 1. Tectonic map of the Aegean regionshowing main tectonic zones above Hellenicsubduction zone, major low-angleextensional detachments and high-anglenormal faults; note Naxos Island andthermal anomaly in central Aegean; Ionianand Tripolitza zones form ExternalHellenides.

© 2016 John Wiley & Sons Ltd

2 R . BOLHAR ET AL .

(north) to bottom (south) into: (i) the Internal zone,(ii) the Vardar-_Izmir-Ankara suture zone, (iii) thePelagonian zone, (iv) the Cycladic zone and (v) theExternal Hellenides (D€urr et al., 1978; Robertsonet al., 1991; van Hinsbergen et al., 2005). The Inter-nal zone consists of continental fragments of the Eur-asian plate, underneath which oceanic crust of theNeotethys was subducted during Cretaceous

convergence (Robertson et al., 1991). The relatedsuture is the ophiolitic Vardar-_Izmir-Ankara zone,which in part was metamorphosed under blueschistfacies conditions in the late Cretaceous (Sherlocket al., 1999). The underlying Pelagonian zone is athrust belt that was in part also metamorphosedunder high-P conditions (Franz & Okrusch, 1992).The Pelagonian was also overthrust by Jurassic

Normal fault(oblique-slip component)

Amphibolite sample

Upper unit

Passive-margin sequence

Basement

Tertiary sediments

Extensionaldetachment

I-type granodiorite

S-type granite

Schist and marble

Migmatite andleucogneiss

37°00’25°35’

37°05’

25°20’

37°10’

25°35’

0 1 2 3 4

N

Na50

Na49 Na48

Na41

km

Fig. 2. Geological map of Naxos showing sample locations in the northwestern portion of the island. The migmatite dome isshown in white; the western part of the island is made up by the Naxos granodiorite.


Z IRCON IN AMPHIBOL I TES 3

ophiolites of the Vardar-_Izmir-Ankara zone in theearly Cretaceous (e.g. Tremblay et al., 2015). TheCycladic zone consists of continental fragments ofthe Adriatic plate and can be further subdivided intothree tectonic units (Ring et al., 1999), which arefrom top to bottom: (i) the non- to weakly metamor-phosed ophiolitic Upper unit, (ii) the high-P rocks ofthe CBU, which is subdivided into three separatemembers: (a) an ophiolitic m�elange, (b) a Permo-Car-boniferous to uppermost Cretaceous passive-marginsequence and (c) a Carboniferous basement nappe,which also occurs as slices in the passive-marginsequence. (iii) The Basal unit as part of the ExternalHellenides consists of Mesozoic and lower Cenozoicplatform carbonates found in several tectonic win-dows (Avigad & Garfunkel, 1989).

The different zones represent, at least in part, dis-tinct palaeogeographic entities that formed duringTriassic rifting and Jurassic drifting within Tethys(Robertson, 2002; Tremblay et al., 2015) and createdribbon-like zones of thinned continental and oceanic

crust with intervening ridges on which carbonateplatforms developed in the Mesozoic. Reischmann(1998) provided evidence for a period of mid-Triassicgranitic magmatism prior to late-Triassic to Jurassicocean formation. Mid-Triassic granites are knownfrom the islands of Samos (Ring et al., 1999), Naxos(Reischmann, 1998), and Evia and Nikouria (U. Ring& R. Bolhar, unpublished data). Br€ocker & Pidgeon(2007) reported mid-Triassic ages for magmatic pre-cursors of meta-tuffaceous and meta-volcanic rocksfrom the islands of Andros, Sifnos and Ios. All theseages attest to a regional mid-Triassic magmatic event.There is also evidence of Jurassic ophiolitic rocksfrom the islands of Andros and Crete (Koepke et al.,2002; Br€ocker & Pidgeon, 2007) and late Cretaceousmafic magmatism (Br€ocker et al., 2014) that is mostlikely related to a late Cretaceous ophiolite belt thatextended from eastern Greece via Turkey, Cyprus,Syria into Oman (Parlak & Delaloye, 1999; Robert-son, 2002; Hinsbergen et al., 2016). Subsequently,rocks from various palaeogeographic zones were suc-cessively subducted and metamorphosed from thenorth to the south from the Cretaceous until therecent (Robertson et al., 1991; Ring & Layer, 2003).Some debate persists as to how many oceanic

basins existed in the Hellenides. There is widespreadevidence that the Vardar-_Izmir-Ankara zone was partof Neotethys (Robertson et al., 1991; Tremblay et al.,2015). Whether or not the Pindos basin was oceanicremains controversial. The widespread occurrence ofradiolarites may suggest that parts of the Pindosbasin were floored by oceanic crust, but structuraland stratigraphic constraints do not support an ocea-nic basin (Tremblay et al., 2015). Jurassic ophiolitesextend much farther north than the Pindos basin, thelatter of which ends in northern Albania. The age ofrifting in the Adriatic units and the Pelagonian ismid-Triassic, and rifting is associated with wide-spread mafic magmatism, including magmatic activityin the Pindos basin. Oceanic sediments in the mel-anges below the Jurassic ophiolites, which formed ata mid-ocean ridge at c. 170 Ma (Maffione et al.,2015), date back to the late-Triassic, showing thatthe onset of oceanization already occurred duringthis time (e.g. Schmid et al., 2008).

GEOLOGY OF NAXOS AND AMPHIBOLITES

Naxos Island occurs within the central Cycladic zoneand is primarily made up by the passive-marginsequence and basement nappe of the CBU. The con-tact between these two units has lenses of serpentiniteand is considered to be a large-scale thrust (Jansen,1973). The basement nappe is largely made up ofCarboniferous granitic gneiss (Reischmann, 1998),which is now migmatitic. Triassic gneiss dated at233 � 2 Ma (single zircon Pb/Pb dating; Reis-chmann, 1998) also occurs in the passive-marginsequence. The basement/passive-margin sequence is

(a)

(b)

Fig. 3. (a) Strongly foliated amphibolite layer in east Naxos;note that amphibolite is fully transposed into foliation-parallelorientation. (b) Amphibolite body from which sample Na49was collected; the amphibolite is foliated at its margins; notethe much stronger foliation development in surrounding schist.



intruded by granitic dykes between c. 15 and 12 Ma(Keay et al., 2001) and a large-scale I-type granodior-ite in the western part of the island between c. 13and 12 Ma (Bolhar et al., 2010). The CBU is sepa-rated in western and eastern Naxos from the Upperunit by the top-NNE displacing Naxos extensionaldetachment (Buick, 1991). Extensional deformationcommenced at c. 23–21 Ma and lasted until c. 10–8 Ma (John & Howard, 1995; Brichau et al., 2006;Seward et al., 2009).

Numerous amphibolite layers and lenses occurespecially within the passive-margin sequence(Fig. 2). The amphibolites are heterogeneouslydeformed, and have been fully transposed into a foli-ation-parallel orientation (Fig. 3a). Commonly, theyhave a mylonitic foliation, a NNE-trending stretchinglineation and top-NNE kinematic indicators thatformed under amphibolite/greenschist facies meta-morphism. However, we also observed relativelyweakly deformed amphibolites (Fig. 3b), which arealso in a foliation-parallel orientation. However, thefoliation is more weakly developed in the central por-tions of these amphibolites (Fig. 3b). The generalmineralogy is epidote, hornblende, plagioclase, chlo-rite, rutile, ilmenite and titanite. X-ray diffractionanalysis indicates that all amphibolites contain garnetand biotite.

The amphibolites underwent high-P metamorphismfollowed by a greenschist to upper amphibolite faciesoverprint (Buick & Holland, 1989; Avigad, 1998;Martin et al., 2006). Published ages for the high-Pevent scatter between c. 70 and 40 Ma (Wijbrans &McDougall, 1986; Martin et al., 2006). MinimumU–Pb ages of zircon rims by Martin et al. (2006) arec. 42 Ma, but these authors also reported olderzircon rim ages of up to c. 69 Ma. Recent Rb–Srmineral data for blueschist facies metamorphic mobi-lizates provide an age for late prograde metamorphicdehydration reactions at 40.5 � 1.0 Ma, representinga maximum age for the peak of high-P metamor-phism on Naxos (A. Peillod, U. Ring, J. Glodny &A. Skelton, unpublished data). Therefore, the olderages of up to c. 69 Ma by Martin et al. (2006) areprobably best interpreted as mixed ages between themetamorphic rims and magmatic cores. The youngestgenerations of rims at c. 42 Ma could be regarded asthe best estimate of an ‘unbiased’ age for the high-Poverprint on Naxos. The subsequent greenschist toupper amphibolite facies overprint occurred betweenc. 20 and 14 Ma (Wijbrans & McDougall, 1986,1988). The S-type granites intruded during and afterthe waning stage of this metamorphic overprintbefore the magmatic arc of the southward retreatingHellenic subduction zone swept through causing theintrusion of the I-type granodiorite and associatedcontact metamorphism (Jansen, 1973). Fission-trackand (U–Th)/He ages constrain the arrival of theCBU in the brittle crust at c. 25–10 Ma (Brichauet al., 2006; Seward et al., 2009). The older fission-

track ages of c. 25–20 Ma are from the south ofNaxos and show that the rocks in the south did notexperience the c. 20–14 Ma metamorphic event.Martin et al. (2006) performed U–Pb zircon dating

on a single amphibolite sample from the passive-mar-gin sequence and obtained ages of c. 270–200 Ma forzircon cores, c. 69–42 Ma for a first generation of zir-con rims, and c. 19–14 Ma for a second rim genera-tion. Zircon cores and first-generation rims havenearly identical d18O values (6.2 � 0.8& and7 � 1& respectively), whereas d18O values of7.8 � 1.8& for the outer rims were interpreted byMartin et al. (2006) as being higher, thereby suggest-ing distinct differences between the two rim genera-tions. They further concluded that identical d18Ovalues of the outer zircon rims and garnet indicatethat both grew together during Miocene high-T meta-morphism. Growth of the first-generation rims at c.69–42 Ma was not in equilibrium with garnet growthin the amphibolite (Martin et al., 2006, p. 187).

METHODS

Whole-rock geochemical analysis

X-ray fluorescence

Major and trace element analysis were determined byX-ray fluorescence (XRF) at the PetroTectonics Ana-lytical Facility, Stockholm University. Samples werepulverized using a stainless steel swing mill for 2 min.Two grams of the resulting powder were mixed with5 g of flux (66% lithium tetraborate: 34% lithiummetaborate), weighed to a precision of �0.0002 g.The stirred mixture was fused in platinum cruciblesfor 10 min at 1100 °C using a Phoenix VFD auto-mated fusion machine to obtain a homogeneous32 mm glass disc with a lower surface of analyticalquality. The concentrations of 10 major elementswere determined relative to X-ray intensities for eachelement derived from analysis of 24 international ref-erence materials with known concentrations. Operat-ing conditions are similar to other facilities (e.g.Johnson et al., 1999). USGS reference material BCR1, AGV 2 and RGM 1 were analysed as unknownsduring the analytical session for data quality control(Table S1). Analytical precision for most major ele-ments is better than 1%.

