Post-collisional Cenozoic extension in the northern Aegean: The high-K to shoshonitic intrusive rocks of the Maronia Magmatic Corridor, northeastern Greece

Mountain-building events can generate thousands of kilometers of high elevation over periods of tens to hundreds of millions of years; the development of such high relief during orogenesis is supported by the development of a thick continental root (e.g., Dewey et al., 1993). In some orogens this crustal thickness is preserved long after collision, while in others post-collisional collapse of the orogen thins the crust and results in widespread extensional structures that dissect the region (e.g., Le Pichon and Angelier, 1979; Dewey, 1988; Wang et al., 2014). In extended orogens, strain localization can drive the formation of trans-crustal normal detachment faults that transfer material from the ductile, lower crust to the surface and can aid the exhumation of metamorphic core complexes from >45 km depth (e.g., the North American Cordillera: Coney, 1987; Wernicke et al., 1987; Camp et al., 2015; and the Aegean: Lister et al., 1984; Jolivet and Brun, 2010; Burg, 2011; Kydonakis et al., 2015). Another feature of highly extended orogenic terranes is the ubiquitous presence of post-collisional magmatism. The conductive transfer of heat from upwelling hot, asthenospheric mantle beneath highly extended crust can generate small volume, partial melts that exploit crustal pathways developed during extensional deformation, leading to intrusive and extrusive post-collisional magmatism (e.g., McKenzie and Bickle 1988; von Blanckenburg and Davies 1995). Feedbacks between metamorphic core complex exhumation and magmatism are numerous and include magmatic volatiles facilitating crustal deformation and crustal thinning and resultant decompression and mantle upwelling, driving lithospheric melting (Teyssier and Whitney, 2002; Whitney et al., 2013; Platt et al., 2015). As a result, both metamorphic core complexes and post-collisional magmatic suites can be used to offer insights into the tectonic processes controlling crustal extension; here we focus on the petrogenesis of post-collisional magmatism.

In collisional orogenic belts, subduction-related calc-alkaline magmatism shows a trend of increasing K with maturity of the magmatic system (e.g., Lipman et al., 1972). The transition from high-K calc-alkaline to shoshonitic magmatism is common in post-collisional tectonic settings where the inherited metasomatized mantle wedge is depleted at the onset of magmatism (e.g., Aldanmaz et al., 2000; Seghedi et al., 2004). This high-K calc-alkaline to shoshonitic evolution in magma geochemistry is observed throughout the Aegean—for example, in the Eocene to middle-Miocene Biga orogenic rocks in NW Turkey and the Oligocene and Miocene magmatic rocks of NE and central Greece (e.g., Maronia, Samothraki, Lesbos and Limnos) (Del Moro et al., 1988; Pe-Piper et al., 2009; Ersoy et al., 2017).

The generation of potassic melts from the mantle is widely ascribed to be the result of anomalously hot mantle melting (Gill 1981). Source mantle geochemistry is also widely considered to be key in the generation of potassic to ultra-potassic fractionation trends in subduction-related settings (e.g., Foley et al., 1987). However, the origin of a K-rich source mantle is debated; current hypotheses include, but are not limited to, (i) intense mantle metasomatism by K-rich fluids from the dehydrated, downgoing slab (e.g., Poli and Schmidt, 2002; Condamine and Médard, 2014); and (ii) contamination and physical mixing of mantle melts and lower crustal or subducted sedimentary melts (e.g., Beccaluva et al., 2013; Wang et al., 2017). In addition to source mantle geochemistry, conditions of magma generation and fractionation are also considered to be important in the production of potassic to ultra-potassic magmas (e.g., Peccerillo et al., 2013). Experimental and geochemical studies have shown that magmas generated at high pressures and temperatures, in the absence of abundant H₂O, can produce shoshonitic liquid lines of descent (LLDs) by stabilizing pyroxene and anorthite-rich plagioclase over olivine in the fractionating assemblage, and that the same magmas generated under shallower or more fluid-rich conditions would fractionate a high-K calc-alkaline trend (Meen, 1987; Freise et al., 2009; Lanzo et al., 2016; Beermann et al., 2017). While all of these models require an elevated mantle geotherm to generate high-temperature mantle melts, the mechanism by which a shoshonitic fractionation trend is achieved has important implications for the conditions of magma petrogenesis.

The Hellenic Orogen (or Hellenides) is a 500-km-long mountain belt that stretches across the Aegean and is bound by the Balkanides in the north and the Hellenic trench, the point of present subduction of the African plate beneath Eurasia, in the south (Smith and Moores, 1974; Mountrakis, 2006). Resulting from closure of the Tethys Ocean and subsequent collision of the African and Eurasian plates, the Hellenic orogen is a well-studied example of post-collisional extension. Predominantly attributed to retreat of the Hellenic trench and roll back of the African Slab, Hellenic extension is also thought to incorporate elements of lithospheric delamination and slab break-off beneath Anatolia to the east of the Aegean (Jolivet and Brun, 2010; Ring et al., 2010; Burg, 2011; Brun et al., 2016, and references therein). Terrane accretion and subduction of intermediate oceanic crust (the Vardar-Izmir and Pindos basins) during the Hellenic Orogen began in the late Cretaceous and continued into the Cenozoic (e.g., van Hinsbergen et al., 2005; Schmid et al., 2008). Post-collisional orogenic collapse of the Hellenides began in the late Mesozoic to Cenozoic, contemporaneous with continued north-dipping subduction further toward the south, and has been irregular both spatially and temporally ever since (e.g., Jolivet and Brun, 2010; Ring et al., 2010).

A two-stage model for Aegean extension was developed to accommodate localized Paleogene extension in NE Greece and distributed Neogene extension across the rest of the region (Jolivet and Brun, 2010; Brun et al., 2016, and references therein). This paper focuses on localized tectonic extension in NE Greece where an early phase of trans-crustal extension accommodated exhumation of the northern Rhodope core complex (RCC) from mid- to lower-crustal levels (e.g., Burg, 2011; Kydonakis et al., 2014). Later Neogene extension in the northern RCC was restricted to the upper crust and the formation of normal fault-bound sedimentary basins (e.g., Burchfiel et al., 2003; Kilias et al., 2013). The Maronia Magmatic Corridor (MMC), an 150-km-long NE-trending belt of post-collisional Oligocene calc-alkaline to shoshonitic plutons, intrudes the Kechros dome at the eastern margin of the northern RCC (Del Moro et al., 1988). In this paper, we present detailed field and petrographic observations and new whole-rock geochemistry, including a full suite of rare earth elements (REEs), of the MMC to better constrain the petrogenesis of the potassic post-collisional magmatism. We use field observations coupled with chemical abrasion–isotope dilution–thermal ionization mass spectrometry (CA-ID-TIMS) U-Pb zircon geochronology and whole-rock geochemistry of the plutons to constrain crystallization to within tens of thousands of years, to investigate the temporal evolution of the MMC, and to discuss how geochemical changes in magma petrogenesis relate to the timing of Cenozoic post-collisional orogenic collapse in the northern RCC.

