Variscan ultra‐high‐pressure eclogite in the Upper Allochthon of the Rhodope Metamorphic Complex (Bulgaria)

The Rhodope Metamorphic Complex (RMC) in Bulgaria has been established as a Mesozoic ultra‐high‐pressure metamorphic province by findings of microdiamond in gneisses. Additionally, Variscan ultra‐high‐pressure metamorphism has been proposed for the Ograzhden/Vertiskos Unit in the Upper Allochthon of the RMC, based on findings of coesite, graphite pseudomorphs after diamond and indirect age constraints. We confirm ultra‐high‐pressure metamorphism of eclogites in this unit using thermobarometry, phase‐equilibrium modelling and the Variscan age of metamorphism using Lu–Hf garnet–whole‐rock dating. In Belica (southern Rila Mountains), kyanite‐ and phengite‐bearing eclogite enclosed in high‐grade gneisses records P‐T conditions of 3.0–3.5 GPa and 700–750°C. Lu–Hf dating of eclogite samples from Belica and Gega (Ograzhden Mountain), where coesite was found, yielded ages of 334.1 ± 1.8 and 334.0 ± 2.2 Ma, respectively, interpreted as the age of garnet growth during post‐collisional subduction of continental crust after closure of the Rheic Ocean.

It either reflects subduction of the European margin under an island arc (Bonev et al., 2015) or closure of the Palaeotethys (Petrík et al., 2016).
This study deals with the Ograzhden Unit, the structurally highest thrust sheet of the Upper Allochthon in the Bulgarian part of the western RMC, equivalent to the Vertiskos Unit in Greece. It comprises orthogneiss and to a minor extent paragneisses, marble and metamafic rocks. U-Pb zircon dating of orthogneisses yielded Ordovician (~462-452 Ma) and Silurian protolith ages (~443-426 Ma;Himmerkus et al., 2009a;Macheva et al., 2006). Syn-to post-tectonic granite intrusions, which crosscut the main foliation

Statement of Significance
The article for the first time demonstrates the existence of Variscan ultra-high-pressure metamorphism in southeast Europe by phase-equilibrium modelling and Lu-Hf garnet dating of eclogites. The dating yielded identical, highly precise results for two localities of 334.0 ± 2.2 and 334.1 ± 1.8 Ma. This is evidence for post-collisional subduction of continental crust south of the Rheic ocean suture. This result is highly significant for the reconstruction of Carboniferous tectonics in Europe.
An outlier of the Ograzhden Unit occurs at Obidim on the eastern slopes of the Pirin Mountains ( Figure 1). Zircons from two metagranites in this area yielded Ordovician protolith ages of 456.1 ± 1.8 and 452 ± 14 Ma and zircons from a metagabbro 454.1 ± 8.3 Ma   (Peytcheva et al., 2015). Thermobarometry of a kyanite eclogite from Obidim (Pirin Mountains) yielded UHP conditions of ~3 GPa/700-750°C . The nearby occurrence of a 321 ± 19 Ma migmatite  again suggests a Variscan age. Scattered relics of eclogite have also been described from the Vertiskos Unit in Greece (e.g. Dimitriadis & Godelitsas, 1991). Kydonakis et al. (2015) studied garnet-kyanite mica schists from the Vertiskos Unit and demonstrated eclogite-facies conditions. They suggested the schists to originally represent the Mesozoic cover of the Vertiskos basement and that therefore the eclogite-facies metamorphism is Mesozoic in age. Kostopoulos et al. (2000) described graphite pseudomorphs after diamond in an amphibolite xenolith within the Triassic Arnea granite intruding the Vertiskos Unit ( Figure 1) and argued that the diamond-forming UHP metamorphism must have been pre-Triassic, probably Carboniferous.
In summary, the age of HP/UHP metamorphism in the Vertiskos/ Ograzhden basement and related units is still unclear: In the Bulgarian part, a Variscan age appears more likely; in Greece, both Alpine and Variscan ages have been suggested. In order to clarify the age of this HP/UHP metamorphism, we determined age and metamorphic conditions of eclogites from the Ograzhden Unit, with two samples from Belica (BEL-1 used for thermobarometry and BEL-2 for geochronology) and one sample from Gega (NF17-5 for geochronology).  Table S1).

| ME THODS
The methods are specified in Supplementary Information (Methods S1).