Laser ablation inductively coupled mass spectrometry

Cores of 3–5 mm diameter were drilled out of theXRF glass discs and embedded in epoxy resin. Afterpolishing, the glasses were analysed by laser ablationinductively coupled mass spectrometry (LA-ICP-MS)using a New Wave 193 excimer laser coupled to aThermo Xseries2 quadrupole ICP-MS at thePetroTectonics analytical facility, Department ofGeological Sciences, Stockholm University. Each



sample was analysed at three spots with a 150 lmspot size. Analytical conditions included a laser pulsefrequency of 10 Hz and a laser energy density of7.5 J cm�2. Each analysis consisting of 30 s back-ground acquisition followed by 40 s data acquisitionand 15 s wash-out. Analysed isotopes (and oxides)were, with dwell times in ms in brackets: 29Si (5),45Sc (10), 51V (15), 53Cr (10), 60Ni (15), 65Cu (15),71Ga (20), 85Rb (10), 88Sr (10), 89Y (10), 90Zr (10),93Nb (10), 133Cs (20), 137Ba (8), 139La (10), 140Ce(10), 141Pr (10), 146Nd (10), 147Sm (20), 153Eu (20),138Ba16O (10), 140Ce16O (10), 157Gd (20), 159 Tb (10),163Dy (20), 165Ho (15), 166Er (20), 169Tm (10), 172Yb(20), 175Lu (30), 178Hf (20), 181Ta (10), 208Pb (10),232Th (10), 238U (10) and 232Th16O (10). AdditionalN2 between 0.8 and 1.2 ml min�1 was added toincrease signal stability and allow for tuning tohigher sensitivity. The signal was tuned for maximumsensitivity for the complete range of masses whilekeeping the ThO/Th ratio below 0.5%. External stan-dardization was performed using NIST 612, analysedat the start and end of the analytical sequence andbracketing a maximum of 20 analyses by two mea-surements of reference material glass. Additionally,BCR-2 and SARM-1 were analysed as unknowns tocontrol accuracy of the analyses. Data reduction wasperformed off-line through the software ‘Iolite’ (Hell-strom et al., 2008; Paton et al., 2011), using the‘Trace_Elements_IS’ routine (Woodhead et al., 2007)with Si as the internal standard. Long-term repro-ducibility of BCR-2 and SARM-1 show that concen-trations for all elements correspond to the publishedvalues, with an accuracy of 5–10% within theGeoREM preferred values (2r). Concentrations of<1 ppm show a precision of <10% (2r).

Zircon geochronology and geochemistry

Laser ablation inductively coupled mass spectrometry

Following cathodoluminescence (CL) imaging, repre-sentative grains with variable morphologies andmicrostructures from each sample were targeted formicro-chemical analysis (Fig. 5), including cores andrims. Zircon grains were analysed by excimer laserablation inductively coupled plasma mass spectrome-try at the Research School of Earth Sciences, ANU,Australia, following procedures reported by Ballardet al. (2001). U–Th–Pb isotope ratio and trace ele-ment concentration data were simultaneouslyacquired by depth profiling the grains. This methodallowed elemental concentrations to be correlatedwith growth zones and U–Pb isotope data. The laserablation spots were carefully selected to avoid min-eral/glass inclusions and cracks. Uncertainties in the206Pb*/238U ages of individual spot analyses take intoaccount errors for each of the isotope ratios used forgeochronology in the same isotope ratios measuredin the TEMORA zircon reference material over the

course of each analytical session (typically 1–2%).Uncertainties for individual spot analyses arereported at the 1 standard error level, and uncertain-ties in weighted mean 206Pb*/238U ages are reportedat the 2 standard error (se) level. During three ana-lytical sessions, 45 spot analyses of the reference zir-con 91500 were made as unknowns. All analyseswere concordant to near-concordant (97–103% con-cordance), and yielded an error weighted mean206Pb/238U age of 1056.4 � 6.2 (2r), identical withinerrors to the 206Pb/238U age of 1062.4 � 0.4 Maobtained by TIMS (Wiedenbeck et al., 1995). Calcu-lated ages from each session lie within uncertainty ofeach other: 1055 � 14 Ma (N = 15), 1052 � 16 Ma(N = 12) and 1059 � 8 (N = 18) Ma. Thus, in thisstudy, the 206Pb/238U age of the reference zircon91500 was determined at levels of accuracy and preci-sion of <1%. Scatter in ages for 91500 may be attrib-uted to inhomogeneity of the reference materials usedfor depth-related fractionation correction and instru-mental drift.Only a subset of rare earth elements (REEs) was

acquired (La, Ce, Sm, Eu, Dy, Lu), and remaining val-ues were calculated through interpolation (Bolhar et al.,2008a). Analytical errors (expressed as relative standarddeviation calculated from the 45 analyses of 91500) are�6% (HfO2), �11% (P, Y), �8% (Ti), �17% (La),�9% (Ce), �13% (Sm, Eu), �12% (Dy) and �10%(Lu). Further details on accuracy and precision werereported previously in Bolhar et al. (2008a).Apparent zircon crystallization temperatures (Ti-

in-zircon: Table S2a) were calculated using the ther-mometer of Watson et al. (2006), assuming unitactivities of TiO2 and SiO2, without applying anycorrection for pressure (Fu et al., 2008). Uncertaintypertinent to calculating Tcrystallization (‘calibrationuncertainty’) has been estimated at <10 °C (2r) forthe temperature range observed in this study (Watsonet al., 2006). External reproducibility of Ti concentra-tion measurements by ICP-MS translates into a tem-perature range of ~10–30 °C.Time-resolved spectra produced during ICP-MS

analysis allowed the minimization of effects from theanalysis of mineral inclusions such as apatite (moni-tored using P) and monazite (monitored using cou-pled Th-La). The ICP-MS raw data were alsomonitored to distinguish growth domains (i.e. core,rim) as laser excavation proceeded.

Secondary ion mass spectrometry

Zircon U–Th–Pb analyses were performed onSHRIMP RG at RSES, ANU following proceduressimilar to Ireland & Williams (2003). Zircon wasanalysed in two modes: depth profiling of minimallypolished samples and analysis of zircon polished tomid section.Zircon was mounted in epoxy but was initially only

polished minimally prior to analysis. The mount was



imaged by reflected light microscopy. Zircon was anal-ysed through a depth-profile of the exposed surface,and essentially on the prism faces of the crystals. Theadvantage of this mode is that, at a sputtering rate ofroughly 3 nm s�1, rims of <1 lm in width could beanalysed as discrete domains. This technique did notpermit site selection based on petrological characteris-tics from CL imaging because only the surface expo-sure was available for analysis. Subsequently, the discwas polished to mid section and imaged using CL,reflected light and transmitted light microscopy. Thissecond analytical session targeted zircon tips that rep-resented significant overgrowths.

Analysis followed standard operating procedures onSHRIMP RG. Spot size of ~30 lm was used fordepth-profiling mode and ~10 lm for analysing thesectioned zircon. A single analysis of the SL13 refer-ence material was used for U concentration calibra-tion. Analyses of the c. 417 Ma Temora referencematerial (Black et al., 2003) were interspersed through-out the analytical session for U–Pb calibration. Datawere calibrated through the SQUID data reductionpackage. Analytical precision of the Temora analyseswas typically 1% for the calibrated Pb+/U+ ratio(normalized to the mean UO+/U+ of the session); fivereference materials in the depth-profiling session gavean excess scatter of 1%; nine reference materials in thesecond session showed no excess scatter.

GEOCHEMISTRY OF AMPHIBOLITES

Results

Major and trace element compositions for fiveamphibolites (including one duplicate of Na49) arereported in Table S1, and graphically presented inFig. 4. The samples contain 48.5–59.4 wt% SiO2, andtotal alkalis range from 3.4 to 6.4 wt%. In the TASdiagram (Fig. 4a; Le Bas et al., 1986) the samples areclassified as subalkali/tholeiitic andesite (Na49),basaltic andesite (Na41) and basalt (Na48, Na50).Mg-numbers of 28–36 are lower than the referencevalue for primary basaltic melts (Frey et al., 1978).In N (normal)-MORB-normalized multi-element dia-gram (Fig. 4b), all samples show variable enrichmentin the most incompatible elements, coupled with sig-nificant negative anomalies for Nb and Ta. SampleNa49 also shows a distinct relative depletion of Ti.Chondrite-normalized REE patterns (Fig. 4c) aresmooth and show variable enrichment in the lightREE. No significant Eu anomaly is observed. Sam-ples Na49 and Na49 (duplicate analysis: dubl.) showremarkably similar patterns, attesting to the excellentreproducibility between ICP-MS analyses.

Interpretation

Mg-numbers lower than the reference value for pri-mary mantle-derived melts imply that precursors to

the amphibolites had undergone differentiation. Thisis consistent with a systematic variation from low tohigh values in SiO2 v. Na2O+K2O, which suggeststhat mafic precursors evolved from a common sourcethrough fractional crystallization, possibly involving

0

2

4

6

8

10

12

14

16

35 40 45 50 55 60 65 70 75

TAS diagram

Na 2O

+K2O

(wt%

)SiO

2 (wt%)

And

esite

Bas

altic

an

desi

te

Bas

alt

Na49

Na48Na50 Na41

Na49 dubl.

0.1

1

10

100

Th Nb Ta La Ce Pr

Nd Zr Hf

Sm Ti Gd

Tb Dy Y Ho Er

Tm Yb

N-M

OR

N-n

orm

aliz

ed

1

10

100

1000

La Ce Pr

Nd

Pm

Sm Eu

Gd

Tb Dy

Ho Er

Tm Yb Lu

Na41Na48Na49

Na49 dubl.Na50

Cho

ndrit

e-no

rmal

ized

(a)

(b)

(c)

Fig. 4. (a) TAS diagram; (b) MORB-normalized multi-elementdiagram; (c) chondrite-normalized rare earth element (REE)diagram. Normalizing values for MORB and chondrite fromHofmann (1988) and Boynton (1984).



crustal assimilation. The sample with the highest levelof REE contains the highest SiO2 and alkali ele-ments, also consistent with igneous differentiationand incorporation of crustal material. The presenceof Ti, and the absence of Eu, anomalies indicate frac-tionation by Ti oxides (ilmenite, magnetite), but noinfluence of feldspar fractionation on REE.

CL IMAGING AND U/Pb GEOCHRONOLOGY

During analysis, emphasis was placed on selecting awide variety of zircon grains with differing sizes,morphologies and internal make-up in order to mini-mize sampling bias. U–Th–Pb isotope and U, Th, Pbelemental data are reported in Table S2a,b. Core andrim domains were identified on the basis of CL imag-ing (Fig. 5). Reverse Concordia diagrams show U–Pbisotope data for individual zircon spot analyses(Fig. 6). Different symbols are used to distinguishcore, rim and unidentified domains and discordantanalyses.

Sample Na41

Zircon varies in length from ~90 to ~150 lm, andshapes are sub- to anhedral, short prismatic (Fig. 5a).Some grains display distinct core–rim microstruc-tures. Cores are variable in CL response, rangingfrom dark to bright and show faint oscillatory zon-ing. Rims tend to be more clearly oscillatory zoned,which mimics the geometry of the core, and their CLresponse is grey in various shades. One core displaysconvolute zoning. Rims are frequently truncated byirregular zones that show blurred oscillatory zoningand/or distinctly higher CL response, resemblingreaction fronts and/or resorption features. Randomlyoriented fracturing and mineral inclusions are visibleas small black specks. In total, at least three distinctzircon growth stages can be resolved (Fig. 5a), theoldest possibly representing inheritance (i.e. derivedfrom a granitoid/sedimentary source) and the young-est probably representing subsolidus modification asthe latter typically discordantly overgrows oscilla-tory-zoned rims.

Fifteen spot analyses (≥95% concordance) yield206Pb*/238U dates of c. 250–223 Ma (Fig. 6a). Noinheritance is detected based on U–Pb geochronology(defined as a date considerably older than the mainpopulation). Dates for cores, rims and undifferenti-ated domains overlap. Near concordant spot analysescombine to a weighted mean age of 234.7 � 4.3(MSWD = 2.8, N = 15). One rim analysis produced adate of 38.9 � 0.8 Ma. The significance of this youngdate will be discussed below.

Sample Na48

This sample is virtually devoid of any extractable zir-con, despite a whole-rock Zr content that is higher

than for samples Na41 and Na50 (145 v. 44 and91 ppm). Only one grain was available for U–Pb andtrace element analysis (Fig. 5d). The subhedral grainrepresents a fragment of an originally larger zirconand the current length is ~130 lm. The grain displaysno clear core–rim relationship. Oscillatory zoning, iforiginally present, is diffuse and disturbed. Crosscut-ting relationships are likewise difficult to discern.Cracks, minerals inclusions and an outer metamor-phic rim are not observed. Three analyses obtainedfrom three different locations yield dates of c. 245,254 and 262 Ma, with uncertainties of c. 2–3 Ma(Fig. 6b). All dates have discordance of >10%, andcombine to a weighted mean age of 254.0 � 9.6 Ma(N = 3, MSWD = 2.3). Due to the discordant natureof all three individual dates, the mean age is not con-sidered very reliable.

