Abstract

Current models for Miocene backarc extension of the Aegean region generally suggest that stretching was accommodated mainly by NE-dipping low-angle normal faults with N to NE sense of shear. A crustal-scale low-angle normal fault system trending over a length of more than 200 km forms the North Cycladic detachment system, which records a NE-directed normal shear sense separating the Cycladic Blueschist unit in the footwall from the Upper Cycladic unit in the hanging wall. Based on new structural field data, we propose the existence of another large-scale low-angle normal fault system, the West Cycladic detachment system, which is exposed on Kea, Kythnos, and Serifos, strikes over a length of at least 100 km, and has a possible extension to the SE, where the existence of a South Cycladic detachment system has been recently postulated. The West Cycladic detachment system shares many similarities with the North Cycladic detachment system, with the notable exception that the structure dips toward the SW with top-to-the-SSW kinematics. New 40Ar/39Ar and U-Th/He thermochronological data suggest that the West Cycladic detachment system accommodated extension throughout the Miocene. Since both the North and the West Cycladic detachment systems were active until the late Miocene but exhibit opposing shear sense, we propose that a large part of the stretching of the Aegean crust was accommodated by these two bivergent crustal-scale detachment systems.

INTRODUCTION

The pattern of lithospheric extension, whether it is symmetric or asymmetric, is a key factor in understanding of the dynamics of rifts, passive margins, backarc extension, and metamorphic core complex formation. Numerous kinematic models suggest either symmetric extension under pure shear far-field strain (McKenzie, 1978) or asymmetric simple shear (Wernicke, 1985; Govers and Wortel, 1993), or combinations of both models (Lister et al., 1986). More complex dynamic models are able to predict both symmetric and asymmetric extension, emphasizing the importance of the initial geometry, preexisting anisotropies, thermal histories, rheology, and shear localization (e.g., Bertotti et al., 2000; Huismans and Beaumont, 2002; Gessner et al., 2007; Regenauer-Lieb et al., 2008). In order to compare such models with natural examples of crustal extension, researchers have been investigating the Cyclades of the Aegean region for the past three decades (e.g., Lister et al., 1984; Buick, 1991; Gautier and Brun, 1994; Tirel et al., 2009; Huet et al., 2011), where Miocene backarc extension has led to the formation of several Cordilleran-type metamorphic core complexes (Lister et al., 1984), which were exhumed along low-angle normal faults. The timing, orientation, and kinematics of the low-angle normal faults, however, have been a matter of debate (cf. Jolivet and Brun, 2010; Ring et al., 2010), with a general consensus that the Miocene low-angle normal faults dip toward the N-NE, recording a top-to-the-N or -NE sense of shear (e.g., Avigad and Garfunkel, 1989; Buick, 1991; Faure et al., 1991; Lee and Lister, 1992; Gautier et al., 1993; Gautier and Brun, 1994; Jolivet and Patriat, 1999; Vanderhaeghe, 2004; Mehl et al., 2007). Recently, it has also been suggested that the low-angle normal faults in the northern Cycladic islands link up to form the crustal-scale North Cycladic detachment system, with a strike length of more than 200 km (Jolivet et al., 2010). Other models favor a more complex evolution, suggesting that an earlier Oligocene S-directed sense of shear has been overprinted by Miocene N-directed shear sense (Forster and Lister, 2009), or Oligocene N-directed shear, which has been overprinted by Miocene S-directed shear (Ring et al., 2011). Structural data from Serifos, located in the western Cyclades, document the S-directed kinematics of a low-angle normal fault, which partly cuts a ca. 10 Ma late-tectonic granodiorite intrusion (Grasemann and Petrakakis, 2007; Iglseder et al., 2009; Tschegg and Grasemann, 2009; Brichau et al., 2010). Complementary studies on other western Cycladic islands such as Kea to the north (Iglseder et al., 2011) have also documented pervasive S-directed kinematics across the island, but the regional correlation of these fault systems, their lateral extent, and the onset of extension are unknown. In this contribution, we present, for the first time, evidence for a top-to-the-SW detachment on the island of Kythnos, which is located between Kea and Serifos. We also complement the existing database in the western Cyclades with new 40Ar/39Ar mica and (U-Th)/He zircon and apatite thermochronological data combined with structural field data from the islands of Serifos, Kythnos, and Kea. The objective here is to expand upon the detailed island-specific investigations and integrate those data and observations with the recent and succinct summaries of Jolivet and Brun (2010) and Ring et al. (2010) and with recent observations of major detachment systems (Jolivet et al., 2010; Ring et al., 2011). Our new data allow us to suggest that the S-directed kinematics of the low-angle normal faults identified on the islands are mechanically linked and form the West Cycladic detachment system, which, together with the North Cycladic detachment system, accommodated Miocene bivergent extension in the Aegean. The existence of a South Cycladic detachment system (Ring et al., 2011) is critically discussed as a possible extension of the West Cycladic detachment system.