The Rhodope Massif is located in the northern Aegean, straddling the border between northern Greece and southern Bulgaria (Fig. 1). The rocks of the Rhodope Massif have a long and complex tectonic history associated with terrane accretion during the Hellenic orogeny and subsequent core complex formation and exhumation (e.g., Ricou et al., 1998). A series of stacked nappes, the Rhodope Massif can be split into two tectonic domains, the northern RCC and the southern RCC, separated by the Nestos thrust (Fig. 1A). The northern and southern RCC have a shared history during terrane accretion in the Mesozoic, but the evolution of the tectonic units diverged during Cenozoic exhumation (e.g., Kydonakis et al., 2014).

The Rhodope Massif has a broadly anticlinal structure and can be subdivided into three lithological units: (i) the lower Pangaion (or Thracia or Drama) terrane of ortho- and para-gneisses, exposed in the core of the fault-bound anticlinal domes (Krohe and Mposkos, 2002; Turpaud and Reischmann, 2010; Burg, 2011; Bonev et al., 2013); (ii) the overlying, upper Rhodope unit, a metaophiolite-gneiss-schist assemblage in the hanging wall of the Rhodope anticline (Krohe and Mposkos, 2002; Burg, 2011); and (iii) the highly deformed imbricate unit sandwiched along the Nestos suture zone between the upper and lower terranes (Bauer et al., 2007; Krenn et al., 2010). In the southern RCC the upper terrane is preserved in the hanging wall of the Kerdyllion fault (Fig. 1A). In the northern RCC lower terrane gneiss domes dominate; in the east, the greenschist facies Mesozoic schist units of the Circum Rhodope Belt (CRB) are considered to be equivalents of the upper Rhodope terrane hanging wall (Bonev and Stampfli, 2008; Krenn et al., 2010; Meinhold et al., 2010; Bonev and Stampfli, 2011; Meinhold and Kostopoulos, 2013).

Considerable geochronological effort, in tandem with detailed microstructural and tectonic contextual study, has been put into unravelling the tectono-metamorphic evolution of the Rhodope Massif (synthesized in Fig. 1B; Table 1). Variscan-age rift-related igneous protoliths in the lower, Pangaion, and upper, Rhodope, terranes indicate a shared, European basement affinity of the tectonic units that make up the Rhodope Massif (Turpaud and Reischmann, 2010; Burg, 2011, and references therein). Jurassic–Cretaceous igneous protoliths in the imbricate unit record subduction-related magmatism on the Rhodopean margin as the rift-basin that separated the Rhodope and Pangaion terranes in the early Mesozoic was closed (e.g., Schmid et al., 2008; Turpaud and Reischmann, 2010; Burg, 2011). This period of subduction is also recorded by diamond- and coesite-bearing ultra-high pressure (UHP) metamorphism, as lenses of the imbricate unit were buried along the Nestos suture zone between 180–160 Ma (Mposkos and Kostopoulos, 2001; Bauer et al., 2007; Krenn et al., 2010); this was co-incident with obduction of the Evros ophiolite onto the Rhodope terrane (Bonev et al., 2015). Greenschist facies metamorphism of the CRB was contemporaneous with the accretion of the Pangaion terrane and ophiolite obduction (Bonev et al., 2013, 2015). In the late Cretaceous, following the underthrusting of the Pangaion terrane beneath the Rhodope terrane, subduction of the Vardar-Izmir Ocean initiated to the south of the Rhodope Massif (e.g., van Hinsbergen et al., 2005; Schmid et al., 2008). Aegean extension initiated in the Rhodope Massif in the Cenozoic during a transitional tectonic period from subduction-related convergence to subduction-related slab rollback and trans-crustal extension (Burchfiel et al., 2008).

The Rhodope Massif was exhumed during late Cretaceous–Cenozoic trans-crustal extension driven by Aegean slab rollback (e.g., Jolivet and Brun, 2010; Ring et al., 2010). Paleocene–Oligocene localized extension formed a series of gneiss domes across the northern RCC and initiated exhumation along the Kerdyllion Detachment in the southern RCC (e.g., van Hinsbergen et al., 2008; Kydonakis et al., 2014, 2015). Dehydration partial melting during core complex unroofing and exhumation is recorded in temporally sporadic Paleocene–Eocene pegmatites and leucogranites across both the northern RCC and southern RCC (e.g., Del Moro et al., 1990; Mposkos and Wawrzenitz, 1995; Wawrzenitz and Krohe, 1998; Lips et al., 2000; Liati et al., 2002). Maastrichtian–Paleocene to early Eocene sediments unconformably overlie the gneiss domes of the northern RCC, indicating that these domes must have been exhumed, at least in part, before the early Eocene (Boyanov et al., 1982; Burg, 2011).

Exhumation in the northern RCC ceased abruptly in the Oligocene, likely with the onset of post-collisional volcanism across the northern RCC (e.g., Rohrmeier et al., 2013), and resulted in an E-trending belt of small (∼30 km diameter), unconnected gneiss domes (Kydonakis et al., 2015). This was contemporaneous with a change in extension style across the Aegean. Early Eocene–Oligocene extension was localized in the northern Aegean, restricted to the northern limit of the hinterland, the northern and southern RCC (Kydonakis et al., 2015). From the mid-Miocene onward, extension was distributed across the Aegean with the formation of the Central Cyclades Core Complex further south and continued exhumation and SW-rotation of the southern RCC (van Hinsbergen et al., 2008; Kydonakis et al., 2014, 2015). The change in extension style is attributed to an acceleration in trench retreat related to the increased rate of convergence between Africa and Eurasia and slab tearing beneath western Turkey in the mid-Miocene (e.g., Jolivet and Brun, 2010; Jolivet et al., 2013; Brun et al., 2016).