Sample Na49

Zircon grains vary in length from ~150 to 210 lm,are euhedral, subrounded and short to long prismatic(Fig. 5c). All grains exhibit a distinct core–rimmicrostructure. In some cases, an inherited core isidentified on the basis of its bright and well-roundedappearance. Interior domains mantling cores showdull CL response, diffuse fine to coarse oscillatoryzoning, which is sometimes sigmoidal in shape. Inte-rior domains tend to be surrounded by darker innerrims along an irregular contact, which points to cor-rosion. The outer rim is very bright, typically thinand most prominently developed in prism termina-tions. This outer rim, which was targeted for SIMSanalysis, discordantly overgrows both grey oscilla-tory-zoned cores and darker, inner rims and is inter-preted as metamorphic recrystallization orcrystallization of newly formed metamorphic zircon.Judging from CL response, zoning and crosscuttingrelationships a minimum of four zircon growthstages are identified: inherited core, magmatic core,inner rim (metamorphic?) and an outer metamorphicrim.A total of 25 LA-Q-ICPMS spot analyses were

obtained, 16 of which have >10% discordance. Fivecore analyses yield dates of c. 248–238 Ma (<5% dis-cordance) and combine to a weighted mean age of244.2 � 3.7 Ma (n = 5, MSWD = 0.8; Fig. 6b). A sec-ond zircon population comprises 11 dates rangingfrom c. 47–38 Ma (Fig. 6b), combining to a weightedmean age of 39.6 � 1.1 Ma (N = 11, MSWD = 12.3).The significance of this age requires assessment withinthe context of available geochronological data, as it isbased on individual dates which have mostly >10%discordance. One date of 14.7 � 0.4 Ma was obtainedby LA-Q-ICPMS. Given microstructural evidence fora late-stage overprinting/recrystallization event, andthe difficulty in analysing thin outer zircon rims usinga laser, selected grains from this sample were also anal-ysed by SHRIMP.



(a) (b)

(c) (d)

(e)

Fig. 5. Cathodoluminescence (CL) images of representative zircon grains extracted from Naxos amphibolites Na41 (a), Na48 (d),Na49 (c: analysis by ICPMS, e: analysis by SIMS) and Na50 (b); white bars in (a, b, c, d) represent 30 lm, white bar in (e)represents 200 lm; circles represent U–Pb dates (206Pb/238U by LA-ICPMS, 1 se), numbers in (e) refer to grains used for analysis.



Two analytical approaches were adopted for SIMSanalysis. Twenty-six analyses were performed in‘depth-profiling’ mode and 28 in ‘mid-section’ mode.The depth-profiling analyses show a range in datesfrom 247 � 12 to 14.9 � 1.4 Ma. The youngest dateat c. 15 Ma (grain 3; Fig. 5e; Table S2b) is associatedwith zircon that has low U (96 ppm) and extremelylow Th (1 ppm) and Th/U of 0.01. This domainappears to be equivalent to the youngest dateobtained by ELA-ICPMS. The chemistry is also con-sistent with it being associated with the CL-brightoutermost rims. Low Th/U (<0.1) is also associatedwith several other grains (e.g. grain 4; Fig. 5e;Table S2b), which have ages ranging from 56 to32 Ma. These grains show high U concentrations of1000–2200 ppm. This chemistry would suggest thatthis zircon appears dark in CL imaging. Grains 7and 16 (not shown) also show high U, ~1600 ppm,

and low Th/U (~0.12) but have older dates of c.70 Ma. Three other grains have dates in the 120–80 Ma range with Th/U of 0.11–0.23. The remainingdates show a broad range from c. 250 to 140 Ma.The chemistry is quite variable in terms of U and Thconcentrations, but the Th/U is limited to a range of0.51–0.78.The Th/U ratio shows a systematic variation from

0.5 to 0.6 at 250 Ma down to near 0.01 at 15 Ma.Such a trend could be construed as a progressivechange in chemistry related to the fluids associatedwith the formation of the late-stage rims. Althoughall of these analyses may reflect mixed componentswithin a single analysis, this is not evident in any co-variation between U concentration and age. Onlysome rims at c. 56–32 Ma have pronounced U con-tents. The depth-profiling technique does not allowselection of discrete domains as with the mid-section

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

10 20 30 40 50 60 70 80

(a) Na41

ConcordiaCoreRimUndifferentiatedDiscordant

207 P

b/20

6 Pb

238U/206Pb

234.7±4.3 Ma(n = 15, MSWD = 2.8)

120 Ma

260 Ma

0

0.05

0.1

0.15

0.2

10 15 20 25 30 35 40

(b) Na48ConcordiaDiscordant

207 P

b/20

6 Pb

238U/206Pb

254.0±9.6 Ma(n = 3, MSWD = 2.3)

260 Ma 240 Ma

0

0.05

0.1

0.15

0.2

0 5 10 15 20 25 30

(d) Na50

ConcordiaCoreRimDiscordant

207 P

b/20

6 Pb

207 P

b/20

6 Pb

238U/206Pb

260 Ma

780 Ma

0

0.1

0.2

0.3

0.4

0.5

0.6

0 100 200 300 400 500 600

ConcordiaCores: LA-ICPMSRims: LA-ICPMSDepth-profiling: SIMSMid-section: SIMS

238U/206Pb

outer rims:14.5±0.5 Ma (n = 8, MSWD = 1.78)

14 Ma17 Ma

(c) Na49

0

0.05

0.1

0.15

0.2

20 25 30 35 40 45 50

SIMS & ICPMS combined: 244.8±2.2 Ma (n = 12, MSWD = 21)

170 Ma200 Ma

Na49

260 Ma

0

0.05

0.1

0.15

0.2

100 120 140 160 180 200

inner rims:39.6±1.1 Ma (n = 11, MSWD = 12.3)

160 Ma

Na49

50 Ma

Fig. 6. Reverse Concordia diagrams summarizing U–Pb geochronological data for zircon from four Naxos amphibolites; (a) Na41,(b) Na48, (c) Na49, (d) Na50; U–Pb data corrected for common lead using 208Pb technique; 1r error bars for individual datapoints; errors on weighted averages quoted at 2r level.



approach. It is possible that some zones are discon-tinuous at the surface.

In order to explore the potential of mixed analysesand the lack of consistency in the determination ofthe youngest component(s) grains were targeted formid-section analysis based on CL imagery. Of partic-ular interest are the CL-bright overgrowths on tips,which were quite evident and could be targeteddirectly. Twenty-eight analyses were carried out on15 of the grains. Sixteen analyses are of rims, and 12of cores. The cores range in age from c. 259 to133 Ma. Three of the rims analysed are also in thisrange (e.g. grain 11; Fig. 5c), while the remainderrange in age from c. 53 to 12 Ma. Nine analyses ofCL-bright overgrowths with thicknesses of 10 lmrange in age from 16.1 to c. 11.8 Ma. Excluding theyoungest analysis (4.3r below the mean), these con-stitute a coherent age group, with a weighted meanage of 14.5 � 0.5 Ma (2r, n = 8, MSWD = 1.8;Fig. 6c). For these nine analyses the U concentra-tions range from 70 to 1390 ppm, Th concentrationsare all <4 ppm, and Th/U <0.01.

Three rims with ages of c. 37–30 Ma (e.g. grain 3,analysis 2; Fig. 5e) also have similar chemistry withU from 180 to 1200 ppm, Th <3 ppm, and Th/U<0.01. One further rim analysis (grain 4, analysis 1;Fig. 5e) is c. 53 Ma with 31 ppm U, 2 ppm Th andTh/U of 0.06. It is possible that these analyses repre-sent mixtures of older core and younger rims, andthese analyses all have quite thin rims in close prox-imity to core material. The chemistry of the spotscan be used to place limitations of core material con-tribution because the Th concentrations are generallyquite high (typically 200–400 ppm). As such, evenonly a small contribution of core material wouldchange the Th concentration of the low Th rims.Considering only c. 15 Ma rims and c. 250 Ma cores,~5% core contribution would be required to pull theanalysis from c. 15 to c. 30 Ma. This would changethe Th concentration to 10 ppm even for the lowerTh concentrations. This is even less attractive aproposition for analysis 4.1 (c. 53 Ma, Th = 2 ppm)where the core contribution would need to be higher(~20%).

All analytical methods (SIMS: depth-profiling andmid-section analysis and LA-ICPMS) reveal the sameage range from c. 250 to 15 Ma, and significant clus-ters at c. 15, c. 40 and c. 250 Ma. Differences are evi-dent in the proportion and the distribution of ages.Targeting the tips after polishing to mid sectionreveals a major overgrowth contribution at c. 15 Ma.However, this age is only evident in a single depth-profiling analysis. These overgrowths have evidentlyaffected the tips (pyramidal faces) in preference tothe prism faces. Consequently, it must be concludedthat analyses in depth-profiling mode may not neces-sarily reveal metamorphic overgrowths on zirconprisms, and the lack of an overgrowth cannot beused as evidence that zircon lacks overgrowths of this

age. While compositional information suggests thatthe c. 50–30 Ma rims are not mixtures, they do notform a coherent group and translating this age rangeto a discrete event from the U–Pb systematics aloneremains problematic.

Sample Na50

Zircon extracted from this rock is short prismaticand subhedral, with an average length of 120–150 lm. Some grains contain inherited cores, butmost show faint, grey oscillatory-zoned interiors.Interior domains are typically truncated by grey andbright discordant and lobate inner rims, with in turnare overgrown by fine darker rims (Fig. 5b). Mineralinclusions occur as black specks in CL imaging.Incursions and embayments of inner and outer rimsinward indicate corrosion. Microstructural evidenceis consistent with a minimum of four distinct zircongrowth stages: inherited core, weakly zoned mag-matic core and two stages of zircon overgrowth fol-lowing corrosion.A total of 34 spot analyses were conducted by LA-

ICPMS on zircon from this sample. The dates rangefrom c. 2046 to 52 Ma, with broad clusters at c. 700and 550 Ma, although the majority of analyses plotalong the Concordia over a wide age range (Fig. 6d).Most analyses are concordant within 10%. Neverthe-less, this sample appears to contain no zircon thatprovides meaningful information on the timing ofcrystallization of the host rock. Instead, zircon agedata are interpreted to reflect inheritance solely, andhence the geochronological make-up of the sourcematerial.

DISCUSSION

Petrogenesis and tectonomagmatic setting

In normalized multi-element diagrams (Fig. 4b), pro-nounced enrichment of highly incompatible elements(Th), coupled with anomalously low high fieldstrength (HFSE) elements (Nb, Ta), is considereddiagnostic of subduction zone or continental intra-plate magmatism. In the latter case, enrichment ofincompatible elements either reflects an enrichedmantle (E-MORB or OIB) source or incorporation ofevolved crust during shallow-level emplacement.Crustal assimilation complicates the distinction intoenriched and depleted mantle sources as small quanti-ties of highly enriched crustal material can effectivelymask primary melt signatures. Nevertheless, negativeHFSE anomalies require some type of magma-crustinteraction, either in the source region (‘source con-tamination’), by addition of a subduction component(fluids, melts), or through assimilation of countryrock during ascent (‘crustal contamination’). While adetailed investigation of the magma genesis of amphi-bolite protoliths, and quantification of mantle and


Z IRCON IN AMPHIBOL ITES 11

crustal contributions, is beyond the scope of thisstudy, it is helpful to define the broad petrogeneticframework and likely tectonic setting.