GEOLOGICAL SETTING

The Cyclades are part of the Hellenides, which are made up of a tectonic pile of several thrust units, consisting of, from top to base (and north to south), the Pelagonian, Pindos, Gavrovo-Tripolitza, Phyllite-Quartzite, and Ionian thrust units (Jacobshagen, 1986; Bonneau, 1984; van Hinsbergen et al., 2005). The dominant sequence is the Cycladic Blueschist unit, which is considered to be a paleogeographic equivalent of the Pindos unit (Bonneau, 1982), and which consists of a polymetamorphic Carboniferous–Permian to latest Cretaceous passive-margin sequence that tectonically overlies the Cycladic Basement unit, a Paleozoic basement diversely metamorphosed and intruded by Triassic granitoids (e.g., Dürr et al., 1978; Blake et al., 1981). In the Lavrion area and in some tectonic windows of the Cyclades, the Basal unit, which represents a remnant of a Late Triassic to Late Cretaceous carbonate platform (Avigad and Garfunkel, 1989), is exposed below the Cycladic Blueschist unit. The Basal unit shows evidence of high-pressure–low-temperature metamorphism (Shaked et al., 2000). Early Miocene isotopic ages have been interpreted as indicating the timing of the high-pressure metamorphism or of the greenschist-facies overprint (Bröcker and Franz, 1998; Ring et al., 2001; Ring and Reischmann, 2002; Bröcker et al., 2004). The Basal unit is considered to be part of the External Hellenides (Avigad et al., 1997).

The rocks of the Cycladic Blueschist unit were exhumed in two successive stages corresponding to two metamorphic episodes (Jolivet and Brun, 2010, and references therein), with the first stage occurring during Eocene Hellenic subduction, exhuming blueschist- and eclogite-facies assemblages (e.g., Altherr et al., 1979; Wijbrans and McDougall, 1988; Wijbrans et al., 1990; Bröcker et al., 1993; Trotet et al., 2001; Ring and Layer, 2003), likely in an extrusion wedge (Ring et al., 2007, 2010; Huet et al., 2011). During subsequent tectonism in the Oligocene and Miocene, exhumation occurred as Cordilleran-type metamorphic core complexes (Lister et al., 1984) with mainly NE-dipping low-angle normal faults (i.e., “detachments”) showing a dominant top-to-the-N or -NE sense of shear (e.g., Buick, 1991; Gautier et al., 1993; Vanderhaeghe, 2004). Latest extension in what is now the backarc region developing above the retreating African slab (Le Pichon et al., 1981) generated middle to late Miocene, predominantly I-type plutonism on several of the islands where final intrusion stages are syn- to postkinematic with respect to the pervasive deformation fabric (Altherr et al., 1982; Keay et al., 2001; Pe-Piper and Piper, 2002; Iglseder et al., 2009). Although most of the hanging wall of the low-angle normal faults has been eroded, isolated klippen are preserved on several Cycladic islands. These rocks, which are part of the Pelagonian unit, consist of a Paleozoic basement with a Paleozoic and Mesozoic carbonate cover overlain by a Jurassic ophiolite (Bonneau, 1982), and they have locally endured a Late Cretaceous high-temperature metamorphic event (e.g., Reinecke et al., 1982; Altherr et al., 1994; Zeffren et al., 2005). Next, we briefly describe the geology, new structural field observations, and new thermochronological data from the islands of Kea, Kythnos, and Serifos in the western Cyclades (Fig. 1).

Kea

The rocks on Kea, the northwesternmost of the Cycladic islands, consist mainly of greenschists, quartz-rich mica schists, and horizons of marbles. Based on the rare preservation of glaucophane, Kea has been suggested to be part of the Cycladic Blueschist unit (Davis, 1982; van der Maar and Jansen, 1983), and in scarce, isolated blueschist lenses, a relic mineral lineation (lm1) trending roughly W-E is preserved (Iglseder et al., 2011). The island was strongly overprinted by SW-directed shearing under greenschist-facies conditions, forming a strong stretching lineation (lm2) trending NE-SW (Fig. 1A). Pervasive foliation (sm1+2) across the island defines a dome-like structure with several meter-long subhorizontal shear zones, with C′-type and shear band fabrics indicating top-to-the-SW shearing. High-angle, SW-dipping ductile-brittle faults display a strong displacement gradient, resulting in ductile drag of the main foliation. The high-angle faults mechanically interact with subhorizontal shear zones indicating SW-directed shear sense (Fig. 2A). Discrete horizons of marbles reveal that the whole sequence is intensely folded and sheared, recording sheath folds at smaller scales (Fig. 2B) and forming type 2 and 3 refolded structures at larger scales. Open upright folds with fold axes parallel to lm2 but with subvertical NE-SW–trending axial planes overprint sm1+2 and are responsible for the dome shape of the island. The fold limbs are partly overprinted by the SCC′ fabric; therefore, the upright folds indicate a shortening component perpendicular to the extension direction also described on other Cycladic islands (Avigad et al., 2001).

