Low-grade retrogression of a high-temperature metamorphic core complex: Naxos, Cyclades, Greece

Retrogressive deformation and metamorphism are often reported from the main low-angle shear zones and detachments of metamorphic core complexes, but their importance is not sufficiently emphasized for the footwall interior. In order to contribute to a better understanding of exhumation-related retrogression processes within and at the top of metamorphic core complexes, an integrated detailed microstructural, textural, 40 Ar/ 39 Ar geochronological, and thermobarometric study on the Naxos metamorphic core complex within the Aegean Sea is presented that provides a new perspective on low-grade retrogression during exhumation through shallow ductile levels. We found variable retrogressive deformation within the Naxos metamorphic core complex, which even pervasively affected significant portions of the migmatite-grade metamorphic core and remnant high-pressure areas of the metamorphic core complex, where retrogression led to pervasive formation of new fabrics within greenschist-facies metamorphic conditions during brittle-ductile transition. Within a continuum of retrogression, 40 Ar/ 39 Ar white mica dating allowed us to deduce three retrogressive ages at 16.52 ± 0.39 Ma (within the Naxos metamorphic core complex), 12.6 ± 0.28 Ma (Moutsounas detachment shear zone on the eastern boundary of the metamorphic core complex), and 10.43 ± 0.44 Ma to 8.40 ± 0.76 Ma (last ductile activity along the Naxos-Paros shear zone to the north of the metamorphic core complex). A further stage of retrogression at 12–11 Ma occurred along distinct low-angle normal faults within the middle Miocene Naxos Granite. Retrogressive microstructures, low-temperature calcite fabrics in marbles, and chloritization in metapelites (at temperatures of ~350–130 ° C) in the metamorphic core complex core resulted mainly from late-stage E-W shortening and folding. Late-stage flow of hydrous fluids resulted in resetting of fabrics and enhancement of ductile deformation. The middle–late Miocene retrogression events are also reflected by a similarly aged tectonic collapse basin in the hanging-wall unit above the detachment. The wide temporal range of retrogression within the Naxos metamorphic core complex coincides in age with retrogressive deformation within other metamorphic core complexes of the Aegean Sea. We interpret the long temporal range of retrogression to reflect outward, southwestward retreat of the subduction and sequential activation of major detachment zones.


INTRODUCTION
In tectonic reconstructions, the recognition of exhumed crust is critically important because such rocks provide information on the tectono-thermal history of the crust. This is particularly the case for metamorphic core complexes with plastically deformed rocks exhumed from middle-to lower-crustal levels to the surface (e.g., Whitney et al., 2013, and references therein;Platt et al., 2015, and references therein). Many details are known concerning the exhumation history and structures related to exhumation, juxtaposing the typically high-temperature metamorphic core complex against older metamorphic and sedimentary upper-plate rocks. Deformation stages within metamorphic core complexes are the result of various superimposed processes, which range from initial viscous deformation to brittle deformation and also include synkinematic fluid flow (e.g., Lister and Davis, 1989;Verdel et al., 2007;Kargaranbafghi et al., 2012). Exhumation can channelize the flow of hydrous fluids and limit pervasive retrogres-sion of metamorphic complexes along the upper margins, particularly close to the brittleductile boundary (e.g., Siebenaller et al., 2013;Whitney et al., 2013;Gébelin et al., 2014;Methner et al., 2015, and references therein).
Low-temperature retrogressive deformation and metamorphism of high-grade metamorphic fabrics are often reported from detachments of metamorphic core complexes (e.g., Mehl et al., 2007;Harigane et al., 2008;Hetzel et al., 2013;Whitney et al., 2013). The detachment fault is characterized by retrogressive shear fabrics and forms under decreasing temperature-pressure conditions from usually ductile deformation within amphibolite-facies conditions (>500 °C) through the brittle-ductile transition (250-400 °C) to purely brittle conditions (e.g., Whitney et al., 2013). This is expressed by chloritization of mafic minerals, sericitization of feldspars, and formation of chlorite breccia at the top by pervasive fluid flow (e.g., Cathelineau and Nieva, 1985;Cathelineau, 1988;Kirschner et al., 1996;Dunlap, 1997;Reddy and Potts, 1999). In some metamorphic core complexes, low-grade retrogression may also occur in the interior during exhumation, and several studies have reported local retrogressive shear zones (e.g., Urai et al., 1990;Urai and Feenstra, 2001;Parra et al., 2002). Although previous workers have already noted this retrograde deformation in the footwall of metamorphic core complexes, they have not sufficiently emphasized its importance. The importance of widespread retrogression in the interior of metamorphic core complexes and its temporal and structural relationships to detachments are the focus of this study.
In this study, based on structural field work, we completed detailed microstructural and textural investigations, thermobarometric calculations, and 40 Ar/ 39 Ar white mica dating of retrogressive fabrics to reveal the significance of lowgrade retrogression of high-and medium-grade metamorphic and granitic rocks. These results, combined with previous studies, provide new insights into the regional retrogression associated with ductile to ductile-brittle deformation and cooling of the Naxos metamorphic core complex. The new data also allow fresh insights into the mode and history of tectonic extension in the Aegean region during Neogene times, and the methodology could possibly be transferred to other metamorphic core complexes.

GEOLOGY OF THE SOUTHERN AEGEAN SEA REGION
One of the most striking features within the south-central Aegean region is the sequential north to south formation of numerous metamorphic core complexes along the North Cycladic (e.g., Andros, Tinos, Mykonos, and Ikaria), Central Cycladic (including the Naxos-Paros detachment), and West Cycladic (Sifnos, Serifos, Kea) detachments ( Fig. 1; Jolivet et al., 2010;Grasemann et al., 2012). The development of these metamorphic core complexes has been linked to N-S extension of the Aegean lithosphere in the backarc domain of the Hellenic subduction zone (Lister et al., 1984;Jolivet et al., 2003;Krohe et al., 2010, and references therein) in the early Neogene (ca. 23 Ma; Gautier et al., 1993;Tirel et al., 2008;Ring et al., 2007Ring et al., , 2010. Extension was and is triggered by the southward retreat of the African slab (e.g., Le Pichon and Angelier, 1981;Jolivet and Faccenna, 2000;Huet et al., 2009Huet et al., , 2011Jolivet et al., 2010;Ring et al., 2010;Grasemann et al., 2012).

GEOLOGY AND EXHUMATION HISTORY OF THE NAXOS METAMORPHIC CORE COMPLEX
Since its recognition as a metamorphic core complex (Lister et al., 1984), extensive research on the classic Naxos metamorphic core complex has provided a wealth of information on its structure, thermal and geomorphologic evolution, and tectonic implications (e.g., Jansen and Schuiling, 1976;Duchêne et al., 2006;Schenk et al., 2007;Vanderhaeghe et al., 2007;Seward et al., 2009;Kruckenberg et al., 2011;Cao et al., 2013a;Siebenaller et al., 2013). Here, we focus on a summary of the main points needed for the purpose of this study.

