Subduction, Underplating, and Return Flow Recorded in the Cycladic Blueschist Unit Exposed on Syros, Greece

Abstract

Exhumed high-pressure/low-temperature (HP/LT) metamorphic rocks provide insights into deep (∼20–70 km) subduction interface dynamics. On Syros Island (Cyclades, Greece), the Cycladic Blueschist Unit preserves blueschist-to-eclogite facies oceanic- and continental-affinity rocks that record the structural and thermal evolution linked to Eocene subduction. Despite decades of research, the metamorphic and deformation history (P-T-D) and timing of subduction and exhumation are matters of ongoing discussion. We suggest that Syros comprises three coherent tectonic slices and that each slice underwent subduction, underplating, and syn-subduction return flow along similar P-T trajectories, but at progressively younger times. Subduction and exhumation are distinguished by lineations and ductile fold axis orientations, and are kinematically consistent with previous studies that document top-to-the-S-SW shear (prograde-to-peak subduction), top-to-the-NE shear (blueschist facies exhumation), and then E-W coaxial stretching (greenschist facies exhumation). Amphibole zonations record cooling during decompression, indicating return flow above a cold slab. Multi-mineral Rb-Sr isochrons and compiled metamorphic geochronology show that the three slices record distinct stages of peak subduction (53–52, ∼50, and 45 Ma) that young with structural depth. Retrograde blueschist and greenschist facies fabrics span ∼50–40 and ∼43–20 Ma, respectively, and also young with structural depth. Synthesized data sets support a revised tectonic framework for Syros, involving subduction of structurally distinct coherent slices and simultaneous return flow of previously accreted tectonic slices in the subduction channel shear zone. Distributed, ductile, dominantly coaxial return flow in an Eocene-Oligocene subduction channel proceeded at rates of ∼1.5–5 mm/yr and accommodated ∼80% of the total exhumation of this HP/LT complex.

Key Points

• Syros is a tectonic stack composed of three slices constructed by subduction and underplating; peak subduction ages young with structural depth

• The subduction-to-exhumation transition is marked by kinematic rotation and cooling during decompression

• Metamorphic geochronology indicates syn-subduction exhumation occurred continuously in an Eocene-Oligocene subduction channel

1 Introduction

The mechanical and thermal properties of the subduction interface control the internal structure, kinematics, and dynamics of a subduction zone (e.g., Agard et al., 2018; Cloos, 1982; Gerya & Stöckhert, 2002; Behr & Becker, 2018). Along the shallow interface (≤20 km), direct observations of the megathrust and accretionary wedge are possible through high-resolution seismic reflection imaging, ocean bottom seismometers, and ocean drilling projects (e.g., Fagereng et al., 2019; H. Kimura et al., 2010; Park et al., 2002). However, seismic tomography and earthquake seismology have limited spatial and temporal resolution (e.g., Calvert et al., 2011; Rondenay et al., 2008) so the geometry and internal structure of the deep interface (∼20–70+ km) remain poorly understood. The deep interface can be studied through geologic observations of exhumed high-pressure/low-temperature (HP/LT) metamorphic rocks. Some of the most spectacular examples – for example, the Franciscan Complex (e.g., Cloos, 1986; Wakabayashi, 1990), Japan (Aoki et al., 2008; G. Kimura et al., 2012), and the Mediterranean region (e.g., Brun & Faccenna, 2008; Jolivet et al., 2003) – have profoundly shaped our understanding of subduction and exhumation processes. Specifically, field studies provide constraints on the structural and kinematic evolution, interface geometry, metamorphic pressure-temperature (P-T) trajectories and thermal structure, and timing and rates of subduction and exhumation (e.g., Agard et al., 2018; Angiboust et al., 2016; Behr & Platt, 2012; Dragovic et al., 2015; Kotowski et al., 2021; Tewksbury-Christle et al., 2021). Geologic observations can validate or challenge the results of geodynamic simulations that model the kinematics and dynamics of plate boundary shear zones (e.g., Cloos, 1982; Gerya et al., 2002; Gerya & Stöckhert, 2002; Penniston-Dorland et al., 2015; Warren et al., 2008).

Syros Island, located in the central Aegean Sea (Figure 1), is an ideal locality to study deep subduction interface processes due to its exceptional preservation and exposure of HP/LT blueschist-to-eclogite facies assemblages (Dürr et al., 1978; Okrusch & Bröcker, 1990; Ridley, 1982, 1984). Despite decades of research on Syros, numerous disagreements persist regarding the structural evolution, metamorphic conditions, and timing and mechanisms of subduction and exhumation on the island (e.g., Aravadinou & Xypolias, 2017; Bröcker et al., 2013; Keiter et al., 2004; Laurent et al., 2018, 2016; Lister & Forster, 2016; Ridley, 1982; Ring & Layer, 2003; Rosenbaum et al., 2002; Schumacher et al., 2008; Skelton et al., 2019; Soukis & Stockli, 2013; Trotet, Jolivet, & Vidal, 2001). Furthermore, crustal-scale extensional detachments that accommodated the latest stages of post-orogenic exhumation are well-documented across the Cyclades (Avigad & Garfunkel, 1989, 1991; Gautier et al., 1993; Grasemann et al., 2012; Jolivet, L., Brun, & J. P, 2010; Jolivet & Brun, 2010; Schneider et al., 2018; Soukis & Stockli, 2013), but workers continue to debate the relative importance of major detachments during syn-orogenic exhumation from peak conditions, and whether the strain was distributed or highly localized on Syros (Bond et al., 2007; Keiter et al., 2004; Laurent et al., 2016; Lister & Forster, 2016; Rosenbaum et al., 2002).

Figure 1

Regional tectonic map of the Cyclades, modified from Grasemann et al. (2012). Syros is outlined by the yellow box. North Cycladic (NCDS), West Cycladic (WCDS), Paros-Naxos (PNDS), and Santorini (SDS) Detachment Systems are outlined in white. Kinematic directions are from Aravadinou et al. (2016), Forster et al. (2020), Grasemann et al. (2012), Huet et al. (2009), and references therein.

In this work, we present new structural and petrologic data and Rb-Sr geochronology, and integrate our results with a synthesis of previously published geochronology, to propose a new model for the evolution of the CBU on Syros. Our results refine the island's deformation-metamorphism history, and shed light on the kinematics, metamorphic conditions, and timing of subduction and return flow in the Hellenic subduction zone. This work has implications for rates and mechanisms of HP/LT rock exhumation and provides a broader framework for regional construction of the Attic-Cycladic Complex.

2 Regional Geologic Setting

The Cycladic Islands and parts of mainland Greece comprise the Attic-Cycladic Complex (ACC), which is divided into three units according to depositional age and metamorphic history. From structural top to bottom, the units are: (a) the Upper Cycladic Unit, (b) the Cycladic Blueschist Unit, and (c) the Basal Unit (e.g., Altherr et al., 1994; Avigad & Garfunkel, 1989; Dürr et al., 1978; Jacobshagen, 1986; van der Maar & Jansen, 1983) (Figure 1). The Upper Cycladic Unit is a suite of ophiolitic slivers, altered carbonates ± serpentinites, Late Cretaceous (70–100 Ma) amphibolite-facies orthogneisses, and Miocene greenschist-facies meta-basalts, and correlates with the Pelagonian realm exposed on mainland Greece (Papanikolaou, 1987). The Upper Unit was the upper plate during Late Cretaceous-Paleogene subduction and crops out above the Cycladic Blueschist Unit (CBU) in the hanging wall of crustal-scale, Miocene detachment faults on several Cycladic Islands (Jolivet et al., 2010, 2013; Soukis & Stockli, 2013).

The majority of the ACC is composed of the Cycladic Blueschist Unit (CBU) (Figure 1). The CBU comprises poly-metamorphosed tectonic slices (Dürr et al., 1978; Forster & Lister, 2005, 2008; Jolivet & Brun, 2010) of the following protoliths: (a) (Jurassic?-to-) Cretaceous (∼80 Ma) mafic igneous crust with enriched-MORB and back-arc geochemical signatures ± serpentinized mantle (Bonneau, 1984; Bulle et al., 2010; Cooperdock et al., 2018; Fu et al., 2015; Seck et al., 1996; Tomaschek et al., 2003), (b) Triassic (∼240 Ma) bimodal rift volcanics (Bolhar et al., 2017; Bröcker & Keasling, 2006; Bröcker & Pidgeon, 2007; Keay, 1998; Robertson, 2007) blanketed by Triassic-to-Cretaceous, locally-sourced (e.g., from Triassic volcanics), rifted and passive continental margin siliciclastic and carbonate rocks (Löwen et al., 2015; Papanikolaou, 2013; Poulaki et al., 2019; Seman, 2016; Seman et al., 2017), and (c) peri-Gondwanan basement cross-cut by Carboniferous calc-alkaline granitoids (Flansburg et al., 2019; Keay, 1998; Keay & Lister, 2002).

Regionally, CBU lithologies record evidence for HP/LT metamorphism under blueschist-to-eclogite facies (‘M1’) conditions between ∼53–40 Ma (Cliff et al., 2016; Dixon, 1976; Gorce et al., 2021; Lagos et al., 2007; Laurent et al., 2017; Okrusch & Bröcker, 1990; Ring, Glodny, et al., 2007; Schliestedt, 1986; Tomaschek et al., 2003; Wijbrans et al., 1990). The CBU was exhumed first within the subduction channel, leading to blueschist and greenschist facies overprinting (e.g., Cliff et al., 2016; Kotowski & Behr, 2019; Laurent et al., 2018; Ring et al., 2020), and then in the footwalls of crustal-scale, low-angle normal faults of the North, West, and South Cycladic (Grasemann et al., 2012; Jolivet et al., 2003, 2010; Jolivet & Brun, 2010; Ring & Layer, 2003, 2011; Roche et al., 2016; Soukis & Stockli, 2013), the Paros-Naxos (Bargnesi et al., 2013; Gautier et al., 1993; Linnros et al., 2019), and the Santorini Detachment Systems (Schneider et al., 2018). Exhumation beneath ductile and semi-brittle detachments led to the development of Metamorphic Core Complexes (MCCs) that locally also produced a greenschist-facies (‘M2’) overprint (Bröcker, 1990; Bröcker et al., 1993). As slab rollback initiated and the arc migrated southward through the former forearc, Miocene I-type and S-type plutons intruded the exhuming CBU, and MCC formation led to a local high-temperature, amphibolite-facies (‘M3’) overprint on some islands (e.g., Paros and Naxos, Mykonos, and Ikaria) between ∼21–17 Ma (Andriessen et al., 1979; Brichau et al., 2007; Lister et al., 1984; Pe-Piper et al., 2002; Rabillard et al., 2018; Vanderhaeghe & Whitney, 2004; Wijbrans & McDougall, 1988).

3 The CBU on Syros Island

3.1 Rock Types and Tectonostratigraphy

Syros is a small island (∼84 km²) in the central Cyclades and is dominantly composed of CBU with a klippe of UU in the southeast in the hanging wall of the Oligo-Miocene Vari Detachment (Keiter et al., 2011; Ridley, 1984; Ring et al., 2003; Soukis & Stockli, 2013) (Figure 1). In the context of the Cyclades, Syros preserves some of the most pristine HP/LT metamorphic rocks, in some places even recording peak assemblages with little to no retrogression (Kotowski & Behr, 2019; Okrusch & Bröcker, 1990; Ridley, 1982); similar assemblages are preserved on the island of Sifnos (Aravadinou et al., 2016; Roche et al., 2016).

Within the CBU on Syros, mafic blueschists and eclogites crop out along three tectonostratigraphic horizons: Kampos Belt, Kini-Vaporia-Kalamisia, and Galissas-Fabrikas (Figures 2 and 3). Each horizon exposes ∼300–500 m (structural thickness) of blueschist-to-eclogite facies meta-basalts and gabbros, serpentinites, and interlayered, foliated felsic gneiss/schist and glaucophane schist sequences (metamorphosed bimodal volcanics) in varying proportions (Bröcker & Keasling, 2006; Dixon & Ridley, 1987; Keiter et al., 2011). Along Kampos Belt (i.e., Kampos mélange), eclogitic meta-gabbros, blueschist facies bimodal meta-volcanics, and serpentinite/chlorite-talc schists are most abundant. Meta-gabbro pods (varying in size from ∼1 to at least 1,000 m³) are commonly veiled by metasomatic reaction rinds that developed at contacts with thin serpentinite carapaces, but clastic meta-sediments and bimodal meta-volcanics comprise the volumetric majority of the surrounding matrix (Dixon & Ridley, 1987; Keiter et al., 2011; Ridley, 1982) (Figure 2). Kini, Vaporia (north of Ermoupoli), and Kalamisia are primarily composed of fine-grained mafic blueschist, and contain pods and lenses of eclogite (centimeters-to-decimeters in diameter) and meters-thick layers of serpentinite/talc schist (Keiter et al., 2011; Kotowski & Behr, 2019). Fabrikas comprises coarse-grained glaucophane-bearing eclogite pods (centimeters to meters in diameter) within a fine-grained matrix of mafic blueschists and quartz-mica schists, capped by meta-carbonate (Kotowski & Behr, 2019; Ring et al., 2020; Skelton et al., 2019). Keiter et al. (2011) suggested that mafic blueschists and eclogites are genetically related, and changes in volume proportions of lithologies reflect primary lateral and/or vertical ‘facies changes’ of an enriched-MORB or back-arc igneous suite. Throughout the structural section on Syros, the CBU is internally coherent but commonly contains outcrop-scale block-and-matrix structures reflecting competency contrasts between different protoliths (e.g., basalts vs. gabbros) and metamorphic rock types (e.g., blueschists vs. eclogites) (Keiter et al., 2011; Kotowski & Behr, 2019).

Figure 2

Geologic and structural map of Syros Island, modified from Keiter et al. (2004, 2011). Structural elements and locations of the Syringas Marker Horizon are from Keiter et al. (2011). Constraints on protolith ages are from the references discussed in Section 3.1. Protolith ages are color coded according to rock type. Localities discussed in this study are shown in bold italics, new Rb-Sr sample names and locations are in italics. The thick dashed black line marks a hypothesized location of the lower nappe-bounding shear zone; its location is primarily constrained by detrital zircon maximal depositional ages and metamorphic geochronology.

Figure 3

Condensed lithostratigraphical column modified from Keiter et al. (2011). Our structural, petrologic, and geochronologic synthesis supports dividing the CBU into three sub-units comprising different proportions and distributions of lithologies. Ages of peak subduction metamorphism young with structural depth.

The majority of the CBU comprises a ∼6–8 km section of intercalated meta-volcanic and meta-sedimentary schists, and calcite- and dolomite-marbles with Jurassic-to-Cretaceous depositional ages (Keiter et al., 2004; Löwen et al., 2015; Papanikolaou, 2013; Seman et al., 2017) (Figures 2 and 3). Keiter et al. (2004, 2011) documented a series of boudinaged marbles, cherts, and albite-bearing quartzite, which they named the Syringas Marker Horizon and interpreted as primary sedimentary layering (orange dots shown in Figure 2). The sequence crops out at three or four structural levels suggesting it marks several km-scale thrust sheets (Dixon & Ridley, 1987; Keiter et al., 2011; Ridley, 1982). Repetition of the Syringas Marker Horizon by km-scale folding is unlikely because the largest observable upright folds within this sequence have amplitudes of several hundreds of meters and the marker horizon never appears to be overturned (Keiter et al., 2011). Furthermore, Keiter et al. (2011) documented the repetition of distinct packages of bimodal, rift-related meta-volcanics (also mapped as “banded tuffitic schists”) that have Triassic magmatic protolith ages (Bröcker & Keasling, 2006; Keay, 1998; Pe-Piper et al., 2002; Seman, 2016) (Figure 2), and Seman (2016) presented detrital zircon (DZ) Maximum Depositional Ages (MDAs) for meta-sedimentary rocks that may point to old-on-young tectonostratigraphic inversions. Both results appear to support imbrication (cf. Figure 3).