At structurally higher levels, the SCC′ fabric in phyllonites becomes more intense and is gradually overprinted by cataclastic deformation. Above the cataclasite zone, an up to 50-m-thick ultramylonitic marble discordantly overlies the greenschists, appearing as isolated klippen, which can be traced over the island (Fig. 2C). Observations of sheath folds and pervasively ultrafine-grained (<10 μm) fabric suggest that the marble ultramylonites are in fact a ductile high-strain zone. Our interpretation is supported by abundant fold axes of isoclinal folds that are parallel to the mylonitic stretching lineation (lm2), with the axial planes parallel to the mylonitic foliation (sm1+2). Such folds have been shown to form during shearing and layer-parallel thinning of the whole sequence, indicating high-strain shear deformation (e.g., Morales et al., 2011). Although the ductile strain is localized in the marbles under greenschist-facies conditions, the marble ultramylonites likely behaved as an extremely rigid slab during exhumation through the upper crust, where cataclastic deformation is localized on both the upper and lower contacts (Fig. 2C). Polyphase cataclasites are commonly overprinted by dissolution-precipitation creep and vice versa. Ultracataclasites with rounded and partly polished clasts are separated along detachment-parallel slickensides from protocataclasites, which are both strongly overprinted by stress-induced dissolution and mass transfer processes (Fig. 2D). Continuous, anastomosing foliation formed by pressure solution in the velocity-strengthening regime is overprinted by typical microstructures in the velocity-weakening regime, such as chaotic, unfoliated cataclasite with a large variation in grain size (Niemeijer and Spiers, 2007). Separated by a zone of intense ultracataclastic deformation, ankeritized dolostones and limestones cap some of the klippen representing the nearly unmetamorphosed protocataclastic remnants of the Upper Cycladic unit (Iglseder et al., 2011).

We carried out a suite of 40Ar/39Ar and (U-Th)/He thermochronologic investigations to complement the age data for Kea reported in Iglseder et al. (2011) and to further understand the exhumation history of the western Cyclades. White micas from the basement schists on Kea yield flat 40Ar/39Ar age spectra with well-defined plateaus and significant 39Ar gas release (>50%) that gave ca. 20–19 Ma ages. In the initial and final heating increments, apparent ages increase slightly, but otherwise the spectra remain flat (Table A11; GSA Data Repository has complete methodology [see footnote 1]). K/Ca ratios follow a pattern that shows a rise in the initial segments, which mimics the apparent age trend, relatively constant values across the plateau segments, and a change in the final steps to a decrease in K/Ca ratio antithetical to the apparent age trend. Notably, somewhat younger samples lie close to (ultra)mylonitic marbles of low-angle normal fault zone or detachment phyllonites, which recrystallized the fabric-forming micas (Iglseder et al., 2011). We also performed (U-Th)/He analyses on zircon and apatite to understand the lower temperature-time history of the western Cyclades. Single-grain triplicate analyses of the accessory minerals were conducted, and results are presented in Tables 1 and 2 (GSA Data Repository has complete methodology [see footnote 1]). Apatite (U-Th)/He ages from the Cycladic Blueschist unit basement range between 8 and 5 Ma, while apatite from two samples of the structurally uppermost levels yield slightly older ages (14 Ma), recording mid-Miocene exhumation. Surprisingly, zircons from the Cycladic Blueschist unit samples often yield anomalously young ages (younger than AHe ages), likely due to high-U overgrowths, a mineralogical phenomenon that is not uncommon within the Cycladic domain (Keay et al., 2001; Schneider et al., 2011).

Kythnos

Kythnos Island is predominantly composed of greenschist sequences intercalated with decimeter-thick gray marbles, which belong to the Cycladic Blueschist unit (De Smeth, 1975). Metabasic rocks occur as isolated lenses within the greenschists, similar to Kea, and are composed of hornblende, magnetite, orthopyroxene, and minor relics of sodium amphibole (Schliestedt et al., 1994). Relics of high-pressure metamorphism are manifested by glaucophane and pseudomorphs of jadeite and garnet (Schliestedt et al., 1994). Deformation under these conditions is dominated by complex refolded structures with dominantly NW-SE–trending fold axes and subhorizontal axial planes (Keiter et al., 2008). The stretching lineation (lm1) generally trends WSW-ENE with a top-to-the-WSW shear sense (Fig. 1B).