Post-subduction magmatism was contemporaneous with trans-crustal extension, punctuating across the Rhodope Massif from the middle Eocene (42 Ma, Smiliyan and Yugovo Plutons; Table 2; Rohrmeier et al., 2013) to the mid-Miocene (11.32 Ma, Strymon Valley volcanics; Table 2; Georgiev et al., 2013). Post-subduction magmatism is closely linked to extension, with trans-crustal structures providing pathways for magmas, evidenced by the spatial association of plutonic rocks and detachment fault systems (e.g., Jones et al., 1992; Pe-Piper and Piper, 2006; Fig. 1). This is noticeable in the distinction between the northern RCC and southern RCC, where magmatism and concurrent volcanism cease in the late Oligocene in the northern RCC (e.g., Márton et al., 2010; Marchev et al., 2010) but continue into the Miocene in the southern RCC (Table 2; Georgiev et al., 2013).

Unconformably overlying the northern RCC metamorphic basement, syn-extensional sedimentation started in the mid-late Eocene as a transgressive sequence of conglomerates, limestones, sandstones, and marls filled the Thrace supra-detachment basin, formed in accommodation space to the SE of the exhuming northern RCC (Kilias et al., 2013). The switch from trans-crustal, ductile extension to upper-crustal, brittle extension in the northern RCC is recorded by the change in depositional environment from a supra-detachment basin in the Eocene-Oliogene, to a series of cross-cutting N-S– to NNE-SSW–trending fault-bound basins in the Oligocene-Miocene (Burchfield et al., 2003, 2008; Kilias et al., 2013).

The MMC is a NE-trending magmatic belt, comprising at least six Oligocene gabbroic to granodioritic intrusions (Fig. 2): the Tris Visses, Halasmata, Leptokaria, Kirki, Kassiteres, and Maronia plutons. The MMC is hosted in the easternmost dome of the northern RCC, the Biala-Rika-Kechros dome (from here onward referred to as the Kechros Dome). First described by Del Moro et al. (1988), the MMC plutons all intrude the northern RCC and associated lithologies, hosted either within the Kechros gneiss dome to the NE (Halasmata and Tris Visses), straddling the Kechros basal detachment in the center of the corridor (Leptokaria), or within the CRB greenschist basement to the SW (Kirki, Kassiteres, and Maronia; Fig. 2). Later Oligocene intercalated sedimentary and volcanic units onlap onto the MMC intrusions. Both the Kassiteres and Maronia plutons are cross-cut by younger porphyritic microgranite intrusions and host occurrences of porphyry–epithermal-style Au-Mo-Cu mineralization (e.g., Melfos et al., 2002; Voudouris et al., 2006). Rb-Sr biotite–whole-rock isochron dating by Del Moro et al. (1988) suggests that the MMC plutons were emplaced between ca. 32 and 28 Ma, younging progressively from NE to SW along the belt.

Field observations, with particular focus on relationships between lithologies, were made over multiple summer field seasons in northeastern Greece. We have prepared our geological map, Figure 2, from existing geological maps of the area supplemented by our own field observations. The spatial extent of the plutons, in particular Kirki, Leptokaria, Halasmata, and Tris Visses, are poorly constrained due to difficult field conditions. The lack of exposure, steep topography, dense vegetation, and poor road access bias the outcrop localities of both this work and previous geological mapping by Del Moro et al. (1988), with most mapped outcrops found along small river channels. However, based on our field observations, we consider it more likely that the observed plutons represent small outcrops of a larger magmatic system interconnected at depth that is poorly exposed, and that additional outcrops could be found in more inaccessible areas.

A suite of 36 samples was collected from the MMC, representing each of the observed intrusive lithologies and alteration styles. At least eight samples were taken from each of the Maronia and Kassiteres plutons, and from across the Kirki–Leptokaria–Halasmata–Tris Visses outcrops to reflect changes in mineralogy between samples. Standard polished thin sections and epoxy mounts were prepared for a representative selection of samples from each pluton, and detailed petrographic observations were made under both transmitted and reflected light. Point counting of the least altered thin section(s) of each of the plutons was undertaken to obtain representative modal compositions of the samples. Scanning Electron Microscope (SEM) analysis was undertaken at the University of Bristol on a Hitachi S-3500 N instrument with 20 kV accelerating voltage and a 15 mm working distance. Phase identification was performed using energy dispersive spectra (EDS), and back-scatter electron (BSE) imagery was used to establish key textural relationships.

Powders of <25 µm grain size were prepared at the University of Bristol and analyzed in the Inorganic Geochemical Laboratories at the British Geological Survey (BGS), Keyworth. Powders were prepared for each of the igneous lithologies collected, with a total of 36 samples and eight repeats analyzed. Weathered outer rims of individual samples were removed such that only the freshest material from the cores of the samples were used for analysis. Samples were then crushed and milled in a ball mill using an agate jar with 12 agate balls. Due to the large grain size (>1 mm) and relative heterogeneity of the samples, >1 kg of rock was used to ensure a homogenous powder was produced; a greater weight of sample was used for coarser-grained samples from the Maronia pluton. Samples were analyzed following the UKAS Accreditation Schedule using wavelength dispersive–X-ray fluorescence spectrometry (WD-XRFS) on fused glass beads for major and minor element oxides, Na₂O₂ fusion inductively coupled plasma–mass spectrometry (ICP-MS) for minor and trace element concentrations, and an aqua regia digest ICP-MS for Sc and Te.

New geochemical data presented in this study supplements previously published work by Del Moro et al. (1988), Papadopoulou et al. (2004), and Mavroudchiev et al. (1993), presenting for the first time a complete suite of trace element and REE analyses for the entire magmatic belt. The results have been filtered by loss on ignition (LOI) values such that any results with LOI >2 wt% have been omitted, removing any bias in the results from heavily clay-altered or hydrated samples.

A subset of nine samples from the MMC sample suite were selected for high-precision U-Pb zircon geochronology (Maronia, n = 4; Kassiteres and Leptokaria n = 5). The MMC zircons have broadly similar euhedral morphologies with many inclusions of apatite, titanite, and melt. Zircons from Maronia were typically larger than those from Kassiteres and Leptokaria and were less abundant in the mineral separate. Inclusion- and defect-free crystals with well-preserved euhedral crystal faces were preferentially selected from the mineral separate for analysis (further details of sample preparation and zircon selection are included in GSA Data Repository File DR2¹). The remaining zircon crystals were mounted in epoxy, polished, and imaged at the University of Bristol using a Centaur Cathodoluminescence (CL) detector on a Hitachi S-3500N Scanning Electron Microscope at 10 kV and 23 mm working distance. CL imagery of the zircons was used to investigate the internal morphology of the crystals, with particular focus on growth and resorption features in the zircons.