3.2 Previously Proposed P-T-D-t Paths

Previously published P-T-D evolutions for Syros fall into two categories. Some workers have argued that the majority of deformation and metamorphism on the island is exhumation-related, following peak pressure conditions of ∼20–24 kbar (Laurent et al., 2016; Lister & Forster, 2016; Trotet, Jolivet, & Vidal, 2001) (Figure 4a). These studies interpret mafic blueschists and eclogites to occupy the top of the structural pile and separate them from underlying meta-sedimentary rocks along extensional shear zones (Forster & Lister, 2005; Laurent et al., 2016, 2018; Trotet, Vidal, & Jolivet, 2001). This model implies that lithologically distinct rock packages were juxtaposed during syn-orogenic exhumation (Forster & Lister, 2005; Laurent et al., 2016). Unaltered and retrogressed eclogite has been documented throughout the structural section on Syros, which is considered evidence that all rocks experienced high-pressure conditions during subduction. However, lithologic packages that currently occupy different structural depths could have followed different P-T paths during exhumation (cf. Laurent et al., 2018; Trotet, Jolivet, & Vidal, 2001; Trotet, Vidal, & Jolivet, 2001), and/or could have been subducted and accreted/underplated at different times (Laurent et al., 2017; Lister & Forster, 2016). Such a model could explain reported differences in P-T estimates across Syros; mafic blueschists and eclogites may have been subducted slightly deeper, earlier, compared to meta-sedimentary lithologies (as discussed by Schumacher et al. [2008]).

Figure 4

(a) Compilation of proposed P-T-D histories for the Cycladic Blueschist Unit on Syros. (b) Closure temperature versus time for compiled metamorphic geochronology listed in Table S2 in Supporting Information S1. This dataset comprises 127 datapoints made up of 208 individual ages (some data points are weighted means), from 18 studies and 5 chronometers, from work published between 1987-2022. The black arrows at the bottom labeled ‘structural development’ shows that the timing and tectonic significance of progressive Eo-Oligocene deformation events and corresponding fabric development is contentious.

Alternatively, other authors have suggested that prograde deformation and metamorphism reached ∼16–18 kbar and is locally preserved, but the exhumation-related strain was partitioned into weaker lithologies (Bond et al., 2007; Cisneros et al., 2021; Keiter et al., 2004, 2011; Ridley, 1982; Rosenbaum et al., 2002) (Figure 4a). These studies interpret mafic blueschist and eclogites to record primary relationships with surrounding schists and marbles or to have been juxtaposed with the schists and marbles during early underthrusting (Blake Jr et al., 1981; Hecht, 1985; Keiter et al., 2004; Ridley, 1982). For either of those cases, map-scale lenses of mafic blueschists and eclogites at Vaporia, Kalamisia, and Fabrikas (Figure 2) need not be separated from surrounding CBU by faults or shear zones (i.e., the structurally highest Kampos sub-unit of Laurent et al. (2016)), but instead could occupy a range of structural depths throughout the tectonostratigraphic pile (Keiter et al. [2004], Figure 3). This model implies that meta-mafic and meta-sedimentary rocks that occupy similar structural levels were subducted together and experienced similar P-T histories during subduction and subsequent exhumation (Cisneros et al., 2021; Keiter et al., 2011; Schumacher et al., 2008).

Although existing metamorphic ages help to roughly distinguish prograde from retrograde fabrics and the timing of subduction versus exhumation, differentiating between these P-T-D models has been challenging because of the difficulty in assigning geologic significance to ages (Figure 4b). Two age clusters are most commonly cited for the timing of peak subduction on Syros: ∼53–50 Ma (U-Pb zircon, Ar/Ar and Rb-Sr white mica, Lu-Hf garnet; Cliff et al. [2016]; Lagos et al. [2007]; Lister & Forster [2016]; Tomaschek et al. [2003]), and both ∼52 Ma and ∼45 Ma for different underplated slices (Ar/Ar white mica; Forster & Lister [2005]; Glodny & Ring [2022]; Laurent et al. [2017]; Lister & Forster [2016]). Recently, Uunk et al. (2022) suggested that Syros is composed of three lithologically distinct sub-units that reached similar peak P-T conditions, but were underplated at progressively younger times, as recorded by garnet-whole rock Lu-Hf isochrons (cf. Figure 3). However, the timing of retrogression recorded by Ar/Ar and Rb-Sr ages span the entire Eocene. Maximum CBU temperatures do not appear to have exceeded those required for diffusional resetting of the Ar/Ar and Rb-Sr systems, therefore it is unclear whether retrograde blueschist-to-greenschist facies white mica ages record incomplete isotopic mixing, and/or partial or continuous recrystallization during exhumation, beneath the isotopic closure temperature of the Ar/Ar and Rb-Sr systems (Figure 4b) (e.g., Bröcker et al., 2013; Cliff et al., 2016; Laurent et al., 2017; Rogowitz et al., 2014; Uunk et al., 2018). An additional challenge is that many geochronologic data points in Figure 4b were collected without a clear framework for linking the ages to specific deformation fabrics.

4 Approach and Methodology

4.1 Structural and Microstructural Analysis

Following detailed mapping by Keiter et al. (2004) and Keiter et al. (2011) (map in Figures 2, 5-7), we collected new structural data at several localities from Northern Syros (Figure 5), Central Syros (Figure 6), and Southeastern Syros (Figure 7). We measured planar and linear structural elements, including foliations and cleavages, axial planes to folds, fold axes, and mineral growth, crenulation, and stretching lineations. We constructed π circle diagrams to constrain fold orientations by plotting poles to metamorphic foliation planes. Each color on a given stereonet in Figures 5-7 corresponds to poles to foliations of a specific rock type, or poles to foliations defining single outcrop-scale folds (e.g., Figures 8i and 8k). Our measurements used to produce π diagrams were all derived from cylindrical structures; even if folds have curved hinge lines on larger scales, we measured folds in locations where hinge lines are locally straight. We calculated poles to mean circles to determine fold axis orientations (bold circles) and compared them to fold axes that could be directly measured (diamonds) and mineral lineations (open circles). We documented minerals defining lineations and fold axes, porphyroblast stability and kinematic context (i.e., pre-, syn, post-kinematic with respect to surrounding fabric), and break-down and replacement textures (Figures 8-12) to constrain metamorphic conditions of deformation. Fold axis orientations and mineral lineations do not provide a sense of shear, but rather stretching and/or transport directions during shearing. However, we also documented many outcrop-to micro-scale shear sense indicators and supplemented our kinematic interpretations with literature constraints. Microstructural analysis (Figure 9) (139 total samples, 21 studied in detail) and quantitative EMPA analyses of zoned minerals (6 samples) refined our interpreted P-T-D history (Figures 10 and 12).

Figure 5

Geology and structural elements of Northern Syros. Base map and foliation orientations are from Keiter et al. (2011). Black fold axes are Keiter et al. (2011)'s intrafolial ‘F2 shear folds.’ The spread of orientations is the result of superposed folding, as older folds were progressively reoriented by S-vergent simple shear during subduction (cf. Keiter et al., 2004). Average lineation orientations are shown in white arrows outlined in blue and green for glaucophane and actinolite measurements, respectively; all measured lineations are plotted on the stereonets. Colored planes are best-fit π circles fitting poles to foliation planes defining single outcrop-scale folds. Cross section A1-A2-A3 is modified from Keiter et al. (2011). Topographic contours are 20 m.

Figure 6

Geology and structural elements of Central Syros. Average lineation orientations are shown in white arrows outlined in blue and green for glaucophane and actinolite measurements, respectively; all measured lineations are plotted on the stereonets. Topographic contours are 20 m. See Figure 5 caption for further details.

Figure 7

Geology and structural elements of Southeast Syros. Black arrows with the circles are upright fold axes in the Vari Unit. Average lineation orientations are shown in white arrows outlined in blue and green for glaucophane and actinolite measurements, respectively; all measured lineations are plotted on the stereonets. Cross section A6-B2-B3 is modified from Keiter et al. (2011); compare with Laurent et al. (2016). Topographic contours are 20 m. See Figure 5 caption for further details.

Figure 8

Selected field photos showing prograde (a)–(d) and retrograde (e–l) deformation and metamorphism. (a) Preservation of primary igneous breccias at Grizzas. (b) Right-side-up sequence of oceanic lithosphere at Kini. (c), (d) S_S at Kini contains lawsonite pseudomorphs and omphacite with glaucophane- and garnet-filled pressure shadows. Black arrows in the close-up photo of (c) point to pseudomorphs with garnet inclusions. (e) D_T1 retrogression under blueschist-facies conditions produced local static glaucophane coronas formed around pinched eclogite lenses at Vaporia. (f) Coaxial E-W stretching of calcite clasts in meta-conglomerate at Delfini during D_T1−2. (g), (h) S_S is cut by S_T1 crenulation cleavage at Azolimnos. (h) Two glaucophane lineations record transposition of S_S (black arrow, parallel to the pen) into alignment with crenulation hinges (white arrow) during D_T1. (i)–(k) D_T2 greenschist facies retrogression and upright folding at Delfini (i), Fabrikas (j), and Lotos (k). (i) White arrows point to F_S folds along the limbs of F_T fold. Dashed white lines mark the axial planar S_T cleavage. (j) Non-coaxial, top-to-the-E extensional shear under retrograde blueschist-to-greenschist facies conditions. (k) S_S cross-cut by D_T folding; fold axes trend E-W. (L) Coaxial, lineation-parallel D_T2 brittle boudinage of epidote-rich lenses in greenschists.

Figure 9

Selected photomicrographs showing prograde (a)–(f) and retrograde (e, g–k) deformation and metamorphism. (a),(b) Internal S_R inclusion trails from Lia Beach (A, PPL; B, XPL). (c) S_S contains syn-kinematic garnet porphyroblasts with foamy quartz inclusion trails that are rotated but continuous with respect to the dominant external foliation. (d) D_S garnets include pseudomorphs after lawsonite (comprising epidote and white mica). (e), (f) S_S in mafic blueschists from Kini. (e) S_S is cut by D_T1 crenulation under glaucophane-stable conditions in mafic blueschists from Lia Beach. (f) Omphacite and garnet in D_S Kini blueschists have asymmetric pressure shadows filled with high-pressure minerals. (g)–(i) D_T1 retrogression in bimodal meta-volcanics at Kampos (g), Azolimnos (h) and Kalamisia (i). (h) D_T2 crenulation transposes S_S and strengthens as albite, chlorite, and actinolite form. (i) Omphacite and paragonite break down to epidote, blue amphibole, and albite. (j) D_T2 brittle micro-boudinage of epidote porphyroblasts in Lotos greenschists. (k) Final stages of D_T2 are characterized by post-tectonic albite growth at Lotos.

Figure 10

False-colored X-Ray maps and representative BSE images of D_S in Kini blueschists (a),(b), a composite D_S-D_T1 fabric in Azolimnos bimodal meta-volcanics (c),(d), D_T1 in Kalamisia blueschists (e),(f), and D_T2 in Fabrikas quartz-mica schists (g),(h). Quantitative analyses of sodic amphiboles in (b) KCS53 and (f) KCS12B are shown in Figure 12; white mica analyses from (h) KCS65 are shown in Figure 11 and Figure S2 in Supporting Information S1.

Figure 11

White mica chemistry and micro-textures. Samples from Azolimnos, Lotos, Kini, and Delfini were targeted for Rb-Sr geochronology. (a) Quantitative phengite and paragonite EMPA analyses. (b) Composite D_S-D_T1 texture in Azolimnos retrograde blueschist, characterized by intergrown phengite and paragonite. (c) D_T2 texture in a greenschist facies meta-volcanic from Lotos. Well-aligned white mica defines the foliation, containing relict amphibole porphryoblasts breaking down to quartz, albite, and mica. Phengite is enriched in Fe near reaction zones, but exhibits only slight reduction in Si a.p.f.u. relative to darker cores. (d) D_T2 texture in Delfini meta-volcanics. Phengite-paragonite pairs are in textural equilibrium. Phengite locally occurs as lineation-parallel overgrowths on paragonite cores.

Figure 12

Amphibole mineral chemistry and micro-textures. (a) Quantitative amphibole EMPA analyses (Leake et al. [1997] classification scheme). All analyses have Na_B > 1.5 apfu except for those indicated with an asterisk. (b) D_T1 static growth zonations in glaucophane contained in retrogressed eclogite pod. (c) D_T1 lineation-parallel zonations developed in glaucophane-filled strain shadow fringing garnet porphyroclasts. (d) Greenschists preserve relict D_T1 sodic amphibole as inclusions in epidote, and matrix amphibole records lineation-parallel compositional changes during D_T2 retrogression.

4.2 Rb-Sr Geochronology

We selected five samples for multi-mineral Rb-Sr geochronology. This technique has been applied to exhumed HP/LT metamorphic rocks to date deformation and metamorphism with considerable success (Angiboust et al., 2016; Cliff et al., 2016; Freeman et al., 1997; Glodny et al., 2008, 2005; Kirchner et al., 2016; Ring, Will, et al., 2007). The primary assumption required to construct a multi-mineral isochron is that the phases defining the isochron were co-genetic, and thus share the same initial Sr composition. We separated and selected minerals that we hypothesized were co-genetic based on our structural and microstructural results, and quantitatively tested this hypothesis by identifying phases that were in isotopic disequilibrium (i.e., fall off the isochron) (Cliff & Meffan-Main, 2003). Penetrative foliations lend support to the assumption of syn-kinematic recrystallization of selected minerals, which can reset the Sr isotopic signature between mica and co-genetic phases to temperatures as low as 300°C (Müller et al., 2000). Furthermore, diffusional resetting of the Rb-Sr system is thought to occur at temperatures >550–600°C (Glodny et al., 2008; Inger & Cliff, 1994), which exceeds maximum temperatures in the CBU. Therefore, we interpret our Rb-Sr ages as (re-)crystallization ages associated with deformation.

