In situ ultraviolet (UV) laser-ablation 40Ar/39Ar dating, microstructural analysis, and stable O, H, and C isotope analyses were performed on white mica–bearing calcite– and quartz–mica schists of the West Cycladic detachment system footwall in order to resolve outstanding uncertainties about the timing of deformation and the role of rock rheology on 40Ar/39Ar dating systematics. In both quartz-rich and calcite-rich samples, deformed and chemically zoned white micas form two chemical populations: (1) a high component of Al-celadonite in undeformed portions of grains (high-pressure remnants), and (2) enrichment in muscovite in deformed portions (low-pressure neocrystallization). Micas in the quartz-rich rocks record higher internal strain, illustrated by elongated, sheared grains and boudinaged mica-fish structures. In this lithology, quartz formed a load-bearing framework that transferred strain to the muscovite packets and facilitated the formation of mica-fish structures. Recrystallization was promoted by coeval fluid infiltration, supported by stable isotope analyses and indented boundaries on bulging quartz grains. In rocks containing calcite-muscovite aggregates, the calcite formed an interconnected weak layer, with strain being accommodated by dislocation creep. In these rocks, micas were only partially neocrystallized. Prismatic white micas, largely unaffected by boudinage or kinking, yielded 40Ar/39Ar ages that are up to 10 m.y. older than deformed (kinked or sheared) portions of the same grains. Overall, the ages attest to strong lithological control on deformation- and fluid-controlled white mica neocrystallization. The oldest, undeformed grain ages in the calcite-rich rocks are consistent with the timing of Eocene metamorphism, with the deformed grain ages interpreted as representing the transition to lower-pressure conditions during nascent extension. Completely neocrystallized grains in the quartz-rich rocks are interpreted as defining the minimum age of Miocene ductile extension along the detachment system. The new data show the power of combining in situ laser-ablation 40Ar/39Ar dating, microstructural analysis, mineral chemistry, and stable isotope data for unraveling the timing and time scales of complex deformation histories.

Resolving the timing and time scales of mid- to upper-crustal deformation processes has long been problematic. Multigrain 40Ar/39Ar incremental heating experiments commonly result in disturbed spectra, with relatively large variations in apparent ages from successive heating steps (e.g., Wijbrans and McDougall, 1986; Maluski et al., 1987; Zeffren et al., 2005; Bröcker et al., 2013). These disturbances have been proposed to be due to excess argon (40Ar decoupled from its parent 40K) or mixed mica populations (either different mica types or multiple generations of the same type), resulting in differential mixing of Ar from different chemical or microstructural reservoirs at each temperature step (e.g., Forster and Lister, 2004; Kula et al., 2010). Nowhere has the complexity of different 40Ar/39Ar experiments been better documented than in the Cycladic region of the Aegean Sea.

In the Cyclades, major crustal extension was accommodated along low-angle normal faults during the Miocene, resulting in the exposure of blueschist and eclogite assemblages in the footwall (Jolivet and Brun, 2010; Ring et al., 2010, and references therein). The West Cycladic detachment system is an example of a large-scale, low-angle normal fault system with a top-to-the-SW-SSW displacement, exposed on the islands of Kea, Kythnos, and Serifos (Grasemann et al., 2012). Deformation imparted a pervasive ductile to brittle structure on the footwall rocks, but it was also accommodated by significant, yet localized shear zones within the footwall.

Thermochronologic ages along shear zone exposures and within the footwall of the West Cycladic detachment system have recently been collected in an attempt to resolve the timing of faulting and displacement (Brichau et al., 2010; Iglseder et al., 2011; Schneider et al., 2011; Grasemann et al., 2012; Berger et al., 2013). Apatite and zircon (U-Th)/He data document Miocene activity across the shear zone, whereas multigrain white mica step-heating 40Ar/39Ar results are less conclusive. Specifically, 40Ar/39Ar spectra from schists in the footwall of the shear zone are commonly characterized by hump-shaped patterns, whereas rocks that exhibit considerable strain and recrystallization (e.g., mylonites, phyllonites) commonly yield flat plateaus (Iglseder et al., 2011; Schneider et al., 2011). This difference in spectral shapes may be partly due to the number of mica crystallization “episodes” recorded in the various lithotypes.

On a regional scale, micas initially crystallized during Eocene high-pressure–low-temperature metamorphism associated with subduction, and subsequently (partially) recrystallized under greenschist-facies to lower-amphibolite-facies conditions during Miocene extension (e.g., Jolivet et al., 2010; Ring et al., 2010; Iglseder et al., 2011; Grasemann et al., 2012). Deformation mechanisms and metamorphic assemblages indicate (re)crystallization and equilibration at maximum temperatures of 350–400 °C (Iglseder et al., 2011; Grasemann et al., 2012; Grasemann and Tschegg, 2012), i.e., the nominal Ar diffusion closure temperature for white mica (e.g., McDougall and Harrison, 1999; Harrison et al., 2009). The 40Ar/39Ar ages cannot, therefore, be interpreted as relating to a simple cooling scenario (e.g., Warren et al., 2011).

In an attempt to improve the geological interpretation of the complex 40Ar/39Ar age populations derived from the Western Cyclades in particular, and complex step-heating spectra from tectonites in general, we conducted in situ laser-ablation 40Ar/39Ar analyses on white micas from within and below a major low-angle normal fault zone exposed on Kea and Serifos Islands (Fig. 1). This method allows us to link age variation recorded within single grains to both microstructures and mineral chemistry, thus providing the possibility of identifying physicochemical and mechanical relationships. In situ 40Ar/39Ar methods have previously been shown to successfully date deformation-related neo- and recrystallization in cases where chemically or microstructurally distinct populations of micas exist (e.g., Kramar et al., 2001; Mulch and Cosca, 2004; Mulch et al., 2002, 2005; Putlitz et al., 2005).

In this study, we have additionally taken the approach of assessing the 40Ar/39Ar ages with respect to the degree of deformation of the sample lithotype. Our results indicate that internally deformed and completely neocrystallized micas in quartz-rich rocks yield the youngest ages, whereas calcitic rocks, which contain micas that are only marginally neocrystallized due to the strong strain partitioning between calcite and mica, preserve older ages. In this paper, we use the term “neocrystallized” to denote micas that recrystallized with a new composition. Stable isotope data moreover indicate moderate fluid-rock interaction, suggesting that fluid infiltration and lithological response to strain are both important for encouraging mica recrystallization and resetting the 40Ar/39Ar chronometer.

Kea and Serifos lie among the islands of the Western Cyclades, in the Aegean Sea (Fig. 1), and they belong to the Attic-Cycladic crystalline complex. The Attic-Cycladic crystalline complex is divided into three units: (1) the Cycladic basement unit, which is a diversely metamorphosed Paleozoic basement intruded by Triassic granitoids, (2) the overlying Cycladic blueschist unit, which consists of a polymetamorphic Carboniferous–Permian to latest Cretaceous passive-margin sequence (e.g., Dürr et al., 1978; Blake et al., 1981), and (3) the Cycladic upper unit, which forms the hanging wall to the Cycladic extensional detachments and includes various klippen of Late Cretaceous granitoids (Altherr et al., 1994), Permian to Tertiary sediments, Mesozoic ophiolites (Katzir et al., 1996; Robertson, 2002), and pre-Eocene metamorphic rocks covered by Late Cretaceous unmetamorphosed carbonates (Reinecke et al., 1982; Patzak et al., 1994; Trotet et al., 2001; Zeffren et al., 2005; Skarpelis et al., 2008).