Structure of Naxos
The classic Naxos metamorphic core complex lies in the Cyclades in the western Aegean Sea and is associated with the Naxos-Paros ductile low-angle fault (e.g., Lister et al., 1984;Gautier et al., 1993) along the northern margins of Naxos and Paros, which is part of the Central Cycladic detachment system ( Fig. 1; Jolivet et al., 2010). The main structure of the Naxos metamorphic core complex is a migmatite-cored gneiss dome with marbles, calc-schists, phengite-schists, orthogneisses, migmatites, and amphibolites, and this is structurally overlain by a low-angle normal fault in the east and a steep strike-sliptype fault zone in the west (Fig. 2). The western part of Naxos exposes the Miocene Naxos Granite (e.g., Pe-Piper et al., 1997;Koukouvelas and Kokkalas, 2003). The hanging-wall unit of the Naxos metamorphic core complex and Naxos Granite consists of remnants of unmetamorphosed Mesozoic ophiolites and overlying Miocene and Pliocene sedimentary rocks (Kuhlemann et al., 2004), which form a synform between the Naxos metamorphic core complex and the Naxos Granite.

Metamorphic Stages and Subsequent Cooling
Two main successive Alpine metamorphic events affected the metamorphic zoning within the Naxos metamorphic core complex (e.g., Andriessen et al., 1979;Urai et al., 1990;Keay et al., 2001). The earlier HP-LT Eocene metamorphism (M1) phase is recorded by relics of blueschist-facies assemblages preserved in metamorphic rocks in the southern part of the island of Naxos (~12 kbar, 470 °C, ~13 °C/km; Wijbrans and McDougall, 1986;Avigad and Garfunkel, 1991;Avigad, 1998). No field gradient is known because of overprint of the M1 phase by M2 metamorphism in central and northern Naxos. The 40 Ar/ 39 Ar white mica ages of ca. 45 ± 5 Ma document an Eocene age for blueschist metamorphism on Naxos (Andriessen et al., 1979;McDougall, 1986, 1988). Martin et al. (2006) presented evidence for M1 high-pressure metamorphic conditions within the high-grade rocks affected by M2 metamorphism.
After initial attempts with K-Ar ages giving results of ca. 15 Ma (hornblende) to ca. 11 Ma (mica; Jansen and Schuiling, 1976;Jansen, 1977), the timing of peak Barrovian-type M2 metamorphism is dated now by Rb/Sr mica and garnet ages  and U-Pb zircon rim ages to range from 19 to 14 Ma (Keay, 1998;Keay et al., 2001;Martin et al., 2006Martin et al., , 2008. An age of 20-25 Ma for the M2 greenschist-facies retrogression has been proposed (Andriessen et al., 1979;Andriessen, 1991), whereas Wijbrans and McDougall (1988) reported 40 Ar/ 39 Ar white mica ages of 27 and 19.9 Ma, which they interpreted to possibly relate to partial retrogression after M1 metamorphism. Duchêne et al. (2006) dated recrystallized phengitic white mica in a sample from zone III (Fig. 2) and interpreted the 29.3 Ma age to represent the retrogressive evolution from peak M1 to M2 recrystallization, which is dated at 22.7-19.8 Ma by K/Ar, 40 Ar/ 39 Ar, and Rb/Sr on muscovite.
Recently, Siebenaller et al. (2013) studied fluids from quartz veins, mainly from the southern part of the M2 migmatite core. They distinguished immiscible CO 2 -rich fluid and a high-salinity brine generated by metamorphic reactions and magma crystallization, respectively. Fluids were initially trapped at 625 °C and 400 MPa and then remobilized during ductile deformation at ~350 °C and 140 MPa. Brittle microcracks contain aqueous meteoric fluids trapped at 250 °C and 80 MPa. This peculiar retrograde pressure-temperature (P-T ) pathway implies that the brittle-ductile transition zone separated two fluid reservoirs, namely (1) the ductile crust, into which fluids originating from crystallizing magmas and fluids in equilibrium with metamorphic rocks circulated at a geothermal gradient of 30 °C/km at lithostatic pressure, and (2)  meteoric fluids percolated along a high geothermal gradient of 55 °C/km at hydrostatic pressure (Siebenaller et al., 2013).

MAIN RETROGRESSIVE DEFORMATION STRUCTURES WITHIN THE NAXOS METAMORPHIC CORE COMPLEX
In this section, we describe the retrogressive deformation structures within the Naxos metamorphic core complex from field observations and macroscopic structural analysis. All microstructural details from samples in the interior and boundary shear zones of the Naxos metamorphic core complex and the Naxos Granite are compiled in Table 1 and the section titled "Retrogressive Microfabrics." In the field, zones of retrogression are characterized by the presence of quartz and calcite veins and variable chloritization of otherwise chlorite-free lithologies.
Mineral deformation features and rock microstructures were characterized with special reference to feldspar, quartz, white mica, biotite, calcite, and chlorite. Texture measurements were performed on quartz and calcite (section on "Crystallographic Preferred Orientations Related to Retrogressive Fabrics"). The deformation and metamorphism conditions were estimated using the microstructure, texture, and mineral thermometry (section on "Conditions of Metamorphism during Retrogression") followed by white mica 40 Ar/ 39 Ar dating results ("White Mica 40 Ar/ 39 Ar Dating" section).

Structures of the Internal Naxos Metamorphic Core Complex
Although high-pressure mineral relics are preserved in the M1-dominated zone, our detailed ductile structural analysis, particularly of calcite marbles and phyllitic rocks, reveals that there is no structural difference between these two M1-and M2-dominated domains. Both units are pervasively affected by retrogression and show similarly oriented ductile fabrics, including the NNE-trending stretching lineation. All rocks have a pervasive foliation (S) and a similarly oriented NNE-or SSW-plunging stretching lineation (Fig.  2E). Bookshelf domino-type structures ( Fig. 3A) indicate top-to-the-NNE transport. Boudinage parallel to the NNE-SSW-trending stretching lineation of rheologically stiff layers within calcite marbles is a common phenomenon (Fig. 3B). In the center of the migmatite dome, S-type granites intruded, as well as aplite and granitic dikes. These, as well as the penetrative foliation, are folded around N-S-trending fold axes (Figs. 3C and 3D) and therefore postdate peak conditions of S and L formation. The folded aplitic dikes and country rocks often reveal a steep approximately N-S-striking axial plane foliation (Fig.  3E), which is less intense in the south.
Mostly subvertical granite dikes are variably deformed and folded (see previous) or postdate the penetrative foliation (Fig. 3F). In the latter case, these dikes often strike N-S and bear a weak subvertical foliation similar in strike to the axial plane foliation. Subvertical, mostly E-W-striking, calcite-filled veinlets and extensional veins filled with coarse epitaxial (rim) and isometric calcite crystals occur within the marble sometimes (Figs. 3G-3I). They document approximately N-S stretching and fluid flow during late stages of the exhumation of the Naxos metamorphic core complex.

Western Boundary Shear Zone at Melanes
In the surroundings of Melanes, at the western border of the Naxos metamorphic core complex, a subvertical belt of fine-grained N-S-striking mylonites is exposed (here termed the Melanes shear zone; Figs. 2B, 2D, and 4), and these form the western, structurally upper boundary of the Naxos metamorphic core complex, against which the unmetamorphic ophiolites are juxtaposed to the west (Fig. 2B). The Melanes shear zone is characterized by mylonitic rocks (amphibolite, paragneiss, and marble) with a gently W-dipping foliation and a penetrative subhorizontal N-S-trending stretching lineation. The microboudins and late extensional veins (quartz, chlorite) and microfaults are filled with chlorite (Figs. 4A and 4B).