Following Glodny et al. (2003, 2008), we cut out ∼5 cm³ cubes of rock from hand samples to isolate specific fabrics corresponding to different stages of the deformation history. Samples were crushed with a small hammer between sheets of paper and sieved and separated by grain size. Grain size fractions 125–250 μm and 250–500 μm were separated based on magnetic susceptibility using a Frantz magnetic separator. Mineral separates were picked by hand under a microscope, and white mica separates were cleaned of inclusions by gently smearing them in a mortar and pestle and washing them through a sieve with ethanol. All Rb and Sr isotopic separation and analyses were conducted at the University of Texas at Austin in the Radiogenic Isotopic Clean Lab. All separates (except apatite) were cleaned in 2 N HCl to remove surficial contamination and spiked with mixed high Rb/Sr and low Rb/Sr spikes. We followed methodology for mineral dissolution, isotope column chemistry, Thermal Ionization Mass Spectrometry (Sr analyses), Solution Inductively Coupled Plasma Mass Spectrometry (Rb analyses), and estimating uncertainties in isotopic ratios as described in Kirchner et al. (2016). Reproducibility on replicate USGS Standard Hawaiian Basalt (BHVO) Rb measurements determines the uncertainty of the Rb-Sr ratio, and long-term reproducibility on the NBS987 Sr standard determines the uncertainty of the Sr ratio (Table 2). Ages were calculated using the IsoplotR toolbox (Vermeesch, 2018) with the ⁸⁷Rb decay constant of 1.3972 ± 0.0045 × 10⁻¹¹ year⁻¹ (Villa et al., 2015).

5 Structures and Deformation Fabrics

The CBU on Syros records evidence for three main phases of deformation and metamorphism herein referred to as D_R, D_S, and D_T1−2 (Table 1). Subscripts follow an alphabetical order according to the relative age of deformation, that is, D_R is the oldest observed deformation, and D_T1−2 is the youngest. Each phase led to spaced to penetrative foliation development, and/or ductile folding of older fabrics. Kinematic indicators, metamorphic mineral assemblages, and porphyroblast zonations described herein demonstrate that D_R and D_S developed on the prograde path and are best preserved in mafic blueschists and eclogites (but are locally preserved as textural relicts in bimodal meta-volcanics and meta-sediments), and D_T developed on the retrograde path and is best recorded by meta-volcanic and meta-sedimentary schists.

Table 1. Summary of Interpreted Deformation-Metamorphism Events in the CBU on Syros

• Note. Where D_R is preserved, it is recorded as inclusion trails within garnet; however, garnet growth occurred at different times in different lithologies (i.e., before, during, and after D_R). Some garnets lack an internal foliation, and others contain inclusion trails that are oblique to (pre-D_S) or continuous with S_S (syn-D_S).

5.1 D_R — Prograde Fabric Development During Subduction Under Blueschist Facies Conditions

D_R is the earliest recognizable prograde event but it is not visible at the outcrop-scale. D_R likely formed a strong, penetrative S_R foliation that is locally recorded as inclusion trails in garnet porphyroblasts at Kampos (Figures 9a and 9b) and is tightly folded during D_S. D_R inclusion trails are commonly oblique to the external foliation and are defined by glaucophane, omphacite, and white mica. However, some garnets contain an internal foliation that is continuous with the external foliation, and others preserve no internal foliation. This suggests that garnet growth occurred prior to, during, and after D_R (during D_S) in different lithologies.

5.2 D_S — Prograde-to-Peak Fabric Development During Subduction Under Blueschist to Eclogite Facies Conditions

5.2.1 D_S Structures

D_S is best recorded at Grizzas and Kini (Figure 6b), with relicts preserved on Kampos Belt (Figure 5c), at Lia Beach, and at Azolimnos (Figure 7b). D_S produced a dominant S_S foliation in mafic blueschists, meta-cherts, and bimodal meta-volcanics at Grizzas that is parallel to the axial planes of intrafolial folds (F_S), and rotated and boudinaged quartz veins. This folding event is characterized by shallowly to moderately plunging SW-trending fold axes clustering around 205–251°/15–35°; glaucophane mineral lineations are similarly oriented (Figure 5b). In rare cases, outcrop-scale prograde metamorphism was not associated with penetrative deformation, indicated by the preservation of igneous protolith features such as pillow lavas (Grizzas, cf. Keiter et al. [2011]), intrusive relationships (Kini, cf. Kotowski and Behr [2019] and Laurent et al. [2016]), and magmatic breccias (e.g., at Grizzas, Episkopi, Figure 8a).

Kini dominantly records D_S deformation-metamorphism; it is bounded by high-angle normal faults and is structurally discordant with respect to the surrounding CBU (Figure 6b; cf. Keiter et al. [2011]). In one location, serpentinite wraps around the base of massive meta-gabbros, which transitions upward into fine-grained blueschists, suggesting local preservation of an attenuated section of metamorphosed oceanic lithosphere (Figure 8b). Similar to Grizzas, the D_S fabric in Kini blueschists contains isoclinal folds (F_S) with shallowly south-plunging fold axes. This fold generation is recorded by a 182°/33° fold axis in Kini schists (Figure 6b; Figure 8d). The S_S axial planar cleavage seen in Kini mafic blueschists (e.g., Figures 8c and 8d) is also seen as textural relicts in quartz-mica rich lithologies, as at Azolimnos (Figure 8g). In some localities, blue amphibole lineations define great circles, likely reflecting folding of earlier (D_R) fabric during D_S (Figures 5c and 6b; relicts at Azolimnos in Figure 7b). In other localities, blue amphibole lineations appear to be reoriented into moderately S- or SW-plunging clusters (e.g., Grizzas and Kini, Figures 5b and 6b). Similarly, Keiter et al. (2004, 2011) documented a significant spread of fold axis orientations which they attributed to superposed folding that systematically reoriented older fold hinges via S-vergent simple shear during prograde-to-peak subduction (i.e., their D₂, black fold axes in Figure 5).

Locally, centimeter-sized, prismatic pseudomorphs after lawsonite indicate that lawsonite grew at the culmination of D_S but did not survive peak conditions. Syn-to-post-kinematic porphyroblasts overgrow the mafic blueschist foliation at Grizzas and Lia, decorate foliation-parallel compositional layers at Kini (Figure 8c), and commonly contain inclusions of garnet, and are included by garnet (Figure 8c, closeup). Pseudomorphs are weakly attenuated along the limbs of folds, but preserve their diamond-like shapes in fold hinges (Figure 8c).

5.2.2 D_S Microstructures and Mineral Chemistry

D_S micro-textures in meta-sedimentary rocks are characterized by strong quartz-mica cleavage-microlithon S_S fabrics and locally record rotated inclusion trails in garnets that are mostly continuous with external foliations (Figure 9c). Quartz-rich microlithons have strong lineation-parallel shape-preferred orientations, and mica-rich cleavages comprise intergrown phengite and paragonite (Figure 9c and Figure 10c). Lawsonite pseudomorphs preserved as inclusions in garnet comprise intergrown epidote and white mica (Figure 9d). Garnet compositional zoning varies between samples; some record complex pulses of Ca-enrichment (e.g., Figures 10c and 10d) while others record Mn-rich cores and Fe-rich rims (Figure 10g).

D_S micro-textures in mafic blueschists are characterized by compositional segregation defined by glaucophane-rich and epidote-rich layering alternating on the mm-scale (∼50–200 μm grain size) (Figure 9e). The S_S foliation contains syn-kinematic porphyroblasts of garnet and omphacite (∼300 μm-5 mm), and contains rutile with minor titanite overgrowths (Figures 9f and 10a). Syn-kinematic phengitic white mica is chemically homogeneous and has 3.35–3.45 Si atoms p.f.u (Figures 11a and Figure S2 in Supporting Information S1). Omphacite and garnet deflect local foliations, and have pressure shadows and strain caps composed of glaucophane, phengite, and paragonite, and/or more omphacite (Figures 8d and 10a). Omphacite porphyroblasts in Kini blueschists have cores of low-Na, high-Mg omphacite, fringed by asymmetric, syn-kinematic pressure shadows of high-Na, low-Mg omphacite (Figure 10a). D_S amphibole is glaucophane (Figures 10a and 12a). Rare examples reveal glaucophane cores with thin, patchy rims (Figure 10b) that trend toward lower Al^iv/(Al^iv + Fe_tot) values and higher (Na + K)_A (Figures 12a and Figure S2 in Supporting Information S1). In the samples studied, garnet compositional zoning is less pronounced than in meta-sedimentary rocks; we observed weak Mn-enrichment in some garnet cores. See Laurent et al. (2018) for more details on garnet zoning.

5.3 D_T — Retrograde Fabric Development, Crenulation, and Re-Folding Through Blueschist-to-Greenschist Facies Conditions

5.3.1 D_T Structures

D_T1 is best recorded at Kampos Belt and Palos (Figures 5a and 5c), Azolimnos (Figure 7b), Kalamisia (Figure 7a), and locally at Kini (Figure 6b). D_T1 structures refold older S_S foliations into inclined-to-upright, open-to-tight, shallowly to moderately N- and NE-plunging folds (Figures 5c and 6d, 6e, 7a, 7b; Figure S4 in Supporting Information S1). Glaucophane, calcite, and quartz mineral and stretching lineations are oriented parallel to F_T fold hinge lines (Figures 5c and 7a, 7b). Along Kampos Belt, D_T1 fold axes span ∼335–055°/15–45°, with a cluster of moderately N-plunging folds (e.g., Figure 5c). At Azolimnos, D_T1 folding locally develops an upright crenulation cleavage (S_T) that cuts the S_S foliation (Figures 7a and 7b, 8g). Cm-scale spaced cleavages are parasitic to larger open folds with 045°/5–10° fold axes and steep axial planes. At Azolimnos, glaucophane lineations define a great circle and swing from N to NE into alignment with F_T1 crenulation hinge lines (Figure 8h). Crenulation of Kini rocks is defined by a vertical, NE-striking S_T1 cleavage that cross-cuts mafic blueschists (Figure 6b).

D_T2 is characterized by E-W oriented mineral and stretching lineations that are primarily indicative of greenschist facies conditions (e.g., Lotos, Delfini; Figures 6a and 6c) but locally preserve blueschist facies conditions where strain was highly non-coaxial (i.e., Fabrikas; Figures 7c and 8j), and can be seen in a wide range of rock types throughout central and southern Syros. At Vaporia, the mafic blueschists and eclogites and the surrounding meta-sedimentary rocks develop identical D_T2 structures (Figures 6d and 6e). Single greenschist facies F_T2 folds range in geometry from open to tight and have near-vertical, E-NE-to E-W striking axial planes. F_T2 fold axes cluster strongly around ∼070–110°/5–30° (Figures 6a, 6c and 8i, 8k), and mineral and stretching lineations defined by actinolite, quartz, calcite, and relict glaucophane are oriented parallel to F_T2 hinge lines (Figures 6a, 6c and 6e). Older S_S foliations are progressively reworked during D_T2 creating a composite retrogressed foliation that is visible as S- and Z-folds (e.g., Figures 8i and 8k) with hinge-limb layer thickness variations locally exceeding 20:1 (Figure S4 in Supporting Information S1). F_T2 folds have axial planar cleavages decorated with actinolite, epidote, and chlorite. Coaxial stretching parallel to F_T2 fold hinges is common, resulting in semi-brittle to brittle boudinage of epidote-rich lenses visible from the meso-to the micro-scale, as competent lithologies become brittle during exhumation (Figure 8l). At Delfini, shear sense clast counting of a carbonate meta-conglomerate (GPS: 37°27’36” N/024°53’46” E) reveals conflicting and/or ambiguous shear sense. This is indicative of dominantly coaxial strain during reworking of a composite foliation that develops syn-kinematically with respect to upright folding (cf. Figure S4 in Supporting Information S1). Although D_T strain is primarily coaxial, strongly asymmetric strain occurs locally on the E-SE side of the island. Non-coaxial D_T1−2 is best preserved at Kalamisia and Fabrikas, respectively. At Fabrikas for example, outcrop-scale extensional top-to-the-E shear bands and boudinage cross-cut eclogite pods are decorated by glaucophane (partially replaced by actinolite) and quartz (Kotowski & Behr, 2019; Laurent et al., 2016).

5.3.2 D_T Microstructures and Mineral Chemistry

D_T1 microstructures transpose and retrogress older S_S foliations (creating a composite, or reworked, S_S foliation), record geochemical evidence for retrogression through primarily blueschist facies conditions, and are primarily coaxial. Crenulation hinges that record D_T1 in mafic blueschists are defined by high-Si white mica and glaucophane that has an identical composition to glaucophane defining the S_S foliation (Lia Beach, Figure 9e; Figure S2 in Supporting Information S1). Coaxial D_T1 deformation in mafic blueschists is evidenced by symmetric strain shadows around partially chloritized garnets. During D_T1, S_S-defining blue amphibole grows in the symmetric strain shadows and records lineation-parallel growth zonations trending from glaucophane to magnesio-riebeckite (Vaporia, Figures 12a and 12c) and locally becomes actinolitic (e.g., Kampos, Figure 9g). Some static textures record the same compositional trend (e.g., Figure 12a and 12b). At Kalamisia, extensional C-C’ fabrics are well-developed in thin sections, and C’ top-to-the-ENE shear bands are decorated with albite, paragonite, and phengite (Figures 10e and 10f). C’ cleavages are also defined by finely recrystallized blue amphibole that records lineation-parallel core-to-rim zonations from high-Al riebeckite to low-Al (and lower (Na + K)_A) riebeckite (Figures 9i, 10f, and 12a). Omphacite and paragonite porphyroblasts record the breakdown reaction omphacite + paragonite + H₂O = sodic amphibole + epidote + albite (Figure 9i), and rutile is overgrown by syn-kinematic titanite (Figure 10e). In quartz-mica schists, the retrogressed S_S foliation comprises alternating glaucophane-rich and quartz-mica ± albite-calcite layering. Bimodal meta-volcanics at Azolimnos exhibit strong foliations defined by intergrown phengite and paragonite that are mostly in textural equilibrium. Phengite is compositionally homogeneous with consistent Si values of 3.33 ± 0.01 atoms p.f.u. (n = 24). Paragonite grains locally contain flakes of phengite that may indicate partial or incomplete recrystallization during shearing (Figure 11b). Locally, a syn-D_T1 axial planar cleavage, S_T1, is defined by actinolite, albite, phengite, and paragonite in the cores of upright F_T1 folds (Figure 9h).

D_T2 microstructures transpose and retrogress older S_S foliations, and are primarily coaxial and record geochemical evidence for retrogression under greenschist facies conditions (e.g., Delfini and Lotos). Locally D_T2 was non-coaxial and developed under blueschist facies conditions (e.g., Fabrikas). Mafic greenschists that record D_T2 comprise strongly retrogressed S_S foliations that are defined by fine-grained white mica, albite, epidote, actinolite, chlorite, calcite, and titanite (∼50–500 μm grain size), and contain lineation-parallel epidote porphyroblasts (∼2–5 mm) and unoriented, mat-like albite porphyroblasts (∼1–5 mm) (Figures 9j and 9k). Amphibole occurs in two distinct contexts: as inclusions in epidote and albite porphyroblasts, and as a dominant S_S foliation-forming phase. Amphibole inclusions record core-rim zonations evolving from magnesio-riebeckite to winchite, and matrix amphibole record core-rim zonations evolving from ferro-winchite to actinolite (Figures 12a and 12d). S_S-defining, syn-D_T2 epidote porphyroblasts have pressure shadows filled with white mica, calcite, and albite, and are boudinaged with necks filled with quartz and calcite (Figures 9j and 12d).