The tectonometamorphic evolution of the Western Cyclades involved at least two major events: (1) M1/D1 is a high-pressure–low-temperature metamorphic episode at 52–40 Ma (Wijbrans et al., 1990; Bröcker et al., 2004; Putlitz et al., 2005; Schneider et al., 2011; Lagos et al., 2007) with a preserved ENE-WSW lineation in the Cycladic blueschist unit exposed on Serifos and Kea (Schneider et al., 2011; Iglseder et al., 2011; Grasemann et al., 2012). These structures likely record extrusion from subduction to midcrustal conditions (Jolivet and Brun, 2010). (2) M2/D2 is characterized by a lower-amphibolite- to greenschist-facies metamorphic overprint during the Miocene. Unroofing and cooling during crustal-scale extension occurred at ca. 21–13 Ma, pervasively deforming the footwall greenschists and marbles, which record a top-to-the-SW-SSW displacement (Iglseder et al., 2011; Grasemann et al., 2012). Tectonism related to M2 is attributed to back-arc extension occurring on the African retreating slab, which also generated Miocene and younger magmatism across the Cycladic domain (Jolivet and Brun, 2010; Ring et al., 2010).

Kea (Fig. 1) is situated 60 km southeast of Athens in the Aegean Sea. The island contains three structural domains: a footwall composed of the Cycladic blueschist unit, a hanging wall composed of the Cycladic upper unit, and a low-angle normal fault zone separating the two units composed of ductile-brittle fault structures (Iglseder et al., 2011; Rice et al., 2012). The structure of the island is domed, as defined by a pervasive foliation across the island (Grasemann et al., 2012). A >450-m-thick footwall composed of intensely folded schists with varying amounts of calcite alternates with marbles (Iglseder et al., 2011). The marbles are frequently (ultra)mylonitic and sometimes contain boudins of dolomite and quartzite. The schists exhibit a common greenschist-facies paragenesis of chlorite + epidote + plagioclase + white mica + quartz ± amphibole. Relict glaucophane present in the schists suggests these rocks originally crystallized under high-pressure conditions (Iglseder et al., 2011). Accessory minerals include sphene, graphite, apatite, tourmaline, late calcite, and magnetite. Albite and epidote porphyroblasts contain mica inclusions.

The low-angle normal fault zone is ∼60 m thick and includes a mixture of rocks derived from the footwall and the hanging wall that show evidence of ductile to brittle deformation (Iglseder et al., 2011). The base of the low-angle normal fault is highly strained and is composed of an up to 20-m-thick zone of phyllonite- and schist-derived proto-cataclasites and cataclasites. The cataclasites contain fragments of gray-blue marble mylonites and phyllonites from the footwall. In the north, south, and southeast parts of the island (Fig. 1), a distinct, up to 30-m-thick, zone of fine-grained ultramylonitic marble is exposed; elsewhere, this zone is much thinner or eroded away.

Decimeter-scale layers of intense ankeritization, limonitization, dolomitization, and silicification occur in the carbonate rocks due to leaching and hydrothermal activity during and after retrograde greenschist-facies metamorphism. There is also evidence of Fe-Mn-Pb-Ag mineralization, probably related to fluid activity derived from magmatic intrusions, such as at Lavrion (Skarpelis, 2007; Iglseder et al., 2011; Berger et al., 2013). Above the marble ultramylonites, several meters of cohesive calcite and dolomite cataclasites are present, which grade into the rarely preserved hanging wall. The hanging wall of the low-angle normal fault is composed of cohesive sedimentary breccias and conglomerates of dolostones and limestones. It is mostly unmetamorphosed, but it experienced cataclasis and significant fluid activity during the formation of the detachment (Iglseder et al., 2011; Grasemann et al., 2012).

Although the island was broadly metamorphosed under greenschist-facies conditions during midcrustal extension, six mica populations have previously been identified and divided into four groups on the basis of their composition (Iglseder et al., 2011): (1a) inclusions in large epidote and albite porphyroblasts; (1b) D1 micas folded in a crenulation cleavage within D2 folds; (2) micas within strain shadows and associated recrystallization trails; (3) syntectonic grains aligned parallel to the main D2 foliation; (4a) grains in SC and SCC′ fabrics; and (4b) post-tectonic porphyroblasts overgrowing or crosscutting the main foliation. Pressure-temperature (P-T) estimates obtained from chlorite–white mica pairs illustrate a progression of cooling and decompression (Iglseder et al., 2011). The highest P-T conditions (7–5.5 kbar, 450–360 °C) were obtained from chlorite–white mica inclusions in albite and epidote porphyroblasts. Shallower conditions ranging from 5.5 kbar and 400 °C to 3 kbar and 350 °C were found in mineral pairs parallel to the main D2 foliation, and conditions of 3–2 kbar and 350–280 °C were obtained from mineral pairs oblique to D2 foliation and in SC-SCC′ fabrics.

Previous multigrain 40Ar/39Ar analyses from Kea have yielded three different types of step-heating spectra (consistent with the identification of multiple generations of mica): type 1—disturbed spectra with apparent ages of 24–20 Ma; type 2—relatively flat spectra (plateaus) with apparent ages of 18–13 Ma; and type 3—significantly disturbed spectra yielding no reliable age information (Iglseder et al., 2011; Grasemann et al., 2012). Type 2 spectra, defined by relatively well-behaved heating increments, were interpreted as representing the maximum ages for the latest displacement of normal-sense ductile shear. Notably, these spectra type are from high-strain rocks such as mylonitic footwall marbles and low-angle normal fault phyllonites.

Serifos Island (Fig. 1) is located 100 km southeast of Athens in the Western Cyclades. Similar to Kea, Serifos has a domal structure and is divided into three tectonic units: a footwall consisting of the basal parts of the Cycladic blueschist unit, the Cycladic blueschist unit, and a hanging wall represented by the Cycladic upper unit. The three units are separated by low-angle normal faults (Schneider et al., 2011; Grasemann et al., 2012).

The footwall consists of a >200-m-thick sequence of mylonitic orthogneiss from the basal Cycladic blueschist unit, which was deformed at temperatures above ∼450 °C (Grasemann and Petrakakis, 2007; Grasemann et al., 2012). The orthogneiss is generally composed of alternating quartzofeldspathic and biotite-amphibolite bands, and I tis associated with an overlying sequence of impure dolomite and calcite marble mylonites. The impurities consist of disseminated quartz grains with quartz and potassic feldspar– or clinopyroxene-rich gneiss layers. The basal complex is overlain by intercalated amphibolites and gneisses, and felsic greenschists and carbonate schists alternating with marble metaconglomerates and calcite marbles, preserving relicts of glaucophane in epidote and albite poikiloblasts (Salemink, 1980; Schneider et al., 2011). The occurrences of glaucophane suggest that these rocks originally experienced high-pressure metamorphism. The low-angle normal fault is characterized by a thin zone of talc and marble cataclasites, and a several-tens-of-meters-thick zone of gneisses/amphibolites proto-cataclasites (Grasemann and Tschegg, 2012). Localized ductile deformation, which occurred at temperatures of ∼300 °C, is present as ultrafine-grained mylonitic marble (Grasemann et al., 2012). The hanging wall consists of undeformed conglomerate marbles and marble-dominated proto-cataclasites that are strongly ankeritized by fluid infiltration (Grasemann et al., 2012).