A total of 42 zircon crystals were analyzed at the NERC Isotope Geosciences Laboratory (NIGL) using the CA-ID-TIMS technique. Three to seven individual zircon crystals from each sample were thermally annealed and chemically abraded to reduce Pb loss (Mattinson, 2005). Isotope dilution and mass spectrometry protocol followed the techniques and instrumentation described by Sahy et al. (2015), and further details are included in GSA Data Repository File DR2.

We present primary field and petrographic observations of the MMC plutons. This work builds upon that of previous authors, in particular Del Moro et al. (1988), and focuses on the most voluminous intrusions along the belt: Maronia, Kassiteres, and Leptokaria (Fig. 2). Two additional, later phases of magmatism are found intruding each of the plutons: an array of aplitic dykes and small (∼500 m²), silicic porphyry intrusions. The felsic units are volumetrically insignificant compared to the host plutons and associated with post-pluton hydrothermal activity, including porphyry-style alteration and mineralization in the area, leading to intense alteration of these rocks (e.g., Melfos and Voudouris, 2016, and references therein). We consider the small-volume, evolved lithologies to be less important to the discussion on the changing geochemistry of the mantle magmatic system in the northern RCC, as they are highly fractionated and likely represent a mixed geochemical signature, reflecting both crustal and mantle melt sources.

The Leptokaria, Kirki, Halasmata, and Tris Visses plutons were first described individually by Del Moro et al. (1988). However, in this paper, we group these plutons together on the basis of similar geochemistry and age, and we refer to them collectively as the Leptokaria complex, after the largest pluton in the group (Fig. 2).

The equigranular (0.2–1.0 mm) Leptokaria complex ranges from monzogabbro to granite in composition (Fig. 3A). A primary assemblage of phenocrystic plagioclase + magnetite + hornblende + biotite is found in all samples of the Leptokaria pluton; clino- and ortho-pyroxene are present in the dioritic Leptokaria samples, while quartz + K-feldspar and accessory zircon + apatite are also present in the more evolved samples. Interstitial, granophyric quartz and K-feldspar intergrowths are found throughout the more evolved samples. Weak alteration, principally chlorite + epidote ± biotite ± orthoclase, is pervasive across Leptokaria, with breakdown of mafic phases and sericitization of plagioclase. Small (0.5–1-cm-wide) veinlets of albite ± chlorite ± epidote are also common. Five of the Leptokaria samples have a cumulate-like texture with interlocking phenocrysts of plagioclase, amphibole (replacing relict clinopyroxene), and magnetite and the development of interstitial, poikolitic amphibole and plagioclase and visible plagioclase rims (Figs. 3B–3C). These cumulate-like rocks likely characterize the mineral phases present during earliest magmatic crystallization and differentiation. Alteration is continuous through the cumulate samples with chlorite-biotite-epidote alteration of mafic phases the most common reactions observed.

The Leptokaria plutons are intruded into ortho- and para-gneisses of the Kechros Dome within the northern RCC. Around Leptokaria village, the pluton is hosted in a two-mica pegmatitic leucogranite, although the contact between the leucogranite and Leptokaria complex is not seen. Contacts between the plutons and the gneisses elsewhere in the Leptokaria Complex are undulose, with mixing of globules of gneiss in the pluton and vice-versa, this suggests that the host gneisses were not completely solid during magma emplacement and accumulation.

Kassiteres is a medium-grained (1–3 mm), clinopyroxene-hornblende diorite and is the next pluton moving SW along the MMC from Leptokaria (Fig. 2). The smallest of the named intrusions, Kassiteres has an aerial extent of ca. 1.8 km². Kassiteres intrudes into metapelitic greenschists of the CRB basement and is unconformably overlain by Oligocene volcaniclastic units. The primary mineralogy of the Kassiteres diorite is clinopyroxene + hornblende + plagioclase + biotite + magnetite phenocrysts and interstitial K-feldspar + quartz with accessory apatite + titanite + zircon (Fig. 3E). Mixing of mafic, clinopyroxene-rich, and felsic, biotite-rich, end members are visible on the macroscopic scale, suggesting periodic injection of new melt into the crystallizing magmatic system (Fig. 3D). Interstitial granophyric quartz-feldspar intergrowths suggest rapid cooling of interstitial melt during the later stages of pluton crystallization (Fig. 3F).

The Kassiteres pluton is variably altered due to hydrothermal activity related to the later intrusion of a porphyritic stock and aplite dykes. Breakdown of clinopyroxene and hornblende to biotite, chlorite, and plagioclase is characteristic of potassic alteration at Kassiteres, pervasive to varying degrees of intensity throughout the pluton. Advanced clay alteration is found in discrete fault zones within the diorite where the fault planes provide a pathway for enhanced fluid-wallrock interaction. This alteration grades from silicified, advanced argillic alteration (silica + low-temperature clays) proximal to the fault plane, to argillic alteration (high-temperature clays) over 5 m from the fault. A diffuse gradation into potassically altered diorite occurs over a greater distance of 15–20 m from the faults.

Maronia is the southernmost pluton of the MMC, separated by ∼30 km from its nearest neighbor, Kassiteres. Outcropping as a stock with an area of ∼3.2 km², it is intruded into metapelites and marbles of the CRB. Three intrusive lithologies were recognized at Maronia by Papadopoulou et al. (2004): a voluminous monzonite–quartz monzonite unit (Fig. 4A), a minor gabbroic unit, and a minor porphyritic granitic unit. The gabbro is texturally and mineralogically indistinguishable from the monzonite–quartz monzonite unit in all but whole-rock geochemistry, while the granite porphyry shows sharp, cross-cutting contacts with the rest of the pluton (Fig. 4F). In this paper, we consider the gabbro–monzonite–quartz monzonite as a single unit (referred to herein as the monzonite unit) with subdivisions arising from varying phenocryst proportions.