The dominant foliation (sm1+2) on Kythnos is deformed into two open elongated NE-SW–trending domes, giving the island its distinctive shape (Fig. 1B). At the highest structural levels, along the culmination of the domes and at the S headland, the schists are intercalated with blue-gray calcite marbles that are deformed into an ultrafine-grained, pink to brown impure calcitic marble ultramylonite. Millimeter-thin quartz layers show evidence for low-temperature grain boundary migration indicative of lower-greenschist-facies conditions during deformation. A clearly developed stretching lineation (lm2) trends NNE-SSW to NE-SW, parallel to isoclinal intrafolial fold axes. The ultramylonitic marble is spatially associated with cataclasite layers, ranging from proto- to ultracataclasites, localizing along a knife-sharp low-angle plane above the marble mylonites. The cataclasite layers show a variation of compositions from marble ultracataclasites to protocataclastic quartzites, likely derived from a hanging wall that is dominated by strongly hydrothermally altered quartzites with abundant iron oxide and barite mineralization. Hence, both the underlying marble and the overlying quartzite act as source rock for the cataclasite sequence. Brittle and ductile fault rocks record a strong shortening that is perpendicular to lm2, with fold axes parallel to this direction and upright to slightly NW-vergent axial planes. The low-angle normal fault on Kythnos is characterized by abundant kinematic indicators in the marble mylonites and interlayered greenschists, all of which record a clear top-to-the-SW to -SSW shear sense (Fig. 2E). High-strain pinch-and-swell shear-band (C′) boudinage of rotated quartz veins in coarse-grained (0.5 mm) blue marbles is overprinted by localized finer-grained milky white shear zones and pervasive brittle fracturing and bookshelf boudinage of the quartz at low-grade (i.e., <300 °C) metamorphic conditions (Fig. 2F). The contact of the marble mylonites to the overlying cataclasites is represented by slickensides, with slickenlines indicating S-directed shear. Dolomitic layers mainly deform by dilational fracturing and pervasive precipitation of calcite veins. Calcite deforms by dissolution-precipitation creep forming SCC′ structures. The C-planes are parallel to the main detachment surface (Fig. 3A). In general, the cataclasites are associated with massive vein formation of different generations filled with calcite, quartz, or iron minerals.

Stretching parallel and shortening perpendicular to the shearing direction resulted in a strong l-fabric of the footwall rocks below the low-angle normal fault. A conglomerate marble, which represents a typical marker horizon in the Cycladic Blueschist unit on Kythnos, but also on Kea and Serifos, records a prolate fabric with a weak foliation and a dominant ENE-WSW–trending lineation (lm1), which is parallel to the long axes of the white deformed conglomerate components (Fig. 3A). Because the marble is statically recrystallized and does not record any fine-grained shear zones, typical for the extensional deformation along the detachment, the deformation is considered to belong to the Eocene exhumation.

Exhumation of the footwall rocks is associated with a strong greenschist-facies overprint throughout the island. White mica K-Ar dating constrains the greenschist-facies conditions to ca. 26–20 Ma (Schliestedt et al., 1994). Our new 40Ar/39Ar white mica ages from footwall schists are consistently younger, yielding ages of ca. 20 Ma, coeval with the tectonism on Kea (Fig. 4; Table A1 [see footnote 1]). The spectra possess a slight convex-upward shape and early Miocene apparent ages. The general hump-shape of these spectra is defined by heating increments that initially monotonically increase in age, approach a quasi-plateau (20%–40% of total 39Ar) during moderate experimental heating stages, and decline again as the final heating stages are approached. The K/Ca ratios associated with these increments closely mimic the initial rising trend while maintaining reasonably high values for the remainder of gas release. The final increment is typically anomalously old, with an antithetic relationship with the K/Ca ratio. Zircon and apatite (U-Th)/He ages record middle to late Miocene cooling, spanning 10 m.y. from 18 Ma to as young as 8 Ma. Older ages are clustered in the southern portion of the island near the detachment at Aghios Dimitrios; similar to Kea, ZHe ages are younger then AHe ages (Tables 1 and 2).

Serifos

The geology of Serifos (Fig. 1C) is dominated by an I-type granodiorite pluton that was intruded into shallow crustal levels above the brittle-ductile transition zone (Marinos, 1951; Stouraiti et al., 2010). Intrusion ages of the pluton and its associated dikes, which invaded W-E–striking brittle high-angle faults, lie between 11.6 and 9.5 Ma (Altherr et al., 1982; Iglseder et al., 2009). In this work, we report new 40Ar/39Ar white mica ages from rocks adjacent to the pluton, within its thermal aureole, which yield ca. 9 Ma ages, and a few older ages (ca. 13–11 Ma) on mylonites away from the contact (Fig. 4; Table A1 [see footnote 1]). These youngest 40Ar/39Ar ages of the three islands are also the most well-behaved spectra, with 90% of the 39Ar released defining the plateau ages. In the eastern part of the island, the structurally highest branch of a low-angle normal fault system cuts the roof of the granodiorite pluton, recording progressive deformation of the undeformed intrusion at lower structural levels to mylonitic deformation within the fault zone. The higher temperatures achieved during deformation, as indicated by recrystallized K-feldspar (Tullis and Yund, 1987), have been interpreted as the result of the synkinematic intrusion of the granodiorite into an extensional shear zone (Tschegg and Grasemann, 2009). Along its western margin, the pluton discordantly crosscuts the sm1+2 foliation of the metamorphic host rocks, composed of three distinct units, all of which are separated by two individual branches of the low-angle normal fault system with similar top-to-the-SSE–directed kinematics:

(1) The tectonically lowermost unit consists of massive >200-m-thick mylonitic gneisses and schists, which are under- and overlain by calcite/dolomite marble mylonites having a dominant W-E to WSW-ENE striking lineation (lm1) and a clear top-to-the-W shear sense (Fig. 3C). The mylonitic gneiss was deformed at temperatures above ∼450 °C and exhibits striped-gneiss microstructures preserving coarse-grained (>1 mm) quartz grains with straight grain boundaries and dynamically recrystallized K-feldspar (Passchier and Trouw, 2005; Tullis and Yund, 1987). Some porphyroclasts record magmatic relics, such as perthitic K-feldspar and sharply zoned plagioclase. Secondary ion mass spectrometry (SIMS) depth-profiling U-Pb geochronology on zircons separated from this unit yields an age of ca. 40 Ma, with rare earth element (REE) signatures and Th/U values indicative of (high-pressure?) metamorphism; cores of the zircons yield a mixed age spectrum with age populations between Carboniferous and Triassic (Schneider et al., 2011). Analogous to the lithotypes below the Upper Tectonic unit on Attica (equivalent to the Cycladic Blueschist unit; Baziotis and Mposkos, 2011) and the structural level of the granodiorite intrusion into the Lower Tectonic unit (i.e., the Basal unit; Skarpelis et al., 2008), these rocks are suggested to represent a part of the Basal unit. Toward higher structural levels, the gneisses and the marble mylonites are increasingly overprinted by a noncoaxial, greenschist-facies, low-angle normal fault with NNE-SSW–oriented stretching lineation (lm2) and SSW-directed shear sense. The fault contact to the overlying Cycladic Blueschist unit is marked by a decimeter-thick zone of talc and marble cataclasites and a several-tens-of-meters-thick zone of gneisses/amphibolites protocataclasites, derived from the hanging wall (Grasemann and Tschegg, 2011). This branch of the low-angle normal fault is cut by the granodiorite intrusion, which was also the likely source for fluid infiltration along the fault zone.

(2) The overlying Cycladic Blueschist unit is characterized by basal amphibolites and greenschist sequences intercalated with marbles, which dominate northern Serifos. Relic occurrences of glaucophane suggest that these rocks have experienced metamorphism related to the widespread high-pressure event typical for the Cycladic Blueschist unit (Salemink and Schuiling, 1987). Deformation is characterized by mesoscopic folds with subhorizontal axial planes and fold axes parallel to the WSW-ENE–trending stretching lineation (lm1).

In the northern part of Serifos, 10 m below the low-angle normal fault at Platy Yialos, the conglomerate marble marker horizon is exposed. The marbles have an almost horizontal foliation, and the prolate-shaped components record a long axis that trends 065°–245°. In a section parallel to the lineation and perpendicular to the foliation, the components have σ-type geometries with clear stair-stepping indicating WSW-directed shear (Fig. 3D). Interestingly, the conglomerate marble is almost unaffected by the SSW-directed shear of the nearby low-angle normal fault, confirming the observation of extremely localized deformation of the ductile-brittle transition zone. The 40Ar/39Ar analyses of white mica of the Cycladic Blueschist unit preserve ages of 38–35 Ma (Schneider et al., 2011), demonstrating that the WSW-trending lineations are the result of an earlier phase of deformation within the western Cyclades, predating the NNE-SSW–directed extension. The tectonic contact to the overlying Upper Cycladic unit is another branch of the low-angle normal fault, which arches over the island and consistently records a top-to-the-SSW–directed shear sense. Ductile deformation is extremely localized within only a few-meter-thick ultrafine-grained marble mylonite (Fig. 3E). The typical stretching lineation associated with this event (lm2) trends NNE-SSW and has been interpreted as the result of Miocene crustal-scale extension (Grasemann and Petrakakis, 2007; Iglseder et al., 2009; Brichau et al., 2010). Quartz in the marble mylonites records microstructural evidence for dislocation glide and brittle fracturing but no evidence for dislocation creep, suggesting deformation conditions within the brittle-ductile transition zone (e.g., Stipp et al., 2002). The low-grade mylonites are cut by knife-sharp slickenside planes (Figs. 3E and 3F), above which centimeter-thick multistage generations of unfoliated and foliated ultracataclasites record SSW-directed shearing in the brittle-ductile and brittle regime. Both brittle deformation and ductile deformation record folds with upright axial planes and NNE-SSW–striking fold axes, indicative of a strong shortening component perpendicular to the shear (extension) direction (Fig. 3F). Interestingly, the low-angle normal fault interacts with, and postdates, sets of WNW-ESE–striking, conjugate high-angle normal faults, which were also exploited during emplacement of the granodiorite intrusion-related dikes (Grasemann and Petrakakis, 2007; Iglseder et al., 2009), indicating deformation during NNE-SSW extension of the crust under subvertical maximum principal stresses. Notably, no 40Ar/39Ar age gradient is observed across the fault zone, and, except for younger ages close to the contact of the granodiorite pluton, only Eocene ages are preserved (Schneider et al., 2011). All zircon and apatite (U-Th)/He ages from the Cycladic Blueschist unit range from 8 to 5 Ma and record late Miocene cooling (Tables 1 and 2), in accord with published fission-track data (Hejl et al., 2002; Brichau et al., 2010). These observations are indicative of a temperature of ∼300 °C during ductile subhorizontal movement of the low-angle normal fault, consistent with the observed deformation mechanisms in calcite and quartz (Passchier and Trouw, 2005).