Naxos Granite
At the southern end, the contact between the hornblende-biotite granite and schists in the Naxos metamorphic core complex is exposed (Fig. 4C), with the foliation of the country rocks predating granodiorite intrusion. There, the granite and mafic enclaves exhibit a magmatic and/or high-temperature N-S-striking solidstate foliation and a stretching lineation defined by elongated feldspars, amphibole, and biotite (Figs. 4D-4G). The stretching lineation plunges moderately to the south. Internally, undeformed granite dikes postdate the foliation. Along its northern and eastern margin, the granodiorite shows abundant signs of retrogression, mainly chloritization along shear zones and faults and chloritized protocataclasites, with some ductile overprint during retrogression at the eastern margin (Figs. 4D and 4E), as well as formation of abundant partially deformed quartz veins.

RETROGRESSIVE MICROFABRICS
More than 50 samples from different sections across Naxos were analyzed (Table 1; for locations, see Fig. 2). All thin sections of samples were cut parallel to the kinematic x-z section (that is, parallel to the stretching lineation and normal to the foliation). Here, we distinguish retrogressive microfabrics within (1) the relict high-pressure rocks in the southeastern and the interior parts of the Naxos metamorphic core complex, (2) the Melanes shear zone, and (3) the Naxos Granite.

Retrogressive Microfabrics within Relict High-Pressure Rocks of the Southeastern Region and in the Internal Naxos Metamorphic Core Complex
At least two deformation episodes are well preserved in the mica schists of SE Naxos. Early deformation is characterized by elongated garnet and feldspar grains and quartz grain aggregates, which imply deformation at relatively high-temperature conditions (Figs. 5A-5D). These rocks also underwent high-pressure metamorphism (e.g., detailed white mica chemistry can be found in Cao et al., 2013a). Retrogression, during a subsequent stage of shearing, is superimposed on the earlier high-temperature microstructures.  Stretched early garnet grains with intragranular fractures are filled with irregular biotite, calcite, quartz, and chlorite (Figs. 5A and 5B). The quartz microstructures were affected by lowtemperature deformation, as evidenced by fine recrystallized grains formed by bulge recrystallization and rare occurrences of small grains (new grains or subgrains) at the triple junctions of the larger quartz grains (Figs. 5C and 5D). A second generation of recrystallized white mica includes fine grains generally concentrated in microscale shear zones (Figs. 5C and 5D). Calcite grains in marble are characterized by coarse grains (Fig. 5E) formed by high-temperature growth that are often dominated by thin twins. However, the coarse grains show serrated and irregular boundaries and recrystallized small grains around the elongated relict grains.
Microstructures show early high-temperature plastic deformation of the main mineral phases, which still are locally preserved in these mylonitic rocks (Fig. 5F) and in coarsegrained marbles (e.g., Figs. 5G and 5H) located near migmatites. Quartz grains, often in ribbon and elongated grain aggregates, are the most common high-temperature microstructures in the sheared schists. Ribbon or coarse-grained quartz is replaced by bulged recrystallized fine grains during retrogression. Some relict early coarse biotite grains surrounded by fine grains of quartz and biotite show sigma fabrics indicating a top-to-the-N shear sense. Early host white mica grains have been progressively deformed by recrystallization to an aggregate of finegrained mica neoblasts oriented parallel to the mylonitic foliation and lineation in shear zones (Fig. 5F). Coarse calcite grains formed during high-temperature deformation exhibit serrated and irregular boundaries, with recrystallized grains showing straight to diffuse and irregular grain boundaries (Figs. 5G and 5H). These larger grains bear several sets of twins or show polyphase twinning (twins in twins) at different angles, possibly indicating that they were caused by multiple deformation stages. Type III and IV twins (according to Ferrill et al., 2004) only occur in large grains. Most calcite twins are thin and straight, similar to types I and II of Ferrill et al. (2004), and are related to low-temperature deformation. Shear bands, sigma-type mica porphyroclasts, and mica fish consistently indicate bulk top-to-the-N shearing related to low-temperature retrogressive deformation.

Retrogressive Microstructures within the Melanes Shear Zone
Mylonitic amphibolite of the Melanes shear zone (Fig. 4) shows relict lens-or fish-shaped amphibole porphyroclasts, homogeneously fine-grained amphibole, biotite, and quartz aggregates, constituting the foliation and stretching lineation (Figs. 6A and 6B). Undulatory or inhomogeneous extinction occurs in relict porphyroclasts. A few relict elongated feldspar grains show cataclastic fabrics (Fig. 6A). Dramatic grain-size reduction of the main minerals (0.1 mm) is observed microscopically within mylonitic amphibolite and even in mica schist (Figs. 6A-6B) and is interpreted to be the result of a late-stage regional low-temperature retrogressive event. The microstructure of mylonitic marble is dominated by elongated, undulose, twinned, and kinked calcite porphyroclasts surrounded by dynamically recrystallized grains producing a typical core-mantle structure. Thin twins are widespread, and undulose extinction is conspicuously absent in these calcite grains. In marbles, further retrogressive microfabrics are present in the extensional veins filled by chlorite, quartz, and/or calcite (Figs. 6C and 6D). Most sheared chlorite aggregates form a new continuous or discontinuous foliation (Fig. 6D).

Retrogressive Microstructures within the Naxos Granite
The granodiorites at margins of the Naxos Granite body experienced shear deformation to form protomylonitic rocks (Fig. 6E) containing K-feldspar, plagioclase, amphibole, quartz, and chlorite. Incipient S-C fabrics are present in the mylonitic granodiorite, with the S plane defined by grain aggregates of K-feldspar and plagioclase, amphibole, quartz, biotite, and chlorite, and the C plane defined by strongly elongated K-feldspar and plagioclase grains (Fig. 6E). The acute angle between S and C is generally 10°-25°. The protomylonitic S-C fabric relates to late solid-state low-temperature deformation conditions. The alignment of feldspar crystals suggests that the fabric partly represents a magmatic foliation, but recrystallization of quartz indicates that solid-state deformation has been superimposed on the magmatic grains. Microcracks infilled by new-formed chlorite (Fig. 6E) in some feldspar porphyroclasts formed semibrittle to brittle microstructures during retrogression.