Foliation-forming white mica in meta-volcanics at Lotos is phengitic, with average Si atoms p.f.u. of 3.41 ± 0.04 (n = 63). Locally, white mica exhibits core-rim zonations characterized by a subtle increase in Fe, decrease in Al, and decrease in Si; Si atoms p.f.u. decrease slightly from core to rim but still overlap with cores within error (dark cores = 3.42 ± 0.04, bright rims = 3.37 ± 0.02; Figures 11a and 11c). Bright zones are also concentrated around reaction pockets where relict amphibole breaks down to albite, quartz, and white mica (Figure 11c). At Delfini, greenschist facies meta-volcanics exhibit strong foliations defined by intergrown, texturally-equilibrated phengite and paragonite. Rare phengites preserve core-rim zonations characterized by decreasing Si from 3.45 ± 0.05 (n = 7) to 3.33 ± 0.2 (n = 31). More commonly, paragonite grains develop syn-kinematic, lineation-parallel phengitic tails that have Si values similar to foliation-forming phengites (3.35 ± 0.04, n = 11) (Figures 11a and 11d).

In blueschist facies fabrics at Fabrikas, the retrogressed S_S foliation comprises syn-D_T2 epidote porphyroblasts that contain rotated inclusion trails of quartz and glaucophane and inclusions of garnet that preserve syn-D_S spessartine-to-almandine zonations (Figures 10g and 10h). Phengite and paragonite define C- and C’-planes of an extensional, top-to-the-E shear fabric. Phengitic white mica reveals a tight range of Si atoms p.f.u (∼3.33–3.39 a.p.f.u, Figure S2 in Supporting Information S1), and Si content of C- and C’-defining phengite is identical (Figures 10g and Figure S2 in Supporting Information S1). Lineation-parallel brittle micro-boudinage of epidote and amphibole porphyroblasts is common; epidote boudin necks are filled with quartz, and blue amphibole boudin necks contain green amphibole needles. A planar S_T2 fabric that cuts S_S is only found in the core of F_T2 folds (i.e., S_T2 crenulation cleavage at Delfini, Cisneros et al. [2021]).

6 Multi-Mineral Rb-Sr Isochron Petrochronology

Results from the five samples selected for Rb-Sr geochronology are shown in Figure 13 and Table 2. The samples include a meta-basalt, two meta-volcanosedimentary rocks, a greenschist facies reaction rind around an epidote pod, and greenschist mineralization in an epidote lens boudin neck, and record distinct stages of the structural evolution as outlined in Section 5 (i.e., D_S, D_T1, and D_T2). All of the isochrons described herein have Mean Standard Weighted Deviations (MSWDs) greater than 1, which suggests that the data dispersion exceeds that predicted by analytical uncertainties (Wendt & Carl, 1991). However, we consider our Rb-Sr ages reliable records of true deformation-metamorphism events (see Table S1 in Supporting Information S1). This is because the isochrons were constructed from mineral suites that our structural and petrographic observations suggest are co-genetic, the co-linearity of the data is striking (with some justifiable exceptions discussed below), and in constructing an isochron we are directly testing which minerals are cogenetic and which are not. The high MSWD values may reflect an underestimation of our analytical uncertainties, or minor Rb-Sr disequilibrium during progressive metamorphism (perhaps due to incomplete recrystallization, e.g., Halama et al. [2018]) that does not significantly affect our Rb-Sr ages (Table S1 in Supporting Information S1).

Figure 13

Multi-mineral Rb-Sr isochrons from a Kini omphacite-epidote-glaucophane schist (SY1616), Azolimnos glaucophane-mica schist (KCS1617), and Delfini actinolite-mica schist (KCS1621). Gray insets are zoom-ins of low Rb/Sr separates outlined in black boxes. Faded gray symbols were excluded from isochron calculations. Multiple white mica separates for each isochron are shown in black symbols. Sample SY1616 records D_S in the northern nappe, KCS1617 records D_T1 in the central nappe, and KCS1621 records D_T2 in the central nappe. D_T retrogression pre-dates the onset of regional core complex capture. Mineral abbreviations: ep = epidote, glc = glaucophane, om = omphacite, grt = garnet, wm = white mica, chl = chlorite.

Table 2. Summary of Rb and Sr Concentrations and Measured Ratios From Analyzed Samples

Sample ID and summary	Mineral	Grain size (μm)	Rb (ppm)	Sr (ppm)	87Rb/86Sr	87Sr/86Sr	+2σ
SY1616: Kini omphacite-epidote-glaucophane schist	epidote (L18-001)	75-125	0.12	1008	0.00035	0.703224	0.000014
Solution on 10 points: 53.5 + 0.6 Ma	glaucophane (L18-010)	75-125	0.15	143	0.00301	0.703225	0.000014
Initial 87/86 Sr: 0.703209 + 0.000012	omphacite (L19-099)	75-125	0.28	31.3	0.02571	0.703235	0.000014
MSWD = 4	white mica (L19-097)	125-250 (0.5 A)	0.31	18.8	0.04765	0.703244	0.000014
	white mica (L19-093)	250-500 (0.2 A)	0.31	15.5	0.05829	0.703234	0.000014
	white mica (L18-009)	125-250 (0.5 A)	0.37	16.6	0.06439	0.703284	0.000020
	garnet #1 (L18-011)	250-500	0.29	13.0	0.06562	0.703261	0.000014
	white mica (L19-094)	250-500 (0.3 A)	1.16	43.8	0.07644	0.703248	0.000014
	white mica (L19-096)	250-500 (0.4 M)	6.05	142	0.12305	0.703289	0.000014
	white mica (L19-095)	250-500 (0.5 M)	34.9	23.7	4.26296	0.706398	0.000014
	removed from isochron
	garnet #2 (L19-098)	250-500	0.35	12.2	0.08338	0.703353	0.000014
KCS1617: Azolimnos glaucophane-mica schist	white mica (L19-103)	250-500 (0.6 M)	9.15	199	0.13309	0.706681	0.000014
Solution on 7 points: 45.5 + 0.3 Ma	white mica (L19-102)	250-500 (0.5 M)	14.7	179	0.23810	0.706776	0.000014
Initial 87/86 Sr: 0.706592 + 0.000022	glaucophane (L18-002)	250-500	0.31	2.68	0.33478	0.706783	0.000014
MSWD = 8	white mica (L19-100)	250-500 (0.3 M)	33.8	178	0.55161	0.706927	0.000014
	white mica (L18-007)	125-250 (0.8 M)	4.66	13.9	0.97194	0.707200	0.000014
	white mica (L19-101)	250-500 (0.4 M)	112	112	2.87985	0.708433	0.000014
	white mica (L18-005)	250-500 (0.7 M)	219	42.5	14.89794	0.716067	0.000014
	removed from isochron
	epidote (L18-003)		0.70	1486	0.00136	0.706668	0.000014
	garnet #1 (L19-004)		0.69	7.53	0.26530	0.706583	0.000014
	garnet #2 (L19-104)		0.87	3.97	0.63469	0.706733	0.000014
KCS1621: Delfini actinolite-mica schist	epidote	125-250	0.97	1961	0.00143	0.706597	0.000014
Solution on 7 points: 37.1 + 0.1 Ma	white mica (L19-225)	250-500 (0.6 M)	54.0	258	0.60594	0.706951	0.000014
Initial 87/86 Sr: 0.706626 + 0.000033	white mica (L19-222)	125-250 (0.5 M)	326	38.6	2.37806	0.707878	0.000014
MSWD = 13	chlorite (L19-226)	250-500 (0.25 M)	9.40	11.3	2.41488	0.707852	0.000014
	white mica (L19-224)	125-250 (0.6 M)	150	155	2.79655	0.708052	0.000014
	white mica (L19-223)	125-250 (0.5 M)	142	173	24.45665	0.719330	0.000014
	white mica (L19-221)	250-500 (0.4 M)	369	20.7	51.64803	0.733354	0.000015
	removed from isochron
	phengite (L19-220)	125-250 (0.4 M)	343	46.8	21.16233	0.717173	0.000014
SY1644: Delfini mineralization in epidote boudin neck	epidote (L19-041)	>2000	0.23	2170	0.00031	0.706608	0.000014
Solution on 3 points: 36.1 + 2.6 Ma	actinolite (L19-042)	∼250-1000	16.9	123	0.39700	0.706899	0.000014
Initial 87/86 Sr: 0.706655 + 0.00058	white mica (L19-040)	>1000	303	29.9	29.24565	0.721388	0.000014
MSWD = 82
SY1402: Lotos reaction rim around epidote pod	apatite	<100	2.49	726	0.00992	0.704969	0.000008
Solution on 5 points: 34.9 + 5.8 Ma	white mica (L19-029)	<125	204	12.5	47.20997	0.724376	0.000014
Initial 87/86 Sr: 0.70455 + 0.00363	white mica (L19-030)	125-250	227	9.54	68.82184	0.739526	0.000015
MSWD = 76,000	white mica (L19-031)	250-500	234	7.13	95.01809	0.753424	0.000015
	white mica (L19-032)	250-500	203	8.65	67.84958	0.736426	0.000015

Sample SY1616 is an omphacite-epidote-glaucophane schist collected at Kini Beach and records D_S. The minerals selected for isotopic analysis define the fabric shown in Figures 8d, 9f and 10a. This sample yielded an age of 53.5 ± 0.6 Ma (MSWD = 4) based on a 10-point isochron defined by epidote, glaucophane, omphacite, six white mica separates, and garnet (Table 2, Figure 13). To test the robustness of the isochron, several two-to five-point isochrons were calculated from combinations of the co-genetic phases; the age does not change but the MSWD is reduced (=1 for 2-pt isochrons by definition; <1 for 3- and 4-pt, and 1.4–1.7 for 5-pt).

Sample KCS1617 is a glaucophane-mica schist collected at Azolimnos and records D_T1. The minerals selected for geochronology define a fabric identical to that shown in Figures 10c and 11b, corresponding to blue sub-horizontal layering as seen in Figure 8g (i.e., a composite S_S fabric reworked and recrystallized during D_T1 retrogression and crenulation). Similar rocks in the CBU on Syros that comprise cm-to-dm-scale intercalations of glaucophane-epidote and quartz-mica-rich layers have been interpreted as bimodal meta-volcanics (Bröcker & Keasling, 2006; Keiter et al., 2011). This sample was collected from a glaucophane-rich layer and yielded an age of 45.5 ± 0.3 Ma (MSWD = 8) based on a 7-point isochron defined by glaucophane and six white mica separates (Table 2, Figure 13). Our microstructural and petrologic characterization of phengite-paragonite intergrowths from this sample are consistent with the colinearity of white mica separates defining the KCS1617 isochron; phengite compositions are homogeneous, and phengite-paragonite pairs are mostly in textural equilibrium. However, two garnet separates fell off of the isochron and are discarded in the age calculation. We justify this based on microstructural observations shown in Figure 10d; garnets preserve complex Ca-zonation patterns and may record pulsed growth. Furthermore, garnets are probably D_S porphyroblasts and are not expected to be in isotopic equilibrium with the D_T1 fabric during crenulation cleavage development and incipient retrogression. Previous work suggests that Sr isotopic zoning in garnets (Sousa et al., 2013) and/or isotopic inheritance from earlier stages of metamorphism (Romer & Rotzler, 2011) tend to make garnets poor candidates for Rb-Sr isochrons. Recent Lu-Hf garnet geochronology from Azolimnos confirms that garnet growth is older than the fabric we dated (Uunk et al., 2022). Adding epidote to the isochron does not change the age (45.43 ± 0.46 Ma, n = 8), but increases the MSWD to 23. Epidote is stable throughout subduction and exhumation and could record subtle zonations that grew during subsequent deformation events and therefore may not be co-genetic (see Cisneros et al., 2021).

Sample KCS1621 is an actinolite-mica schist collected from Delfini and records D_T2. It was collected from a fold limb of a structure like the one in Figures 8i and is interlayered on the decimeter-scale with actinolite-epidote-chlorite schists, meta-cherts, and mica-schists that locally preserve blue amphibole lineations (see Figure S4 in Supporting Information S1, photos R-T for examples of such structures). This sample yielded an age of 37.1 ± 0.1 Ma (MSWD = 13) based on a 7-point isochron defined by epidote, chlorite, and 5 white mica separates (Table 2, Figure 13). For this sample, various combinations of 2- to 6-pt isochrons all yield ages of ∼35–37 Ma with MSWD varying from ≪1 (e.g., 3-pt epidote-chlorite-white mica), to 1 (e.g., 2-pt white mica-chlorite) to 21 (e.g., 4-pt epidote-chlorite-white mica-white mica). Even isochrons that are not well-defined in high-Rb space (i.e., do not contain phengitic white mica) yield nearly identical ages to the 7-point isochron (Table S1 in Supporting Information S1). Our microstructural observations demonstrate that core-rim zonations in paragonite and phengite are syn-kinematic and cogenetic, and that the dominant fabric is retrograde. Rare, relict higher-Si phengite cores testify to an earlier higher-pressure history that was pervasively recrystallized under lower-P conditions. Relict high-Si cores do not appear to impact our age calculations, but it is possible that the one mica separate we discarded from the KCS1621 isochron could reflect incomplete recrystallization.

Sample SY1644 is a collection of minerals precipitated in the neck of a boudinaged brittle epidote-rich lens from Delfini, and sample SY1402 is a greenschist facies reaction rind at the margin of an epidote-rich lens from Lotos. These samples are representative of semi-brittle boudinage associated with D_T2 stretching (e.g., Figure 8l), which are genetically related to the white mica-rich fabric shown in Figure 11c. These samples yield ages with reasonable uncertainties, but extremely high MSWDs. Sample SY1644 yielded an age of 36.1 ± 2.6 Ma (MSWD = 82) based on a 3-point isochron defined by epidote, actinolite, and phengite, and sample SY1402 yielded an age of 34.9 ± 5.8 Ma (MSWD = 76,000) based on a 5-point isochron defined by apatite and 4 phengites (Table 2). For both samples, 2-pt isochrons yield ∼36 Ma and ∼29–36 Ma, respectively (MSWD = 1; Table S1 in Supporting Information S1). Based on our microstructural and petrologic observations of metamorphic replacement reactions, Si- and Fe-zoning in phengites, and pervasive chloritization, it is unsurprising that these samples yielded poorer isochrons. Therefore, we consider these data qualitative. However, these ages are similar to and trend slightly younger than KCS1621, which is consistent with our structural observations.

7 Discussion

7.1 Synthesis of Structural and Petrologic Data and Interpreted Deformation-Metamorphism History

7.1.1 D_R Deformation and P-T Conditions

We interpret the D_R fabric as the oldest recognizable in the CBU, formed under lawsonite-blueschist facies conditions based on several lines of evidence: (a) D_R inclusion trail mineralogy (e.g., glaucophane, omphacite, and phengite), (b) pseudomorphs of D_R−S lawsonite included in D_S garnets from metabasites on Syros (also seen on Sifnos) (Okrusch & Bröcker, 1990; Ridley, 1982), and (c) syn-kinematic D_R−S omphacite blasts recording up-pressure, core-to-rim zonations marked by increasing jadeite component (Figure 10a) (cf. Thompson, 1974). Lawsonite and epidote appear to have both been stable in mafic bulk compositions during D_R, with lawsonite growing later on the prograde path under higher-pressure conditions (cf. Ballevre et al., 2003). This is consistent with textural observations of lawsonite growing both syn- and post-tectonic with respect to the S_R foliation, incorporating inclusions of garnet (which also grows near peak pressures, cf. Baxter & Caddick [2013]; Dragovic et al. [2012, 2015]), and being included by garnet.

