Detailed field investigations were carried out in the study area, including systematic documentation of lithological features and deformation structures at all accessible outcrops, given that the terrain is covered by thick soil and dense vegetation. Thirty-five representative rock samples were selected for petrographic analyses (Appendix A).

Mineral phase chemistry of two representative samples was analyzed at the University of Geneva's Department of Earth Sciences using a JEOL JXA 8200 Superprobe equipped with five wavelength-dispersive spectrometers. Operating parameters included an accelerating voltage of 20 kV, a 15 nA beam current, and a defocused beam of ∼10 µm. Counting times of 30 s on peak and 15 s on background on both sides of the peak were used for all elements, except for Na and K, which were measured 20 and 10 s on peak and background. For this reason, Na and K were also measured first. Limits of detection were calculated as the minimum concentration required to produce count rates three times higher than the square root of the background (3σ; 99 wt % degree of confidence at the lowest detection limit). Concentrations below the limits of detection are reported as not detected. Raw data were corrected for matrix effects using the φρZ method of Armstrong (1991). Natural minerals, oxides (corundum, spinel, hematite, and rutile), and silicates (albite, orthoclase, anorthite, and wollastonite) were used for calibration. Mineral formulas were calculated using a software package MINPET, written by Linda R. Richard (Gatineau, Québec, Canada).

X-ray powder diffraction (XRD) was performed on the global particle fraction on a set of eight selected samples. Sample preparation included material powdering in an agate mortar prior to whole rock measurements. The measurements were undertaken at the Geosciences Clay Laboratory of Texas Tech University using a Bruker D8 Advance diffractometer. This instrument features a horizontal goniometer axis and synchronized rotation of both the X-ray source and detector arms. Measurements consisted of a step scan in the Bragg-Brentano geometry with CuKα radiation (40 kV and 40 mA). Sample mounts were scanned for 1.8 s per 0.02°, from 3 to 70° 2θ. XRD traces interpretation was accomplished using Bruker EVA software and comparison against the PDF4 database issued by the International Centre for Diffraction Data.

Bulk-rock powders from 11 samples were analyzed for major elements by X-ray fluorescence (XRF) at the Department of Geosciences, Texas Tech University, and for trace elements by inductively coupled plasma mass spectrometry (ICP-MS) at Actlabs (Canada). XRF analyses were performed on fused glass beads using international mafic rock standards for calibration. Analytical accuracy and precision are better than ±1% for major elements and ±5% for trace elements. Trace element concentrations were determined by ICP-MS following a near-total four-acid digestion (HF–HNO3–HClO4) with subsequent dissolution in aqua regia. The resulting solutions were analyzed by ICP-MS. Detection limits are typically on the order of 0.01–0.1 ppm for most trace elements, depending on the element and analytical package. Detection limits are defined as three times the standard deviation (3σ) of the blank and are reported at 10 times this value.

Neodymium isotopic compositions of four bulk rock samples were measured at the Noble Gas Laboratory Pacific Centre for Isotopic and Geochemical Research, University of British Columbia using a Triton Plus mass spectrometer. Normalizing ratios of ¹⁴⁶Nd / ¹⁴⁴Nd =0.7219 were assumed. The ¹⁴³Nd / ¹⁴⁴Nd ratio for the La Jolla standard was 0.5118451±0.000010 (2σ). Total procedural blanks were ∼150 pg. Python™ programing language (numpy package) was used to calculate the Monte Carlo propagation error through 10000 iterations for ¹⁴³Nd / ¹⁴⁴Nd(t).