Quartz CPOs
Quartz fabrics are generally demonstrated by c-axis crystallographic preferred orientation (CPO) patterns or c-axis pole figures (Fig. 7). Quartz CPOs analysis was conducted on 18 mylonitic quartz-rich rocks from the Naxos metamorphic core complex (Table 1), of which six samples are from an earlier paper (Cao et al., 2013a). Four basic types of quartz c-axis fabrics are recognized, which reflect different quartz microstructures: type I, close to y maxima; type IIa, crossed girdles with maxima close to the y axis; type IIb, crossed girdles with maxima between the x and z axes; and type III, single girdles (Schmid and Casey, 1986).
Fabrics of type I are characterized by one or two maxima close to the y axis (Fig. 7, e.g., N52, N57, N24), which resulted from prism <a> slip, a medium-temperature deformation mechanism (Stipp et al., 2002). However, some maxima are elongated, which is likely the result of a combination of rhomb <a> slip and low-temperature basal <a> slip (Schmid and Casey, 1986;Stipp et al., 2002). The low-temperature overprint on minerals in these samples is also indicated by the microstructure, including bulging recrystallization of quartz, extreme grain-size reduction, and formation of white mica fish and of a second generation of fine-grained white mica. Krabbendam et al. (2002) claimed that some fine-grained quartz-rich shear zones were actually high grade, and the grain size was pinned by sec ondary phases (graphite). The maxima are primarily distributed asymmetrically around the y axis, which may be related to a late stage of noncoaxial shear, although the relationship between the quartz crystallographic fabric geometry and finite strain geometry is not straightforward (Sullivan and Beane, 2010). Type II fabrics are characterized by complex asymmetrical superposition fabrics with one or two maxima close to the y axis and asymmetrical girdle axis (type IIa: N64, N65, N33, N29A, N1C) and crossed girdles with maxima between x and z (type IIb: N60, N58D, N33B, N7C, N32, N9C). Type IIa fabrics present the maximum in the center (y) and the weaker submaxima distributed on a great circle (x-z), between the maxima, with a large opening angle of 65° and a smaller opening angle between 20° and 45°, indicative of high-to middle-temperature deformation with dominant combined prism <c> and rhomb <a> slip (e.g., Schmid and Casey, 1986). The weak submaxima in the fabrics considered here were attributed to a basal <a> slip low-temperature superimposition during subsequent deformation (Stipp et al., 2002), consistent with microscopic observations. Type IIb fabrics shows a weak maximum close to y, and one or two maxima are distributed in the x-z plane. The patterns indicate activity of basal <a> slip and weak rhomb <a + c> slip systems, with less frequent prism <a> slip, suggesting a strong low-temperature deformation overprint during medium-temperature deformation. These asymmetric girdle fabrics passing through the y-direction are observed in the mylonitic rocks, which are ascribed to coaxial strain (Passchier and Trouw, 2005). Type III fabrics are characterized by asymmetric single girdle fabrics passing through the y-direction, and the concentrations of c-axes are close to the z axis (N66, N28, N11A, N1A). The single girdle patterns show clockwise rotation with respect to the x-z plane, suggesting dominant activation of <a> basal slip at lower temperatures, possibly within lower-greenschistfacies conditions. The slightly asymmetric pat-terns record a significant component of noncoaxial top-to-the-NNE sense of shear deformation (Fig. 7), consistent with information presented by Krabbendam et al. (2002).

Calcite CPOs
Calcite CPOs analysis was conducted on 24 mylonitic marbles from the Naxos metamorphic core complex. All calcite c-axis textures and sample locations are presented in Figures 8 and 9.
Calcite microstructures of these samples from the interior of the Naxos metamorphic core complex show protomylonitic fabrics with coarse calcite porphyroclasts surrounded by later recrystallized smaller grains. In most samples, coarse grains show serrated and irregular boundaries and are elongated slightly oblique to the foliation. The CPO is characterized by

(C) Chlorite (Chl)-filled vein crosscutting the foliated marble. (D) Coarse-and fine-grained chlorite in the mylonitic marble. (E) S-C fabric in the mylonitic granodiorite, where S plane is defined by grain aggregates of two feldspars (K-feldspar [Kfs] and plagioclase [Plag]), quartz, and chlorite, and C plane is defined by strongly elongated grains of two feldspars. (G) Sheared chlorite indicating shear sense.
a broad c-axis <0001> maximum close to the z plane and a weak girdle in y-z, which is subnormal to the foliation plane. The <a> axis is distributed within a girdle in the x-y shear plane, and the e planes are relatively weakly distributed on a similarly oriented axis. This supports the fact that the calcite texture presented in this paper developed by intracrystalline slip (dominated by c slip) and e-twinning under low-temperature conditions (Cao et al., 2013a). In calcite, the c-axis maximum rotated against the shear sense with respect to the main shear plane observed in porphyroclasts and recrystallized grains. The asymmetry indicates top-tothe-N-directed shear. Most samples from the eastern boundary are ultramylonites with a fully recrystallized microstructure. The small calcite grains from most samples show strong CPOs with a single c-axis maximum subperpendicular to the foliation. The a planes form a girdle subparallel to the foliation, and e planes exhibit a weak single maximum similar to the c axis and are orientated nearly perpendicular to the foliation plane.

Hornblende-Plagioclase Thermometry of Granodiorite (Naxos Granite)
The Al content in hornblende (aluminum-inhornblende barometry) is commonly used with coexisting hornblende and plagioclase (horn-blende-plagioclase thermometry) in a paragenesis also containing quartz and K-feldspar to constrain the pressure and temperature during the crystallization of an intermediate magma or subsequent metamorphism. Chemical compositions of plagioclase and hornblende of sample N72 from a granodiorite of the Naxos Granite are given in Table DR1. 1 Based on hornblende solid-solution models and well-constrained natural and experimental systems, the temperature was calculated by application of the hornblende-plagioclase geothermometer of  Holland and Blundy (1994) and the pressure by the Al-in-hornblende geo barometer of Schmidt (1992), Anderson (1996), and Ridolfi and Renzulli (2012). The calculated temperature is in the range from 661 °C to 583 °C, and the average is 621 ± 19 °C. The calculated pressures mainly range from 1.70 to 1.24 kbar, and the average is 1.5 ± 0.2 kbar, representing the lithostatic pressure during emplacement of the granodiorites of the Naxos Granite (Gautier et al., 1993).

Two-Feldspar Thermometry
A few samples from the Naxos metamorphic core complex and foliated Naxos Granite contain both sodic and potassic feldspar. The thermometry for samples N71, N53, and N61 was done using microprobe analyses of two coexisting feldspars according to the method of Perchuk et al. (1991). In the case of the Naxos Granite, we used the pressure during intru sion of ~1.5 kbar to calculate the disequilibration of temperature (Table DR2 [see footnote 1]). For sample N71, with coarse-grained colorless plagio clase and flesh-colored K-feldspar, we calculated a temperature range from 539 °C to 238 °C, which implies reequilibration under variable temperatures after solidification (Table DR2 [see footnote 1]). Samples N61 and N53 are orthogneisses from the core of the Naxos metamorphic core complex, for which the calculated equilibration temperature ranges from 438 °C to 258 °C (N61), and from 334 °C to 214 °C (N53), suggesting low-temperature mineral equilibration.

Chlorite Thermometry-Related Microfabric Types
High-resolution backscatter electron (BSE) analyses allow better definition of the details of microfabrics in various types of chloriterich sheared rocks, discussed earlier (Table 1). Two deformed microfabric types were observed for the occurrences of chlorite and white mica from the samples (Fig. 10), (1) type I sheared, coarse-grained chlorite and white mica (D 1 ); and (2) type II sheared, fine-grained chlorite and white mica in the microshear zones (D 2 ).