Effective geochemical fingerprinting is possibleusing Th/Nb/Yb and Ti/Nb/Yb systematics (Fig. 7),allowing distinction between OIB and MORB derivedmafic rocks (Pearce, 2008). Displacement from themain OIB–MORB array towards elevated Th/Nbsuggests magma–crust interaction (subduction or con-tinental intraplate). More complex situations, involv-ing interaction with suboceanic lithosphere or exoticmantle heterogeneities, are not accounted for. Whenplotted in the Th/Yb v. Nb/Yb projection (Fig. 7a),samples Na50 and Na48 plot within the mainMORB–OIB array, but are shifted away from the N-MORB, and towards the E-MORB section. Sinceboth Th and Nb behave as incompatible elementsduring mantle melting, this shift along the mainMORB–OIB array away from N-MORB indicatesinvolvement of a mantle plume, or generally lowerdegrees of partial melting when compared to N-MORB basalts. Elevated Th/Yb ratios along verticaltrajectories from the main MORB-OIB array pointto involvement of crustal material, either by subduc-tion or crustal contamination. A conclusive separa-tion of both processes is difficult by using Th/Nb/Ybrelationships alone, and will be discussed in moredetail below.

Overall, it appears that protolithic melts of at leasttwo Naxos amphibolites (Na48, Na50) formed in anoceanic intraplate environment involving smallerdegrees of melting than prevalent in a mid-oceanicridge setting. Elevated Th concentrations (Na41,Na49) point to input from crustal sources, probablyvia crustal assimilation (see below for supporting evi-dence).

Subduction v. crustal contamination signatures

Several geochemical criteria aid in distinguishingbetween incorporation of crustal material in a sub-duction zone environment and in an intraplate settingwhere hot, mafic melts percolate upwards throughthe crust. Principally, crustal incorporation into maficmelts can be modelled using coupled recharge–assimi-lation–fractional crystallization (Spera & Bohrson,2001), assimilation during turbulent magma ascent(Huppert et al., 1985), coupled melting–assimilation–storage–homogenization (Hildreth & Moorbath,1988), and coupled assimilation and fractional crys-tallization (DePaolo, 1981). Any of the above modelsrequires adequate knowledge of input parameters,such as the rate of assimilation (DePaolo, 1981),which is hardly ever achievable with existing datasets, and typically requires radiogenic and stable iso-tope data. However, the latter model predicts thatcrustal input increases as fractional crystallizationproceeds, detectable by correlated proxies for crustalinput and fractional crystallization. When plotted in

Th/Yb and La/Smchond v. SiO2 diagrams (Fig. 7b)amphibolites display a positive correlation, providingsupport for a coupled process of crustal incorpora-tion and fractional crystallization. In the case ofandesite sample Na49, it appears plausible that suffi-cient quantities of crustal material were assimilatedto induce saturation of Fe–Ti oxides. This is sup-ported by the negative Ti anomaly in the normalizedmulti-element diagram (Fig. 4b) and a Fe2O3 valueof ~6 wt%, lower than in the other mafic samples(~10–14 wt%).In the Th/Yb v. Nb/Yb diagram, subduction-

related mafic rocks plot above the main MORB-OIB

0.01

0.1

1

10

0.1 1 10 100

(a)

Th/Y

b

Nb/Yb

E-MORB

N-MORB

OIB

Crustal contaminationor fluid transfer of Th

Na48

Na50

Na41

Na49 dublNa49

0

0.5

1

1.5

2

2.5

3

0

1

2

3

4

5

6

45 50 55 60 65

(b)

Th/Yb

La/Smchond

SiO2

0.1

1

10

0.1 1 10 100

(c)

TiO

2/Y

b (w

t%/p

pm)

Nb/Yb

E-MORBN-MORB

OIB

Na48

Na50

Na41 Na49 dubl

Na49

Fig. 7. Geochemical binary plots: (a) Th/Yb v. Nb/Yb; (b)TiO2/Yb v. Nb/Yb; (c) Th/Yb v. La/Smchond; trace elementratios act as monitor for crustal input; SiO2 contentconventional proxy to whole-rock igneous differentiation.



array (Fig. 7a), and therefore cannot be distin-guished readily from crustal contaminated basalts.The fact that two samples (Na48, Na50) plot insidethe MORB-OIB field in this Pearce-type projectiondoes not contradict the view that Naxos amphibo-lites were emplaced into continental crust since ‘con-tamination-free pathways’ are feasible (Pearce,2008). It rather confirms that all samples formed bymantle melting at degrees higher than normallyencountered in MORB settings, but lower than inOIB settings, and that some mafics were morestrongly influenced by interaction with evolved crustthan others. A subduction origin for sample Na49can be excluded because this samples does not plotinside the MORB-OIB array in TiO2/Yb v. Nb/Ybspace (Fig. 7c). Subduction-influenced mafics format temperatures, and encounter degrees of melting,that are very similar to those prevalent in MORBsettings (Pearce, 2008). For this reason, subduction-related mafics would occupy identical compositionalfields as MORBs (Pearce, 2008). Since felsic, evolvedcrustal material is typically depleted in TiO2 (due tooxide fractionation), it plots below the MORB arrayof Fig. 7c and, hence, any oceanic mafic melts thathad assimilated crustal material of this compositionwould be transposed towards lower TiO2/Yb values.This observation applies to sample Na49, and ittherefore appears that this amphibolite sample mayhave incorporated noticeable quantities of felsic orgranitic crust.

Zircon inheritance or saturation?

Having established a significant role of crustal con-tamination in generating the geochemical signaturespreserved in the amphibolites from Naxos, the natureand origin of zircon extracted from these rocksrequire assessment. Specifically, did zircon crystallizefrom mafic precursors during cooling or immediatelyafter emplacement, once sufficient quantities of Zrwere acquired to reach saturation, or were theyentrained as xenocrysts in the course of country rockassimilation?

Zircon saturation

The stability of zircon in crustal melts can be assessedusing zircon solubility behaviour as a function of sat-uration temperature, Zr content and host melt com-position (approximated by M = Na+K+2Ca/AlxSi;Watson & Harrison, 1983). Recently, the originalconcept of zircon solubility was refined to make itapplicable to mafic host melts (Boehnke et al., 2013).Only for sample Na49 is the calculated M valuewithin the original calibration range of 1.65–1.70(Watson & Harrison, 1983), the other three sampleshave M values of 2.63–2.75. Therefore, sample Na49is likely to provide the most reliable constraints (satu-ration temperature of ~770 °C). Inferred saturation

temperatures in this study range from ~610 to770 °C, well below solidus temperatures(860 � 30 °C) for MORB gabbros (Coogan et al.,2001) and dry/water-saturated liquidus temperatures(1170–1060 °C) of ferrobasalts (Botcharnikov et al.,2008). When applied to amphibolites Na48 and Na50with basaltic compositions (SiO2 <52 wt%), the abovefinding implies that saturation temperatures are wellbelow empirical temperatures at which completesolidification of the host rock occurs. In other words,Naxos mafic rocks would have reached a state of fullcrystallization well before zircon precipitated. Conse-quently, Naxos metabasites are unlikely hosts of pri-mary magmatic zircon. Assuming that some Zr wasacquired from extraneous sources, original Zr concen-trations, and hence pre-contamination saturationtemperatures, would be even lower, further diminish-ing the likelihood of in situ crystallization of zircon inmafic melts with a whole-rock composition similar tothose of Naxos amphibolites. Boehnke et al. (2013)examined solubility of zircon in mafic melts at tem-peratures of 1225–1175°, and concluded that basalticmelts require Zr concentrations >5000 ppm, whichare only possible to attain in late stage, evolved melts.The required Zr concentration is considerably higherthan observed in Naxos amphibolites (Zr: 44–169 ppm; Table S1). We note, that the relatively widerange in Ti-in-zircon temperatures (650–850 °C) alsoprovides evidence against in situ crystallization of zir-con from basaltic to andesitic rocks (Fig. 4a;Table S1).

Microtextural evidence

Cathodoluminescence imaging documents a markeddiversity in the microtextural make-up of amphibo-lite-hosted zircon. Core–rim relationships are com-mon, and interior domains are typically truncated byinner and outer rims along irregular contacts (Fig. 5).Several features combine to suggest a xenocrystic ori-gin for the majority of zircon:

1 With the exception of sample Na50, Naxos amphi-bolites yield zircon that show grey to dark CL col-ours across multiple broad-scale domains. Broadzonation is consistent with formation of the zirconin a plutonic environment (e.g. Corfu et al., 2003).

2 The CL response in zircon is associated with therelatively high concentration of the trace elementsDy and Tb, while suppressed CL response is typi-cally associated with U causing metamictization orvacancies in the structure (e.g. Nasdala et al.,2003). Zircon precipitation from primitive meltswould be expected to have low U concentrationsand hence bright CL response. Indeed, sampleNa49 contains zircon with thin bright rims, sug-gesting formation under subsolidus conditions inequilibrium with a trace element-poor source, suchas a mafic host rock. In addition, the rims have



extremely low Th/U ratios, consistent with a meta-morphic origin (Ireland & Williams, 2003).

3 Classification of zircon morphologies according toemplacement sites, speed of crystallization, magmacomposition and temperature (the latter two fac-tors combining to zircon saturation) is of generalnature. Nevertheless, sub-volcanic intrusions andhigh-level granites and gabbros tend to form nee-dle-shaped, acicular zircon crystals (Corfu et al.,2003). In contrast, deep-seated, slowly coolingintrusions tend to form zircon with short prismaticto equant shapes (Corfu et al., 2003), althoughexceptions exist (e.g. dioritic zircon: Tapster et al.,2016) Most, if not all, zircon examined in thisstudy belong to the second type of morphology,and therefore are more likely to have crystallizedin deeper environments. While it could be arguedthat zircon formed late in pockets of highlyevolved melts, such minerals often (but not always)display only partly developed crystal faces (Scoates& Chamberlain, 1995; Corfu et al., 2003). Giventhe shallow-level, intrusive emplacement of Naxosamphibolites, zircon would have crystallizedrapidly and developed skeletal or incomplete crys-tal shapes (Bossart et al., 1986; Corfu et al., 2003),contrary to what is observed in this study.

4 Fundamentally, the diversity and complexity ofmicrostructures and the presence of multiplegrowth generations points to complex crystalliza-tion histories, contrary to relatively simple crystal-lization expected from a mafic, relatively primitivemelt, even if zircon saturation did indeed occur(but see above). In particular, multiple events ofcorrosion (by a fluid) or resorption (by a melt) asmanifested by transgressive zircon-regrowth alongirregular contacts are most consistent with re-equi-libration of, and new crystallization onto, pre-exist-ing grains (subsequent to entrainment fromextraneous sources) in Zr-undersaturated magmatichosts.