(3) Remnants of the Upper Cycladic unit are exposed above the cataclastic fault core of the branch of the low-angle normal fault separating the Cycladic Blueschist unit in the footwall. The rocks consist mainly of marble-dominated protocataclastic lithologies that are strongly ankeritized by massive fluid infiltration (Fig. 3E). In the SW peninsula of Serifos, a highly altered several-tens-of-meters-thick serpentinite lens associated with talc schists is exposed. Since only remnants of the Upper Cycladic unit are preserved, and these marble-rich protocataclasites are strongly altered, it was difficult to separate suitable material for geochronological dating. However, whereas most apatite (U-Th)/He analyses from Serifos yielded late Miocene ages, one sample from the Upper Cycladic unit at Platy Yialos and another at Kavos Kiklopas gave 12–15 Ma ages (Fig. 1; Table 2).

DISCUSSION

Significance and Timing of the West Cycladic Detachment System

Our tectonic and thermochronological analyses of the western Cyclades presented here, combined with previous work (Hejl et al., 2002; Grasemann and Petrakakis, 2007; Iglseder et al., 2009, 2011; Tschegg and Grasemann, 2009; Brichau et al., 2010; Schneider et al., 2011), allow us to consider several detachments observed on the islands of Kea, Kythnos, and Serifos. Each of the low-angle normal faults bears many similarities in terms of their tectonometamorphic evolution in exhuming the Cycladic Blueschist unit with Miocene top-to-the-SW to -SSW–directed kinematics, and an arching of the ductile to brittle shear zone over the islands that defines the carapace of the exhumed domes. These observations are in marked contrast to the long-standing view that extension in the northern and western Cyclades is dominated by a single NE-directed displacement or that top-to-the-S kinematics represent only local brittle-ductile deformation (e.g., Avigad and Garfunkel, 1989; Buick, 1991; Faure et al., 1991; Lee and Lister, 1992; Gautier et al., 1993; Gautier and Brun, 1994; Jolivet and Patriat, 1999; Vanderhaeghe, 2004; Mehl et al., 2007; Jolivet et al., 2010). Whether or not extension in the western Aegean crust has been accommodated by bivergent low-angle normal faults depends on the regional importance given to the top-to-the-S kinematics and their timing with respect to movement along the North Cycladic detachment system.

It is evident that the low-angle normal faults in the western Cyclades, similar to the North Cycladic detachment system, were active at very shallow dip angles, because they mechanically interacted with conjugate high-angle normal faults and record brittle-ductile vertical vein formation typical for a vertical maximum principle stress direction (Mehl et al., 2005, 2007; Lecomte et al., 2010; Collettini, 2011). Whereas ductile extensional deformation is pervasive in the footwall of the low-angle normal fault on Kea and partly also on Kythnos, the low-angle normal fault between the Cycladic Blueschist unit and the Upper Cycladic unit is extremely localized along the brittle-ductile transition zone rocks on Serifos. Therefore, the earlier ENE-WSW–oriented lineation is completely overprinted on Kea and best preserved on Serifos, where ca. 38–35 Ma white mica 40Ar/39Ar ages (Schneider et al., 2011) record exhumation from earlier tectonometamorphic conditions. The Eocene ENE-WSW–trending lineations could be evidence for the basal thrust of a high-pressure extrusion wedge that brought up the Cycladic Blueschist unit during subduction, which is complemented by top-to-the-NE normal sense displacement on Sifnos and Syros (Trotet et al., 2001) illustrating the top of the wedge (Ring et al., 2007, 2011; Huet et al., 2009).

The exact onset of latest extension is unclear, but ca. 20–17 Ma 40Ar/39Ar ages on fabric forming white mica from Kea and Kythnos suggest that ductile movement along the low-angle normal fault was active in the early Miocene. Although the onset of backarc extension has been suggested to have started between 35 and 30 Ma (Jolivet and Brun, 2010), a comparison of Aegean-wide extensional structures revealed that there was an important phase of extensional deformation starting around 23 Ma (e.g., Altherr et al., 1982; Wijbrans and McDougall, 1988; Bröcker et al., 1993). The thermochronological and structural data are supported by the observation that the oldest sediments in the extensional basins of the Aegean are also early Miocene in age (Buttner and Kowalczyk, 1978), suggesting that a Miocene regional extension resulted in the opening of the Aegean Sea basin. Additionally, deposition of shallow-marine detrital sediments on the Upper Cycladic unit started around 23 Ma (Angelier et al., 1978; Sánchez-Gómez et al., 2002; Kuhlemann et al., 2004). Based on thermochronological data, extensional faulting on Serifos was active between 15 and 6 Ma (Iglseder et al., 2009; Brichau et al., 2010; this study). Further, an unambiguous timing constraint for the extensional displacement is the emplacement of the Serifos granodiorite between 11.6 and 9.5 Ma (Iglseder et al., 2009), which crosscuts a structurally deeper branch of the low-angle normal fault between the Basal unit and the Cycladic Blueschist unit, but its roof is cut by a branch of the low-angle normal fault, postdating the intrusion. Although the postintrusive displacement has been suggested to exceed 1.5 km (Tschegg and Grasemann, 2009), the exhumation of rocks from midcrustal levels along very shallow-dipping normal faults is indicative of total displacements on the order of several tens of kilometers (Jolivet et al., 2010). Quantitative constraints of natural fault systems demonstrate that faults with strike lengths on the order of 100 km would be required to achieve maximum displacements on the order of tens of kilometers (Walsh and Watterson, 1988; Cowie and Scholz, 1992; Dawers et al., 1993). In order to accommodate the displacements necessary to exhume these rocks from such depths, a more laterally extensive fault system is necessary. We therefore propose that the low-angle normal faults on Kea, Kythnos, and Serifos form a mechanically linked single crustal-scale structure, which we term the West Cyclades detachment system (Fig. 5).