Serifos is further characterized by a syntectonic granodioritic intrusion that occupies half of the island (Iglseder et al., 2009; Tschegg and Grasemann, 2009). This pluton intruded into the extensional shear zone during M2, at ca. 10 Ma (Iglseder et al., 2009), and resulted in extensive metasomatism and fluid infiltration along the fault zone (Iglseder et al., 2009; Tschegg and Grasemann, 2009; Schneider et al., 2011; Grasemann et al., 2012). The Cycladic blueschist unit rocks surrounding the intrusion display increasing metamorphic grade toward the contact from actinolite-bearing schists to garnet-bearing gneisses. The host rocks that are close to the granodiorite contain skarn veins that have Ca- and Fe3+-rich garnet with hedenbergite and magnetite (Salemink, 1980; Schneider et al., 2011).

Previously published 40Ar/39Ar data from Serifos samples also yield evidence for mixed age populations. The oldest mica ages, 38–32 Ma, are derived from the schists and amphibolites of the Cycladic blueschist unit and are interpreted as recording the greenschist-facies conditions that followed Eocene high-pressure metamorphism (Schneider et al., 2011; Grasemann et al., 2012). Zircon rims from the Cycladic blueschist unit, recording U-Pb ages of ca. 40 Ma, provide supporting evidence for an Eocene tectonothermal event (Schneider et al., 2011). The intrusion and related metasomatism of the granodiorite pluton have locally reset mica chronometers to reflect late Miocene cooling after M2; Grasemann et al. (2012) reported 40Ar/39Ar white mica ages of ca. 9 Ma within the thermal aureole of the intrusion.

In an attempt to document the mechanisms driving the variation in 40Ar/39Ar ages across the islands and between different lithotypes, we collected samples from Kea and Serifos (Fig. 1), targeting the footwall of the detachment system and the low-angle normal fault zone. Criteria for the selection of the samples included visible white mica grains within a tectonic fabric and the presence of calcite or dolomite for the application of O and C stable isotope analyses. Our previous mapping efforts on the islands (Iglseder et al., 2009, 2011; Schneider et al., 2011; Grasemann et al., 2012) allowed us to strategically identify locations in which to sample. Samples with an EC-prefix consist of schists, calcite schists, and a few mylonitic marbles and are all from the footwall rocks, except EC-14, which is from just a few meters below the ultramylonites of the low-angle normal fault zone. Samples with an MPV-prefix, comprising calcite/dolomite (ultra)mylonitic marbles, were collected at the same outcrop as the EC suite and were analyzed for O and C stable isotopes. Thirteen samples were used for stable isotope analyses, and six were selected for chemical characterization and 40Ar/39Ar dating (Fig. 2).

For 40Ar/39Ar dating purposes, we targeted white mica populations that were identifiably in equilibrium with chlorite, thus providing samples suitable for dating the latest phase of deformation/equilibration under greenschist-facies conditions. The white mica–chlorite associations were found (1) parallel to the main foliation, (2) as inclusions in albite porphyroblasts, and (3) in strings (veinlets) of chlorite. Geochemical and 40Ar/39Ar analyses in this paper focus on micas in the main foliation, where micas were generally larger (100–200 μm in width), and on micas in veinlets of chlorite (only sample EC-10).

Mineral assemblages as well as major deformation microstructures for each sample are compiled in Table 1. Quartz recrystallization is evident in all the samples to variable extent, with no clear correlation to lithotype, geography, or proximity to the detachment system. Quartz crystals in all samples show subgrain rotation and bulging of grains, indicated by lobate grain boundaries. Samples EC-4, ΕC-9, and EC-16 exhibit more quartz subgrain formation than other samples, as well as deeply sutured boundaries of the bulging grains. These same rocks contain interconnected elongated mica crystals (Fig. 2A), which are internally deformed by C′-type shear zones characterized by larger grain aspect ratios than the other samples (EC-10, EC-13, and EC-14). Generally, strain in the rocks is recorded by the formation of mica-fish structures, mantled albite porphyroblasts, undulose extinction in quartz, and biaxial quartz. Samples EC-9, EC-10, and EC-16 also show crystallographically controlled planar indentations on the bulging quartz boundaries (Fig. 2B).

The six samples that were selected for chemical characterization and 40Ar/39Ar dating show a correlation between mineralogy and degree of deformation (Table 1). We use the term calcite-rich to denote a sample with more calcite than quartz within the total volume of the rock, and quartz-rich to denote a sample with more quartz than calcite. Samples that contain a higher concentration of calcite, especially samples EC-13 and EC-14 (both calcite schists), contain micas that are less deformed than the samples containing more quartz. White micas in these samples are blocky and prismatic clasts, as compared to quartz-rich sample EC-4 (0 vol% calcite), where the deformation is more pervasive (Fig. 2). In these quartz-rich rocks, mica-fish structures are very elongated and strongly sheared, and they partly deform into foliation-parallel mica layers with internal SC/SCC′ fabric. Schist sample EC-9 (35 vol% calcite) also contains elongated micas, but mica-fish structures are less common. Kinked micas were observed mostly in calcite-rich samples (EC-10: 59 vol%, EC-13: 78 vol%, and EC-14: 65 vol%). In those samples, the calcite exhibits curved and tapered twins, which could correspond to type III twinning (Burkhard, 1993; Ferrill et al., 2004). Mica crystals in EC-10 are mostly found in strings or veinlets of chlorite (Fig. 2C) and have a more equant shape (strained but still blocky), whereas small (<50 μm) crystals in the calcite matrix are prismatic. Sample EC-16, which has similar concentrations of calcite and quartz, contains elongated mica crystals, few of which are kinked. Throughout this paper, we use the term “deformed white mica” to denote the strained and elongated crystal shape (high aspect ratio) of the micas, the presence of sheared mica-fish structures, and evidence for internal deformation like SC or SCC′ fabrics. The term “undeformed white mica” is used to characterize the prismatic, blocky (low aspect ratio), and unsheared crystal shape of micas, but minor internal kinking may occur.

The disturbed 40Ar/39Ar spectra presented in the literature for the Western Cyclades may be a consequence of mixed populations of micas. Since more than one microstructural population of mica has already been distinguished (Iglseder et al., 2011; this study), it is important to also characterize variations in mineral chemistry. The chlorite–white mica pairs of the samples in this study are of identical composition to those reported in our previous study (Iglseder et al., 2011), where we calculated mineral pair equilibrium temperatures of 280–400 °C for the rocks on Kea; we therefore infer similar temperatures for our new samples.