Zircon Zr/Hf and Th/U ratios as magmatic indicators

Zircon is the principal repository of both Zr and Hfin a wide range of igneous and metamorphic rocks(e.g. Bea et al., 2006), and is expected to stronglyinfluence whole-rock Zr–Hf concentrations duringigneous differentiation. Fractionation of Zr and Hfby clinopyroxene crystallization (e.g. David et al.,2000) can be largely ruled out for the Naxos amphi-bolites because of the lack of correlation between(chondritic: ~37) Zr/Hf ratios and Sc concentration(sample Na41 is an exception with Zr/Hf = 32). Inline with Linnen & Keppler (2002) who argued thatmineral fractionation in depolymerized (mafic) meltswould exert little influence on whole-rock Zr/Hf, thelimited range in whole-rock ratios seen in this studyis also inconsistent with significant fractionation by

zircon. In contrast, granitoids with low zircon solu-bility typically document a wide range in Zr/Hfratios. Owing to high but unequal partition coeffi-cients of Zr and Hf for zircon in granitic rocks(KDZr ~ 2520; KDHf ~ 2420; Wang et al., 2010), zir-con crystallization imprints low Zr/Hf on coexistingmelt (e.g. Claiborne et al., 2006). This progressivereduction of the remaining melt in Zr/Hf is accompa-nied by precipitation of zircon with correspondinglylower Zr/Hf ratios (e.g. Bolhar et al., 2008a). Inevolved rocks, Zr/Hf ratios range from 60 to as lowas 15–20 for late-stage granitic and pegmatitic zircon(Wang et al., 2010). Consequently, Zr/Hf ratios inzircon serve as a fingerprint for crystallization histo-ries and, thus, can assist in determining whether zir-con formed in primitive, basaltic or evolved, graniticmelts. With reference to Naxos amphibolites, Zr/Hfratios in zircon formed from basaltic to intermediatemelts would be consistently high and restricted inrange (due to small degrees of differentiation),whereas granitic zircon would show a considerablerange in values from high to low, with a preponder-ance of lower values.Zr/Hf in all zircon grains, including core and rim

domains, are depicted in histograms of Fig. 8. Insamples Na41 and Na50, zircon data define unimodaldistributions with abundance peaks at 55–60 and 45–50 respectively. Sample Na49 reveals a bimodal dis-tribution with abundance peaks at 35–40 and 55–60,while Na48 provides only sparse data. Both the rangeand the presence of low zircon Zr/Hf values contrastwith uniformly chondritic whole-rock values, whichcannot be reconciled with significant mineral fraction-ation involving zircon. Observed zircon Zr/Hf alsocontradicts the prediction of high values of zirconthat would crystallize from a relatively primitive meltthat had not undergone extensive fractionationcaused by crystallization and removal of zircon.Complementary to Zr/Hf, values of Th/U in zirconcan also serve as monitor for magmatic processes(e.g. Bolhar et al., 2010, 2012). Like Zr and Hf, Thand U strongly partition into the zircon crystal lat-tice, causing a reduction in both elements in zirconand coexisting melt as magmatic differentiation pro-ceeds. Empirically determined partitioning coefficientsare KDTh[zircon-rock] of ~10 and KDU[zircon-rock] of~100, translating into KDTh/U[zircon-rock] of ~0.1(Zartmann & Richardson, 2005). This KDTh/U valueis similar to the value estimated for gabbroic zircon(~0.14; Kirkland et al., 2015). Thus, partition coeffi-cients of 0.1–0.15 would apply to zircon that crystal-lized in amphibolitic rocks. A comparison ofmeasured Th/U in zircon (Fig. 9) and host rockdemonstrates that zircon could not have crystallizedin mafic melts similar in composition to Naxosamphibolites. This is because Th/U ratios of 2–6 inwhole rocks (Table S1) require ratios in zirconformed under equilibrium conditions of 0.2–0.6,



which is exceeded by the range observed in Naxosamphibolite zircon. Indeed, zircon precipitated ingabbroic melts under equilibrium conditions is char-acterized by Th/U ratios of 0.9 � 0.1 (n = 40) (Kirk-land et al., 2015). Ignoring sample Na50 (withxenocrystic zircon based on the spread of ages) andfocusing on samples Na41 and Na49, which showone and two distinct age populations, the variabilityin Th/U (from ~1 to 0) is far greater than predictedfrom empirical knowledge (e.g. partition coefficient).In summary, neither Zr/Hf nor Th/U systematics inamphibolitic zircon from Naxos favour precipitationfrom a mafic melt, supporting microstructural evi-dence and zircon saturation data.

Three zircon ages in one amphibolite?

Below, timing and conditions of zircon growth areevaluated by integrating geochronological and geo-chemical information within the context of available

age data. Two analytical methods were applied,namely LA-ICPMS and SIMS depth-profiling andmid-section analysis. ICPMS allowed the simultane-ous measurement of trace elements, including REEand Ti, while SIMS permitted measurement of ultra-thin zircon rims, which are not sufficiently wide forLA-ICPMS or conventional SIMS techniques. Sam-ple Na49 is particularly suitable to decipher the tec-tono-magmatic history of the Naxos amphibolites,since three resolvable zircon age populations can beassigned to distinct zircon growth stages (Figs 5c,e &6c). Combined ICPMS and SIMS data provide aweighted average of 244.8 � 2.2 (n = 12,MSWD = 21: Fig. 10a). High Th/U ratios (>0.5;Fig. 10b) and remnant oscillatory zoning imply thatthis age records magmatic crystallization, presumablyin a granitoid environment sensu lato.An intermediate age of 40.7 � 2.0 Ma is obtained

from LA-ICPMS analysis of inner rims. ICPMS andSIMS dates combine to an age of 39.6 � 1.1 (n = 11,

01234567

20 30 40 50 60 70 80

Cou

nt

Zr/Hf

Na-49

02468101214

20 30 40 50 60 70 80

Cou

nt

Zr/Hf

Na-50

0

1

2

3

4

5

Cou

nt

Na-48

0

2

4

6

8

10

12Coresrims

Undifferentiated

Cou

nt

RimsNa-41

Fig. 8. Histograms showing the variabilityin zircon Zr/Hf for Naxos amphibolitesamples Na41, Na48, Na49 and Na50;analytical data for core, rim andunidentified domains shown in differentcolours.

0

0.5

1

1.5

2Na41

0 50 100 150 200 250 300

Th/U

0

0.5

1

1.5

2Na48

0 50 100 150 200 250 300

Th/U

0

0.5

1

1.5

2NA49

0 50 100 150 200 250 300

Th/U

206Pb/238U age 206Pb/238U age

0.5

1

1.5

2

0 500 1000 1500 2000 2500

Na50

Th/U

Fig. 9. Th/U variability in zircon asfunction of U–Pb age; only core (filled) andrim (white circles) data shown; grey barsindicate main age populations.



MSWD = 12.3). Inner rims post-date corrosion/re-sorption of interior domains, which is evident fromCL imagery, while low Th/U ratios (<0.06) and alack of clear oscillatory zoning argues against precipi-tation in a magmatic environment.

The youngest age of 14.5 � 0.5 Ma (n = 8,MSWD = 1.78) is confined to analyses of thin outerrims, using both SIMS mid-section and depth-profil-ing methods. One LA-ICPMS analysis also yieldedthis young age and allowed simultaneous measure-ment of REE (Fig. 10c). Outer rim analyses yieldextremely low Th/U ratios (≤0.01). Th/U

concentration ratios have been frequently applied todiscriminate between igneous and metamorphic zir-con (Williams & Claesson, 1987; Ireland & Gibson,1998; Rubatto, 2002; Mojzsis & Harrison, 2002).Commonly, low to very low Th/U ratios (<0.1–0.01)have been cited as a criterion to identify metamor-phic and hydrothermal zircon, in contrast to igneouszircon (Th/U > 0.5). One explanation for the decou-pling of Th and U is that Th4+ is more preferablyrejected from the zircon lattice than U4+ due to itslarger ionic radius during recrystallization undermetamorphic conditions (Hoskin & Schaltegger,2003). Alternatively, Th availability for zircon maydepend on breakdown and formation of other Th–U-bearing minerals in the course of metamorphicrecrystallization of the host rock. In any case, Th/Uin newly grown zircon is also likely controlled byfluid-chemistry and prevalent temperature.In addition to differences in microtextures (Fig. 5)

and Th/U ratios (Fig. 10b), a distinction between zir-con populations from Naxos is also possible usingchondrite-normalized REE patterns (Fig. 10c), imply-ing different modes of zircon formation. Grainsattributed to the c. 245 Ma event display consistentlyhigher REE abundances across all REE when com-pared to younger zircon. Normalized HREE abun-dances define flatter patterns, while anomalies for Euand Ce are more pronounced in c. 245 Ma zircon.REE patterns belonging to this group broadly matchthe REE compositional range displayed by magmaticzircon derived from Mesozoic I-type granodioritesfrom New Zealand (Bolhar et al., 2008a,b), consis-tent with a plutonic origin. In contrast, c. 40 andc. 15 Ma zircon grains are characterized by steeperpatterns, especially across the MREE to HREE. Insome cases, no values for La and Eu are displayeddue to concentrations below detection limits by LA-ICPMS. Generally lower REE abundances ofyounger zircon are consistent with formation undermetamorphic conditions. Hoskin & Black (2000)observed that chondrite-normalized REE patterns inmetamorphic zircon, when compared to the originalpre-metamorphic patterns, become flatter in theLREE, accompanied by reduction of Ce, Eu anoma-lies and development of concave HREE. The similar-ity in REE patterns between zircon with a well-established metamorphic origin (Hoskin & Black,2000) and c. 40 and c. 15 Ma zircon rims from thisstudy suggests that Naxos zircon rims are metamor-phic in origin. Reduced REE, especially light andmiddle REE, may reflect exchange between originaltrace element-rich (granitoid) zircon and a mafic,trace element-poor micro-environment, such as theamphibolitic host rock, after entrainment from anextraneous source. In any case, reduced REE, con-cave normalized HREE and low Th/U (<0.01) in c.40 and c. 15 Ma zircon collectively point to meta-morphic zircon formation, consistent with microtex-tural evidence.

0

50

100

150

200

250

300(a)

206 P

b/23

8 U a

ge (M

a)

14.5±0.5 Ma (n = 8, MSWD = 1.78)

39.6±1.1 Ma (n = 7, MSWD = 12.3)

244.8±2.2 Ma (n = 12, MSWD = 21)

Inheritence or mixed ages?

0

0.2

0.4

0.6

0.8

1

0 50 100 150 200 250 300

(b)Cores: LA-ICPMSRims: LA-ICPMSDepth-profiling: SIMSMid-section: SIMS

Th/U

206Pb/238U age (Ma)

Mixed ages: incomplete metamorphic recrystallization

0.1

1

10

100

1000

104

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

(c)

NA49_13NA49_05.NA49_06NA49_11NA49_07NA49_20NA49_15NA49_25NA49_26NA49_19NA49_04NA49_03

Cho

ndrit

e-no

rmal

ized

245

Ma

40 M

a

15 Ma

Fig. 10. (a) Weighted averages of 208Pb-corrected 206Pb/238Udates from sample Na49; 1r errors for individual ages. (b)Zircon Th/U ratios v. 206Pb/238U dates for sample Na49; SIMSand LA-ICPMS presented by different symbols; rim and coreanalyses also distinguished; semi-continuous range in Th/Uratios with ages from 240 to 40 Ma may reflect incompleterecrystallization of primary magmatic zircon. (c) Chondrite-normalized REE diagram showing core and rim (outer, inner)analyses of zircon from Na49.



Principally, metamorphic zircon can form by pre-cipitation from fluids using ions liberated from degra-dation of unstable Zr-bearing minerals and sub-solidus recrystallization of protolith zircon (Hoskin &Schaltegger, 2003). Fluid-aided recrystallization hasbeen previously described for metamorphic zirconfrom Syros (Tomaschek et al., 2003) and Serifos(Schneider et al., 2011). Both studies document vari-able degrees of recrystallization of xenocrystic zircon,resulting in neoblastic zircon with a spongy or porousappearance. This newly formed metamorphic zirconis characterized by very low Th/U, low trace elementconcentrations, including REE and correspondinglybright CL response. Low trace element abundanceswere explained by preferential expulsion duringrecrystallization, a process probably accompanied byhigh mobility of HFSE under high pressures(Tomaschek et al., 2003). However, in the presentstudy, no zircon was observed with spongy or con-spicuously porous outer rims. Therefore, microstruc-tural evidence for fluid-aided dissolution–re-precipitation is lacking, and we propose that meta-morphic recrystallization occurred in solid-state modein the case of both rim generations, involving grain-boundary migration and defect diffusion (Hoskin &Black, 2000). Observed internal features, diagnosticfor solid-state recrystallization, include blurred andconvoluted primary oscillatory zoning as well astransgressive recrystallization (Fig. 5). The semi-con-tinuous range from c. 250 to 15 Ma (Fig. 10a,b) isunlikely an artefact from binary mixing during analy-sis, since most dates intermediate between c. 240 andc. 40 Ma were acquired by conventional SIMS mea-surement on defined zircon domains. Instead, therange in ages probably reflects incomplete resettingof pre-existing zircon, whereby a memory of the zir-con’s pre-metamorphic chemical and isotopic compo-sition is partly retained due to inefficient subsolidusrecrystallization (Hoskin & Black, 2000). Mixed ageswere also previously reported in variably and partlyrecrystallized zircon from glaucophane-bearingmetasedimentary rocks from Serifos (Schneider et al.,2011), which yielded U–Pb ages of c. 40 Ma, identi-cal to those measured for zircon rims from Naxos.There is independent evidence for an age of c. 40 Mafor the high-P overprint on Naxos (Peillod et al.,unpublished data). Therefore, it appears likely thatzircon rims older than c. 40 Ma may reflect incom-plete subsolidus recrystallization, which in turn wouldalso explain the extensive, geologically unsoundspread in U–Pb zircon ages of c. 69–42 Ma (Martinet al., 2006) and c. 47–38 Ma (this study).