White Mica Compositions and Barometry
According to microstructures based on BSE analysis (Fig. 11), intensely recrystallized fine-grained white micas were observed along discrete microshear zones in our dated samples (see next section). The compositional results of representative analysis of white mica from the 39 Ar/ 40 Ar dated samples (N33B, N1A, N1C, N24) are given in Table DR4 (see foot-note 1). The analysis of white mica from the four samples indicates the compositions of various types of white mica reflect diverse origins and times of growth. White micas in sample N33B have a dominance of low Si at ~6.08-6.17 a.p.f.u. and, therefore, muscovite composition. One grain presented a high Si value at ~6.69 a.p.f.u. with phengitic composition. White micas from samples N1A and N1B yielded a wide range of compositions of Si from ~6.23 to 6.40 a.p.f.u. and have, therefore, muscovite composition with a low Si a.p.f.u. and phengite composition with a low Si a.p.f.u., respectively Coarse-grained white micas in sample N1C yielded a dominance of high Si at ~6.55-6.97 a.p.f.u. and have, therefore, phengite composition. Some phengitic compositions with high-Si white micas also occur in the sheared zones. Fine-grained white micas in sample N24B yielded a dominance of low Si at ~6.14-6.18 a.p.f.u. and, therefore, muscovite composition. We note that phengite is, therefore, inherited from high-pressure metamorphism, and it was only rarely reset to low-pressure muscovite during shearing and exhumation. The muscovite grains with a low Si value at ~6.2 a.p.f.u. (Table  DR4 [see footnote 1]) yielded mainly lowpressure conditions (<2.5 kbar).

WHITE MICA 40 Ar/ 39 Ar DATING
White mica concentrates were prepared from four samples from the southern relict high-pressure rocks (sample N33B), the eastern boundary (samples N1A and N1C), and the northern part (sample N24). The 40 Ar/ 39 Ar analytical technique is described in Appendix A3, and the analytical results are given in Table DR5 (see footnote 1) and Figure 12. The compositions of white micas in dated samples were also analyzed with electron microprobe (see previous sections).
Sample N33B from the southern relict high-pressure belt, south of the corundum-in isograd of M2 metamorphism, is a deformed fine-grained mica schist (Figs. 12A and 12B). Small, newly formed white mica grains (<250 µm) that crystallized during retrogressive deformation were sampled to investigate the extent of resetting of the Ar isotopic system in the preexisting white mica. The Ar release pattern resulted in a plateau age of 16.52 ± 0.39 Ma (steps 5-10 including 87% of 39 Ar released; Fig. 12A). Isotope inversion resulted in an age of 16.3 ± 1.2 Ma (with a 40 Ar/ 36 Ar initial value of 324 ± 120 Ma; Fig. 12B). We therefore consider the plateau age of 16.52 ± 0.39 Ma to be geologically significant and to date the formation age during retrogression within greenschist-facies conditions after peak M2 metamorphic conditions. Sample N1A is a mylonitic quartzite close to the upper boundary of the Moutsounas shear zone along the eastern boundary of the Naxos metamorphic core complex. In this sample, mica grains are fully recrystallized and mostly fine grained (<100 µm), and almost no porphyroclasts are present (Figs. 12C and 12D). Polycrystalline and elongated quartz aggregates are common due to recrystallization. The recrystallized quartz and white mica aggregates with opaque minerals form a continuous or discontinuous foliation and stretching lineation. The argon release pattern of the white mica concentrate of rare larger grains results in a pattern with two age groups. A plateau age of 14.34 ± 0.36 Ma (steps 2-6 including 65.1% of 39 Ar released; Fig. 12C) can be juxtaposed with another mean age of 20.1 ± 1.4 Ma (steps 7-11 including 30% of 39 Ar released). Isotope inversion of all steps resulted in an age of 12.6 ± 0.28 Ma (with a 40 Ar/ 36 Ar initial value of 713 ± 240 Ma; Fig. 12D), showing significant excess argon. We think that the sample consists of micas of two different age populations, although we cannot exclude the presence of excess Ar in the older population. We consider that the to plateau age of 14.34 ± 0.36 Ma of the low-temperature steps approximately dates Ar loss of older grains during shearing.
In the thin section of sample N1C, we observed that the mica grains are substantially more affected by recrystallization than micas from adjacent rocks (e.g., sample N1A). The age spectrum of sample N1C shows a staircase pattern ranging from an age of ca. 5 Ma to a maximum of 50 Ma, all with high analytical uncertainties because of the low gas yield due to the small grain size (Fig. 12E). We interpret this pattern as mixture between two or three age components, including a possibly untrustworthy young age component at 5-14 Ma (with a too low radiogenic component of <12% 40 Ar*), one at ca. 50 Ma, and a third that could be deduced from a mean age of 30.7 ± 3.5 Ma from the last four steps. To assess potential age components, we performed single-grain total fusion experiments on several further single grains (with only one or five steps because of the low gas yield due to the small grain size and young age). The experiments N1C-exp1 and N1C-exp2 (Figs. 12F and 12G) show staircase patterns, and ages as low as ca. 13.9 ± 0.9 Ma and 11.9 ± 0.8 Ma in the initial steps of the experiments and wellconstrained ages of 24.5 ± 2.0 Ma and 19.3 ± 1.0 Ma. Finally, a total fusion experiment (experi ment 3) on a small single grain yielded an age of 38.3 ± 4.2 Ma. The staircase patterns clearly show Ar loss of older mica grains, which possibly formed during M2 or M1 (experiment 3) metamorphism (Figs. 12E and 12F) during final shearing at elevated temperatures. This means that this sample includes a variety of age populations with old ages and two younger populations, which is the same as the isochrones of the nearby sample N1A at 12.6 ± 0.28 Ma and 20.1 ± 1.4 Ma. Together, the experiments show a mixture of inherited high-pressure grains (38.3 ± 4.2 Ma) and of newly recrystallized grains during shearing. The age of the latter is reasonably constrained by Ar loss between 13.9 ± 0.9 Ma and 11.9 ± 0.8 Ma. Sample N24 is a mylonitic aplite gneiss from the northern part close the Naxos-Paros detachment. Microstructures show strong plastic deformation of the main mineral phases (e.g., quartz, white mica, feldspar) in the sample. Sheared porphyroclasts and recrystallized white mica are surround feldspar porphyroclasts. Two experiments on multiple grains were performed. White mica from sample N24-exp1 gave a mean age of 6.71 ± 0.34 Ma for steps 2-4 and a plateau age of 9.37 ± 0.37 Ma (steps 5-8, 50% of 39 Ar released; Fig. 12H). Isotope inversion resulted in an age of 9.2 ± 4.1 Ma (with a 40 Ar/ 36 Ar initial value of 330 ± 310 Ma; Fig. 12I). We consider the plateau age of 9.37 ± 0.37 Ma to be geologically significant and to date the time of shearing. The largest white mica grains were selected to perform another experiment on sample N24. The experiment N24-exp2 gave a somewhat scattered Ar release pattern, which allowed the calculation of a mean age of 8.40 ± 0.76 Ma (steps 2-6) and a plateau age of 10.43 ± 0.44 Ma (steps 7-12; Fig. 12J). Isotope inversion resulted in an age of 11.56 ± 0.82 Ma (with a 40 Ar/ 36 Ar initial value of 140 ± 58 Ma; Fig. 12K). We think that the results of both experiments are geologically meaningful, showing shearing, which started at 10.43 ± 0.44 Ma and possibly lasted until 6.71 ± 0.34 Ma.