After backscatter electron (BSE) imaging with a scanning electron microscope (SEM), mineral chemical compositions were determined in six samples (EC-4, EC-9, EC-10 from Kea; EC-13, EC-14, EC-16 from Serifos; Fig. DR11) using the JEOL JXA-8230 SuperProbe at the University of Ottawa (Ottawa, Canada). Analyses were conducted with an accelerating voltage of 15 kV and probe current of 20 nA to focus and concentrate the beam to a minimum of 5 μm. In total, 4–5 grains of mica per polished section were analyzed by conducting traverses across single grains to explore intragrain chemical variations. Results of the white mica analyses are given in Table 2, and the supplementary material contains the results of the chlorite analyses (Table DR1 [see footnote 1]) and location of traverses (Fig. DR1 [see footnote 1]).

Cation site distribution was calculated according to Vidal et al. (2001) and Parra et al. (2002a), and the end-member decomposition lists the four principal end members. Chlorite analyses exhibiting >0.5% (Na2O + K2O + CaO) and mica analyses exhibiting >0.5% (MnO + TiO2 + Cl) were rejected. Chlorite has compositions varying between clinochlore, daphnite, and Mg-amesite, with clinochlore being the dominant phase, especially in EC-10 and EC-13.

All white micas have more than 3.15 atoms per formula unit of Si, with an average of 3.4 on both islands. White mica has compositions varying between muscovite and Fe-Al celadonite. The end-member decomposition indicates that all samples have an average of ∼77% muscovite, except one sample (low-angle normal fault sample EC-14) that has 66% muscovite and rather more Fe-Al celadonite. Ternary diagrams (Fig. 3A) indicate that samples on both islands yield two mica populations, defined by an Al-rich cluster and a second cluster that has less Al but more Fe, Mg, and K. Figure 3 illustrates the Tschermak and pyrophyllitic substitutions, which also suggests two different populations: one being more evolved toward a more Al-celadonite–rich composition, and the other more muscovite-rich.

In addition to BSE images for all the samples, chemical maps of Fe, Mg, Mn, Na, Ca, Si, O, C, Ti, K, and Al in white mica were produced to better resolve the distribution of the chemical variation within single mica crystals. Mapping was conducted using both the wavelength dispersive spectrometer (WDS) as well as the energy dispersive spectrometer (EDS) of the JEOL JXA-8230 SuperProbe at the University of Ottawa (Ottawa, Canada). An accelerating voltage of 20 kV and a beam current of 200 nA were combined with a spot size of 1 μm to produce a 300 × 600 pixel map with a 20 μm pixel size and 100 ms dwell time on each pixel. The maps of Mg, Fe, and Al, presented for EC-14 and EC-16 in Figure 4 (and the remaining samples in Fig. DR2 [see footnote 1]), attest to a clear and distinct pattern in the major-element chemistry. The rims of the white mica are deformed and contain more Al but less Fe and Mg. There also appears to be patchy zoning in the core of the crystals. Zoning of an Al-rich rim and Al-poor core in micas helps to further demonstrate the two-cluster geochemical pattern exhibited on the white mica ternary diagram (Fig. 3).

Brittle-ductile shear zones are thought to operate as pathways for extensive fluid flow (e.g., Gébelin et al., 2011; Grasemann and Tschegg, 2012; Berger et al., 2013). This fluid could affect the 40Ar/39Ar system by introducing excess argon (40Ar decoupled from its parent 40K) or by facilitating recrystallization and Ar removal. In order to resolve the degree of fluid-rock interaction within the West Cycladic detachment, we analyzed δD on mica, and δ18O and δ13C on coexisting calcite and dolomite. An altered rock appearance (yellow-red color) and abundance of localized mineralized zones (ankerite) suggest the tectonites of the West Cycladic detachment system witnessed fluid infiltration, and the footwall experienced alteration as well, by leaching and hydrothermal activity during and after greenschist-facies metamorphism (Iglseder et al., 2011; Berger et al., 2013).

The isotopic compositions were analyzed at the G.G. Hatch Isotope Laboratories, at the University of Ottawa (Ottawa, Canada). Calcite and dolomite were extracted via a Dremel tool; mica was separated via common mineral separation techniques. Following rock staining to identify carbonate mineralogy, calcite and dolomite δ18O and δ13C values (±0.1‰; Table 3) were determined using a Thermo-Finnigan Gasbench II Delta Plus XP Isotope Ratio Mass Spectrometer (IRMS). Muscovite δD values (±3‰; Table 3) were determined using a Thermo-Finnigan IRMS with the interface Conflo IV, and a Thermo-Finnigan High Temperature Conversion Elemental Analyzer (TC/EA). Oxygen and hydrogen isotope values were measured relative to Vienna standard mean ocean water (VSMOW), and carbon isotope values were measured relative to Vienna Peedee belemnite (VPDB).

Samples from Serifos display the largest variation in oxygen isotopes (∼15‰), from 13.9‰ to 28.9‰. In comparison, samples from Kea show a variation of ∼10‰ in oxygen isotopes, from 19.8‰ to 30.1‰. A conspicuous diminution trend in δ18O can be observed on both islands, though more clearly evident in Serifos samples (Fig. 5), which also exhibit a weaker coupled reduction in O and C isotope values. There is no obvious coupled depletion visible in the results from Kea; a reduction in δ18O is noticeable, but the δ13C values are rather scattered.

On average, samples from the footwall possess a more depleted carbon isotope signature compared to samples from the low-angle normal fault zone on both islands. The results suggest that unmineralized samples and those that are dominated by a metamorphic signature trend toward samples that are more obviously mineralized and commonly contain ankerite (Fig. 5A). There is a visible correlation between the δ18O and δD values, where samples from both islands exhibit a negative trend (Fig. 5B). Kea displays higher δD values than Serifos, and the overall variation of δD is slightly higher on Serifos: ∼7‰, from −66.2‰ to −59‰, compared to ∼5‰ on Kea, from −61.3‰ to −56.1‰.

Five samples that contained large (100–200 μm in width), distinct mica crystals were selected for in situ 40Ar/39Ar dating to explore age variations across single grains and within and between samples (Figs. DR3 and DR4 [see footnote 1]). Rock slabs were initially mounted on a glass slide with cyanoacrylate glue (“superglue”) and ground to 80 μm thickness before polishing. After imaging, the polished slabs were removed from the glass slides by soaking in acetone, then cut into 5 mm squares to isolate the micas of interest, and washed in acetone, methanol, and water before packing into foil packets for irradiation. All samples were irradiated at McMaster University (Hamilton, Canada) and analyzed at The Open University 40Ar/39Ar Laboratory (Milton Keyes, UK). Irradiation flux was monitored using the GA1550 biotite standard with an age of 98.79 ± 0.54 Ma (Renne et al., 1998). The following correction factors were applied to the biotite standards: (39Ar/37Ar)Ca: 0.00065, (36Ar/37Ar)Ca: 0.000265, and (40Ar/39Ar)K: 0.0085 (±0.5%) based on analyses of Ca- and K-salts; only the K-correction was applied to the white mica analyses, since they contain no Ca. Sample J-values (±0.5%) were calculated by linear interpolation between two bracketing standards; a standard was placed between 8 and 10 samples in the irradiation tube (distance ∼2–3 cm). Results were corrected for background concentrations, 37Ar decay (where 37Ar/39Ar > 1), and neutron-induced interference reactions. Background measurements bracketed every sample measurement. Analyses were also corrected for mass spectrometer discrimination, using a value of 295 based on long-term analysis of the instrumentation. The decay constant of Steiger and Jäger (1977) was used in order to compare values with other published data from the Cyclades; this may introduce a systematic age bias of up to 1% (Renne et al., 2010). One sample (EC-13) of the original six that was characterized in terms of its mica chemistry deteriorated during sample preparation and irradiation, and we could not conduct 40Ar/39Ar analyses.