A feature common to all amphibolite samples fromNaxos is the presence of resolvable Triassic age com-ponents at c. 235 Ma (n = 15; Na41), c. 239 Ma(n = 2; Na50), c. 245 Ma (n = 12; Na49) and c.255 Ma (n = 3; Na48) (Figs 6 & 10). For reasonsoutlined above, Triassic zircon in all samples is unli-kely to be co-genetic with the host rock, but was

entrained during assimilation of crustal rocks, eitherduring ascent and emplacement at shallow-crustallevels or at the level of melt generation during thevery early stage of magmatism. Triassic granites arecommon in the Cyclades (Reischmann, 1998; Ringet al., 1999; Br€ocker & Pidgeon, 2007). For instance,Reischmann (1998) provided evidence for graniteemplacement in Naxos at 233 � 2 Ma. Sample Na50contains zircon ages ranging from c. 2406 to 52 Ma,unambiguously attesting to the xenocrystic nature ofthe zircon. On the basis of REE compositions similarto those observed for granodioritic zircon from theSeparation Point Suite in New Zealand, elevated Th/U and sector and oscillatory zoning in zircon cores itis argued that c. 250 Ma zircon was likely derivedfrom (a) granitoid source(s). Age variability in otheramphibolite samples is more limited, with each sam-ple showing evidence for material with a very homo-geneous zircon age population, possibly reflectinglocalized contamination. Resolvable age populationscan be used to characterize crustal protoliths, pro-vided that a sufficient number of grains inheritedfrom source regions provide reliable dates (e.g. Keay& Lister, 2002). This is not the case in the presentstudy, although sample Na50 gives some clues to thesedimentary provenance. A smear of ages from 1000to 250 along the Concordia (Fig. 6d) points to a verymixed inheritance. This range in ages overlaps withpre-Carboniferous age peaks obtained from detritaland inherited zircon grains from seven differentCycladic islands, including Naxos (Keay & Lister,2002). Considering all samples, the youngest clusterof inherited zircon ages at c. 260–220 Ma representsthe maximum crystallization age for the amphibo-lites.As discussed above, the trend in SiO2 v.

Na2O+K2O space (Fig. 4a) defined by all four sam-ples, along with subparallel normalized REE andmulti-element patterns (Fig. 4b,c), strongly suggestthat at least three amphibolite samples (excludingNa49) are genetically related, that is, they werederived from similar sources and underwent similarmagmatic differentiation, despite resolvable(geochronological) heterogeneity of the contaminantsdetectable in the amphibolites. Their close spatialassociation on the island of Naxos would furthersupport a close genetic relationship (Fig. 2). In viewof this, the nature of zircon inheritance requires someconsideration; namely (i) why are averaged Triassicages for each sample not identical within errors giventhe relatively small region of emplacement, and (ii)why did one sample preserve a wide range of inher-ited U–Pb dates (Na50), but apparently did not wit-ness any of the metamorphic overprinting events thataffected other amphibolites. Curiously, the preserva-tion potential for inheriting zircon is, in fact, the low-est in sample Na50, when compared to all samplesstudied, inferred from the low saturation temperature(~660 °C). In the absence of other supporting



evidence, the simplest explanation to question (ii)would be that sample Na50 did not reach conditionsconducive for new zircon growth or recrystallizationduring metamorphism at c. 40 and c. 15–14 Ma.This, in turn, could be a function of proximity toheat sources and/or depth of emplacement, durationof exposure to sufficiently elevated P–T conditionsand, finally, nature and composition of fluids andhost rock. In relation to question (i) it is feasible thatthe distinct and tight age ranges of inherited grains inamphibolites, other than sample Na50, may relate tovery local country rock with homogeneously old zir-con, so mafic dykes only sampled very limited coun-try rock during emplacement. Alternatively, samplesmay have experienced variable Pb loss associatedwith high-grade metamorphism later in their geologi-cal history.

Zircon inheritance in mafic rocks

A major finding of this work is that magmatic zirconin the Naxos amphibolites did not crystallize in situin the mafic dykes. This is because: (i) The geochem-istry of the zircon is variable and akin to zircon fromgranitoid rocks. Mafic rocks can be expected to pre-cipitate compositionally more homogeneous zircongiven the considerably lower degrees of magmatic dif-ferentiation until the point of hypothetical zircon sat-uration when compared to granites. This conflictswith highly variable values of Zr/Hf and Th/U seenin both zircon cores and rims. (ii) Complicated inter-nal structures of the zircon, including corrosion andresorption. (iii) The occurrence of xenocrystic zirconin sample Na50 with inherited cores ranging in agefrom c. 1000 to 250 Ma. (iv) Whole-rock geochem-istry of the amphibolites requires igneous differentia-tion coupled with assimilation of felsic material.

As discussed above, there is evidence for mid-Triassic granitoids across the Cyclades. It is likelythat the amphibolite precursors on Naxos, and prob-ably elsewhere in the Aegean Sea region, assimilatedzircon from these Triassic granitoids when theyintruded. However, sample Na50 must have alsosampled other material during ascent and crystalliza-tion of its precursor rocks as evidenced by thexenocrystic cores. We suggest that those cores werederived from schists of the passive-margin sequence(Keay & Lister, 2002).

First generation of zircon rims at c. 40 Ma

Having concluded that the magmatic zircon in theamphibolites represents assimilated foreign material,the question arises whether the two rim generations(re)crystallized in situ in their current host rocks. Thefirst-generation rims with ages of c. 42–38 Ma aredark greyish and have highly variable Zr/Hf ratios.Variable zircon rim geochemistry arising from exten-sive differentiation and concomitant fractionation of

Zr and Hf in zircon crystallizing melt appears unli-kely. In other words, first-generation rims were prob-ably not formed in a mafic rock (i.e. in situ in theamphibolite). This inference is in line with Martinet al. (2006), who also found first-generation zirconrims in one single amphibolite sample with U–Pbages of c. 69–42 Ma and d18O values of 6.2 � 0.8&.Judging from differences in 18O values for inner zir-con rims and co-existing garnet (5.8 � 1.0 to8.0 � 1.0&) in the amphibolite, Martin et al. (2006,p. 187) concluded that the first-generation rims didnot grow in equilibrium with garnet. Outer zirconrims, on the other hand, are characterized by d18Ovalues of 7.8 � 1.8&. Contemporaneous formationof zircon and garnet is also not supported by the lackof HREE depletion in zircon. Martin et al. (2006)also reported zircon rims from one calcic gneiss (c.57–41 Ma) and one metapelite (c. 50 Ma). In thesetwo cases d18O values in zircon and those in garnetare identical (Martin et al., 2006), implying that gar-net rims and c. 40 Ma zircon overgrowths stem fromhigh-P metamorphism on Naxos.Our interpretation is that the younger first-genera-

tion zircon rim ages of c. 40 Ma date the high-Poverprint, although determination whether the first-generation rims grew in situ in the amphibolites or ina former granitic host rock remains inconclusive.

The outer metamorphic rims at 15–14 Ma

The interpretation of the Miocene age of14.5 � 0.5 Ma of the bright outer rims of zirconfrom sample Na49 is more straightforward. Very lowTh/U ratios (≤0.01) of these rims strongly suggest ametamorphic origin of the zircon overgrowths. Thelow REE concentrations of the outer zircon rimscould be attributed to crystallization in equilibriumwith the mafic host rock and thus in situ growth.Martin et al. (2006) concluded that identical d18Ovalues of outer zircon rims and garnet from theiramphibolite sample indicate that both formedtogether during Miocene high-T metamorphism at19–14 Ma. The latter, and our, ages of c. 15–14 Mathen require that high-T conditions on Naxos lasteduntil c. 14 Ma.

Two possible tectonic scenarios

We provided evidence that the mid-Triassic magmaticzircon cores are not co-magmatic to the host amphi-bolites, and thus do not date the emplacement of themafic dykes. However, because our data and those ofMartin et al. (2006) do not allow for the unambigu-ous discrimination whether the first-generation zirconrims grew in situ or not, two tectonic scenarios arepossible for the emplacement of the mafic dykes. (i)Intrusion during late-Triassic to Jurassic oceanizationof the Aegean realm, in which case the c. 40 and c.15–14 Ma zircon rims would have grown in situ in



the amphibolites. (ii) Emplacement in the Miocene asa result of mafic underplating during large-scaleextension in the central Aegean. In this case, only thec. 15–14 Ma old metamorphic outer rims would rep-resent in situ growth.

(1) Remnants of late-Triassic to Jurassic oceaniclithosphere are exposed in probably one ormaybe two subparallel, NNW–SSE trending beltsstretching across mainland Greece, Albania,Croatia and Serbia and document the formerexistence of the Vardar-_Izmir-Ankara ocean andthe Pindos basins (Robertson, 2002; Bortolottiet al., 2013). In the central Aegean, the passive-margin/basement succession of the CBU is com-monly regarded as a lateral equivalent of the Pin-dos basin, and Jurassic ophiolitic rocks have beenreported from Crete and Andros (Koepke et al.,2002; Br€ocker & Pidgeon, 2007). However, gab-bros and plagiogranites from the ophiolitic mel-ange at the top of the CBU in Syros, Tinos andSamos yielded late Cretaceous ages (Tomascheket al., 2003; Br€ocker et al., 2014).

The tectonic position and the lithological associa-tion of the Naxos amphibolites place them into thepassive-margin sequence. It would then be conceiv-able that the mafic bodies intruded in the passivemargin setting in the late-Triassic or early/middle-Jurassic when the continental crust was stretched andthe Vardar-_Izmir-Ankara ocean and Pindos basinformed. Since the mafic dikes intruded into continen-tal crust, they may record an early rifting/break-upstage before the oceanic crust of the Vardar-_Izmir-Ankara ocean was fully developed.

(2) In the second feasible tectonic scenario the maficintrusives were emplaced in the early-Miocene.The c. 15–14 Ma old metamorphic outer rimsgrew in situ in the amphibolites and indicate thatemplacement of the mafic dykes must haveoccurred before c. 15 Ma.

Large-scale continental extension in the centralAegean commenced by c. 23 Ma (e.g. Lister et al.,1984; Ring et al., 2010). The degree of lithosphericstretching was highest in the central Cyclades and itis spatially associated with the thermal anomaly inthe Naxos-Paros-Ios region (Thomson et al., 2009;Ring et al., 2010). A major problem in Aegean tec-tonics is understanding the heat source driving thethermal anomaly. Kincaid & Griffiths (2003) showedthat subduction flow is driven around and beneaththe sinking plate, and velocities increase within themantle wedge and are focused towards the centre ofthe overriding plate. Therefore, the overriding plateheats more along the centreline, which may explainthe thermal anomaly in the Naxos-Paros-Ios region.In other words, the resulting lithospheric thinningand increased flow in the asthenosphere were greatestin the centre of the extending domain, and may have

caused the thermal anomaly in the Naxos/Paros-Iosregion a few million years later. Magmatic underplat-ing to the base of thinned continental crust wouldhave caused elevated heat and fluid flow. This sce-nario would be testable through the discovery ofearly-Miocene mafic melts on Naxos or through adeep geophysical survey aimed at finding a maficunderplate. It is at least feasible that the Naxosamphibolites represent early-Miocene mafic intrusivesand would then add support to this hypothesis.