7.1.2 D_S Deformation and P-T Conditions

Deformation stage D_S captures peak metamorphic conditions, and produced: (a) an axial plane schistosity, S_S, associated with tight to isoclinal folds (F_S) that fold S_R and have SSW-plunging fold axes, (b) SSW-to-S-plunging mineral lineations, (c) a blueschist-to-eclogite facies fabric containing syn-kinematic garnet, omphacite, and (now pseudomorphed) lawsonite porphyroblasts, and (d) chemical zonations in glaucophane and omphacite that record syn-kinematic increase in pressure and temperature. Our structural observations are consistent with previous studies that attribute top-SSW thrust-sense kinematics to prograde-to-peak subduction (Keiter et al., 2004; Laurent et al., 2016; Philippon et al., 2011). New and compiled metamorphic geochronology demonstrates that different structural levels of the CBU on Syros experienced peak conditions and D_S deformation at different times (i.e., younging with structural depth, cf. Figure 15; discussed further below). However, it appears that each tectonic slice experienced similar P-T trajectories, including peak P-T, despite subducting at different times. Uunk et al. (2022) came to a similar conclusion by combining garnet Lu-Hf geochronology with thermodynamic modeling; they suggested that garnets throughout the Syros CBU grew at similar pressures (19–21 kbar) but at different times.

We do not provide new quantitative constraints on D_S metamorphic conditions, but peak P-T for the D_S fabric shown in Figure 14 are consistent with our petrologic observations and previous studies. Peak temperatures have been calculated from garnet-omphacite major element exchange for mafic blueschists and eclogites (450–550°C) (Laurent et al., 2018; Okrusch & Bröcker, 1990; Rosenbaum et al., 2002; Schliestedt, 1986); the upper limit of glaucophane stability in marble (∼500°C at ∼15–16 kbar; Schumacher et al. [2008]); and calculated lawsonite-out reactions that predict up-temperature, prograde dehydration according to the reaction lawsonite = epidote + paragonite + H₂O) at ∼400–500°C over ∼12–20 kbar (depending on bulk rock and fluid composition) (Evans, 1990; Liou, 1971; Philippon et al., 2011; Schumacher et al., 2008) (Figure 14). Raman Spectroscopy of Carbonaceous Material from graphite schists suggests slightly higher temperatures of ∼540–560°C (Laurent et al., 2018). Observed porphyroblast stability (e.g., lawsonite pseudomorph inclusions in garnets and vice versa), amphibole chemistry, and the volumetric dominance of glaucophane-bearing marbles throughout the CBU on Syros are generally consistent with peak T of ∼500–550°C.

Figure 14

Preferred, schematic P-T-D-t path for the Cycladic Blueshist Unit, consistent with observations and analytical results from this study. The shape of the path is modified from Schumacher et al. (2008). Amphibole stability fields constraining D_T2 temperatures are from Otsuki and Banno (1990). The timing of metamorphism labeled along the P-T loop corresponds to progressive subduction and exhumation of three distinct tectonic slices (the northern, central, and southern nappes; see Sections 6, 7.1 and Figure 15). Mineral abbreviations: crs = crossite (sodic amphibole), mg-rieb = magnesio-riebeckite, wch = winchite, act = actinolite. Facies fields defined in Figure 4.

Reported peak pressures for D_S are variable in the literature and challenging to reconcile. Early conventional thermobarometry suggested peak P of ∼12–18 kbar in mafic blueschists and eclogites (Dixon, 1976; Okrusch et al., 1978; Okrusch & Bröcker, 1990; Schliestedt, 1986). These pressures are supported by solid inclusion quartz-in-garnet barometry constraining garnet growth conditions at Kini, Kalamisia, Delfini, and Lotos to ∼13–17 kbar (Behr & Becker, 2018; Cisneros et al., 2021). Gorce et al. (2021) recently investigated one sample from Fabrikas, and demonstrated that the majority of pressures derived from solid inclusion barometry (∼17–19 kbar) overlap within the error of pressures derived from thermodynamic modeling that account for garnet fractionation, although the P-T models trend slightly higher (∼20–22 kbar). Other thermodynamic models suggested rocks reached ∼19–21 kbar (Uunk et al., 2022) or even as high as ∼22–24 kbar (Laurent et al., 2018; Skelton et al., 2019; Trotet, Jolivet, & Vidal, 2001). We consider this unlikely based on the abundance of S_S paragonite and the absence of kyanite in meta-mafic rocks, which suggests that the upper stability limit of paragonite at ∼20–23 kbar was not reached (Okrusch & Bröcker, 1990; Schliestedt, 1986; Skelton et al., 2019) (Figure 14), although we acknowledge that the kyanite-in reaction is strongly dependent on bulk rock composition (cf. Laurent et al., 2018). Large differences in P-T estimates between traditional phase equilibria and recent thermodynamic modeling may reflect arbitrary choices of thin section domains selected as representative bulk compositions (e.g., Lanari & Engi, 2017). This effect has been demonstrated for CBU lithologies on Syros (see Figure 15 in Laurent et al. [2018]) and is especially likely in garnet-bearing rocks, due to the strong disequilibrium effect that garnet exerts on local bulk composition (Lanari, & Engi, 2017; Lanari & Engi, 2017; Lanari & Duesterhoeft, 2018). It is also possible that higher-P conditions are real, but have not yet been sampled by solid inclusion techniques.

7.1.3 D_T Deformation and P-T Conditions

D_T represents retrograde deformation under blueschist-to-greenschist facies conditions during exhumation. D_T is distinguished by: (a) transposition of the S_S foliation during the formation of upright, open to tight F_T folds and progressive new (S_T) fabric development, (b) lineation orientations that rotate from N-NE (D_T1) to E-W (D_T2) with progressive strain and (in general) increasing greenschist facies retrogression, (c) dominantly coaxial, but locally non-coaxial deformation, and (d) chemical zonations in amphibole tracking syn-kinematic decrease in pressure and temperature during the development of a composite, reworked foliation (e.g., S_S is locally deformed and metamorphosed during D_T). Our structural data are consistent with previous studies that identified top-NE and top-ENE extensional shear as well as E-W coaxial stretching (at different times and locations, as discussed further below) during exhumation (e.g., Bond et al., 2007; Keiter et al., 2004; Laurent et al., 2016; Rosenbaum et al., 2002; Trotet, Vidal, & Jolivet, 2001).

During D_T, foliation-forming amphiboles transition from glaucophane to (magnesio) riebeckite, to winchite, to actinolite. The progressive decrease of total Al, Na_B, and (Na + K)_A in amphibole indicates that P and T decreased as D_T evolved. Glaucophane coronas that develop around eclogite pods during D_T1 are chemically similar to syn-D_S glaucophane, and retrogressed glaucophane records decreasing Al^vi (KCS53, KCS52B) and Na_B (KCS53) from core to rim, and a minor increase in (Na + K)_A (Figure 12, Figure S2 in Supporting Information S1). These signatures indicate decompression and potentially slight warming (Ernst & Liu, 1998; Laird & Albee, 1981; Moody et al., 1983; Raase, 1974; Robinson, 1982), at the subduction-to-exhumation transition.

D_T2 is characterized by foliation-forming sodic-calcic amphiboles, and local relicts of sodic amphiboles are found as inclusions in porphyroblasts. The chemical transition from sodic-to-calcic amphibole as recorded in Syros CBU rocks indicates cooling during decompression (Brown, 1977; Ernst & Liu, 1998; Laird & Albee, 1981; Maruyama et al., 1983; Moody et al., 1983; Otsuki & Banno, 1990; Schmidt, 1992; Thompson, 1974) through albite-epidote blueschist facies and eventually greenschist facies conditions (Figure 14). This P-T trend is supported by the abundance of albite and titanite overgrowths on rutile (this study), boudin neck quartz-calcite oxygen isotope temperatures and quartz-in-epidote inclusion barometry (Cisneros et al., 2021), and decreases from core-to-rim in celadonite component of foliation-forming white micas (Laurent et al. [2018] and this study). While we cannot rule out an initial phase of isothermal decompression at high pressures, our documented amphibole geochemical zonations are better explained by cooling during decompression at moderate pressures and do not support a positive thermal excursion into the epidote-amphibolite facies field (e.g., edenite, pargasite, crossite), as suggested by Laurent et al. (2018), Lister and Forster (2016), and Trotet, Jolivet, and Vidal (2001) P-T-D paths. Notably, Aravadinou et al. (2016) reported syn-kinematic amphibole zonations from retrograde fabrics in the CBU on Sifnos that also support exhumation along a cooling during the decompression pathway (see also Schmädicke & Will [2003]).

7.2 Synthesis of Previously Published Metamorphic Geochronology

We compiled published metamorphic geochronology from 1987 through 2022 and took inventory of the descriptions of deformation fabrics and metamorphic textures provided by the authors to re-evaluate the significance of Eocene and Oligocene ages in the context of subduction versus exhumation. A full compilation can be found in Supplementary Figure S1 and Table S2. We applied several qualitative filters to the dataset to derive a subset of ages that we can confidently attribute to fabric-forming events. The filters are justified as follows:

Zircon U-Pb ages are robust records of igneous crystallization, but as metamorphic ages they can be difficult to place in pro- or retrograde context (Liu et al., 2006; Poulaki et al., 2021; Tomaschek et al., 2003; Yakymchuk et al., 2017). We include U-Pb ages from Tomaschek et al. (2003) for comparison with other ages, but we do not rely on them for island-scale interpretations.

Garnet Lu-Hf and Sm-Nd are commonly considered reliable indicators of ‘peak’ subduction ages (i.e., maximum depths) (Gorce et al., 2021; Lagos et al., 2007; Uunk et al., 2022), because HP/LT garnets tend to grow rapidly following reaction overstepping (Baxter & Caddick, 2013; Dragovic et al., 2012, 2015). For example, Lagos et al. (2007) reported evidence for rapid, pulsed garnet growth near peak conditions from tight clustering of Lu-Hf ages even though samples exhibited different Lu zoning profiles and distributions between their cores and rims (see also Skora et al. [2006]). In this case, this refutes the possibility that garnet cores grew significantly earlier than their rims somewhere along the prograde path. Uunk et al. (2022) employed a stepwise dissolution technique that may have preferentially removed younger rim ages, and they did not provide constraints on Lu zoning in their samples, but nonetheless succeeded in differentiating several statistically distinct ‘bulk’ Lu-Hf garnet-whole rock isochron ages for rocks occupying distinct structural levels on the island. However, evidence for protracted garnet growth is also locally present. Gorce et al. (2021) reported Sm-Nd ages derived from garnet cores and rims that were statistically distinguisable, and concluded that garnets nucleated near peak conditions and continued to grow during decompression over a span of ∼5 Myr. In this case, coupling geochronology with thermodynamic modeling provided a clear tectonic context for multi-stage or continuously evolving metamorphism.

White mica Ar/Ar has the potential to capture the timing of metamorphism during fabric development. However, this system is highly susceptible to disequilibrium, partial (re-) crystallization and mixed ages, and/or unpredictable loss or gain of radiogenic products, making it difficult to interpret the geological significance of an age (Bröcker et al., 2013; Laurent et al., 2016; Lister & Forster, 2016; Maluski et al., 1987). For the final dataset, we only included five Ar/Ar step-heating ages with strong plateaus from micro-drilled grains which all had clear micro-textural context (Laurent et al., 2017), and one strong plateau age from a well-characterized marble shear zone (Rogowitz et al., 2014). We acknowledge that in other HP terranes, even strong plateau ages have been previously attributed to excess Ar (Sherlock & Arnaud, 1999). However, the Ar ages included in this study overlap within reported error of independent Rb-Sr isochrons from rocks at the same locality and/or similar structural levels, which suggests that at least locally, excess Ar is absent (or apparently absent; cf. Ruffet et al. [1995]).

Rb-Sr isochrons are typically considered reliable constraints on fabric ages when the selected fabrics, and minerals defining them, are well-characterized (Bröcker & Enders, 2001; Bröcker et al., 2013; Glodny & Ring, 2022; Skelton et al., 2019). Furthermore, constructing a Rb-Sr isochron directly tests the assumption that selected minerals were in isotopic equilibrium during metamorphism. This validates interpretation of Rb-Sr ages as deformation-metamorphism events even if whole rock powders serve as Sr anchors (e.g., Bröcker & Enders, 2001). Micro-drilling of white micas and co-genetic Sr-rich phases (epidote or calcite) also provide strong textural context for regressed ages (Cliff et al., 2016).

In some cases, we propose alternative interpretations of published data based on our own structural observations. Skelton et al. (2019), for example, interpreted three of their Rb-Sr isochrons from Fabrikas as peak metamorphic ages (i.e., D_S), but we interpret Fabrikas fabrics as D_T1−2, associated with early exhumation (cf. Figure 7c). This revised interpretation is supported by previous petrologic observations of eclogite breakdown to blueschist, replacement of glaucophane by actinolite, and core-to-rim decrease in celadonite component of foliation-forming phengitic white mica (Kotowski & Behr, 2019; Laurent et al., 2018) (see also Figure 10), and structural studies that report dominantly extensional top-to-the-E, exhumation-related deformation immediately beneath the Vari Detachment (Kotowski & Behr, 2019; Laurent et al., 2016). Top-E extensional kinematics at Fabrikas clearly contrasts with other localities where prograde, top-SSW deformation is preserved. Sm-Nd garnet geochronology coupled with thermodynamic modeling of a sample from Fabrikas further support this interpretation; Gorce et al. (2021) estimated peak conditions were reached at ∼45 Ma (garnet cores) and blueschist facies retrogression occurred through ∼40 Ma (garnet rims). The ∼40 Ma age for blueschist retrogression overlaps with Rb-Sr isochrons for blueschist fabrics presented by Skelton et al. (2019).

In another case, Cliff et al. (2016) analyzed micro-drilled phengites from blueschist-to-greenschist facies (i.e., D_T1 to D_T2) extensional fabrics in calc-schists and quartz-mica schists. Four of their samples from Delfini were described as blueschist-facies (data points marked with black stars in Figure 15); however, we observed penetrative greenschist facies deformation at Delfini (D_T2). Glaucophane is locally preserved in abundance in calc-schists at Delfini, and elsewhere on Syros. Rather than reflecting blueschist facies conditions during deformation, this could be due to a glaucophane-stabilizing, CO₂-bearing fluid under greenschist facies P-T conditions (Kleine et al., 2014), or channelized fluid flow at lithological boundaries leading to heterogeneous retrogression (Breeding et al., 2003).

Figure 15

Metamorphic age versus structural depth along the cross-section line A1-A6 as shown in Figure 2 (modified from Keiter et al., 2011). Locations that crop out on the cross section line are labeled; in the upper panel, other locations are indicated that project into or out of the page at the structural level shown (e.g., Kini is a normal fault-bounded block on the west side of the island and therefore is not shown on the cross section). Only ages that were confidently linked to our proposed deformation scheme are included. Clusters of ages outlined in black boxes are derived from the same locality and collapse onto a single point on the cross section. Delfini symbols marked with stars were reported as blueschist-facies fabrics by Cliff et al. (2016); however, local preservation of glaucophane under greenschist facies conditions can be due to CO₂-bearing fluids. Bold half arrows outlined in red indicate interpreted nappe-delimiting ductile shear zones, which were likely reworked as extensional structures during exhumation.