Hanging Wall of the West Cycladic Detachment System

Due to the level of erosion, only remnants of the hanging wall of the West Cycladic detachment system (i.e., Upper Cycladic unit) are preserved on Kea, Kythnos, and Serifos. Additionally, these remnants have been strongly altered by infiltration of fluid along the low-angle normal faults. Nevertheless, the serpentinite in SW Serifos and some remnants of the unmetamorphosed hanging wall on Kea suggest that these rocks are not part of the Cycladic Blueschist unit. The rocks in the hanging wall of the West Cycladic detachment system should be part of the Upper Cycladic unit representing the lateral continuation of the Pelagonian unit on the Greek mainland (Bonneau, 1982). Interestingly, on the island of Milos, situated 40 km SSW of Serifos and in the hanging wall of the West Cycladic detachment system, a Mesozoic metamorphic basement and an Upper Miocene–Lower Pliocene fossiliferous transgressive marine sedimentary sequence are exposed (Fytikas, 1977); therefore, Milos cannot be part of the Cycladic Blueschist unit. In fact, on Milos, the first marine sediments were deposited on the hanging wall of the West Cycladic detachment system, while the rocks on Serifos were exhuming and cooling from midcrustal level in the footwall position of the West Cycladic detachment system.

Possible Lateral Continuation of the West Cycladic Detachment System

If we accept that the low-angle normal faults exposed in the western Cyclades are indeed linked, the maximum displacement along the West Cycladic detachment system would probably be along the approximate center of the detachment, which is on Kea, where the thickest marble ultramylonites are present (up to several tens of meters). Additionally, ductile deformation associated with extension is much more distributed in the footwall on Kea, suggesting a relatively deeper exposed level of the Miocene low-angle normal fault with respect to the fault system on Serifos. It is on Serifos that a branch of the low-angle normal fault, which separates the Cycladic Blueschist unit from the Upper Cycladic unit, characterized by the marble ultramylonites, reaches only a few meters in thickness. The SE termination of the fault could be located at the southern end of Sifnos, where discrete S-directed, subhorizontal, meter-long brittle-ductile shear bands have been reported (Figs. 6A and 6B; Weil et al., 2010). Although Sifnos has been considered to record penetrative top-to-the-NE sense of shear (Avigad, 1993; Trotet et al., 2001), the recent work of Ring et al. (2011) suggested the existence of a Sifnos detachment with top-to-the-SSW kinematics that operated under brittle conditions with only minor displacement. Low-temperature thermochronological data suggest that movement along the Sifnos detachment largely terminated between 13 and 10 Ma (Ring et al., 2011). These observations together with the fact that the Sifnos detachment does not juxtapose Cycladic Blueschist unit against the Upper Cycladic unit, but operated within the Cycladic Blueschist unit with only minor displacement, corroborate our suggestion that Sifnos may represent the SE termination of the West Cycladic detachment system.

On the other end of the fault system, the NW continuation of the West Cycladic detachment system is probably found in Attica (mainland Greece), where several low-angle normal faults at different structural levels have been reported (e.g., Krohe et al., 2010). Similar to Serifos, close to the SE end of the West Cycladic detachment system, a late-tectonic granodiorite pluton intruded between 11.9 and 8.8 Ma (Skarpelis et al., 2008; Liati et al., 2009) into the low-angle normal fault contact between the Lower Tectonic unit (equivalent of the Basal unit; Baziotis et al., 2009) and the Upper Tectonic unit (equivalent of the Cycladic Blueschist unit; Baziotis and Mposkos, 2011), pinning the latest stage of movement along the West Cycladic detachment system. Similar to the tectonic history on Serifos, the final extensional movements along the low-angle normal faults took place at rather shallow depths, postdating the intrusion (Skarpelis et al., 2008). Abundant shear sense indicators in both ductile marble mylonites and brittle polyphase cataclasites between the Lower and the Upper Tectonic units record SSW-directed kinematics similar to the shear sense of the West Cycladic detachment system (Figs. 6C and 6D).