SUMMARY AND CONCLUSIONS

Zircon in amphibolites from Naxos show a wide rangeof U–Pb ages of c. 700–550 Ma (up to 2046 Ma), c.262–220 Ma, c. 47–38 Ma and c. 15–14 Ma. The old-est age peak >550 Ma represents inherited cores. Allavailable evidence indicates that the c. 240 Ma mag-matic zircon in the Naxos amphibolites did not crys-tallize in situ and therefore does not constrain the ageof intrusion of the mafic dykes. The c. 240 Ma oldmagmatic zircon rather crystallized from felsic/grani-toid melts and was subsequently assimilated into theintrusives during emplacement. The two rim genera-tions are likely metamorphic in origin. While the c.40 Ma zircon overgrowths may not be in situ, the c.15–14 Ma zircon growth event occurred sometimeafter emplacement and solidification.Two tectonic scenarios for emplacement of the

mafic dykes are permissible: (i) Intrusion during late-Triasssic to Jurassic oceanization of the Aegeanrealm, and (ii) emplacement in the Miocene as aresult of mafic underplating during large-scale conti-nental extension of the overriding plate above theretreating Hellenic subduction zone.

ACKNOWLEDGEMENTS

R.B. acknowledges financial support from theUniversity of Canterbury through a postdoctoral fel-lowship. M. Palin is thanked for assistance duringLA-Q-ICPMS analysis and data reduction, and A.Rocholl for useful discussions. D. van Hinsbergen,N. M. W. Roberts and one anonymous reviewer pro-vided constructive criticism, and D. Robinson isthanked for editorial handling.

REFERENCES

Avigad, D., 1998. High-pressure metamorphism and coolingon SE Naxos (Cyclades, Greece). European Journal of Miner-alogy, 10, 1309–1319.

Avigad, D. & Garfunkel, Z., 1989. Low-angle faults above andbelow a blueschist belt—Tinos Island, Cyclades, Greece.Terra Nova, 1, 182–187.

Ballard, J.R., Palin, J.M., Williams, I.S., Campbell, I.H. &Faunes, A., 2001. Two ages of porphyry intrusion resolvedfor the super-giant Chuquicamata copper deposit of north-ern Chile by ELA-ICP-MS and SHRIMP. Geology, 29, 383–386.



Bea, F., Montero, P. & Ortega, M., 2006. A LA–ICP–MSevaluation of Zr reservoirs in common crustal rocks: impli-cations for Zr and Hf geochemistry, and zircon-forming pro-cesses. The Canadian Mineralogist, 44, 693–714.

Black, L.P., Kamo, S.L., Allen, C.M. et al., 2003. TEMORA1: a new zircon standard for Phanerozoic U-Pb geochronol-ogy. Chemical Geology, 200, 155–170.

Boehnke, P., Watson, E.B., Trail, D., Harrison, T.M. & Sch-mitt, A.K., 2013. Zircon saturation re-revisited. ChemicalGeology, 351, 324–334.

Bolhar, R., Weaver, S.D., Palin, J.M., Cole, J.W. & Paterson,L.A., 2008a. Systematics of zircon crystallisation in the Cre-taceous Separation Point Suite, New Zealand, using U/Pbisotopes, REE and Ti geothermometry. Contributions toMineralogy and Petrology, 156, 133–160.

Bolhar, R., Weaver, S.D., Whitehouse, M.J., Palin, J.M.,Woodhead, J.D. & Cole, J.W., 2008b. Sources and evolutionof arc magmas inferred from coupled O and Hf isotope sys-tematics of plutonic zircons from the Cretaceous SeparationPoint Suite (New Zealand). Earth and Planetary Science Let-ters, 268, 312–324.

Bolhar, R., Ring, U. & Allen, C.M., 2010. An integrated zir-con geochronological and geochemical investigation into theMiocene plutonic evolution of the Cyclades, Aegean Sea,Greece: part 1: geochronology. Contributions to Mineralogyand Petrology, 160, 719–742.

Bolhar, R., Ring, U., Kemp, A.I. et al., 2012. An integratedzircon geochronological and geochemical investigation intothe Miocene plutonic evolution of the Cyclades, Aegean Sea,Greece: part 2—geochemistry. Contributions to Mineralogyand Petrology, 164, 915–933.

Bortolotti, V., Chiari, M., Marroni, M., Pandolfi, L., Principi,G. & Saccani, E., 2013. Geodynamic evolution of ophiolitesfrom Albania and Greece (Dinaric-Hellenic belt): one, two,or more oceanic basins? International Journal of EarthSciences, 102, 783–811.

Bossart, P.J., Meier, M., Oberli, F. & Steiger, R.H., 1986.Morphology versus U-Pb systematics in zircon: a high-reso-lution isotopic study of a zircon population from a Variscandike in the Central Alps. Earth and Planetary Science Let-ters, 78, 339–354.

Botcharnikov, R.E., Almeev, R.R., Koepke, J. & Holtz, F.,2008. Phase relations and liquid lines of descent in hydrousferrobasalt—implications for the Skaergaard intrusion andColumbia River flood basalts. Journal of Petrology, 49,1687–1727.

Boynton, W. V., 1984. Chapter 3. Cosmochemistry of the rareearth elements: meteorite studies. In: Rare Earth ElementGeochemistry, Developments in Geochemistry (ed. Henderson,P.), 2nd edn, pp. 115–1522. Elsevier, Amsterdam.

Brichau, S., Ring, U., Ketcham, R.A., Carter, A., Stockli, D.& Brunel, M., 2006. Constraining the long-term evolution ofthe slip rate for a major extensional fault system in the cen-tral Aegean, Greece, using thermochronology. Earth andPlanetary Science Letters, 241, 293–306.

Br€ocker, M. & Pidgeon, R.T., 2007. Protolith ages of meta-igneous and metatuffaceous rocks from the Cycladic Blues-chist Unit, Greece: results of a reconnaissance U-Pb zirconstudy. The Journal of Geology, 115, 83–98.

Br€ocker, M., L€owen, K. & Rodionov, N., 2014. Unravelingprotolith ages of meta-gabbros from Samos and the Attic-Cycladic Crystalline Belt, Greece: results of a U-Pb zirconand Sr–Nd whole rock study. Lithos, 198–199, 234–248.

Buick, I.S., 1991. The late Alpine evolution of an extensionalshear zone, Naxos, Greece. Journal of the Geological Society,148, 93–103.

Buick, I.S. & Holland, T.J.B., 1989. The P-T-t path associatedwith crustal extension, Naxos, Cyclades, Greece. GeologicalSociety, London, Special Publications, 43, 365–369.

Campanha, G.A.C., Faleiros, F.M., Basei, M.A.S., Tassinari,C.C.G., Nutman, A.P. & Vasconcelos, P.M., 2015. Geo-chemistry and age of mafic rocks from the Votuverava

Group, southern Ribeira Belt, Brazil: evidence for 1490 Maoceanic back-arc magmatism. Precambrian Research, 266,530–550.

Claiborne, L.L., Miller, C.F., Walker, B.A., Wooden, J.L.,Mazdab, F.K. & Bea, F., 2006. Tracking magmatic processesthrough Zr/Hf ratios in rocks and Hf and Ti zoning in zir-cons: an example from the Spirit Mountain batholith,Nevada. Mineralogical Magazine, 70, 517–543.

Coogan, L.A., Wilson, R.N., Gillis, K.M. & MacLeod, C.J.,2001. Near-solidus evolution of oceanic gabbros: Insightsfrom amphibole geochemistry. Geochimica et CosmochimicaActa, 65, 4339–4357.

Coogan, L.A. & Hinton, R.W., 2006. Do the trace elementcompositions of detrital zircons require Hadean continentalcrust? Geology, 34, 633–636.

Corfu, F., Hanchar, J.M., Hoskin, P.W. & Kinny, P., 2003.Atlas of zircon textures. Reviews in Mineralogy and Geo-chemistry, 53, 469–500.

David, K., Schiano, P. & Allegre, C.J., 2000. Assessment ofthe Zr/Hf fractionation in oceanic basalts and continentalmaterials during petrogenetic processes. Earth and PlanetaryScience Letters, 178, 285–301.

Deng, H., Kusky, T., Polat, A. et al., 2014. Geochronology,mantle source composition and geodynamic constraints onthe origin of Neoarchean mafic dikes in the Zanhuang Com-plex, Central Orogenic Belt, North China Craton. Lithos,205, 359–378.

DePaolo, D.J., 1981. Trace element and isotopic effects ofcombined wallrock assimilation and fractional crystalliza-tion. Earth and Planetary Science Letters, 53, 189–202.

D€urr, S., Altherr, R., Keller, J., Okrusch, M. & Seidel, E.,1978. The median Aegean crystalline belt: stratigraphy,structure, metamorphism, magmatism. Alps, Apennines, Hel-lenides, 38, 455–476.

Franz, L. & Okrusch, M., 1992. Aragonite-bearing blueschistson Arki island, Dodecanese, Greece. European Journal ofMineralogy, 4, 527–537.

Frey, F.A., Green, D.H. & Roy, S.D., 1978. Integrated modelsof basalt petrogenesis: a study of quartz tholeiites to olivinemelilitites from south eastern Australia utilizing geochemicaland experimental petrological data. Journal of Petrology, 19,463–513.

Fu, B., Page, F.Z., Cavosie, A.J. et al., 2008. Ti-in-zircon ther-mometry: applications and limitations. Contributions to Min-eralogy and Petrology, 156, 197–215.

Grimes, C.B., John, B.E., Kelemen, P.B. et al., 2007. Trace ele-ment chemistry of zircons from oceanic crust: a method fordistinguishing detrital zircon provenance. Geology, 35, 643–646.

Hellstrom, J., Paton, C., Woodhead, J. & Hergt, J., 2008.Iolite: software for spatially resolved LA-(quad and MC)ICPMS analysis. Mineralogical Association of Canada ShortCourse Series, 40, 343–348.

Hildreth, W. & Moorbath, S., 1988. Crustal contributions toarc magmatism in the Andes of central Chile. Contributionsto Mineralogy and Petrology, 98, 455–489.

Hinsbergen, D.J., Maffione, M., Plunder, A. et al., 2016. Tec-tonic evolution and paleogeography of the Kırs�ehir Blockand the Central Anatolian Ophiolites, Turkey. Tectonics, 35,983–1014.

van Hinsbergen, D.J.J., Langereis, C.G. & Meulenkamp, J.E.,2005. Revision of the timing, magnitude and distribution ofNeogene rotations in the western Aegean region. Tectono-physics, 396, 1–34.

Hofmann, A.W., 1988. Chemical differentiation of the Earth:the relationship between mantle, continental crust, andoceanic crust. Earth and Planetary Science Letters, 90, 297–314.

Hoskin, P.W.O. & Black, L.P., 2000. Metamorphic zircon for-mation by solid-state recrystallization of protolith igneouszircon. Journal of Metamorphic Geology, 18, 423–439.



Hoskin, P.W. & Schaltegger, U., 2003. The composition of zir-con and igneous and metamorphic petrogenesis. Reviews inMineralogy and Geochemistry, 53, 27–62.

Huppert, H.E., Stephen, R. & Sparks, J., 1985. Cooling andcontamination of mafic and ultramafic magmas duringascent through continental crust. Earth and PlanetaryScience Letters, 74, 371–386.

Ireland, T.R. & Gibson, G.M., 1998. SHRIMP monazite andzircon geochronology of high-grade metamorphism in NewZealand. Journal of Metamorphic Geology, 16, 149–167.

Ireland, T.R. & Williams, I.S., 2003. Considerations in zircongeochronology by SIMS. Reviews in Mineralogy and Geo-chemistry, 53, 215–241.

Jansen, J.B.H., 1973. Geological Map of Naxos (1/50 000).National Institute of Geological Mining Research, Athens.

Jansen, J.B.H. & Schuiling, R.D., 1976. Metamorphism onNaxos: petrology and geothermal gradients. American Jour-nal of Science, 276, 1225–1253.

John, B.E. & Howard, K.A., 1995. Rapid extension recordedby cooling-age patterns and brittle deformation, Naxos,Greece. Journal of Geophysical Research: Solid Earth, 100,9969–9979.

Johnson, D.M., Hooper, P.R. & Conrey, R.M., 1999. XRFanalysis of rocks and minerals for major and trace elementson a single low dilution Li-tetraborate fused bead. Advancesin X-ray Analysis, 41, 843–867.