The primary mineralogy of the monzonite is clinopyroxene + orthopyroxene + amphibole (tremolite-ferrosillite) + biotite + plagioclase + K-feldspar + apatite + magnetite + quartz ± rare olivine with accessory zircon. 1–2-cm-diameter phenocrysts of primary plagioclase (50%–80% by area), clinopyroxene (5%–15%), and minor orthopyroxene (1%–3%) are surrounded by interstitial K-feldspar (up to 20%; Fig. 4C). Primary biotite and amphibole phenocrysts in textural equilibrium with adjacent crystals are very rare and are more commonly found as secondary breakdown rims of primary pyroxene phenocrysts (Fig. 4B). Evidence of post-magmatic hydrothermal alteration is visible in minor sericitization and uralization of plagioclase and pyroxene-amphibole phenocrysts respectively.

Additional igneous facies are observed along the western margin of Maronia where the pluton intrudes CRB marbles. In addition to a thin (∼1 m) endo-skarn boundary along the marble-monzonite contact (Katerinopoulou et al., 2009; Fig. 4E), an orthocumulus facies is exposed in the northwest of the pluton extending for up to 10 m into the monzonite from endoskarn boundary (Fig. 4D). Cumulus pyroxene adacrysts and intercumulus oikocrystic orthoclase-albite suggest that the western margin of the pluton exposes close to the base of the Maronia plutonic system. This orthocumulus facies is not observed on the eastern margin of Maronia where a ∼250-m-wide hornfels aureole extends into the host CRB metapelites.

The NNE-SSW trending Ktismata Fault cuts the pluton separating the eastern and western portions of the pluton (Melfos et al., 2002; Fig. 4F). Differing intrusive relationships between the western (deeper, orthocumulus facies) and eastern (shallow, baked aureole) margins of the pluton suggest an east-dipping normal fault. Minimal right-lateral displacement of < 100 m is observed where the fault cuts the northern pluton margin, which can be accounted for by a minor oblique component of motion. An additional strike-slip fault, the Marmaritsa Fault, cuts the southern margin of the Maronia pluton subparallel to the coastline. This structure is part of a regional crustal trend, potentially related to the North Anatolian Fault Zone (Melfos et al., 2002). The brittle nature of the faulting, cutting Oligocene magmatic features within Maronia, suggest post-emplacement initiation of both structures.

The MMC samples vary in composition from 48 to 68 wt% SiO₂ (Fig. 5; Table 3; GSA Data Repository Table DR1). The Maronia pluton has a more restricted SiO₂ compositional range (50–56 wt% SiO₂), while the Kassiteres and Leptokaria samples cover the full range (48–68 wt% SiO₂). Within the Leptokaria sample suite, a difference in SiO₂ content separates cumulate-like samples (48–53 wt% SiO₂) from the intermediate to felsic plutonic samples (55–68 wt% SiO₂), with all the Leptokaria samples falling on a linear fractionation trend. Samples from the MMC follow a calc-alkaline differentiation trend; however, their similarity to “typical” calc-alkaline magmatism is limited. As SiO₂ increases with differentiation, compatible elements, including MgO, Fe₂O₃^T, TiO₂, and CaO (Figs. 6B–6E), decrease. A highly scattered, negative correlation between Al₂O₃ and SiO₂ is observed in all of the MMC samples (Fig. 6A). The incompatible elements, e.g., Na₂O, K₂O and Ba (Figs. 6F–6G), show a positive correlation, increasing with SiO₂. Sr behaves as a compatible element in all the MMC samples with a steady decrease in Sr with fractionation (Fig. 6H). The MMC shows a spatial and geochemical migration from purely calc-alkaline magmatism in the Leptokaria samples in the north, to high-K calc alkaline magmatism in the central Kassiteres samples, to shoshonitic magmatism in the southern Maronia pluton (Fig. 5A). In the Maronia samples, the absolute concentration of the incompatible elements is 1–1.5 times greater than in Kassiteres and Leptokaria samples, in line with shoshonitic magmatism (Figs. 6G–6H).

Irregular trace element patterns are observed from all plutons along the MMC (Fig. 7A). The MMC samples are enriched in fluid mobile, large-ion lithophile elements (LILEs; e.g., Ba, Rb, Pb) and depleted in fluid immobile, high field-strength elements (HFSEs; e.g., Nb, Ta, Zr) compared to the primitive mantle, in a trace signature typical of a metasomatized, fertile mantle source (Elliott et al., 1997). The plutons are enriched in the light REEs relative the heavy REEs (Fig. 7B) and show decreasing Dy/Yb (middle REE to heavy REE fractionation) with differentiation (Fig. 6J). The mildly concave REE pattern is flatter than the highly convex, amphibole-controlled REE fractionation trend, and is most similar to that of a clinopyroxene-hornblende–controlled REE pattern (Pearce and Peate, 1995; Tatsumi, 1989; Davidson et al., 2013). A pronounced, negative Eu anomaly, which decreases with increasing SiO₂, is observed in all of the MMC samples (Fig. 6I).

CL imaging revealed simple zoning patterns in most all of the zircons from the MMC (further descriptions are included in GSA Data Repository File DR2). Oscillatory and sector zoning in the crystals reflects a simple magmatic history of the zircons with little evidence of zircon inheritance suggesting the zircons experienced a single period of crystallization. Therefore, dissolution of full zircon crystals was justified to use the largest sample volume and reduce analytical uncertainty, offering the highest-precision dates. Two individual zircon analyses were excluded from our interpretations as high common Pb (Pb* > 2.5 pg) was taken as an indicator of a low precision analysis (see GSA Data Repository Table DR3).

Our results were interpreted within a magmatic framework in which it is recognized that zircon crystallization is protracted over time scales of 10⁵–10⁶ yr (e.g., von Quadt et al., 2011). The youngest statistically significant population of zircons was sought to represent the timing of the final stages of crystallization. Our preferred age interpretations are presented in Table 4; alternative interpretations and a discussion of the sample discordance are included in GSA Data Repository File DR2. All ages reported are ²⁰⁶Pb/²³⁸U to minimize the influence of excess ²⁰⁷Pb, the result of initial U/Pa disequilibrium, on our age interpretations (Crowley et al., 2007). Dates are reported in millions of years ago (Ma) to 2σ uncertainty.