Bivergent Extension in the Aegean

We suggest that the low-angle normal fault on Serifos links with S-directed low-angle normal faults on the adjacent islands of Kythnos and Kea, forming the West Cycladic detachment system, which, together with the North Cycladic detachment system, accommodated Miocene extension in the Aegean. The oldest documentation of a low-angle normal fault in the Aegean region with top-to-the-S kinematics comes from the South Cyclades shear zone, exposed on Ios (Lister et al., 1984). The island of Ios notably preserves both S-directed and N-directed shear sense; the latter kinematics are preserved in the northern part of the island and are characterized by greenschist-facies and brittle conditions (Huet et al., 2009). In order to reconcile the bivergence of Ios and the well-constrained top-to-the-N shear sense of the nearby Naxos metamorphic core complex (Urai et al., 1990; Buick, 1991; Gautier et al., 1993), several authors suggested either a complex switch of the shear sense or a bivergent exhumation of Ios during the Miocene (Vandenberg and Lister, 1996; Forster and Lister, 2009; Thomson et al., 2009). Such symmetrically arranged detachment systems, which define a metamorphic core complex and establish a bivergent continental breakaway zone, have been also suggested for the central Menderes in the Anatolide belt of western Turkey (Hetzel et al., 1995; Gessner et al., 2001). Recently, Ring et al. (2011) suggested that the South Cyclades shear zone on Ios could be laterally linked with the low-angle normal fault on Serifos. In conflict with these interpretations, Huet et al. (2009) has shown that top-to-the-S deformation on Ios is recorded in shear bands and pressure shadows associated with blueschist-facies minerals and proposed that the Cycladic Blueschist unit–Cycladic Basement unit contact (i.e., South Cycladic shear zone) is not a normal fault but a thrust, active in the subduction zone at high-pressure–low-temperature conditions. The data of the present paper do not directly contribute to the long-standing debate about the South Cycladic shear zone. However, the existing data from the western Cyclades strongly support a termination of the West Cycladic detachment system to the SE on Sifnos, where only brittle top-to-the-SSW displacement is recorded, and no Upper Cycladic unit is juxtaposed in the hanging wall.

Our work does provide strong evidence that Miocene tectonism in the western Cyclades was dominated by top-to-the-SW extension, and that at least the western and northern parts of the Cyclades represent a bivergent extensional orogen. Most geodynamic numerical models (e.g., Chéry et al., 1992; Bassi et al., 1993) are either intrinsically symmetric or predict symmetric extension, and only models with inherited inhomogeneities developed asymmetric extension (e.g., Govers and Wortel, 1993; Tirel et al., 2008). More complex numerical models considering mechanical feedback processes have shown a strong sensitivity of rift mode to various parameters like the strength of the lithosphere, strain softening processes, crust-mantle coupling, and extension velocities (Huismans and Beaumont, 2002; Huet et al., 2011). Although field data have shown that within the metamorphic core complexes, the dome-shaped foliation envelope can be associated with opposite senses of shear along opposing limbs of the dome (Gautier and Brun, 1994), the majority of metamorphic cores show a uniform sense of shear from one limb to the other, with finite strain intensities usually higher in the low-angle normal fault zones (Davis, 1983; Lister and Davis, 1989).

It is important to note that the North Cycladic detachment system and the West Cycladic detachment system represent individual low-angle normal fault systems with distinct tectonometamorphic evolutions, which form both a number of discrete domes and uniform sense of shear across the dome. Therefore, a discussion about symmetric and asymmetric extension should involve both the scale and the three-dimensionality of the structures. In three dimensions, different fault segments with various spatial sizes and orientations may accommodate crustal extension, and therefore bivergent crustal extension in the western and northern Cyclades does not necessarily require bivergent extension of the whole Aegean block. With some notable exceptions (e.g., Brichau et al., 2006, 2008), the geochronological data from the western Cyclades and from many other parts of the Aegean currently do not allow a precise temporal resolution to determine the period(s) during which the individual low-angle normal fault segments were active in the Miocene, and therefore a temporal and spatial jump of the activity cannot be excluded based on the presented data. Future tectonic and quantitative models for the region now need to be able to resolve higher temporal resolution and three-dimensional mechanics.

CONCLUSIONS

(1) We have identified hitherto unrecognized concomitant Miocene low-angle normal fault systems on three islands of the western Cyclades.

(2) The sense of shear from these Miocene structures consistently is indicative of SW-SSW–directed extension.

(3) We propose the mechanical linking of these structures, resulting in the West Cyclades detachment system, which was broadly synchronous with the North Cyclades detachment system, the timing of which is proposed to be during the peak backarc extension and detachment activity in the Cyclades.

(4) At least in the western and northern Cyclades, these roughly symmetrically arranged detachment systems define bivergent crustal-scale boudinage.

We thank the Austrian Science Fund FWF (grant number: P18823-N19) for supporting project ACCEL (Aegean Core Complexes along an Extended Lithosphere). We also thank IGME, the Institute for Mining and Exploration in Athens, for providing permission and technical support. Stimulating discussions with K. Petrakakis, H. Rice, E. Draganits, B. Huet, N. Mancktelow, C. Tschegg, and K. Soukis are greatly acknowledged. L. Jolivet and G. Axen provided important comments to an earlier version of this manuscript. Two anonymous reviews and handling of the manuscript by R.M. Russo, are gratefully acknowledged.

1GSA Data Repository Item 2012031, 40Ar/39Ar white mica analytical data from the western Cyclades and Ar-Ar methodology and (U-Th-Sm)/He methodology, is available at www.geosociety.org/pubs/ft2012.htm, or on request from editing@geosociety.org, Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301-9140, USA.