Keay, S. & Lister, G., 2002. African provenance for themetasediments and metaigneous rocks of the Cyclades,Aegean Sea, Greece. Geology, 30, 235–238.

Keay, S., Lister, G. & Buick, I., 2001. The timing of partialmelting, Barrovian metamorphism and granite intrusion inthe Naxos metamorphic core complex, Cyclades, AegeanSea, Greece. Tectonophysics, 342, 275–312.

Kincaid, C. & Griffiths, R.W., 2003. Laboratory models of thethermal evolution of the mantle during rollback subduction.Nature, 425, 58–62.

Kirkland, C.L., Smithies, R.H., Taylor, R.J.M., Evans, N. &McDonald, B., 2015. Zircon Th/U ratios in magmatic envi-rons. Lithos, 212, 397–414.

Koepke, J., Seidel, E. & Kreuzer, H., 2002. Ophiolites on thesouthern Aegean islands Crete, Karpathos and Rhodes:composition, geochronology and position within the ophio-lite belts of the Eastern Mediterranean. Lithos, 65, 183–203.

Le Bas, M.J., Le Maitre, R.W., Streckeisen, A. & Zanettin, B.,1986. A chemical classification of volcanic rocks based onthe total alkali-silica diagram. Journal of Petrology, 27, 745–750.

Linnen, R.L. & Keppler, H., 2002. Melt composition controlof Zr/Hf fractionation in magmatic processes. Geochimica etCosmochimica Acta, 66, 3293–3301.

Lister, G.S., Banga, G. & Feenstra, A., 1984. Metamorphiccore complexes of Cordilleran type in the Cyclades, AegeanSea, Greece. Geology, 12, 221–225.

Maffione, M., Thieulot, C., van Hinsbergen, D.J., Morris, A.,Pl€umper, O. & Spakman, W., 2015. Dynamics of intrao-ceanic subduction initiation: 1. Oceanic detachment faultinversion and the formation of supra-subduction zone ophio-lites. Geochemistry, Geophysics, Geosystems, 16, 1753–1770.

Martin, L., Duchene, S., Deloule, E. & Vanderhaeghe, O.,2006. The isotopic composition of zircon and garnet: arecord of the metamorphic history of Naxos, Greece. Lithos,87, 174–192.

Mojzsis, S.J. & Harrison, T.M., 2002. Establishment of a 3.83-Gamagmatic age for the Akilia tonalite (southern West Greenland).Earth and Planetary Science Letters, 202, 563–576.

Muir, R.J., Ireland, T.R., Weaver, S.D. et al., 1998.Geochronology and geochemistry of a Mesozoic magmaticarc system, Fiordland, New Zealand. Journal of the Geologi-cal Society, 155, 1037–1053.

Nasdala, L., Zhang, M., Kempe, U. et al., 2003. Spectroscopicmethods applied to zircon. Reviews in Mineralogy and Geo-chemistry, 53, 427–467.

Parlak, O. & Delaloye, M., 1999. Precise 40Ar/39Ar ages fromthe metamorphic sole of the Mersin Ophiolite (southern Tur-key). Tectonophysics, 301, 145–158.

Paton, C., Hellstrom, J., Paul, B., Woodhead, J. & Hergt, J.,2011. Iolite: freeware for the visualisation and processing ofmass spectrometric data. Journal of Analytical Atomic Spec-trometry, 26, 2508–2518.

Pearce, J.A., 2008. Geochemical fingerprinting of oceanicbasalts with applications to ophiolite classification and thesearch for Archean oceanic crust. Lithos, 100, 14–48.

Reischmann, T., 1998. Pre-Alpine origin of tectonic units fromthe metamorphic complex of Naxos, Greece, identified bysingle zircon Pb/Pb dating. Deksίοm sgς Εkkgmιjής ΓexkοcιjήςΕsaιqίaς, 32, 101–111.

Ring, U. & Layer, P.W., 2003. High-pressure metamorphismin the Aegean, eastern Mediterranean: underplating andexhumation from the Late Cretaceous until the Miocene torecent above the retreating Hellenic subduction zone. Tecton-ics, 22, 1–23.

Ring, U., Laws, S. & Bernet, M., 1999. Structural analysis of acomplex nappe sequence and late-orogenic basins from theAegean Island of Samos, Greece. Journal of Structural Geol-ogy, 21, 1575–1601.

Ring, U., Kr€oner, A., Buchwaldt, R., Toulkeridis, T. & Layer,P., 2002. Eclogite-facies metamorphism and shear-zone pat-terns in the Mozambique belt, northern Malawi, east-centralAfrica: implications for the assembly of Gondwana. Precam-brian Research, 116, 19–56.

Ring, U., Will, T., Glodny, J. et al., 2007. Early exhumationof high-pressure rocks in extrusion wedges: Cycladic blues-chist unit in the eastern Aegean, Greece, and Turkey. Tec-tonics, 26, 1–23.

Ring, U., Glodny, J., Will, T. & Thomson, S., 2010. The Hel-lenic subduction system: high-pressure metamorphism,exhumation, normal faulting, and large-scale extension.Annual Review of Earth and Planetary Sciences, 38, 45–76.

Roberts, N.M., Slagstad, T., Parrish, R.R., Norry, M.J., Mar-ker, M. & Horstwood, M.S., 2013. Sedimentary recycling inarc magmas: geochemical and U-Pb–Hf–O constraints onthe Mesoproterozoic Suldal Arc, SW Norway. Contributionsto Mineralogy and Petrology, 165, 507–523.

Robertson, A.H., 2002. Overview of the genesis and emplace-ment of Mesozoic ophiolites in the Eastern MediterraneanTethyan region. Lithos, 65, 1–67.

Robertson, A.H.F., Clift, P.D., Degnan, P.J. & Jones, G.,1991. Palaeogeographic and palaeotectonic evolution of theEastern Mediterranean Neotethys. Palaeogeography, Palaeo-climatology, Palaeoecology, 87, 289–343.

Rubatto, D., 2002. Zircon trace element geochemistry: parti-tioning with garnet and the link between U-Pb ages andmetamorphism. Chemical Geology, 184, 123–138.

Schmid, S.M., Bernoulli, D., Fugenschuh, B. et al., 2008. TheAlpine-Carpathian-Dinaridic orogenic system: correlationand evolution of tectonic units. Swiss Journal of Geosciences,101, 139–183.

Schneider, D.A., Senkowski, C., Vogel, H., Grasemann, B.,Iglseder, C. & Schmitt, A.K., 2011. Eocene tectonometamor-phism on Serifos (western Cyclades) deduced from zircondepth-profiling geochronology and mica thermochronology.Lithos, 125, 151–172.

Scoates, J.S. & Chamberlain, K.R., 1995. Baddeleyite (ZrO2)and zircon (ZrSiO4) from anorthositic rocks of the Laramieanorthosite complex, Wyoming: petrologic consequences andU-Pb ages. American Mineralogist, 80, 1317–1327.

Seward, D., Vanderhaeghe, O., Siebenaller, L. et al., 2009.Cenozoic tectonic evolution of Naxos Island through a mul-ti-faceted approach of fission-track analysis. Geological Soci-ety, London, Special Publications, 321, 179–196.

Sherlock, S., Kelley, S., Inger, S., Harris, N. & Okay, A.,1999. 40Ar-39Ar and Rb-Sr geochronology of high-pressuremetamorphism and exhumation history of the Tavsanli



Zone, NW Turkey. Contributions to Mineralogy and Petrol-ogy, 137, 46–58.

Spera, F.J. & Bohrson, W.A., 2001. Energy-constrained open-system magmatic processes I: general model and energy-con-strained assimilation and fractional crystallization (EC-AFC)formulation. Journal of Petrology, 42, 999–1018.

Tapster, S., Condon, D.J., Naden, J. et al., 2016. Rapid ther-mal rejuvenation of high-crystallinity magma linked to por-phyry copper deposit formation: evidence from the KoloulaPorphyry Prospect, Solomon Islands. Earth and PlanetaryScience Letters, 442, 206–217.

Thomson, S.N., Ring, U., Brichau, S., Glodny, J. & Will,T.W., 2009. Timing and nature of formation of the Iosmetamorphic core complex, southern Cyclades, Greece. In:Extending a Continent: Architecture, Rheological Coupling,and Heat Budget (eds Ring, U. & Wernicke, B.), GeologicalSociety London, Special Publications, 321, 139–167.

Tomaschek, F., Kennedy, A.K., Villa, I.M., Lagos, M. & Ball-haus, C., 2003. Zircons from Syros, Cyclades, Greece—re-crystallization and mobilization of zircon during high-pressure metamorphism. Journal of Petrology, 44, 1977–2002.

Tremblay, A., Meshi, A., Deschamps, T., Goulet, F. & Goulet,N., 2015. The Vardar zone as a suture for the Mirdita ophi-olites, Albania: constraints from the structural analysis ofthe Korabi-Pelagonia zone. Tectonics, 34, 352–375.

Vanderhaeghe, O., 2004. Structural development of the Naxosmigmatite dome. Geological Society of America SpecialPapers, 380, 211–227.

Wang, X., Griffin, W.L. & Chen, J., 2010. Hf contents and Zr/Hf ratios in granitic zircons. Geochemical Journal, 44, 65–72.

Watson, E.B. & Harrison, T.M., 1983. Zircon saturation revis-ited: temperature and composition effects in a variety ofcrustal magma types. Earth and Planetary Science Letters,64, 295–304.

Watson, E.B., Wark, D.A. & Thomas, J.B., 2006. Crystalliza-tion thermometers for zircon and rutile. Contributions toMineralogy and Petrology, 151, 413–433.

Wiedenbeck, M.A.P.C., Alle, P., Corfu, F. et al., 1995. Threenatural zircon standards for U-Th-Pb, Lu-Hf, trace elementand REE analyses. Geostandards Newsletter, 19, 1–23.

Wijbrans, J.R. & McDougall, I., 1986. 40Ar/39Ar dating ofwhite micas from an Alpine high-pressure metamorphic belton Naxos (Greece): the resetting of the argon isotopic sys-tem. Contributions to Mineralogy and Petrology, 93, 187–194.

Wijbrans, J.R. & McDougall, I., 1988. Metamorphic evolutionof the attic Cycladic metamorphic belt on Naxos (Cyclades,Greece) utilizing 40Ar/39Ar age spectrum measurements.Journal of Metamorphic Geology, 6, 571–594.

Wijbrans, J.R., Schliestedt, M. & York, D., 1990. Single grainargon laser probe dating of phengites from the blueschist togreenschist transition on Sifnos (Cyclades, Greece). Contribu-tions to Mineralogy and Petrology, 104, 582–593.

Williams, I.S. & Claesson, S., 1987. Isotopic evidence for thePrecambrian provenance and Caledonian metamorphism ofhigh grade paragneisses from the Seve Nappes, ScandinavianCaledonides. Contributions to Mineralogy and Petrology, 97,205–217.

Woodhead, J.D., Hellstrom, J., Hergt, J.M., Greig, A. &Maas, R., 2007. Isotopic and elemental imaging of geologicalmaterials by laser ablation inductively coupled plasma-massspectrometry. Geostandards and Geoanalytical Research, 31,331–343.

Zartman, R.E. & Richardson, S.H., 2005. Evidence from kim-berlitic zircon for a decreasing mantle Th/U since theArchean. Chemical Geology, 220, 263–283.

SUPPORTING INFORMATION

Additional Supporting Information may be found inthe online version of this article at the publisher’sweb site:Table S1. Whole-rock major (wt%) and trace ele-

ment (ppm) data.Table S2. Zircon U–Th–Pb data for (a) four

amphibolites by LA-ICPMS and (b) Na49 by SIMS.

Received 27 May 2016; revision accepted 19 November 2016.



Zircon in amphibolites from Naxos, Aegean Sea, Greece ...people.geo.su.se/uwe/publications/Bolhar_et_al... · Zircon in amphibolites from Naxos, Aegean Sea, Greece: origin, signiﬁcance

Documents