Samples were loaded into an ultrahigh-vacuum laser port and placed under a heat lamp for 8 h to reduce atmospheric blank levels. In situ mica analyses were achieved using a Nd-YAG 213 nm ultraviolet (UV) laser coupled to a Nu Noblesse gas spectrometer with an ion-counting detector. Depending on the volume of the gas released, UV analysis consisted of the ablation of a 40-μm-diameter spot or ablating 100–200-μm-long transects with a 40-μm-diameter beam, both for ∼180 s, followed by 120 s of gettering time, before the gas was admitted into the mass spectrometer. Gases were gettered using two SAES getters (at 450 °C and room temperature). Peaks from 36Ar to 40Ar were scanned 10 times each, and amounts were extrapolated back to the inlet time. Data were reduced using an in-house software package (ArMaDiLo) developed by James Schwanethal.

Similar to other small-volume noble gas studies, sample and background 36Ar measurements approached detection limits and were commonly within error of each other. The correction for atmospheric argon magnifies errors on the 36Ar measurement and results in anomalously high analytical errors on the final 40Ar/39Ar age (e.g., Sherlock et al., 2005). Samples were only corrected for atmospheric argon where the 36Ar measurement was more than twice the background signal (Table 4). The uncertainty on the calculated age for uncorrected samples was doubled in compensation (e.g., Sherlock et al., 2005, 2008; Warren et al., 2011, 2012). The isotope measurements are reported to 1σ uncertainty; the calculated ages are reported to 2σ.

Each sample yielded a range of ages (Table 4) that differed between the islands (Fig. 6). Isotopes 38Ar produced from Cl and 37Ar from Ca can both be useful monitors of intergrowths of biotite and intergrowths of chlorite or contamination with calcite, respectively. Measured concentrations of 38Ar and 37Ar were very low and generally showed no correlation with age (Fig. DR5 [see footnote 1]). Higher levels of 37Ar were commonly measured in the calcite-rich samples and suggest that the laser beam ablated some calcite during analysis. The ages were corrected for 37Ar only when counts were >1500 (Table 4), because the ages below this limit are indistinguishable from those with negligible 37Ar.

On Kea, all three samples yielded a similar distribution of single spot ages (Fig. 6): EC-4 with ages between 35.3 ± 0.3 Ma and 16.3 ± 0.3 Ma (n = 24), EC-9 with ages between 35.0 ± 0.6 Ma and 19.1 ± 0.5 Ma (n = 9), and EC-10 with ages between 35.3 ± 0.3 Ma and 21.3 ± 0.7 Ma (n = 8). The youngest ages were yielded by sample EC-4, a quartz mylonite boudin within a marble ultramylonite (Fig. 2), with a few young single ages from EC-9 (schist), as well. The weighted average 40Ar/39Ar ages of these results yield late Oligocene to Miocene dates, similar to the step-heated 40Ar/39Ar results of Iglseder et al. (2011).

In order to detect potential core-rim age differences in the micas, transect and spot ages were measured along the rims and within the cores and tips of large grains from EC-4 and EC-9 (Table 4). This approach yielded scattered results; within a single grain, the oldest and youngest single spot ages, and the mean core and rim ages, are within error. Moreover, the range and mean of ages are similar to analyses yielded from small grains (ca. 18–22 Ma, with a mean of ca. 21 Ma).

As expected, mica ages on Serifos are overall older than those on Kea (Fig. 6). Thirty-five single spot analyses from EC-14 yielded ages ranging from 57.5 ± 0.7 Ma to 21.6 ± 0.4 Ma, and eight analyses of EC-16 yielded ages ranging from 59.6 ± 0.5 Ma to 21.7 ± 0.4 Ma. Notably, the rock containing more calcite (EC-14: 65 vol% calcite) preserved older ages than the rock with less calcite (EC-16: 26 vol% calcite). The weighted mean age of each sample is similar to the late Eocene step-heated 40Ar/39Ar ages reported for Serifos previously (Schneider et al., 2011).

Age Populations

Our overall range of in situ ages yielded by the different samples is broadly consistent with previously published step-heating data (Maluski et al., 1987; Wijbrans et al., 1990; Baldwin, 1996; Tomaschek et al., 2003; Zeffren et al., 2005; Iglseder et al., 2011; Schneider et al., 2011; Grasemann et al., 2012; Bröcker et al., 2013). However the range of 40Ar/39Ar ages yielded from each sample in this study requires interpretation, and the in situ nature possibly offers some insight into the complex spectra pattern in other studies. A probability distribution plot of the ages from Kea (Fig. 6A) shows a distinct population at 18.4 ± 0.5 Ma (n = 13, samples EC-4 and EC-9) with older single spot ages between 20 and 35 Ma that do not define a statistical age population. There is a positive correlation of older age preservation and higher calcite content; for example, EC-10 is a calcite schist with poor schistosity and 59 vol% of calcite that preserves dominantly Oligocene ages. As noted, we observe no clear relationship of age with microstructural position (mica core, rim, tip), although a few grains in EC-4 yielded younger ages in the crystal tips. This spatial variability is consistent with previous reports of heterogeneous 40Ar distributions in naturally deformed muscovite (e.g., Kramar et al., 2001; Mulch et al., 2002; Bröcker et al., 2004; Mulch and Cosca, 2004; Cosca et al., 2011). The dominant ca. 18 Ma age population is consistent with the ages yielded by the “type 2” spectra of step-heated samples from Kea (Iglseder et al., 2011).

A probability distribution plot of the ages from Serifos (Fig. 6B) illustrates the same relationships as those from the Kea data set. In sample EC-14, two age populations can be resolved based on microstructural location of the spot analyses. Deformed portions of grains and whole deformed grains yield younger ages than undeformed portions of grains and whole undeformed grains (Fig. 6B; Table 4; Figs. DR3 and DR4 [see footnote 1]). Fifteen analyses of deformed white mica from EC-14, commonly elongate mica-fish structures and sheared or kinked rims, yield a mean age of 36.7 ± 2.5 Ma, whereas 15 analyses of undeformed white mica, generally prismatic and equant grains, or prismatic cores, yield a mean age of 45.4 ± 1.2 Ma. In several grains, the prismatic core is older than the prismatic rim; however, since both are prismatic and not sheared, they are considered as undeformed grains. Considering the ages from the undeformed mica in both samples, a dominant population at 44.1 ± 1.0 Ma (n = 17) is calculated.