Finally, Rogowitz et al. (2014) dated phengites from a top-E extensional greenschist facies marble shear zone, and hypothesized the ages would be Miocene in accordance with the regional ‘M2’. They interpreted their Eocene ages as evidence that Miocene deformation did not reset the isotopic signature. However, these authors concluded that the microstructures in their marble mylonite sample were consistent with calcite deformation at ∼300°C. This suggests their ages capture a true Eocene recrystallization event (e.g., strong E-W stretching during greenschist facies D_T2) below the Ar/Ar closure temperature in white micas (∼400–450°C, cf. Hames & Bowring [1994]; Harrison et al. [2009]).

In Figure 15, the refined compilation (n = 47) and new Rb-Sr geochronology (n = 5) are projected onto the cross-section line drawn in Figure 2. Ages are labeled according to fabric generation. Faded data points were assigned textural identities but do not record penetrative strain (e.g., randomly oriented, radiating cluster). Key observations from new and compiled geochronology include:

1. D_S, blueschist-to-eclogite facies deformation-metamorphism spans ∼53 to ∼45 Ma, and is captured by a multi-mineral Rb-Sr isochron (this study), and Lu-Hf and Sm-Nd garnet ages.

2. D_S ages are oldest and well-clustered at Grizzas and Kini (∼53–52 Ma), younger at Azolimnos (∼49 Ma), and youngest at Fabrikas (∼45 Ma).

3. D_T1, retrograde blueschist facies deformation-metamorphism spans ∼50–40 Ma (Rb-Sr isochrons and Ar/Ar single grain analyses) and youngs with structural depth, that is, from Kampos, to Azolimnos, to Fabrikas.

4. D_T2, retrograde greenschist facies deformation-metamorphism spans ∼42–20 Ma (Rb-Sr isochrons and one Ar/Ar age) and youngs with structural depth, that is, from Palos (∼43–35 Ma), to Delfini (∼35–28 Ma), to Posidonia (∼28–20 Ma).

5. Rocks that presently occupy different structural levels developed distinct fabric generations contemporaneously. Examples include: Azolimnos D_S and Kampos D_T1 (∼49 Ma), Fabrikas D_S and Kampos D_T1−2 (∼45 Ma), Fabrikas D_T1 and Palos D_T2 (∼43–38 Ma), and Posidonia D_T2 and non-penetrative greenschist metamorphism in the north (faded symbols, ∼25–20 Ma). In other words, retrograde blueschist and greenschist facies deformation-metamorphism occurred first in the structurally highest units and progressed structurally downwards with time.

8 A Revised Tectonic Model for the CBU on Syros

Here we synthesize protolith age constraints and our structural, petrologic, and geochronologic data to propose a revised tectonic model for the CBU on Syros. First, we present a pre-subduction configuration, then discuss a stepwise reconstruction of progressive subduction, underplating, and exhumation, leading to the proposed three-part tectonic stack exposed on Syros today

8.1 Pre-Subduction Configuration

Figure 16 builds on previous work (e.g., van Hinsbergen et al., 2020; Papanikolaou, 1987, 2013; Ring et al., 2010) and illustrates a highly schematic paleogeographic setting for the protoliths of the CBU on Syros and Southern Cyclades immediately prior to subduction at ∼60 Ma. Peri-Gondwanan Cycladic Basement, cross-cut by Carboniferous magmatism (∼315 Ma on Syros, Tomaschek et al. [2008]; 330–305 Ma in Southern Cyclades, Flansburg et al. [2019]), was rifted in the Triassic (∼245–237 Ma, Bröcker & Pidgeon [2007]; Keay [1998]). Syn-rift bimodal volcanic and sediments intruded and blanketed the hyper-extended margin; these will become the diagnostic marker horizons referred to as banded tuffitic schists and bimodal meta-volcanics mapped by Keiter et al. (2011) (orange and dark gray in Figure 16; cf. Figure 2). Rifting was followed by passive margin sedimentation of psammites, debris flows, and carbonates from the Triassic (∼230 Ma) through the Cretaceous (∼75 Ma) (Löwen et al., 2015; Poulaki et al., 2019; Seman, 2016; Seman et al., 2017). Carbonates interbedded with clastic sediments may be the protolith for the Syringas Marker Horizon (Keiter et al., 2011). Cretaceous igneous rocks (∼80 Ma, Tomaschek et al. [2003]) dissected the potentially hyper-extended basement and passive margin sedimentary sequence, forming a small oceanic-affinity (backarc?) basin (Bonneau, 1984; Fu et al., 2012; Keiter et al., 2011). Bröcker and Keasling (2006) also reported ∼80 Ma crystallization ages for zircons in blackwall alteration zones around a jadeitite block in the Kampos Belt and interpreted the ages to record hydrothermal and/or metasomatic processes in a Cretaceous subduction zone, but this has not yet been confirmed (e.g., Bulle et al., 2010). Major and trace elements, REE patterns, and stable oxygen isotopes in serpentinites on Syros are consistent with formation in either an abyssal hyper-extended margin or mid-ocean ridge/fracture zone environment, rather than a mantle wedge source (Cooperdock et al., 2018).

Figure 16

Schematic paleogeographic reconstruction of the protoliths for the Cycladic Blueschist Unit (CBU) at ∼60 Ma. The zoomed-in cross section is modified from Seman (2016). Stepwise evolution of the CBU during subduction is shown in the next figure.

Some previous studies propose that the ophiolite-affinity sequences within the CBU are meta-olistostromes or meta-debris flows, due to the juxtaposition of various rock types and depositional/igneous crystallization ages (Bröcker & Enders, 1999; Dixon & Ridley, 1987; Hecht, 1985). However, we argue that the spatial distributions and contact relationships between different rock types do not require substantial mechanical mixing or submarine landslides; they can instead be explained by the progressive evolution of the hyper-extended margin as described above. Furthermore, the presence of several in-tact young-on-old stratigraphic relationships in the detrital zircon record throughout the meta-sedimentary section (e.g., northern Syros) suggests that the stratigraphic pile was not homogenized by submarine landslide or mélange mixing during subduction (cf. Poulaki et al., 2019; Seman, 2016; Seman et al., 2017).

The most interpretive parts of Figure 16 are the locations of mafic intrusive igneous rocks. These rocks could reflect off-axis, shallow intrusions related to Cretaceous seafloor spreading, or older mafic igneous rocks related to Triassic rifting and represent the protoliths for mappable exposures of blueschist-eclogite lithologies (dark blue in Figure 2, commonly producing block-in-matrix shear zones). Protolith ages have not been determined for Kini, Vaporia, Kalamisia, or Fabrikas mafic rocks on Syros, but zircons in meta-gabbroic pods in other block-in-matrix exposures on Tinos and Samos, and all studied blocks on Kampos Belt on Syros, yield Cretaceous ages (Bröcker & Enders, 1999; Bröcker et al., 2014; Bröcker & Keasling, 2006; Bulle et al., 2010). Regardless of their origin, the key point is that protoliths for mafic blueschists and eclogites were distributed throughout the CBU before subduction, rather than only coming from the small ocean basin in the north. This interpretation is supported by metamorphic geochronology that demonstrates Kampos/Kini and Fabrikas experienced peak metamorphism at different times that do not overlap within statistical uncertainty (see Figure 15 and discussion below).

This paleogeographic interpretation allows us to split the CBU on Syros into three sub-domains characterized by distinct, but related, protolith assemblages (dashed boxes in Figure 16; see also Figure 3). These sub-domains are the precursors to each of the three main tectonic slices that comprise the structural pile on Syros.

8.2 Peak Subduction of the Palos-Gramatta-Kampos Nappe (∼53 Ma)

The Palos-Gramatta-Kampos nappe (northern nappe) comprises remnants of the Cretaceous oceanic lithosphere and associated syn-to-post-magmatic sedimentation over a Triassic-Jurassic rift margin (Figure 16). Seman (2016) estimated a Maximum Depositional Age of 80 ± 6 Ma for the protolith of the Gramatta schists. The zircons were interpreted to be sourced from a Cretaceous volcanic center and deposited locally in a pelagic environment. Seman (2016) suggested that overlapping depositional and crystallization ages between the Gramatta schists and Grizzas/Kampos meta-igneous rocks indicate the two were genetically related. Alternatively, since MDAs do not unequivocally constrain depositional ages, the Gramatta schist protoliths could be significantly younger than 80 Ma. Additionally, coeval ages of metabasic blocks and detrital zircons could suggest both protoliths have the same provenance but the sedimentary material was mixed during transport. The latter scenario requires that there was no active volcanism in the region after 80 Ma. Regardless, these observations suggest that the pre-subduction relationship of sedimentary rocks deposited nearby an active or recently extinct volcanic center was preserved throughout subduction and exhumation.

Our view of this structurally highest subunit differs from that of Laurent et al. (2016)'s ‘Kampos subunit’ in that it does not solely comprise meta-mafic lithologies, and it does not include the map-scale meta-mafic lenses at Vaporia, Kalamisia, and Fabrikas (see also Section 9). Garnet Lu-Hf from Grizzas and Kampos Belt, and new Rb-Sr isochrons from Kini yield identical ages of D_S fabric development within error, suggesting that Kini was originally subducted as part of the northern nappe (Figure 15), and was down-dropped by late-stage, high-angle normal faults to its present position (cf. Keiter et al., 2011; Ridley, 1984). Garnet Lu-Hf geochronology from Uunk et al. (2022) supports our interpretation that Fabrikas and Katerghaki were subducted later than Kampos and Kini.

Prograde-to-peak subduction was characterized by extremely high top-to-the-SSW asymmetric shear strain and at least two stages of foliation development under blueschist-to-eclogite-facies conditions (D_R and D_S; Figure 17a) (Keiter et al., 2004; Laurent et al., 2016; Philippon et al., 2011). Our observations of early prograde SW-plunging fold axes and mineral lineations preserved at Grizzas, Kini, and locally along Kampos Belt, are consistent with previous observations of top-SW prograde shear sense. Girdled glaucophane lineations (e.g., Kini, Kampos) record continuous kinematic rotation from SW to N-S during subduction. Metamorphism led to the formation of blueschists and eclogites under identical P-T conditions, reflecting differences in bulk composition and/or protolith texture (Kotowski & Behr, 2019; Skelton et al., 2019), creating outcrop-scale block-and-matrix structures. As such, subduction-related strain was very heterogeneous. This is evidenced by rheologically strong meta-gabbros at Grizzas and Kini that preserve primary igneous features (Keiter et al., 2004, 2011; Kotowski & Behr, 2019; Laurent et al., 2016).

Figure 17

Block diagrams illustrating the structural evolution and timing of subduction and exhumation recorded by the three tectonic slices in the Syros nappe stack. Compare stepwise subduction of sub-units to the paleogeography in Figure 16. Horizontal scaling is equivalent to subduction rates of ∼2–3 cm/yr and diagrams are roughly 2× vertically exaggerated. The thickness of the interface is exaggerated for clarity. Contacts between sheets are illustrated as thick black lines; we interpret these structures to be distributed shear zones (∼10–100s m), rather than discrete localized structures.

The northern nappe was underplated after D_S and before D_T exhumation, thus removing it from the active subduction interface. Detrital zircon U-Pb data may support independent structural observations that suggest a large thrust-sense shear zone separates the northern nappe from the central nappe beneath it (Keiter et al., 2004; Laurent et al., 2016; Seman, 2016) (Figure 17a; drawn as the structurally highest ‘nappe-bounding’ thrust in cross section in Figure 15). These contacts are not discrete structures, but rather distributed shear zones that ‘smeared’ and locally mixed lithologies along unit contact zones. This thrust-sense shear zone placed Triassic and Cretaceous igneous rocks (Kampos) atop Cretaceous (Syringas) sediments. After it was removed from the interface, the underplated nappe started to exhume, while subduction of the intermediate nappe continued beneath it.

8.3 Subduction-and-Imbrication of the Syringas-Azolimnos Nappe and Blueschist Facies Exhumation of Northern Nappe

The Syringas-Vaporia-Azolimnos nappe (central nappe) occupies the central portion of the island and comprises interbedded Triassic-to-Cretaceous meta-sedimentary schists, meta-volcanic schists, and meta-carbonates (Figure 16). In contrast to Laurent et al. (2016)'s central Chroussa subunit, we suggest that Vaporia and Kalamisia meta-mafic lenses belong to this central slice and record primary intrusive and/or depositional relationships with surrounding CBU meta-sediments that were sheared during subduction (cf. Keiter et al., 2011).

Until recently the timing of peak D_S during subduction of the central nappe was unknown, but we hypothesized that it must have reached peak conditions sometime between 52 and 45 Ma based on well-constrained ages of peak subduction in the northern and southern nappes. A weighted average of four garnet Lu-Hf isochrons from Myttakas, Delfini, Azolimnos, and Chroussa of 48.6 ± 0.6 Ma supports our hypothesis (i.e., the ‘passive margin domain’ of Uunk et al. [2022]). For Azolimnos specifically, Uunk et al. (2022) reported a garnet Lu-Hf isochron age of 49.0 ± 0.5 Ma. This is consistent with our Rb-Sr isochron from the same outcrop, which constrains the timing of incipient blueschist facies retrogression at 45.5 ± 0.3 Ma, thus bracketing the timing of the subduction-to-exhumation transition in an imbricated slice of the central nappe.

D_S in the central nappe is largely overprinted during subsequent exhumation-related deformation, but early fabrics are reminiscent of D_S in the northern nappe and similarly consist of isoclinal folds and strong cleavage development (e.g., textural relicts at Azolimnos). While D_S developed in the central nappe, D_T1 exhumation-related blueschist facies fabrics formed at the same time in the northern nappe (Figures 15 and 17b).

Detrital zircon U-Pb geochronology and Maximum Depositional Ages (MDAs) of meta-sedimentary rocks in the central nappe suggest that several old-on-young stratigraphic inversions may exist, which would imply that imbrication occurred along cryptic ductile thrusts during subduction (Seman, 2016) (thin black thrusts in Figure 15, pink stars in Figure 17b). For example, Seman (2016) documented an old-on-young stratigraphic inversion where Triassic meta-volcanics at Delfini are thrust atop Cretaceous meta-sediments east of Kini (Figure 2). Even though these structures cannot be seen in the field, the presence and locations of inferred thrusts are further supported by the repeated Syringas Marker Horizon, which never appears overturned (orange circles and pink stars in Figures 15 and 17b, respectively). Thus, we propose that the central nappe is bounded by larger nappe-delimiting shear zones to its north and south, and also comprises smaller-scale thrusts accommodating internal imbrication of CBU meta-sedimentary rocks, shown in the cross section in Figures 2 and 15. Imbrication is further supported by garnet Lu-Hf ages which reveal at least two distinct peak subduction events within the meta-volcano-sedimentary rocks that occupy the central nappe of Syros and comparable lithologies in the CBU on Sifnos (Uunk et al., 2022).