Impact of Retrogression on Microfabrics
As described already, in the Naxos metamorphic core complex, retrogression has been documented along some local shear zones (e.g., Urai et al., 1990), but regional retrogression is still rarely discussed in more detail. It is especially noteworthy that the presence of lowtemperature overprinted fabrics in these localities indicates that retrogression was not only restricted to the boundary shear zone of Naxos (e.g., Cao et al., 2013a), but it also affected large domains of the entire metamorphic sequence of the Naxos metamorphic core complex. The metamorphic rocks show abundant microstructures that can be readily related to retrograde processes. These processes include development of fine quartz grains by bulging recrystallization, brittle deformation of feldspar, coarse calcite grains with irregular boundaries and thin mechanical twins, homogeneous grain-size reduction of biotite, and the occurrence of discrete microshear zones with fine-grained white mica, which relate to the regional deformation/ exhumation. Displacement and/or pull-apart of fractured feldspar porphyroclasts in the quartz and biotite matrix indicate that shearing continued under low-temperature (lower greenschist-facies) conditions. Textures of quartzofeldspathic mylonitic gneisses and marbles are consistent with extensive late retrogressive activity. The compositions of muscovite and chlorite reflect a diverse origin and time of growth. Conspicuously, inherited grains from M1 and M2 stages of metamorphism are often preserved in mylonites. The presence of secondary chlorite and albite in boudins and shear bands, and the absence of annealing of the highpressure mineralogy indicate that later stages of deformation also occurred at relatively low P-T conditions. The orthogneisses and aplitic dikes may be explained by a sequence in which early high-temperature mylonitic fabrics and microstructures were overprinted by semibrittle to brittle adjustments during exhumation through the ductile-brittle transition. Shear sense criteria in these rocks indicate top-to-the-N sense of motion, as documented in other parts of the island in relation to extension (Urai et al.,1990;Buick, 1991aBuick, , 1991b, but late shear bands showing top-to-the-S sense of shear are also well developed (Avigad, 1998). The Naxos metamorphic core complex can be interpreted to have developed finally under retrogressive conditions so that the zone of strongest shearing narrowed toward the margins of the Naxos Granite, suggesting that strain localization occurred during cooling of the granitic rocks. Within the Naxos metamorphic core complex, the interior is also strongly affected by retrogression. Retrogression in the migmatite core occurred by inflow of meteoric hydrous fluids along approximately E-W-trending quartz veins, for which Siebenaller et al. (2013) found P-T conditions at 250 °C and 80 MPa. Chlorites within mylonitic phyllites are found to lose stability even within sub-greenschist-facies metamorphic conditions during shearing, and a succession of chlorite formation from higher to lower temperatures can be found. A reason could be the interaction of older mineral assem-blages with hydrous fluids, possibly hydrothermal fluids along the zone of shearing and retrogression during brittle-ductile transition. This is believed to reset the chemical composition of feldspar, chlorite, and white mica. We therefore take the temperatures of the chlorite thermometry as approximate values of the final stages of retrogression, and these values are consistent with temperatures found in the fluid inclusion study of Siebenaller et al. (2013). The data show that earlier microfabrics such as those in microlithons (type 1) yield higher temperatures than the subsequent sheared chlorites (type 2). If the temperatures are approximately correct, this would suggest that ductile shear in phyllosilicate-rich rocks and marbles possibly occurred at sub-greenschist-facies metamorphic conditions (at temperature ranging from 320 °C to 215 °C).

Retrogression and Geological Significance of the Melanes Shear Zone
The W-dipping Melanes zone is a retrogressive shear zone, along which amphibolite-grade metamorphic rocks were retrogressed during NNE-directed shearing. The ductile deformation of the metamorphic rocks in the Melanes shear zone was accompanied by extreme grainsize reduction of minerals during the retrogressive event. An estimate of the temperature conditions during retrogressive deformation can be inferred by the features of the deformed behavior of the minerals in shear zones (Simpson, 1985). The temperature is also estimated, by the chlorite thermometry, at low-grade conditions within marble (300-400 °C). The Melanes shear zone forms the eastern limb of a synform between the Naxos metamorphic core complex in the east and the Naxos Granite in the west, in which the bedding of Lower Miocene clastic rocks is in part in a nearly subvertical position. Consequently, the Naxos-Paros detachment fault, including the Melanes shear zone, seems to have been folded after deposition and after intrusion of the Naxos Granite by a major phase of E-W shortening, as already mentioned. For a few Cycladic islands, such late-stage E-W shortening was also previously postulated by Urai et al. (1990), Avigad et al. (2001), andMarsellos et al. (2010).

Timing of Deformation with Respect to Retrogression within the Naxos Metamorphic Core Complex
The relative succession recorded by microstructures and fabrics and the 40 Ar/ 39 Ar white mica ages allow dating of these events for the whole metamorphic core complex. Isotopic exchange in recrystallized minerals is con-trolled by the relative importance of microstructure development, including grain-size reduction (e.g., Mulch and Cosca, 2004). White mica can grow over a variety of time scales under a large range of metamorphic conditions (e.g., Mulch and Cosca, 2004). Retrogressive mylonitization within the Naxos metamorphic core complex occurred under greenschist-facies conditions during exhumation. The new data from fine-grained sheared rocks also reveal that some earlier-formed, phengite-rich white mica grains possibly remained compositionally stable, although subsequent shearing and other muscovite-rich grains formed new microstructures. These fine-grained white micas were well developed in the mylonitic foliation and in shear bands during the retrogressive deformation (post-M2) evidenced by microstructural and petrographic observations.
Phengites of the M1 metamorphic stage underwent deformation during the interval between the M1 and M2 metamorphic stages. However, the effect of deformation on the Ar isotopic ages seems to have been different in different samples. Instead, crystallization of muscovite during the M2 metamorphism at the expense of phengite appears to be the most important process for resetting of the argon isotopic system. From the results of white mica 40 Ar/ 39 Ar dating, we found three reasonably constrained ages associated with retrogressive growth of white mica within the Naxos metamorphic core complex: (1) A plateau age of 16.52 ± 0.39 Ma in a shear zone in the southern part is interpreted as the formation age of a retrogressive shear zone overprinting earlier fabrics. (2) Another plateau age of 14.34 ± 0.36 Ma and evidence for Ar loss of older grains between 13.9 ± 0.9 Ma and 11.9 ± 0.8 Ma indicate motion along in the Moutsounas shear zone and directly date shearing at the upper margin of the Naxos metamorphic core complex, although some uncertainty remains. This is in agreement with the stratigraphic evidence given by overlying sediments from Naxos and Paros, which range in age between 11 and 5 Ma (Bargnesi et al., 2013). (3) Finally, very young plateau ages of 10.43 ± 0.44 Ma to 8.40 ± 0.76 Ma were found in the northern part of the Naxos metamorphic core complex, which is consistent with the youngest time of ductile shear deformation along the Naxos-Paros ductile normal fault. The age is close to cooling ages also recorded in nearby (U-Th)/He zircon and apatite fissiontrack ages of ca. 12 Ma and ca. 9 Ma (Seward et al., 2009;Brichau et al., 2006). This implies rapid cooling and possibly localized shear heating affecting white mica during recrystallization by thermally resetting the argon isotopic system through loss of radiogenic argon. A high geothermal gradient of 55 °C/km at hydrostatic pressure was assumed for this final stage of exhumation (Siebenaller et al., 2013), supporting the assumption of rapid cooling.