| Eclogite texture and P-T conditions
The Peak metamorphic P-T conditions have been calculated for sample BEL-1 using thermodynamic modelling, and "conventional" geothermobarometry. We used the Perple_X thermodynamic software (Connolly, 2005: version 6.8.6) with the internally consistent thermodynamic database of Holland and Powell (2011). Solid-solution models for garnet, white mica (White et al., 2014), omphacite (Green et al., 2007), plagioclase (Holland & Powell, 2003) and amphibole (Dale et al., 2005) were used, as available from the Perple_X datafile (solution_model.dat). The bulk rock composition was determined from whole-rock analysis.
The calculated phase diagram ( Figure 3) shows that compositional isopleths of garnet (X Grt Mg ), omphacite (X Omp Na ) and phengite (Si a.p.f.u.) matching the measured compositions (Table 1) intersect in the stability field of garnet + phengite + omphacite + kyanite + rutile + coesite, i.e. the peak-pressure metamorphic assemblage, constraining P-T conditions of 3.0-3.5 GPa and 700-750°C. At these conditions, amphibole and zoisite are not stable which suggests that inclusions of amphibole and zoisite in garnet are remnants from a prograde, lower P-T stage. The P-T results obtained by conventional geothermobarometry (Ravna & Terry, 2004) from the garnet with the highest grossular content, omphacite with the highest jadeite content and phengite with the highest Si content (Table 1)  Lu line profiles through this garnet grain (Figure 4b), however, show maxima close to but not directly at the rim of the garnet, and not coinciding with the Mn increase. One of the garnet grains from sample TA B L E 1 Representative compositions of garnet, omphacite and phengite from sample BEL-1 measured by WDS (wavelength-dispersive spectrometer) analysis used for P-T calculations. NF 17-5 has been disintegrated into three pieces by fracturing and resorption ( Figure 5). Abundance of Mn in this garnet is rather uniform but shows a slight increase at the resorbed rims of disrupted garnet fragments. Lutetium increases towards the rims of the original grain and does not correspond to the Mn distribution. There is almost no increase in Lu at the rims between the garnet fragments.

| Lu-Hf geochronology
The Lu-Hf isotopic compositions of the mineral separates and whole rock powders are shown in Table 3. In sample NF17-5, absolute Hf contents in the whole rock vary between 0.526 and 0.556 ppm and 176 Lu/ 177 Hf ratios between 0.05993 and 0.07443. In the garnet F I G U R E 4 Major element and Lu distribution in garnet grains from sample BEL-2 (Belica). (a) Mn concentration map and major-element profile of the garnet composition (find microprobe data in Table S2). (b) Mn concentration map with laser spots and Lu concentration profile.

F I G U R E 5
Major element and Lu distribution in garnet grains from sample NF17-5 (Gega). (a) Mn concentration map and major-element profile of the garnet composition (find microprobe data in Table S3). (b) Mn concentration map with laser spots and Lu concentration profile.
separates of sample NF17-5, the absolute Hf contents range be-

| Garnet composition patterns and age interpretation
Major-element maps and profiles of BEL-2 and NF17-5 show homogeneous distribution, reflecting diffusional re-equilibration. A slight increase in Mn at resorbed garnet rims can be explained by backdiffusion during resorption (Figures 4 and 5).
Distribution of Lu in the dated samples does not show the typical bell-shaped concentration profiles with a central peak as often TA B L E 3 Lu-Hf isotopic compositions of the whole rocks (WR; b = bombed digestion, tt = tabletop) and garnet separates (Grt) of the sample NF17-5 and BEL-2.

F I G U R E 6
Lu-Hf isochrons for the two eclogite samples BEL-2 and NF17-5. Uncertainties are 2σ. The decay constant λ176Lu = 1.865 × 10 −11 a −1 was used . WR b = bomb-digested whole rock, WR tt = tabletop-digested whole rock, Grt = garnet separate observed and interpreted to result from Lu fractionation into garnet (e.g. Otamendi et al., 2002;Skora et al., 2006). Instead, the profile of BEL-2 ( Figure 4) is saddle-shaped with peaks in the outer parts of the garnet crystals. We interpret the marginal peaks as not being due to resorption and back-diffusion (which could lead to a "younging" of the ages; Kelly, Carlson, & Connelly 2011) because (a) they are not directly at the rims but inside the garnet; and (b) they do not correspond with the most resorbed rims. Such secondary peaks are explained by an increase in diffusion rate during garnet crystallization when the temperature increases (Skora et al., 2006). In NF17-5, the Lu contents increases from core to rim with the highest values in the outermost measured spots. Nevertheless, we interpret these profiles as showing growth zoning with rim peaks, where the outermost points were not measured close enough to the rim to see if the Lu content decreases again ( Figure 5). Remnants of a central Lu peak, if they exist, may have been missed by the sections.
The RMC, therefore, experienced UHP metamorphism twice, during the Variscan and Early Alpine Orogeny. This reflects the tendency of continents to break apart and re-collide along earlier collisional orogenic belts, a process known as the Wilson cycle (Wilson, 1966).

CO N FLI C T O F I NTE R E S T
The authors declare that there is no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available in the main text and the Supporting Information of this article.