While we acknowledge that zircon U-Pb MDAs do not provide unequivocal constraints on protolith depositional ages, and additional geochronologic and structural evidence is needed, they do provide intriguing insights into the chronostratigraphy of the tectonic packages corroborated by differences in zircon provenance signatures. Even though imbrication has interesting implications for the CBU's structural history, the present tectonic model of progressive subduction, underplating, and return flow is supported by our data (particularly by the compiled metamorphic geochronology) whether or not the central section is imbricated. Evidence for syn-subduction imbrication has been documented elsewhere in the Cyclades, for example on Evia where lower marbles were thrust over the upper marble sequence during prograde, top-to-the-SE shearing (Gerogiannis et al., 2021); on NE Sikinos and Ios islands where several old-on-young inversions occur throughout the meta-sedimentary strata near the CBU-Cycladic Basement contact (Poulaki et al., 2019); and on Milos (Grasemann et al., 2018). Therefore, if syn-subduction tectonostratigraphic repetition is confirmed on Syros, imbrication may be a common and/or recurring process during Eo-Oligocene subduction across the Cyclades.

During peak subduction of the central nappe (D_S), D_T1 deformation occurred in the northern nappe and was characterized by upright folding, crenulation cleavage development, and NE-trending fold axes and mineral lineations. This structural transition is marked by ∼120–180° rotation in dominant mineral lineations and fold axis orientations from the S-SW to the N-NE. We interpret the crenulation cleavage formed during D_T1 to be a signature of the ‘subduction-to-exhumation transition,’ when rocks ‘turn the corner’ in the subduction channel, based on the observation that crenulation lineations are decorated by high-pressure phases with compositions similar to peak D_S blueschist-to-eclogite facies conditions (Kini, Figure 9e). Xypolias et al. (2012) also documented upright ductile folding at the subduction-to-exhumation transition in the CBU in Evia. They interpreted the structures as cross-folds that formed via constrictional flow under net compression and retrograde blueschist facies conditions. D_T1 and subsequent strain localized in weaker CBU meta-sediments during exhumation (e.g., Palos, Gramatta), whereas prograde subduction-related fabrics are locally preserved in rheologically strong meta-gabbros at Grizzas and Kini. These observations support previous structural studies that suggest exhumation-related deformation progressively localized toward the bottom of the structural pile, leading to more pervasive greenschist facies overprints in the south of the island (Laurent et al., 2016; Lister & Forster, 2016; Ring et al., 2020).

The structural base of the central nappe is difficult to pinpoint. However, metamorphic geochronology suggests that it is somewhere below Azolimnos and is likely above the Fabrikas tectonostratigraphic horizon, which we argue comprises the third and lowermost nappe. The presence of a nappe-bounding shear zone is also consistent with progressive southward facies changes in the rock types, as carbonate horizons thin substantially and paragneissic material crops out at the island's southern tip, as well as the presence of thrust fault-bounded marble klippe exposed locally on the southern portion of the island (Figures 2 and 15). Laurent et al. (2016) proposed a nappe-bounding extensional shear zone across the island based on the observed intensity of greenschist overprinting and the disappearance of marbles and suggested its western extent crops out as splaying shear zones above and below the Delfini peninsula (i.e., their ‘Achaldi-Delfini shear zone’). If this is the case, then new and compiled geochronology suggests that greenschist facies overprinting in the southern slice spanned ∼36-20 Ma. Alternatively, if the nappe-bounding shear zone occupies a slightly deeper structural level (i.e., right beneath Delfini peninsula, such that Delfini represents the lowermost portion of the central nappe and is heavily retrogressed under greenschist facies conditions), then D_T1−2 development in the central slice is slightly older (∼35-30 Ma) than in the southern slice (∼30–25 Ma). Our new Rb-Sr isochrons and recent garnet Lu-Hf ages from Uunk et al. (2022) support the latter interpretation.

We suggest a slightly modified position of the lower-bound of the central nappe, which is shown in Figure 2. The location of this structure is primarily constrained by metamorphic geochronology at Fabrikas, Chroussa, and Delfini; detrital zircon MDAs near Kini; the locations of the Syringas Marker Horizon; and locations of previously mapped marble klippe. We emphasize that the location of this structure is a hypothesis (hence the dashed line and question marks). Further structural observations are needed to refine its location, for example, evidence for a poly-deformed shear zone with an early thrust-sense history overprinted by younger extensional shear (cf. Laurent et al., 2016). Both stages of deformation were likely accommodated across distributed shear zones rather than discrete fault planes. As such, ductile shear zones (or fault planes) may be continuously transposed during exhumation, erasing outcrop evidence of prior thrusting of the CBU. We propose that this nappe-bounding shear zone accommodated the underplating of the central nappe around ∼49 Ma while the southern nappe was still subducting, and was subsequently reworked during exhumation.

8.4 Peak Subduction of the Fabrikas Nappe and Blueschist Facies Exhumation of the Central Nappe (∼45 Ma)

The Fabrikas nappe (southern nappe) comprises Triassic meta-sedimentary schists, meta-volcanic schists, thinner meta-carbonate horizons compared to the central nappe, and continental-affinity material of Posidonia (cf. Keiter et al., 2011). This meta-sedimentary sequence was spatially associated with mafic igneous rocks with unknown crystallization ages (Figure 16). The primary difference between our southern slice and Laurent et al. (2016)'s Posidonia subunit is that it contains the Fabrikas meta-mafic lens, which they placed in the structurally highest Kampos subunit. Otherwise, our structural measurements and metamorphic observations are similar.

A Lu-Hf garnet-whole rock isochron from Katerghaki indicates that subduction of the Fabrikas nappe had already begun by ∼48 Ma (Uunk et al., 2022). The timing of peak subduction is constrained at ∼45 Ma by garnet core Sm-Nd crystallization ages (Gorce et al., 2021). This age is distinctly younger than peak subduction at ∼53 Ma and ∼49 Ma of the northern and central nappes, respectively. The subduction-to-exhumation transition, or underplating event, of the southern nappe, is bracketed by peak subduction as recorded by garnet Sm-Nd ages, blueschist facies metamorphism, and garnet rim growth during decompression at ∼40 Ma (Gorce et al., 2021), and blueschist facies retrogression recorded by Rb-Sr multi mineral isochrons between ∼42–39 Ma (Skelton et al., 2019).

Between ∼48–45 Ma, rocks of the central nappe exhumed in the subduction channel under blueschist facies conditions. Retrograde D_T1 blueschist fabrics at Azolimnos, well-constrained at ∼45 Ma (this study), overlap with garnet Sm-Nd core ages at Fabrikas and therefore trend older than retrograde D_T1 blueschist fabrics at Fabrikas, which supports the separation of the central and southern tectonic slices. At this time, mafic blueschists and eclogites and surrounding meta-sedimentary schists in the central nappe developed identical D_T1−2 structures (e.g., Vaporia and overlying meta-sedimentary rocks, and Kalamisia and Azolimnos; Figures 6 and 7). This indicates that during D_T1−2, mafic blueschists and eclogites, and surrounding meta-sedimentary rocks were exhumed together, and in some places, the strain was partitioned between them. Therefore, even if mafic blueschists and eclogites reached higher pressures on their prograde path, they must have been partially exhumed and juxtaposed with CBU meta-sediments by ∼45 Ma to explain concordant exhumation-related structures.

8.5 Exhumation of the Syros Nappe-Stack in the Subduction Channel Under Greenschist Facies Conditions (Through ∼20 Ma)

Between ∼44–20 Ma, greenschist facies D_T2 fabrics continuously developed throughout the accreted CBU stack, younging systematically with structural depth, as each underplated nappe was exhumed in series from north to south. Retrograde greenschist facies deformation-metamorphism occurred first in the structurally highest northern nappe and migrated structurally downward through time (cf. Glodny & Ring, 2022; Ring et al., 2020; Roche et al., 2016).

Exhumation imparted penetrative deformation that progressively transposed older fabrics under blueschist facies (D_T1) and eventually greenschist facies (D_T2) conditions. Upright folding that initiated under blueschist facies conditions continued through greenschist facies conditions in the subduction channel. Kinematic rotation culminated in strongly E-W oriented fold axes and structures that formed under net compression, but were associated with fold axis-parallel elongation and stretching (cf. Xypolias et al., 2012). Our new Rb-Sr isochron age from Delfini provides precise constraints that exhumation was dominated by net compression under greenschist facies conditions until at least ∼37 Ma. Furthermore, exhumation-related D_T1 and D_T2 strains were dominantly coaxial and well-distributed. This is evident from symmetric strain shadows on garnets, ductile pinching of partially retrogressed eclogites at Agios Dimitrios, and outcrop-scale greenschist facies folds with sub-horizontal E-W trending hinge lines with hinge-parallel symmetric boudinage of competent blueschist and epidote-rich lenses (e.g., Delfini and Lotos; Figure 8, Figure S4 in Supporting Information S1).

The youngest dynamic D_T2 greenschist facies fabrics associated with subduction channel exhumation are ∼25–20 Ma and are recorded in the southern slice (Figure 15). At this time, greenschist facies metamorphism continued in the northern and central nappes, but was not associated with penetrative strain (e.g., random grains, radiating clusters, decussate textures; Cliff et al. [2016]). These observations indicate strain progressively localized toward the base of the stack through time. Patchy, static metamorphism in the northern and central nappes may reflect local fluid availability as deformation migrated structurally downward.

8.6 Upper Plate Extension and Core Complex Capture

Slab rollback accelerated by ∼23–21 Ma, which is constrained by dating of detachment faults and supra-detachment sedimentary basins that developed in response to initial upper crustal extension (Gessner et al., 2013; Ring et al., 2010). Rollback led to core complex capture, the southward migration of the volcanic arc through the former forearc (e.g., the Tinos granite, 14.6 ± 0.2 Ma, Bolhar et al. [2010]), and continuous supradetachment basin development through the late Miocene (e.g., Paros, ∼14–7 Ma; Bargnesi et al. [2013]). CBU rocks were exhumed in the footwall of the North and West Cycladic Detachment Systems and related smaller-scale structures (Brichau et al., 2007; Grasemann et al., 2012; Jolivet et al., 2010; Soukis & Stockli, 2013). On Syros, the Vari Detachment operated as a semi-brittle to brittle extensional structure and accommodated late stages of exhumation (Figure 2).

Soukis and Stockli (2013) presented low-temperature zircon and apatite (U-Th)/He thermochronology and concluded that the southern Syros CBU was juxtaposed with two structurally higher upper-plate units, the Upper Unit (intermediate structural level) and Vari Unit (structurally highest), along at least two semi-brittle detachment faults. While the Tinos Detachment exhumed CBU rocks between ∼22–19 on what would become neighboring Tinos Island, low-angle normal faults juxtaposed the Vari and Upper Units on Syros. Exhumation of the Vari and Upper Units at ∼13–15 Ma was roughly coeval with magmatism on Tinos but the Syros CBU was exhumed later, ∼8–10 Ma, beneath the Vari Detachment (Soukis & Stockli, 2013).

Previous meso- and microscale observations demonstrate those kinematics along the Vari Detachment are top-to-the-ENE (Soukis & Stockli, 2013), consistent with our structural observations. High-angle normal faults dip both SW and NE, indicating overall NE-SW extension (Soukis & Stockli, 2013). The final exhumation of the CBU on Syros occurred in multiple, temporally distinct, rapid episodes of unroofing. Exhumation beneath the Vari Detachment was rapid, but only accommodated the final ∼6–9 km of vertical exhumation (Ring et al., 2003).

9 Implications

9.1 Comparisons With Previous Tectonic Models

The tectonic model described above has several similarities and differences compared to previous models. First, our results agree with previous studies suggesting that Syros is composed of distinct tectonic slices that reached peak conditions at different times (Glodny & Ring, 2022; Laurent et al., 2017; Lister & Forster, 2016; Ring et al., 2020; Skelton et al., 2019; Uunk et al., 2018, 2022). Most of these studies propose two distinct slices in the north and south, but our data suggest the likely presence of an additional third slice in between.

The timing of peak subduction of Kampos/Kini versus Fabrikas compared to early retrograde blueschist deformation at Azolimnos constrains the number of tectonic slices. Our new Rb-Sr isochron from Kini demonstrates the D_S fabric developed at 53.5 ± 0.6 Ma. This overlaps with independent constraints on the timing of garnet and zircon metamorphic rim growth at Grizzas and Kini. Together these ages bracket the timing of peak subduction of the structurally highest slice. This is statistically distinguishable from the peak subduction in Fabrikas rocks at ∼45.3 ± 1.0 Ma (Gorce et al., 2021). Thus, to a first degree, these data lend further support to the presence of two distinct structural slices in the north and south (Glodny & Ring, 2022; Ring et al., 2020; Uunk et al., 2018).

Skelton et al. (2019) interpreted their Rb-Sr isochrons as records of peak subduction at ∼40 Ma at Fabrikas, which is younger than garnet core crystallization ages inferred to record peak metamorphism at ∼45 Ma in the same outcrop. As discussed above, the dominant outcrop-scale structures, metamorphic mineral chemistry, and mineral replacement textures at Fabrikas are consistent with retrograde blueschist facies deformation. However, if Fabrikas did reach peak conditions at ∼40 Ma, this supports the inference of a central slice above Fabrikas, since recent garnet Lu-Hf and our new Rb-Sr isochron from Azolimnos indicate that rocks occupying a higher structural level above Fabrikas experienced peak subduction at ∼49 Ma (Uunk et al., 2022) and blueschist facies retrogression at 45.5 ± 0.3 Ma (this study). Our Azolimnos D_T1 Rb-Sr isochron is statistically distinguishable (i.e., 2σ errors do not overlap) from Fabrikas D_T1 inferred from retrograde garnet growth (∼40.5 ± 1.9 Ma; Gorce et al. [2021]) and blueschist fabric development (∼41.4 ± 0.5, 41.6 ± 1.5, and 39.6 ± 1.2 Ma; Skelton et al. [2019]). Therefore, if the garnet core crystallization ages presented by Gorce et al. (2021) are accurate records of peak subduction of the southern Fabrikas nappe, this also supports the interpretation of an intermediate slice, which was exhuming at the same time as the Fabrikas nappe reached peak conditions.

We argue that mafic blueschists and eclogites do not exclusively occupy the structurally highest tectonic slice, in contrast to Laurent et al. (2016) and Trotet, Jolivet, and Vidal (2001). Rather, our interpretation is that protoliths for mafic blueschists and eclogites were present throughout the CBU before subduction and therefore appear to record primary relationships (cf. Keiter et al., 2011). This implies that the mafic blueschists and eclogites at Vaporia, Kalamisia, and Fabrikas are not separated from surrounding schists and marbles by shear zones and/or detachments, as shown for the ‘Kampos subunit’ of Laurent et al. (2016). The primary observations supporting that Fabrikas cannot belong to the same subducting unit as Kampos are that Fabrikas meta-volcanic record peak metamorphism that is distinctly younger than that of Kampos and Kini, Fabrikas crops out toward the southern end (i.e., bottom) of the dominantly north-northeast-dipping structural pile, and Fabrikas meta-volcanic are associated with more meta-carbonate and meta-clastic sedimentary lithologies than Kampos and Kini suggesting they represent subduction of different protolith assemblages. Moreover, the fact that Fabrikas occupies the immediate footwall of the Vari Detachment does not necessarily imply that it belongs to the structurally highest unit. Even though the Vari Detachment has been interpreted as the paleo-subduction channel roof, continuous ductile extension along a shallowly to moderately dipping structure throughout the Eo(?)-Oligocene, in addition to the proposed 6–9 km of semi-brittle exhumation accommodated by ∼20 km of localized slip in the Miocene (Ring et al., 2003), can easily explain tectonic removal of the uppermost units. (U-Th)/He ages and strong cataclastic reworking at the base of the Vari Unit further attest to this (Soukis & Stockli, 2013). These processes would juxtapose structurally deeper CBU units with the Upper Unit in the hanging wall.