Retrogression within the Granodiorite (Naxos Granite)
Shear deformation within the Naxos metamorphic core complex continued after emplacement of bodies of the Naxos granodiorite to the west (e.g., Faure et al., 1991;John and Howard, 1995). The data presented herein suggest that the penetrative microstructural and fabric development in the granodiorite at eastern margins of the Naxos Granite was associated with early N-directed shearing at relatively high-grade temperature conditions. The calculated temperature of sample N72 is in the range from 661 °C to 583 °C, and the average is 621 ± 19 °C (Table DR1 [see footnote 1]). Since these granitic rocks were not metamorphosed, this deformation phase could have developed during cooling of the granodiorite. Ductile-brittle reactivation, associated with low-temperature hydrothermal alteration and formation of pervasive chlorite, appears to have occurred late in the deformation history of the granodiorite, indicating that the deformation occurred within the greenschist facies, as supported by results of the chlorite thermometry calculations. Consequently, microstructures reflect cooling of the granodioritic rocks from the ductile to semiductile (ductile-brittle) regime (e.g., Sibson, 1977;Shimamoto, 1989) during retrogression. Keay et al. (2001) and Bolhar et al. (2010) found that the I-type Naxos Granite body crystallized episodically between 13 and 11 Ma. Note that the ages of retrogression within the Naxos Granite are, therefore, significantly younger than the age of motion along the Moutsounas shear zone. This observation was attributed to a protracted history involving initial partial melting at deeper crustal levels, followed by crystallization and cooling at progressively shallower crustal levels. All zircon and apatite fission-track ages of the Naxos metamorphic core complex and the Naxos Granite range from 11.8 ± 0.8 to 9.7 ± 0.8 Ma and 11.2 ± 1.6 to 8.2 ± 1.2 Ma, respectively, and consistently decrease northward in the direction of hangingwall transport (Hejl et al., 2002;Brichau et al., 2006;Andriessen, 1991), but Seward et al. (2009) could not find this trend. Brichau et al. (2006) interpreted their results to support the scenario of an identical fault dip and a constant or slightly accelerating slip rate of ~6-8 km/m.y. on the Naxos-Paros extensional fault system across the brittle-ductile transition, which separated the metamorphic core complex from the overlying unit during exhumation. A rolling hinge top-to-the-N displacement was proposed in recent models (Brichau et al., 2006;Jolivet et al., 2010). Interestingly, we can distinguish two stages of tilting. In the south, the intrusional relationships between granodiorite of the Naxos Granite body and the Naxos metamorphic core complex are exposed, which formed during intrusion of the granodiorite between 13 and 11 Ma, and, in the north, a synform with Lower Miocene sedimentary rocks is observed. Consequently, the southern part is from a deeper level, implying tilting to the north. This is in contrast to cooling ages of Brichau et al. (2006), which imply an earlier age of cooling between ca. 11 and 8 Ma in the south and consequently top-to-the-S tilting according to their rolling hinge model.

Southward Shift of Retrogression in the Attic-Cycladic Belt
Retrogression, generally overprinting highpressure rocks, is a common phenomenon in the Attic-Cycladic belt (Fig. 1B). For retrogression, we also recognize a trend from older ages of 29-21 Ma in the north to 11-8 Ma in the south (Fig. 1B). Again, Naxos has in intermediate position but shows a wide temporal range of retrogression within the Naxos metamorphic core complex, consistent with retrogressive deformation within other metamorphic core complexes of the Aegean Sea (Fig. 1B). We interpret these long temporal ranges of retrogression to reflect outward southwestward retreat of subduction and, therefore, sequential activation of major detachment zones .

Generalized Model for Retrogression: Implications for Exhumation
An understanding of the retrogression history of the Naxos metamorphic core complex has helped in deciphering the metamorphic scheme, including the P-T path and exhumation processes of metamorphic rocks. This reequilibration is facilitated by information from fluid-rock interaction, and new mineral assemblages that are often developed in discrete zones of extensive retrogression during ductile-brittle transition (e.g., Kreulen, 1980;Cartwright and Buick, 1995;Buick and Holland, 1989;Cao et al., 2013aCao et al., , 2013b and inflow through subvertical veins (Siebenaller et al., 2013). During low-temperature retrogression, microshear zones appear to have been zones of more focused and discrete deformation and fluid-rock interaction in comparison to earlier high-temperature retrogression. In most structural settings, dynamic retrogression is related to the brittle-ductile transition zone (e.g., Michibayashi and Masuda, 1993), although retrogression in much deeper structural levels could occur, e.g., the amphibolitization of water-free granulite-or eclogite-facies mineral assemblages (e.g., El-Shazly and Sisson, 1999;Xu et al., 2009). At the brittle-ductile transition, retrogressed sheared rocks are most common, and fluid flow is proven by precipitation of minerals such as chlorite along the overlying cataclastic carapace. This also involves leaching of elements and minerals mostly from the footwall. The sources of hydrous fluids could be from three basic sources, which could be possibly distinguished by fluid inclusion and stable isotope (hydrogen, oxygen) studies (Siebenaller et al., 2013). These sources include: (1) descending meteoric water (and possibly seawater) from sedimentary basins or the surface (e.g., Yardley et al., 2000), which infiltrates along brittle normal faults to depth, and (2) ascending magmaticinduced hydrous fluid flow during exhumation of metamorphic rocks, e.g., from synextensional granites (e.g., Zong et al., 2010). (3) The higher-than-normal heat flow also might trigger metamorphic dehydration reactions at depth, like decomposition of chlorite, muscovite, or biotite, which then yields a significant amount of uprising hydrous metamorphic fluids. The flow of hydrous fluids would trigger a number of processes along the zone of shearing and retrogression during ductile-brittle transition, among which the lowering of the shear stress of rock-constituting minerals like quartz (Tullis and Yund, 1989;Chernak et al., 2009), sericitization of feldspar, and chloritization are the most important ones. In the case of the Naxos metamorphic core complex and the Naxos Granite, fluids could be expected to have derived from all three sources (see also Siebenaller et al., 2013): metamorphic dehydration reactions at depth, magmatic fluids from granite intrusions, and meteoric fluids from the hanging-wall unit. Because of the long duration of the metamorphic processes, some continuity of retrogression can be expected between ca. 16 and 9 Ma.
In Figure 13, we describe a generalized model for retrogression related to late-stage ductile-brittle transition, when the metamorphic core complex reached shallow crustal levels during exhumation. The model is based on the geodynamic, structural, and petrologic conditions of the Naxos metamorphic core complex, but it could be applied to other geodynamic settings in which extension prevails. The cooling of the exhuming metamorphic core complex results in ascent of the brittle-ductile transition zone to higher structural levels by cooling and exhumation of formerly deeply buried rocks. In these settings, subvertical gashes and veins can be found, and some of these veins are filled by minerals precipitated from ascending and descending fluids. It is generally at the level of the veins where ascending fluids are overheated with respect to the temperature of the country rocks, representing, therefore, hydrothermal fluids (Scott et al., 2015). In such cases, these fluids can also transform the country rocks by static retrogression. This may be generally the case when these veinlets and veins are generally undeformed and occur in a late stage of the structural succession, as in the case of the Mykonos (Menant et al., 2013) and Sifnos meta morphic core complexes (Neubauer, 2005). Note that such veins are often also mineralized with ore minerals.
In summary, all these minerals lower the differential stress along the zone of shearing, retrogression, and faults and lead finally to shear concentration within such zones at the ductilebrittle transition. This represents and provides a positive feedback mechanism for faulting and stabilization of retrogression immediately below the brittle-ductile transition zone during exhumation of metamorphic core complexes.