The ²⁰⁶Pb/²³⁸U ages form two distinct groups (Fig. 8). We can interpret two geological windows in time for pluton emplacement and crystallization, (a) intrusion of the Kassiteres and Leptokaria plutons over 0.9 Myrs (from 32.96 to 32.02 Ma); and 2.2 Myrs later (b) intrusion of the Maronia pluton over 0.2 Myrs (from 29.78 to 29.57 Ma). Antecrystic zircons, those crystallized prior to final pluton crystallization either during an earlier phase of magmatism or deeper in the magmatic plumbing system, are recognized in the MMC analyses (e.g., z2 RT14_022; Table DR3 [see ^{footnote 1}]). Zircons that are much older that the interpreted ages of crystallization (>5 Myrs) are interpreted as xenocrystic zircons, entrained from the wallrock during magma ascent (e.g., z2 RT14_024; File DR2 [see ^{footnote 1}]). The autocrystic zircons, those crystallized in situ in the magmatic system, show good clustering in seven of the nine samples from the MMC, allowing interpretation of a youngest population of zircons (Table 4). In Leptokaria samples RT14_022 and RT14_024, the youngest zircon is taken as a maximum age of crystallization of the pluton, allowing the possibility of younger zircons in the sample.

Our geochemical and isotopic data strongly suggest heterogeneity between Maronia and the remainder of the MMC. Therefore, the petrogenesis of the Maronia pluton will be discussed separately to the Kassiteres–Leptokaria magmatic suite before the magmatism along the MMC as a whole is discussed in relation to the extensional tectonics in the Kechros dome.

Previous geochronological data from the MMC comprises nine Rb-Sr biotite–whole-rock pairs ranging from 34.9 ± 1.5–28.9 ± 1.5 Ma, which are interpreted to reflect an ∼8 Myr window of magmatism along the MMC, younging from NE to SW (Del Moro et al., 1988). However, the Rb-Sr geochronometer is highly susceptible to disturbance by hydrothermal overprinting (Jenkin et al., 2001), and at least three of the MMC plutons are associated with significant hydrothermal activity (Melfos and Voudouris, 2017, and references therein). Our high-precision zircon CA-ID-TIMS U-Pb geochronological data show two tightly constrained magmatic windows: (i) 0.9 Myrs (from 32.045 ± 0.027–32.932 ± 0.039 Ma) for the Kassiteres–Leptokaria magmatic suite, and (ii) 0.2 Myrs (from 29.596 ± 0.027–29.762 ± 0.024) for the Maronia magmatic suite (Table 4; Fig. 8). These two discrete time periods are consistent with the distinct geochemical signature of Maronia compared to Kassiteres–Leptokaria.

Within the Kassiteres–Leptokaria magmatic suite, our data do not duplicate the younging of magmatic activity from NE to SW concluded by Del Moro et al. (1988). While the youngest and oldest samples, RT14_028 (32.045 ± 0.027 Ma) and RT14_010 (32.932 ± 0.039 Ma), respectively, do follow this SW-trend, the remaining three samples show ongoing crystallization in the intervening time period, consistent with pulsed magma accumulation across the pluton. This agrees with our earlier interpretation that there is potential for the Leptokaria outcrops to be connected at depth by a large, mostly unexposed magmatic system. The inter-sample age range (0.887 ± 0.039 Myrs at Leptokaria and 0.166 ± 0.028 Myrs at Maronia) reflects the time scale of pluton cooling. As the larger intrusive body, the duration of cooling of the Kassiteres–Leptokaria complex is nearly an order of magnitude greater than the Maronia pluton.

We emphasize that the high-precision U-Pb zircon geochronology presented in this paper shows a clear temporal break in exposed, intrusive magmatism in the MMC, reflecting two discrete magmatic periods. The first magmatic period produced the Kassiteres–Leptokaria magmatic suite and was followed, after an apparent 2.2 Myr break in recorded magmatism, by the Maronia plutonic complex.

Equigranular, plutonic samples from the Kassiteres–Leptokaria magmatic suite record the first stage of post-collisional magmatism hosted in the southern Kechros dome. The high-K calc alkaline magmatic suite has trace element geochemistry that is consistent with a sub-continental lithospheric mantle source (Del Moro et al., 1988; Pe-Piper and Piper, 2001). The presence of hornblende phenocrysts in textural equilibrium with surrounding mineral phases (Fig. 3E) indicates crystallization of the magmatic suite at a depth of >∼3 km (Lledo and Jenkins, 2007; Mutch et al., 2016). The intrusions show granophyric interstitial textures reflecting fast cooling of the last vestiges of melt during rapid final crystallization.

The major and minor element geochemistry of the Kassiteres–Leptokaria suite, most notably the steadily depleting Al₂O₃ with fractionation, the consistently negative Eu anomaly, and the compatible behavior of Sr, suggest plagioclase was an important phase during magmatic differentiation. The presence of abundant plagioclase in the Leptokaria samples with a cumulate-like texture further supports the early appearance of plagioclase on the liquidus (Table 3). This is atypical of arc-type magmatism where partial mantle melts with elevated water contents suppress plagioclase saturation (Gaetani et al., 1993). Decreasing Dy/Yb with fractionation (Fig. 6J) suggest clinopyroxene and/or amphibole are the predominant ferromagnesian phases controlling fractionation (Davidson et al., 2013). Amphibole is the most abundant phase in the cumulate-like samples (Table 3); however, relict subhedral clinopyroxene in the core of many of the amphibole phenocrysts (Fig. 3B) suggests clinopyroxene was a precursor to amphibole during fractionation. This can account for the more primitive REE pattern (Fig. 7B; Smith, 2014).

Experimental studies on the effect of intensive parameters: temperature (T), pressure (P), H₂O content, and oxygen fugacity (fO₂), on phase equilibria are commonly used to investigate the physiochemical conditions of magma differentiation, crystallization, and storage (e.g., Scaillet et al., 2016). Naney (1983) produced phase assemblage diagrams, at both 200 MPa (∼6 km) and 800 MPa (∼24 km), of fO₂ controlled granodiorite compositions as a function of T and H₂O content. The synthetic starting composition used by Naney (1983), “R5 + 10MI,” and the natural analogue hornblende, biotite, granodiorite R5 (Nockolds, 1954), are similar in composition to the Leptokaria tonalities that make up the more evolved end member of Kassiteres–Leptokaria magmatism (GSA Data Repository Table DR1). Using the major fractionating assemblage of plagioclase + pyroxene + amphibole and the phase diagrams of Naney (1983), the T and H₂O field of lower crustal fractionation (at P = 800 MPa, 24 km) can be constrained to 880 < T < 980 °C and 3 < H₂O content < 4.75 wt% (Fig. 9). The drier nature of the magmas suggests that the fertile mantle source has been tapped by multiple different melting events during which H₂O has been mobilized, prior to the generation and emplacement of the Kassiteres–Leptokaria magmas between 32.9 and 32.0 Ma. We suggest that the elevated temperature necessary to melt a drier mantle wedge (>880 °C, Fig. 9) was provided by the accumulation of heat during prolonged (more than ∼10 Myr) asthenospheric upwelling and mantle melting in the northern Aegean, which began in the mid-Eocene (e.g., Rohrmeier et al., 2013).