Our observations indicate that prograde textures are locally preserved in mafic blueschists and eclogites (cf. Keiter et al., 2004), but the majority of the Syros CBU has been overprinted during subduction channel exhumation (cf. Bond et al., 2007; Rosenbaum et al., 2002; Trotet, Vidal, & Jolivet, 2001). Heterogeneous rock types that occupy a given nappe were subducted and exhumed together, and therefore experienced identical P-T paths (in contrast to Trotet, Vidal, & Jolivet [2001]; Trotet, Jolivet, & Vidal [2001]). Therefore, differences in strain, metamorphic mineral assemblages, and/or preserved kinematics between mafic blueschists and eclogites and meta-sedimentary rocks can be attributed to relative strengths, bulk composition, and fluid availability (and composition) during metamorphism (see Schmädicke & Will [2003] for a similar discussion of P-T paths and retrogression of the CBU on Sifnos).

9.2 Subduction Kinematics and Exhumation Processes

Kinematics of prograde subduction and the subduction-to-exhumation transition in the CBU are debated but have direct implications for the geometry and mechanics of the subduction channel shear zone. Here, we summarize the shear sense implications of our preferred tectonic and kinematic model, and then we discuss alternative interpretations. This model is supported mostly by age relationships and lineations defined by different mineral assemblages is consistent with previous kinematic studies as discussed below, and implies a systematic progression in a shear sense through time. This model should be viewed as a hypothesis, and we emphasize that it is built on the assumption that subduction occurred along a roughly north-dipping interface. Further structural analysis and evidence of shear sense reversal at the proposed shear zone contacts could test this hypothesis. However, in many cases, we expect that subduction-related shearing will have been obliterated by younger exhumation-related deformation.

1. Subduction along a roughly N-NE-dipping interface produced a prograde, thrust-sense, top-to-the-SSW sense of shear. Although our fold axis orientations and mineral lineations are not kinematic indicators, they are kinematically consistent with this configuration. Previous studies of the CBU on Syros have documented high-P mineral lineations (e.g., glaucophane and phengite) aligned N-S with top-to-the-SSW shear sense (Jolivet et al., 2010; Keiter et al., 2004; Kotowski & Behr, 2019; Laurent et al., 2016; Philippon et al., 2011). Similar prograde kinematics were identified along the contact between the CBU and Cycladic Basement exposures on Sikinos and Ios (Augier et al., 2015; Huet et al., 2009), and in a sliver of what is thought to be CBU on Milos (Grasemann et al., 2018). This prograde shear sense is also consistent with prior observations of top-SSW prograde thrusting observed on mainland Greece (Jacobshagen, 1986) and seismic tomography and palinspastic reconstructions that suggest subduction in the Aegean has occurred along a single north-dipping subduction zone since the Early Cretaceous (van Hinsbergen et al., 2005).

2. At the subduction-to-exhumation transition, underplating requires a balance between down-dip shear traction and up-dip buoyancy forces (Beaumont et al., 2009; Raimbourg et al., 2007; Warren et al., 2008; Xia & Platt, 2017). Velocity vectors across a dipping planar shear zone (i.e., non-downward tapering) can also yield simultaneous subduction and return flow of underplated sheets depending on the relative contributions of down-dip shear traction (Couette flow) and up-dip buoyancy (Poiseulle flow) (Beaumont et al., 2009; Raimbourg et al., 2007; Warren et al., 2008; Xia & Platt, 2017), thus giving rise to the subduction channel. The boundaries of an exhuming sheet are defined by shear zones with opposing kinematics - extensional at the top, and thrust-sense at the base - that switch across the axis of maximum exhumation velocity (Gerya et al., 2002; Raimbourg et al., 2007; Xia & Platt, 2017). Therefore, immediately after underplating, extensional shear at the top of the underplated sheet can accommodate early stages of exhumation. In the proposed kinematic framework, this corresponds to top-NE/ENE extension, roughly aligned with the inferred paleo-subduction direction (cf. Keiter et al., 2004; Laurent et al., 2016; Trotet, Vidal, & Jolivet, 2001). Top-NE extension can produce upright NE-trending folds; we observed N-NE trending fold axes and crenulation and mineral lineations in retrograde blueschists at Kampos, Kini, and Azolimnos that are potential records of this transition.

3. Further exhumation will proceed contemporaneously along the top of the channelized shear zone while subduction of younger units continues beneath. In the center of an exhuming sheet dominated by plug flow (i.e., Poiseuille flow), kinematics are effectively coaxial. Previous kinematic analyses have argued for either dominantly asymmetric extension, or NE-SW and EW-oriented bulk coaxial thinning, during exhumation (Bond et al., 2007; Keiter et al., 2004; Laurent et al., 2016; Rosenbaum et al., 2002; Trotet, Vidal, & Jolivet, 2001). We suggest that both styles may have occurred at different structural levels in exhuming sheets and/or at different times, and now both signatures manifest in different lithologies and locations. This potentially reconciles contradictory interpretations from past studies. Furthermore, if the subduction zone is net compressional and σ₁ is roughly perpendicular to the trench, then upright folds with trench-parallel fold axes and fold axis-parallel stretching may develop when the rocks reach the base of the overriding crust. In the proposed framework, this implies E-W trending fold axes and fold axis-parallel E-W coaxial stretching. If this is the case, then our structural analysis, geochronology, and petrology from Delfini suggest that the subduction zone was net compressional until at least 37 Ma.

Our observations are not consistent with previous studies that propose subduction occurred along a west-dipping slab with a prograde top-to-the-ESE thrust sense of shear in a tapered wedge (Aravadinou et al., 2016, 2022; Xypolias et al., 2012). This geometry would imply that the thrusting direction was nearly parallel to the current structural grain of the Cyclades on a regional scale, as opposed to perpendicular (Ridley, 1982; Xypolias et al., 2012). Furthermore, our kinematic model describes rotation in transport directions during progressive exhumation from NE- (blueschist facies) to E-W (dominantly greenschist facies), supported by amphibole zonations, metamorphic replacement reactions, and isochron geochronology (e.g., Azolimnos vs. Delfini). Aravadinou et al. (2022) interpreted the opposite rotation from E-W to NE-directed transport through time based on the evolution of quartz and calcite petrofabric analysis from two shear zones in Northern Syros (Kastri; this is part of the same structure we interpret as the contact between the northern and central slices) and Northeastern Syros (Agios Dimitrios). However, they noted that amphiboles in both shear zones were all sodic (no calcic amphibole present) and unzoned, making it difficult to discern the relative conditions or timing of operative deformation between exposures. Reconciling these interpretations with our own would require either the simultaneous subduction of at least two slabs with nearly orthogonal polarities in the Eocene or large-scale post-orogenic rotations.

Our kinematic insights are consistent with a recent study on Amorgos Island that combined structural mapping, detrital zircon U-Pb, and (U-Th)/He low-temperature thermochronology to reconstruct the tectonic history of part of the southern Pelagonian domain (Laskari et al., 2022). These authors identified top-to-the-NW syn-metamorphic shortening occurring during the latest Eocene-early Oligocene, followed by early-mid Miocene exhumation. They interpreted this stage of shortening to have occurred in a retro-wedge position. It is possible that this could be associated with short-lived, SE-directed intracontinental subduction within the overriding plate. According to our tectonic model, this shortening event overlaps with the timing of CBU exhumation under greenschist facies conditions and upright folding in the subduction channel. This provides further evidence that the CBU-Pelagonian system in the Cyclades was net compressional through the early Oligocene. Therefore, inferences of multiple subduction directions may indeed be correct, but due to limited preservation of prograde deformation and metamorphism in the Cyclades, a consistent regional-scale framework cannot be drawn.

We have documented upright folding and fold axis-parallel elongation that accommodated net structural thickening during early exhumation, which was also identified in the CBU on Evia (Xypolias et al., 2012). However, while Xypolias et al. (2012) documented parallelism between peak fabric lineations and upright fold axes and interpreted that the two stages were genetically related, our observations on Syros suggest that the subduction-to-exhumation transition was locally associated with a distinct change in a stress state and kinematic rotation. This is because the trends of peak fabric lineations and retrograde fold axis lineations initially diverge by ∼45° and progressively rotate to ∼90° with increasing greenschist overprinting and exhumation-related strain (e.g., Azolimnos). This may reflect lateral and/or temporal changes in subduction and exhumation-related kinematics that are not yet understood or described in our hypothesized model, or changes in the degree of subduction obliquity, along-strike of the paleo-trench. However, these observations are consistent with a punctuated phase of underplating preceding return flow, as opposed to simultaneously-operating opposing thrust- and normal-sense shearing at the nose of a tapered extrusion wedge (Ring et al., 2020).

Our structural observations suggest that non-coaxial deformation on the eastern and southeastern sides of the island can be attributed to proximity to the Vari Detachment, which may have operated as the extensional subduction channel roof (Aravadinou & Xypolias, 2017; Laurent et al., 2016; Ring et al., 2020). The rocks in the immediate footwall of the detachment are retrogressed under blueschist-to-greenschist facies conditions and record strong top-down-to-the-E extensional shear consistent with regional core-complex kinematics. Therefore, if the Vari Detachment did originally operate as a subduction channel roof, it may have been thoroughly reworked and rotated during Oligo-Miocene extension, making it difficult to confidently discern prograde kinematics from the CBU rocks.

Compiled metamorphic geochronology and new Rb-Sr ages allow us to calculate exhumation rates of 1.5–5 mm/yr (=1.5–5 km/Myr) for each underplated nappe. These rates are roughly an order of magnitude slower than subduction for the Hellenides, and are consistent with buoyancy-driven, channelized return flow in a distributed shear zone (Burov et al., 2014; Gerya et al., 2002; Warren et al., 2008). Furthermore, mm/yr exhumation rates are not consistent with fast rates (comparable to subduction rates) predicted for exhumation along deep-reaching, highly-localized detachments in a downward-tapering ‘extrusion wedge’ (e.g., Ring et al., 2020; Ring & Reischmann, 2002), nor with forced return flow and melange-like mixing in a low-viscosity wedge (Cloos, 1982; Gerya et al., 2002). Thus, between ∼50 and ∼25 Ma, return flow in the subduction channel accomplished at least 35 km, and potentially as much as 55 km of vertical exhumation from maximum depths to the greenschist facies middle crust (∼4 kbar, ∼15 km), accounting for ∼75%–80% of CBU exhumation.

On a regional scale, subduction, underplating, and syn-subduction exhumation were fundamental processes during the construction of the greater Attic-Cycladic Complex (e.g., Jolivet et al., 2003; Laurent et al., 2017; Lister & Forster, 2016; Ring & Layer, 2003; Ring et al., 2020; Trotet, Jolivet, & Vidal, 2001; Uunk et al., 2022). CBU rocks on Sifnos have garnet crystallization ages of ∼47–45 Ma in meta-volcanosedimentary rocks (Dragovic et al., 2012, 2015), and ∼43 Ma in continental affinity rocks (Uunk et al., 2022), both of which are comparable to the base of the Syros stack. The Basal Unit exposed on Evia and Samos reached peak conditions at ∼24–22 Ma (Ring et al., 2001; Ring & Reischmann, 2002; Ring & Layer, 2003), contemporaneous with late stages of syn-subduction greenschist facies exhumation at the base of the CBU on Syros (Figure 15). The structurally deeper Phyllite-Quartzite Nappe and Plattenkalk unit exposed on Crete experienced HP/LT metamorphism between ∼24–20 Ma (Seidel et al., 1982; Thomson et al., 1999), which also overlaps with the latest stages of greenschist facies exhumation on Syros (Figure 15). Extension and core complex capture that initiated during trench rollback reworked the Attic-Cycladic Complex to its present configuration, and locally reactivated nappe-bounding thrusts as extensional structures (e.g., Vari Detachment on Syros).

10 Conclusions

Structural analysis, metamorphic petrology, and new and compiled geochronology suggest that exhumed HP/LT rocks on Syros Island (Cyclades, Greece) record progressive subduction, underplating, and return flow of three separate tectonic slices. Each nappe is interpreted to record a similar structural and metamorphic history, despite subducting at different times. Prograde subduction and underplating of each tectonic slice was characterized by asymmetric top-to-the-SSW and top-to-the-S shear strain and was reached at ∼53–52 Ma (northern nappe), ∼50–49 Ma (central nappe), and ∼45 Ma (southern nappe).

Prograde deformation and metamorphism are locally preserved in the northern and central nappes, but the majority of the island's meta-volcano-sedimentary lithologies were retrogressed during syn-orogenic blueschist-to-greenschist facies exhumation. The subduction-to-exhumation transition in each nappe is marked by systematic kinematic changes: dominant transport directions rotated from roughly N-S (syn-subduction), to NE (immediately post-underplating, at the subduction-to-exhumation transition), to E-W (later return flow). Strain geometry during return flow may change as a function of structural depth in a given coherent, exhuming sheet, with non-coaxial shear at the top and coaxial shear in the interiors. Rocks that experienced more penetrative greenschist facies overprinting appear to have experienced dominantly coaxial strain.

Progressive subduction of structurally deeper nappes occurred contemporaneously with exhumation of structurally higher nappes throughout the Eocene and Oligocene, capturing syn-subduction exhumation in the Hellenic subduction channel shear zone. The subducting plate-upper plate system experienced net compression in the Cyclades until (at least) the late Eocene. Subduction channel return flow proceeded at ∼1.5–5 mm/yr, which is an order of magnitude slower than subduction, and accounted for ∼80% of the vertical exhumation of the CBU. Continuous subduction, punctuated underplating, and syn-subduction exhumation appear to be fundamental processes during the construction of the Attic-Cycladic Complex in the Central and Southern Cyclades.

Acknowledgments

This work was funded by an NSF Graduate Research Fellowship awarded to A.K., an NSF Career Grant (EAR-1555346) awarded to W.B., an NSF Grant (EAR-1725110) awarded to W.B., J.B., and D.S., a Jackson School Seed Grant awarded to J.B., W.B., and D.S., Jackson School Graduate Research Fellowships awarded to A.K. and M.C, and Ford Foundation fellowship awarded to M.C. Many thanks to Staci Loewy and Aaron Satkoski (JSG, UT Austin) for help with Rb-Sr chemistry and isotope analyses, James Maner for assistance with the microprobe, and Emily Mixon for help with mineral separation. This project was part of A.K.’s Ph.D. dissertation and benefited from many conversations with Mark Cloos and Spencer Seman. The authors are grateful to Valentin Laurent, Uwe Ring, Michael Bröcker, Paris Xypolias, Laurent Jolivet, and Federico Rosetti for thoughtful and constructive reviews that improved this manuscript. We also thank Laurent Jolivet and Federico Rosetti for editorial handling.

Conflict of Interest

The authors declare no conflicts of interest relevant to this study.

Open Research

Data Availability Statement

The data that support the conclusions of this article are presented in the main text and in Supporting Information S1. Quantitative microprobe analyses are available on the EarthChem repository (https://doi.org/10.26022/IEDA/112282). Structural data and Rb-Sr geochronologic data are available on the ETH repository (https://doi.org/10.3929/ethz-b-000463143).