Two samples (Kjumberk Hill rhyolitic lava: GA-1 and Margečan rhyolitic ignimbrite: MA-11) were collected for zircon U-Pb geochronology. The zircons from the sample were extracted using the standard method of crushing, sieving, heavy liquid separation, magnetic separation and hand-picking at the Croatian Geological Survey. The ∼20–30 zircon grains per sample were then mounted in epoxy, ground, and polished until their crystal centers were exposed. Zircon U-Pb isotopic measurements were performed by laser ablation–inductively coupled plasma mass spectrometry (LA-ICP-MS) at the Arizona LaserChron Center (ALC) in the Department of Geosciences at the University of Arizona following techniques detailed in Gehrels et al. (2006, 2008) and Gehrels and Pecha (2014). Analyses used a Photon Machines Analyte G2 Excimer Laser attached to a Thermo Element2 HR single-collector ICP-MS. The U-Pb age for each analysis was then calculated using the ALC in-house software E2agecalc, and for concordant subsets, a weighted average age was calculated using Isoplot 4.15 (Ludwig, 2011).

Metamorphic pressures were estimated using the phengitic mica barometer of Massonne and Schreyer (1987), which is based on the pressure-dependent incorporation of the Tschermak substitution in white mica. This substitution involves the coupled exchange of Si for Al in the tetrahedral site, accompanied by the substitution of Al for Mg or Fe in the octahedral site, expressed by the reaction: Mg2++ Si4+↔ 2Al³⁺. With increasing pressure, phengitic mica accommodates higher Si contents per formula unit, reflecting enhanced stability of the celadonitic component under high-pressure conditions. For a given temperature, the Si content of phengitic mica therefore provides a quantitative constraint on pressure. Calculations were performed using measured mica compositions normalized to 11 oxygens, applying the Massonne and Schreyer (1987) calibration at fixed temperatures to evaluate the pressure conditions recorded during mica crystallization.

In addition, metamorphic pressures were independently constrained using the b-axis parameter of K-white mica. The b-cell dimension of phengitic muscovite is known to increase systematically with pressure as a result of enhanced celadonitic substitution (Sassi and Scolari, 1974; Guidotti and Sassi, 1986; Kisch et al., 2006; Ferreiro Mählmann et al., 2012). The b-parameter is determined from the position of the d(060,331) reflection, following the relationship b=6×d(060), measured by X-ray diffraction and provides a semi-quantitative estimate of pressure across metamorphic regimes. This method relies on the documented linear relationship between the d(060,331) spacing and the celadonite content of white mica (Ernst, 1963; Guidotti et al., 1989; Rieder et al., 1998). XRD traces were inspected in the 59–63° 2θ range, and b-cell dimensions were calculated using the WIN-METRIC v.3.0.7. refinement software (Bruker-AXS). The resulting b-parameter values provide an independent constraint on pressure conditions and serve to corroborate pressures inferred from phengitic muscovite chemistry based on Tschermak substitution.