Maronia shows a different petrogenetic evolution to Kassiteres–Leptokaria, coincident with a 2.2 Myr temporal, and a 30 km spatial, break in recorded magmatism. The Maronia monzonite has a shoshonitic affinity with significant enrichment in incompatible and volatile elements (e.g., P₂O₅, Pb, Sr, Rb; Figs. 5B, 6G, 6H). Geochemical and isotopic evidence advocate a heterogeneous, highly metasomatized source mantle as the foundation for the generation of ultra-potassic shoshonitic to lamproitic magmas (e.g., Foley et al., 1987; Beccaluva et al., 2013; Condamine and Médard, 2014; Müller and Groves, 2016). Sr isotopic data from the MMC reported by Del Moro et al. (1988; Table 4), show consistently enriched radiogenic ⁸⁷Sr across the Kassiteres–Leptokaria (⁸⁷Sr/⁸⁶Sr = 0.7057–0.7080) and Maronia (⁸⁷Sr/⁸⁶Sr = 0.7061–0.7076) magmatic suites. This is unlike the Sr isotopic data reported from the Eocene–Oligocene Biga orogenic volcanic rocks, NW Turkey, where the magmatic suite becomes increasingly radiogenic and potassic with time, which is interpreted to reflect an increasing crustal contribution (Ersoy et al., 2017). Consequently, we suggest that the MMC source mantle was enriched by a radiogenic source, likely crustal-derived metasomatic fluids, prior to the start of high-K calc-alkaline magmatism at Kassiteres–Leptokaria. Therefore, processes involved in magma generation and fractionation were responsible for the evolution of a shoshonitic trend at Maronia. Further to this, experimental data for the contamination and mixing of mantle and sedimentary melts in the production of a K-rich starting melt generate evolved, peraluminous shoshonitic magmas (Wang et al., 2017) while the Maronia magmatic suite has a highly restricted SiO₂ (51–56 wt% SiO₂) range and is metaluminous.

Melting of a phlogopite-rich veinlet in the highly metasomatized source mantle could be a potential source of an anomalously K-rich melt from which to fractionate a shoshonitic trend (e.g., Condamine and Médard, 2014). Such melts would inherit a diagnostic trace element pattern due to the compatible nature of LILEs (e.g., Rb and Ba) and incompatible nature of HFSEs (e.g., Zr and Y) in phlogopite (LaTourette et al., 1995). However, the Maronia samples show a relative enrichment in Rb and Ba over the Kassiteres–Leptokaria suite and similar values of Zr and Y (Fig. 7A). Direct mantle melts would be expected to have a Mg# = 80 – 85 (Mg# = Mg/(Mg + Fe²⁺) * 100; molar), which will decrease with fractionation (Herzberg et al., 2014). The Mg# of the Maronia shoshonites ranges from 49.4 < Mg # < 60.2 indicating that significant fractionation from direct lherzolitic or harzburgitic mantle would be required to produce the potassic fractionation trend observed at Maronia. Therefore we favor the model of pyroxene + An-rich plagioclase + magnetite fractionation in the absence of major olivine, in an H₂O-poor magma, as a mechanism to produce a shoshonitic LLD (Meen, 1987; Freise et al., 2009; Lanzo et al., 2016; Beermann et al., 2017).

We have modeled fractional crystallization by subtracting varying proportions of the proposed orthopyroxene + clinopyroxene + An-rich plagioclase + magnetite fractionating assemblage (Meen, 1987; Beermann et al., 2017), from the most primitive known sample from Maronia (MP 81; Mavroudchiev et al., 1993). We do not include olivine in our fractionating assemblage as we consider it to be a minor phase in the generation of a LLD (Meen, 1987) and we assume K₂O to be incompatible during fractionation (Freise et al., 2009). The observed fractionation trend from the parent magma can be closely matched by a modeled fractionation assemblage of 15% orthopyroxene (En 0.8), 25% clinopyroxene (Di 0.8), 52% plagioclase (An 75) and 8% magnetite (Mag 0.7; Figure 10A; Table 5). The close agreement between the modeled and observed LLDs suggests that fractionation of pyroxene + plagioclase (+ minor magnetite), which have higher silica content than olivine (OL ∼30 wt% SiO₂; OPX ∼50 wt% SiO₂; CPX ∼45 wt% SiO₂; PLAG ∼50 wt% SiO₂), can sufficiently enrich the residual magma in incompatible elements with small changes in melt SiO₂, to produce a shoshonitic LLD (Meen, 1987; Beermann et al., 2017). Using the predicted fractionating phase assemblage and estimated bulk partition coefficients (Table 6), the observed trace element fractionation trends can be reproduced and are shown in Figure 10B for Ba, Cr, Rb, Sr, Zr, and Y.

A pyroxene + plagioclase + magnetite fractionating assemblage can be produced by an H₂O-poor (<2–3 wt% H₂O) magma at high-T (950–1050 °C) and a range of pressures (200–1,000 MPa; Baker and Eggler, 1987; Meen, 1987; Beermann et al., 2017). This suggests that fractionation of the same parental mantle source region as Kassiteres–Leptokaria could produce a shoshonitic trend with a similar trace element signature as the Maronia pluton by losing H₂O. In a post-collisional setting, H₂O is a finite resource. Thus, if moderately wet magmas (<4.75 wt% H₂O) were produced in the generation of the Kassiteres–Leptokaria magmatic suite driving the H₂O content of the parental source region down, the later Maronia magmas could be generated from an H₂O-poor (<∼2 wt%) source region in which pyroxene + plagioclase are stable instead of olivine. The generation of shoshonitic magmas at Maronia requires a change in mantle melting conditions between the intrusion of the Kassiteres–Leptokaria magmatic complex and intrusion of the Maronia pluton. We suggest that the temporal break (2.2 Myrs) in recorded magmatism in the MMC is the result of H₂O loss in the source mantle driving up the solidus temperature such that, between intrusion of Kassiteres–Leptokaria and Maronia, the source mantle was at sub-solidus conditions. We propose that prolonged heat conduction from the upwelling thermal anomaly beneath the Kechros dome in the intervening 2.2 Myrs is a potential source for the additional heat input required. Resultant melting of the H₂O-poor source mantle stabilized a pyroxene + plagioclase–dominated fractionating assemblage and generated a shoshonitic LLD at Maronia. A similar mechanism of restricted H₂O flux into a metasomatized mantle source is suggested for the generation of the shoshonitic volcanics in the Aeolian arc, Italy (e.g., Peccerillo et al., 2013; Beermann et al., 2017).