Although clear age populations are apparent, the Kea data preserve a hint of the ca. 36 Ma deformed mica age population that is obvious on Serifos; the Serifos data record a trace of the Miocene age population that is present on Kea. Next, we discuss interpretation of the age populations and ranges in terms of fluid-rock interaction (as evidenced by stable isotopes), timing of original metamorphic crystallization (as evidenced by major-element data), and timing of deformation-induced (neo)crystallization (as evidenced by microstructural analyses and lithology).

40Ar/39Ar Ages: Crystallization, Cooling, or Contamination?

White mica 40Ar/39Ar dates from metamorphic and deformed rocks may either represent the timing at which the mica cooled through a nominal closure temperature, Tc (if the grains crystallized at T >> Tc; e.g.; Dodson, 1973), the timing of mica crystallization (if T < Tc), or a “contaminated” date that is not geologically meaningful. Excess 40Ar, which is 40Ar decoupled from parent 40K, resulting in an 40Ar/39Ar age older than the “true” age, may be incorporated into newly formed grains as they crystallize or may diffuse into grains if the grain boundary concentration and temperature are high enough. In many cases, it is difficult to determine the extent of any excess 40Ar contamination, yet it is important to do so in order to link the ages to crystallization or cooling “events.”

In the Western Cyclades, neither the rocks as a whole nor the micas in particular reached temperatures >450 °C (maximum Tc for white mica; Harrison et al., 2009) during the latest episode of ductile extension (Iglseder et al., 2011; Grasemann et al., 2012). The 40Ar/39Ar age ranges therefore represent the timing of white mica (neo)crystallization or reflect variable contamination with excess Ar.

Mica neocrystallization during deformation may lead to significant grain-hosted 40Ar removal or gain, depending on the local fluid 40Ar concentration. Midcrustal shear zones are commonly considered to operate as open systems where fluid residence time is relatively short compared to the time scales of 40K decay and mineral recrystallization reactions (Fourcade et al., 1989; Dipple and Ferry, 1992). Fluids circulating through these shear zones during deformation should promote local removal of Ar and limit its availability for uptake into newly crystallizing grains, as long as the local fluid Ar concentration is low. Any grains crystallizing under these fluid conditions at T < Tc should therefore yield 40Ar/39Ar ages that are representative of the timing of deformation-induced (neo)crystallization (Sanchez et al., 2011). (If this fluid Ar concentration is high, such as fluids from the dehydration of older metamorphic rocks that possess higher 40Ar, there could be excess 40Ar.) Conversely, low fluid volumes or low interconnectivity could lead to high fluid Ar concentrations. This would hinder local Ar removal and encourage incorporation of excess 40Ar into the grain during crystallization or lead to 40Ar becoming trapped in defects, cleavages, or along grain boundaries. For the purposes of interpreting the mica 40Ar/39Ar ages, it is therefore critical to determine whether fluid was present and mobile in the sample during deformation and mica (neo)crystallization.

Evidence for Fluid-Rock Interaction during Deformation

Stable isotope geochemistry is an excellent tool for tracking fluid-rock interaction, as the isotopic signature of rocks can be modified during thermal events and hence may be indicative of devolatilization and fluid flow. The main objectives in collecting stable isotope data during this study were to determine (1) whether the variation in isotopic values was caused by fluid-rock interaction, and (2) whether this occurred during deformation.

Carbonate rocks typically preserve δ18O values ranging between 21‰ and 32‰, and δ13C values ranging between −1‰ and 4‰ (Hoefs, 1997). The calcite and dolomite in Kea samples thus yielded protolith-like δ13C signatures (Fig. 5A), with values similar to massive marbles on other islands of the Cyclades (e.g., Kreulen, 1988; Matthews and Schliestedt, 1984). However, the ∼10‰ variation in δ18O (Fig. 5B) suggests that this isotopic signature has been modified. The δ18O values from Serifos samples, which are lower than the Kea samples, exhibit a larger (∼15‰) variation, with values shifted more toward primary magmatic water values (Fig. 5B). Devolatilization is assumed to only contribute a variation of ∼1‰ (Hoefs, 1997), a much lower amount than the variation observed in δ18O values on Kea and Serifos. The data are therefore suggestive of other processes, such as moderate fluid-rock interaction or primary lithological control.

In addition to lower δ18O values, both islands show depletion in mica δD values and provide direct insight into the nature of fluid-mica interaction during (neo)crystallization. The different variations in δ18O and δD may indicate variable amounts of rock buffering with fluid interaction, where the reservoir of δ18O in calcite/dolomite is more affected by this process than the reservoir of δD in mica. The narrow range of δ13C values in marbles may be a consequence of a weaker fluid-rock interaction that did not completely modify the isotopic system, thus reflecting more or less the primary lithology.

The trend toward more negative values in O and H isotope systematics likely implies an influx of either magmatic/metamorphic water or meteoric water; our data cannot discriminate between these two water compositions. We can, however, examine the local geology and hypothesize its consequences to our data. There is evidence of widespread but localized leaching and hydrothermal activity that resulted in locally intense ankeritization, dolomitization, and silicification of carbonate rock on Kea (Iglseder et al., 2011). The range of δ18O in the marbles, especially from Serifos, is consistent with the effects of the ca. 10 Ma plutons that intruded the West Cycladic detachment system on Serifos and at Lavrion, resulting in Fe-Mn-Pb-Ag-Zn mineralization along the ductile to brittle detachment surfaces (Salemink, 1980; Grasemann and Tschegg, 2012; Berger et al., 2013). Moreover, Siebenaller et al. (2013) demonstrated that during the exhumation of the Naxos metamorphic core complex, fluids in the ductile zone were dominated by magmatic and metamorphic composition, whereas those in the brittle zone were dominated by a meteoric composition. Despite the strong evidence for Miocene magmatically derived fluids, we cannot rule out fluid-rock interaction under greenschist-facies conditions during Miocene exhumation. However, because of the preserved ductile structure of our rocks, we do not think that our stable isotope data are significantly impacted by a meteoric signal.

The deformation mechanisms observed in the recrystallized quartz of our samples are further evidence of fluid-rock interaction during ductile deformation (inset Fig. 2B). Planar indentations on the bulging grains were probably formed by water on the migrating boundaries (Mancktelow and Pennacchioni, 2004). This fast grain boundary migration mechanism indicates that the recrystallization took place in wet quartz, whereas subgrain rotation and slow bulging are more likely to occur in dry quartz. Thus, on the basis of the deformation features displayed by the quartz and the moderate stable isotopic variation of our samples, we conclude that fluids interacted with the West Cycladic detachment system rocks during ductile deformation, possibly more so in quartz-rich rocks than in calcite-rich rocks. This logic provides evidence to suggest that the 40Ar/39Ar ages represent the timing of (neo)crystallization events rather than variable contamination with excess 40Ar.