CONCLUSIONS
The following major conclusions can be drawn from this study: (1) Microstructural and textural analysis of samples and field observations reveals that there is variable retrogression within the detachment shear zone. Retrogression even pervasively affected significant portions of the migmatitegrade metamorphic core and remnant high-pressure areas. There, retrogression led to pervasive formation of new fabrics within greenschistand sub-greenschist-facies metamorphic condi-tions during the exhumation of the Naxos metamorphic core complex.
(2) Detailed analysis of the microstructure, composition, and 40 Ar/ 39 Ar dating of synkinematic white micas constrains retrogressive deformation conditions and change of fabrics within and at the top of the Naxos metamorphic core complex and Naxos Granite. The thermochronologic data record retrogression in late Miocene times (ca. 16.52 ± 0.39 Ma to 8.40 ± 0.76 Ma), consistent with final exhumation of the Naxos metamorphic core complex.
(3) Retrogressive microstructures, low-temperature calcite fabrics in marbles, and chloritization in metapelites (at temperatures of ~350-130 °C) in the Naxos metamorphic core complex core mainly developed during late-stage E-W shortening and folding. The lower-temperature group could be tentatively interpreted to be the result either of a late-stage hydrothermal overprint related to granite/granodiorite intrusions or inflow of meteoric water. The flow of hydrous fluids resulted in resetting of fabrics.
(4) The wide temporal range of retrogression within the Naxos metamorphic core complex may be consistent with retrogressive deformation within other metamorphic core complexes of the Aegean Sea. We interpret this long temporal range to reflect outward southwestward retreat of subduction and therefore sequential activation of major detachment zones.

Backscattered Electron Diffraction Method
Crystallographic preferred orientations (CPOs) were measured using a backscattered electron diffraction (BSED) detector mounted on a tungsten filament Hitachi S-3400N-II scanning electron microscope (SEM) at the China University of Geosciences, Beijing (Liu et al., 2012). Detailed optical microstructural observations and CPO analysis were conducted on thin sections. All thin sections were cut parallel to the kinematic x-z plane (where x is parallel to lineation, y is parallel to foliation and normal to lineation, and z is normal to foliation). Highly polished thin sections were prepared without cover and polished using a colloidal silica suspension with a particle size of 500 nm for 1-3 h. Thin sections were put in the SEM chamber at a 70° tilt angle with the rock lineation (structural x reference direction) parallel to the SEM x axis. The BSE patterns were acquired at a low acceleration voltage of 15 kV and a beam working distance of ~18-20 mm. Conducting resin tapes attached to the sample surface surrounding the measurement area were used to reduce charging effects. The BSED data acquisition was done using both point scan and mapping modes. Indexing was accepted when at least six detected Kikuchi bands matched with those in the standard reflector file for the analyzed mineral phases, and indexed points with a mean angular deviation (MAD) larger than 1.2 (between detached and simulated patterns) were eliminated to avoid suspicious indexing. BSED analysis was completed using the HKL Channel 5 software package. Pole figures were plotted in lower-hemisphere equal-area stereographic diagrams with the trace of the mylonitic foliation (S) and the stretching lineation (L) as reference directions. Systematic misindexing was noted in automated orientation maps, and such data were replaced by zero solution pixels. A detailed description of the BSED technique can be found in, e.g., Prior et al. (1996).

Electron Microprobe Analysis (EPMA) Methodology
EMPA of the mineral chemistry of minerals (white mica, two feldspars, chlorite, garnet, biotite) were carried out on a JEOL electron microprobe (JXA-8600) at the Department Geography and Geology, University of Salzburg, using a wavelength dispersive system. We used an acceleration voltage of 15 kV and a sample current of 40 nA. Natural and synthetic mineral standards were used to calibrate the microprobe, and raw data were reduced using standard ZAF correction. The detection limits (2s) for the elements are: Si = 0.06 wt%, Al = 0.04 wt%, and Na, K, Mg, Mn, and Fe = 0.025 wt%.

Ar/ 39 Ar Method
The 40 Ar/ 39 Ar analytical techniques largely follow descriptions given in Handler et al. (2004) and Rieser et al. (2006). Preparation of the samples before and after irradiation, 40 Ar/ 39 Ar analyses, and age calculations were carried out at the ARGONAUT Laboratory of the Geology Division at the University of Salzburg. The mineral separates were obtained by crushing hand-sized samples and sieving through 200-355 µm fractions. They were further purified by washing with deionized water. Mineral concentrates were packed in aluminum foil and placed in quartz vials. For calculation of the J-values, flux monitors were placed between each 4 or 5 unknown samples. The sealed quartz vials were irradiated in the Rez reactor (Prague, Czech Republic) for 16 h. Correction factors for interfering isotopes were calculated from 10 analyses of two Ca-glass samples and 22 analyses of two pure K-glass samples, and they are: 36 Ar/ 37 Ar(Ca) = 0.00022500, 39 Ar/ 37 Ar(Ca) = 0.00061400, and 40 Ar/ 39 Ar(K) = 0.026600. Variation in the flux of neutrons was monitored with the DRA1 sanidine standard, for which a 40 Ar/ 39 Ar plateau age of 25.03 ± 0.05 Ma was origi- Geological Society of America Bulletin, v. 1XX, no. XX/XX 23 nally reported (Wijbrans et al., 1995). Here, we used the revised value of 25.26 ± 0.05 Ma (van Hins bergen et al., 2008). The 40 Ar/ 39 Ar analyses were carried out using an ultrahigh-vacuum Ar-extraction line equipped with a combined infrared (IR) laser system, and a VG-ISOTECHTMNG3600 mass spectrometer.
Stepwise heating analyses of samples were performed using a defocused (1.5 mm diameter) 25 W CO 2 -IR laser operating in Tem00 mode at wavelengths between 10.57 and 10.63 µm. Gas admittance and pumping of the mass spectrometer and the Ar extraction line were computer-controlled using pneumatic valves. The NG3600 is an 18-cm-radius 60° extended geometry instrument, equipped with a bright Nier-type source operated at 4.5 kV. Measurements were performed on an axial electron multiplier in static mode with peak jumping, and the stability of the magnet was controlled by a Hall probe? For each increment, the intensities of 36 Ar, 37 Ar, 38 Ar, 39 Ar, and 40 Ar were measured, and the baseline readings on mass 34.5 were automatically subtracted. Intensities of the peaks were back-extrapolated over 16 measured intensities to the time of gas admittance either by a straight line or a curved fit, depending on intensity and type of pattern of the evolving gas. Intensities were corrected for system blanks, background, postirradiation decay of 37 Ar, and interfering isotopes. Isotopic ratios, ages, and errors for individual steps were calculated following suggestions by McDougall and Harrison (1999) and Scaillet (2000) using decay factors reported by Steiger and Jäger (1977). Definition and calculation of plateau ages were carried out using ISOPLOT/EX (Ludwig, 2001).