Post-collisional magmatism in the northern RCC began in the mid-Eocene, contemporaneous with a period of trans-crustal extension and core complex exhumation, and continued into the Oligocene (Table 2; e.g., Marchev et al., 2005; Rohrmeier et al., 2013). This phase of post-collisional extension is attributed to the collapse of thickened orogenic lithosphere induced by slab rollback of the N-ward subducting African plate (e.g., Jolivet and Brun, 2010; Ring et al., 2010; Burg, 2011; Kydonakis et al., 2015). Eocene–Oligocene magmatism from across the northern RCC has a strong enriched metasomatic mantle signature, the result of asthenospheric mantle upwelling beneath thinned lithosphere remobilizing and re-melting the subcontinental lithospheric mantle (e.g., Pe-Piper et al., 2009).

In the Kechros dome, Eocene crustal extension was accommodated by tectonic unroofing of the lower-crustal gneisses along the Kechros detachment fault, a trans-crustal structure that would be later exploited by post-collisional magmatism (Marchev et al., 2010; Márton et al., 2010). An Oligocene, high-K calc-alkaline magmatic event at 32.9–32.0 Ma in the southern Kechros dome resulted in the intrusion of the Kassiteres–Leptokaria magmatic suite, a high-K calc-alkaline series of intrusions with a strong subduction-related sub-continental lithospheric mantle signature. Following a 2.2 Myr break in magmatism, the intrusion of the shoshonitic Maronia pluton marks the last magmatic activity in the Kechros dome prior to tectonic change. The potassic intrusion, enriched in incompatible and volatile elements, marks an anomalously hot, short period of magmatism in the Kechros dome. Melting of a now H₂O-poor (<∼2 wt% H₂O) mantle source fractionated plagioclase + pyroxene to produce the Maronia intrusion and marked the culmination of asthenospheric heating of a mature, Cretaceous metasomatized lithospheric mantle. We suggest that the termination of magmatism in the Kechros dome following the intrusion of the Maronia pluton at 29.6 Ma reflects the southward migration of the asthenospheric upwelling, switching off the heat source for mantle melting beneath the Kechros dome. We hypothesize that the asthenospheric upwelling migration is co-incident with initiation of SW-rotation of the southern RCC as described by van Hinsbergen et al. (2008) and others. The continuation of post-collisional magmatism in the southern RCC, such as the Miocene shoshonitic magmatism in Samothraki (e.g., Christofides et al., 2000) and Skouries (e.g., Siron et al., 2016) reflects the persistence of asthenospheric upwelling beneath the southern RCC and melting of the mature, H₂O-poor (<∼2 wt% H₂O) mantle wedge.

The MMC intrusions provide chronological markers and geochemical tracers that record the petrogenetic evolution of magmatism in the Kechros dome, northern RCC. Late Cretaceous to Oligocene extension of the northern Aegean resulted in asthenospheric mantle upwelling beneath the northern RCC, which provided a heat source for re-melting of the Cretaceous metasomatized mantle wedge. High-precision geochronology coupled with a full suite of whole-rock geochemical analyses, detailed field and petrographic observations, and petrological modeling demonstrate that:

1. The MMC does not represent a continuous magmatic period between 35 and 28 Ma (Del Moro et al. 1988) but two, temporally and geochemically distinct, magmatic complexes, the Kassiteres–Leptokaria magmatic suite and the Maronia plutonic complex. Intrusion of Kassiteres–Leptokaria was between 32.05 ± 0.02 and 32.93 ± 0.02 Ma, while magmatism at Maronia lasted from 29.60 ± 0.02–29.76 ± 0.02 Ma. In addition, it is recommended that the MMC zircon U-Pb data presented here are used in preference to Del Moro et al. (1988) to constrain the timing of magmatic activity in the MMC;

2. Extended hydrous (3 < H₂O content < 4.75 wt%) partial mantle melting beneath the northern RCC allowed for the intrusion of batches of lithospheric mantle-source magma over an ∼900,000 year time scale at Kassiteres–Leptokaria;

3. The shoshonitic affinity of the Maronia samples reflects incompatible element enrichment produced by plagioclase + pyroxene–controlled mantle fractionation from a source mantle highly depleted in H₂O (<∼2 wt% H₂O); and therefore,

4. The temperature of mantle melting beneath the Kechros dome increased between intrusion of the Kassiteres–Leptokaria magmatic suite (at 32.9–32.0 Ma) and the Maronia pluton (29.8–29.6 Ma). Consequently, we suggest that the apparent 2.2 Myr break in recorded magmatism could be the result of a sub-solidus lithospheric mantle, and that heat accumulation in the intervening time period provided the heat required for the renewal of mantle melting.

This work was funded by a NERC GW4+ UK Doctoral Training Partnership Studentship (Grant NE/L002434/1) and in part by the BGS University Funding Initiative (S276), which are gratefully acknowledged. Supporting data are provided within this paper and the GSA Data Repository (Item 2018177). Ben Buse and Stuart Kearns are thanked for their help with SEM analysis, and Dr. C. Gowing at the Inorganic Geochemical Facility at the British Geological Survey for the whole-rock geochemistry analyses. We would like to thank Katerina Gkika, Ery Hughes, and Simon Dahlström for assistance in the field, and the petrological modeling benefited greatly from discussions with Robert Loucks. D.J. Condon and J. Naden publish with the permission of the Executive Director, British Geological Survey (NERC). We thank Alexandre Kounov and two anonymous reviewers for their expert reviews, which have greatly improved the manuscript.