Links Among Compositional Variation of White Mica, Deformation, and 40Ar/39Ar Ages

The composition of the white mica is noticeably variable in all samples, and it is likely indicative of changing P-T conditions during neocrystallization events. Despite this, the mica composition data broadly show two distinct mica populations: (1) an Al-celadonite–rich composition, and (2) a muscovite-rich composition (Fig. 3), both clearly identified by chemical mapping (Al-rich, Fe-Mg–poor rims; Fig. 4; Fig. DR2 [see footnote 1]). The Tschermak substitution produces micas that vary in composition between ideal muscovite and (theoretical) Al-celadonite (Parra et al., 2002a). The micas described here contain lower silica concentrations, but they follow the Tschermak substitution trend, suggesting that the overall decrease in Si is due to a different substitution reaction. The data also demonstrate that a moderate pyrophyllitic substitution also occurs (Fig. 3).

The Al-celadonite component, commonly found our mica cores, may be a relict from the M1 peak pressure event (cf. Vidal and Parra, 2000; Parra et al., 2002a; Iglseder et al., 2011). We did not specifically focus on collecting samples that preserved the highest-pressure metamorphic remnants. We note, therefore, that the Al-celadonite compositions in these samples may reflect bulk rock composition rather than crystallization at peak pressure (Bröcker et al., 2004). The muscovite-rich composition recorded in our micas most likely formed during the M2/D2 greenschist-facies overprint.

Al-celadonite is most abundant in Serifos calcite schist EC-14, one of the samples with the least-deformed muscovite, despite being sampled from the low-angle normal fault zone. This sample contains two age populations based on microstructures, at ca. 36 Ma and ca. 45 Ma. Mica grains in Serifos schist EC-16, which has considerably less calcite than EC-14, are more deformed (Fig. 2E), exhibit a greater degree of neocrystallization, and possess a composition closer to ideal muscovite. Micas in this sample yield ages that lie within the two populations defined by EC-14, and more young ages between 21 and 26 Ma.

Kea sample EC-4, effectively the most deformed sample, yields only mica of muscovite-rich composition (Fig. 3B) with no evidence of any remnant Al-celadonite. This sample yields the youngest ages, with a dominant ca. 18 Ma population. Such correlation between age and mineral composition has been documented in neocrystallized mica (e.g., Mulch and Cosca, 2004). In Kea sample EC-9, mica proximal to calcite yields older ages (33 and 35 Ma), although the two-dimensional view of the slabs may be misleading with regards to the mineral surrounding the mica crystals (Fig. DR4B [see footnote 1]). The other, younger ages in EC-9 are all surrounded by quartz. In all the quartz-rich samples, chemical composition clearly relates to visible grain deformation: A decrease in Al-celadonite concentration is measured toward the deformed rims.

The data therefore suggest that mica partially or totally recrystallized with a new composition in zones of localized deformation. Chemical equilibrium was maintained by neocrystallizing grains with a different composition rather than re-equilibrating old grains via intracrystalline diffusion (e.g., Worley et al., 1997; Vidal and Parra, 2000; Parra et al., 2002b).

Lithological Control on Mica (Neo)Crystallization during Deformation

The amount of strain recorded by the mica appears to be directly associated with the amount of calcite in the samples, and it appears to exert a direct control on the 40Ar/39Ar age. The relative impermeability of calcite-rich rocks and especially the rheology of calcite-mica aggregates versus quartz-mica aggregates under brittle-ductile conditions may provide an explanation. A synoptic diagram (Fig. 7) shows these rheology factors in relation to the mica compositions and ages.

Several geochemical and experimental studies have demonstrated that grain boundaries in calcite from marbles are impermeable to carbon isotopes but permeable to oxygen isotopes, and that water-rich fluids interact with marbles mainly along calcite grain boundaries rather than within the calcite grains (e.g., Arita and Wada, 1990; Wada et al., 1998; Schumacher et al., 2008). Therefore, fluidborne carbonate ions are unlikely to penetrate into the calcite, thus limiting carbon diffusion. This may explain the narrow range in our δ13C values. In the Cyclades, the marbles are intercalated with mica schists. On Tinos, Ganor et al. (1991) demonstrated a very low permeability perpendicular to the marble-schist contacts but enhanced layer-parallel permeability within the schists. The calcite-rich rocks are therefore relatively impermeable, especially if they have poor schistosity, and probably experienced less fluid-assisted recrystallization than the quartz-rich rocks. This lower permeability has implications for the evacuation of Ar during deformation-induced mica recrystallization. It is also conceivable that Ar solubility is different (maybe lower?) in calcite-rich rocks, if the CO2 content in the pore fluid is higher than in quartz-rich rocks. Argon solubility has been compared to CO2 solubility in fluids, which is lower than H2O’s solubility (Carroll and Stolper, 1993; Kelley, 2002).

Calcite crystals in EC-10, EC-13, and EC-14 are larger than those in EC-9 and EC-16, and they show curved twins (Fig. 2). The latter two samples also contain more quartz than calcite. Bulging quartz grain boundaries in these samples suggest that neighboring calcite crystals underwent recrystallization in the presence of fluids, resulting in a smaller crystal size.

The mica in calcite-rich samples, although it is apparent it experienced strain such as minor kinking, probably did not undergo significant recrystallization, as curved calcite twins are still preserved in the surrounding matrix. Under greenschist-facies conditions, calcite is weaker than the packets of mica, especially when mica-fish in “hard orientations” effectively behave as rigid clasts (Etchecopar, 1977), thus responding to differential stresses like an interconnected weak layer localizing strain in the calcite matrix (Handy, 1990). High-strain experiments of calcite-muscovite aggregates demonstrated that once a foliation in a rock has developed, strain is concentrated within calcite shear zones rather than within the mica (Delle Piane et al., 2009). Under these high-strain conditions, dislocation creep in calcite controls deformation. This is consistent with the observed type III twins (Burkhard, 1993) in our samples, which occurred under synmetamorphic deformation above 200 °C, by intracrystalline deformation mechanisms (r- and f-glide slip systems).

In contrast, quartz in quartz-rich schists may form a load-bearing framework (Handy, 1990) with a more complex apparent rheological contrast to the highly anisotropic mica packets. Although mica-fish can still stabilize in “hard orientations” (Etchecopar, 1977), we observe that mica-fish are boudinaged into highly elongated shapes, which are eventually sheared parallel to the cleavage planes recording internal SC/SCC′ fabrics. In fact, the deformed micas look geometrically very similar to model results of Dabrowski et al. (2012), where weak inclusions of a two-phase composite in simple shear localize strain in shear zone–parallel layers and as strongly elongated fish-type sigmoids resembling SC/SCC′ microstructures. We therefore speculate that the highly anisotropic mica packages in favorable orientations may be weaker than the quartz matrix. Recrystallization in the Cyclades samples appears to have additionally been enhanced by fluid availability, as indicated by stable isotope and quartz microstructure data.

In summary, deformation in the white micas in the West Cycladic detachment system samples was promoted in the quartz-rich rocks and stunted in the calcite-rich rocks, as suggested by our cumulative data. We suggest that white micas in deformed quartz-rich rocks are more likely to yield 40Ar/39Ar ages that relate to the last-recorded phase of deformation-induced crystallization. In contrast, older, predeformation ages are more likely to be recorded by white micas in calcite-rich rocks. Due to their low permeability, however, the likelihood of 40Ar inheritance during recrystallization is potentially higher in calcite-rich rocks, because of inefficient Ar removal from the local grain boundary network.

Tectonic Interpretation of Ages

On Kea, quartz-rich sample EC-4 yields micas exhibiting the strongest deformation and the youngest age population. We therefore interpret the ca. 18 Ma population to constrain the timing of the latest stages of D2 ductile extension. This interpretation is in agreement with the oldest zircon and apatite (U-Th)/He ages from the West Cycladic detachment system (Grasemann et al., 2012) and other suggestions for the timing of the greenschist-facies overprint on Sifnos at 18.9 ± 0.3 Ma (Wijbrans et al., 1990) and Naxos between 15.0 and 19.8 Ma (Wijbrans and McDougall, 1988).

Due to the complete neocrystallization of the micas on Kea, no evidence for high-pressure Al-celadonite micas remain, and reliable age information from earlier events is considered unresolvable. Kea’s older ages, ranging from 20 to 35 Ma, are recorded by micas that are not Al-celadonite-rich, and these ages cannot therefore be interpreted as high-pressure relics. Rather, these ages may instead be interpreted as preserved artifacts of an earlier deformation event such as the Cycladic blueschist unit wedge deformation (extrusion) during the Eocene. Iglseder et al. (2011) and Grasemann et al. (2012) argued that deformation along the West Cycladic detachment system would have migrated to shallower levels through the Miocene, ultimately defining a sharp brittle detachment. These older ages may suggest the rock witnessed progressive deformation as it was exhumed to shallower levels.

On Serifos, the 40Ar/39Ar data suggest two episodes of Eocene-age deformation, with only weak evidence for the Miocene deformation that is prevalent on Kea. The Serifos samples yield a strong ca. 44 Ma age signature from undeformed (portions of) mica. This age, associated with high Al-celadonite content and little-to-no mica deformation, could reflect an Eocene thermal event, postulated from ca. 40 Ma zircon rims on Serifos (Schneider et al., 2011). The concordance of U-Pb, Rb-Sr, and 40Ar/39Ar ages at ca. 40 Ma in the Cyclades provides cumulative supporting evidence for a regional event at that time (e.g., Altherr et al., 1979; Putlitz et al., 2005; Schneider et al., 2011). The five older ages of ca. 55–59 Ma may be the result of excess 40Ar, although no additional chemical evidence for this is recorded. In addition, these ages lie within the spread of published 40Ar/39Ar ages that are interpreted as recording the timing of the M1 event in the Cyclades (e.g., Maluski et al., 1987; Tomaschek et al., 2003; Putlitz et al., 2005).

The separate and distinct Serifos age population of ca. 36 Ma, resolved from more-deformed parts of grains, is similar to previously published multigrain step-heating 40Ar/39Ar ages from tectonite samples (Schneider et al., 2011). These ages could represent a maximum age for the Miocene (D2) ductile extension, since the dated mineral phases are muscovite-rich and did not crystallize under high-pressure conditions. This timing, however, is significantly older than the ca. 18 Ma M2/D2 recorded on Kea. If deformation occurred synchronously on Serifos and on Kea, then the ca. 36 Ma dates on Serifos record incomplete resetting of the Ar system as the ca. 44 Ma micas deformed and recrystallized at ca. 18 Ma. This explanation is consistent with reduced permeability of calcite-rich rocks during deformation. Alternatively, and preferably, this age could represent a minimum timing for the end of M1/D1 extrusion under low-pressure conditions (i.e., end of D1 ductile deformation), effectively the transition between M1/D1 and M2/D2.

In summary, our in situ data suggest that the previously published complex step-heating spectra from Kea and Serifos are likely the result of mixtures of heterogeneously deformed and neocrystallized micas, consistent with published experimental and natural demonstrations of mixed mica generations (e.g., Wijbrans and McDougall, 1986; Tomaschek et al., 2003; Bröcker et al., 2004; Di Vincenzo et al., 2006; Kula et al., 2010; Cosca et al., 2011). Equally, the flat plateaus previously reported from the more highly strained rocks (Schneider et al., 2011; Iglseder et al., 2011) are explainable as being due to the presence of one dominant mica composition, and hence one age population, as the micas have been reset by deformation and neocrystallization.

The variation of in situ 40Ar/39Ar ages yielded by white micas from Kea and Serifos of the West Cyclades detachment system (ca. 60–16 Ma) is linked to the preservation of at least two distinct mineral chemical populations, which are themselves related to the rheology of the host rock. One mica population is enriched in Al-celadonite and may have been inherited from the high-pressure metamorphism event (M1). The second mica population is enriched in muscovite as a consequence of the Tschermak substitution; this likely occurred during neocrystallization in the course of exhumation-related ductile deformation. Neocrystallization of the white micas was localized due to mineralogical-dependent strain partitioning and was enhanced by fluid availability within the brittle-ductile shear zone. Quartz-rich rocks contain highly deformed micas that show complete neocrystallization and muscovite-rich compositions. In these rocks, the quartz formed a load-bearing framework that transferred the strain to the micas, resulting in either highly elongated mica-fish or foliation-parallel mica layers with SC/SCC′ type microstructures. In contrast, the calcite-rich rocks contain micas that only partially neocrystallized during deformation and, hence, retain both mica compositions. Less-interconnected fluid pathways also appear to have limited mica recrystallization.

One dominant, deformation-related population of 40Ar/39Ar ages was resolved from the Kea quartz-rich samples, and it is ca. 18 Ma. Undeformed white micas from calcite-rich samples collected on Serifos yielded a mean 40Ar/39Ar age of ca. 44 Ma, whereas deformed and neocrystallized mica rims and whole small grains yielded younger ages of ca. 36 Ma. Micas that were not completely neocrystallized (as shown by their chemistry) and that yielded the older ca. 44 Ma ages are interpreted as portions of micas that originally crystallized during Eocene metamorphism under relatively high pressure. The younger Eocene age most likely represents neocrystallization of mica under low-pressure conditions during the transition from crystalline wedge extrusion to the onset of ductile thinning and extension.

Funding for this research was provided by a Natural Sciences and Engineering Research Council of Canada Discovery grant (to Schneider). We would like to thank Paul Middlestead and Marie-Pierre Varin (Ottawa) for assistance with the stable isotope analyses, Glenn Poirier (Ottawa) for guidance with the scanning electron microscope and electron microprobe analyses, and Sarah Sherlock and Alison Halton for assistance in The Open University Ar-Ar laboratory. Discussions with Ben Huet, Hugh Rice, and Marcin Dabrowski are also greatly appreciated. Thoughtful, thorough, and constructive reviews on the current paper were provided by N. Mancktelow and M. Cosca, and comments provided by D. Chew, G. Di Vincenzo, Y. Rolland, and M. Bröcker on previous versions of this manuscript improved the clarity of our arguments.

1GSA Data Repository Item 2015079, SEM-BSE images, wavelength dispersive spectroscopy chemical maps, photos of dated rock slabs, and isotope diagrams, is available at, or on request from, Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301-9140, USA.