The rhyolitic lavas from the Kjumberk Hill area exhibit a porphyritic texture and a homogeneous structure. They are composed of plagioclase phenocrysts, which are commonly altered to fine-grained white mica and clay minerals, whereas the predominantly fresh K-feldspar phenocrysts contain only occasional altered inclusions (Fig. 3a–b). The groundmass is leucocratic and devitrified, made of fine-sized (∼0.1–0.5 mm) quartz and white mica crystals (Fig. 3a–b). Rhyolitic ignimbrites at the investigated locality reach thicknesses of up to 100 m and are represented by welded, laminated, flow foliated units and welded massive ignimbrites. Foliated bands range from 1 to 10 cm in thickness and consist of alternating very thin layers of vitrophyric (Fig. 3c), crystaloclastic (Fig. 3d–e), and lapilli-bearing vitrophyric (Fig. 3f) rhyolitic pyroclastic flow material. Vitrophyric laminas of the rhyolitic pyroclastic flow are made of altered matrix-supported ash-sized pyroclastic detritus (<2 mm in size) and a devitrified glassy groundmass. The vitrophyric lapilli bearing laminae contain altered matrix-supported to clast-supported pumice fragments larger than 2 mm, along with crystaloclasts embedded in a devitrified glassy groundmass (Fig. 3f). The crystaloclastic laminae of the rhyolitic pyroclastic flow are composed of quartz, albite (An0.0-4.5Ab95.1-99.9Or0.1-0.6; Figs. 3e, 4a; Appendix B) and K-feldspar (An0.0-1.0Ab1.8-4.2Or95.0-98.4; Figs. 3e, 4a; Appendix B) crystaloclasts, along with very rare pumice fragments, while the matrix is altered to white mica (phengite) exhibiting a conjugate tectonic cleavage texture (Fig. 3e). Welded rhyolitic lapilli ignimbrites exhibiting massive texture are made of clast-supported to matrix-supported altered lapilli-sized pumice fragments (>2 mm), and less commonly of feldspar and quartz crystaloclasts, all set within a devitrified glassy groundmass (Fig. 3f). The alteration products in all varieties are fine-grained quartz and foliated phyllosilicates (Fig. 3c–e). The alteration is polyphase, having developed during (i) syn-depositional and high-temperature volcanic processes, (ii) subsequent low-temperature post-depositional diagenetic alteration, and (iii) tectonic overprinting expressed as dynamothermal metamorphism, marked by a foliation cleavage along which white mica crystallized (Fig. 3d–e). The white mica may be classified as phengitic muscovite (Fig. 4) with a Si content of 3.4 a.p.f.u. (Appendix B). Measurements of the chemical composition of phengitic muscovite show slight variations in the composition of the core and rims (Fig. 3g; Appendix B), which are reflected in slightly higher Al, Fe and Na contents at the rims. The undulose extinction of quartz crystaloclasts further indicates post-diagenetic dynamo-metamorphism to which the analyzed rocks were subjected to.

X-ray diffraction mineralogy supports the petrographic observations described above, while also providing additional constraints on the extent of post-magmatic alteration. Quartz is the dominant crystalline phase in all investigated rhyolitic ignimbrites and rhyolitic lavas, consistent with the quartz-rich groundmass and abundant quartz crystaloclasts observed in thin section (Appendix C). White mica occurs in all samples and locally reaches moderate relative abundances, reflecting pervasive alteration of the volcanic matrix and feldspar phases, as well as the development of cleavage-parallel phengitic muscovite documented petrographically and by EMPA. In contrast, K-feldspar is detected only in a limited number of samples and is absent from the majority of the investigated material. Its scarcity in the XRD patterns, despite petrographic evidence for apparently fresh K-feldspar phenocrysts, indicates that these phases are at least partially altered, most likely to fine-grained clay minerals such as kaolinite. Minor albite, kaolinite, and subordinate illite–smectite interstratifications occur sporadically, consistent with low-temperature alteration superimposed on the primary volcanic assemblage. Trace amounts of calcite and anatase were identified in a small number of samples and are interpreted as late-stage alteration products. XRD mineral assemblages corroborate the petrographic evidence for extensive devitrification of volcanic glass and confirm a polyphase alteration history involving (i) syn-depositional/high-temperature processes, (ii) subsequent low-temperature post-depositional diagenetic alteration, and (iii) tectonically controlled dynamothermal overprinting, expressed by the formation of foliated white mica.

Based on the white mica barometer of Massonne and Schreyer (1987), and assuming a metamorphic temperature of 300 °C whose value is taken as the mean temperature value of a very low degree of metamorphism according to Winkler (1979), and the conodont alteration index (CAI), vitrinite reflectance, and illite crystallinity data (Árkai et al., 1991; Kukoč et al., 2023; Vukovski et al., 2024), the estimated pressure ranged between 1.1 and 1.2 GPa. These values exceed the pressure conditions inferred from the b0 lattice parameter of white mica (Appendix B), which yields moderate pressure estimates (0.6–0.8 GPa). The b0 lattice parameter data documented herein are, however, consistent with the values reported for a range of high-pressure, low-temperature (HP–LT) terrains (Iwasaki et al., 1978; Okrusch et al., 1978; Altherr et al., 1979; Frey et al., 1983; Schertl et al., 1991; Okay, 2002; Kisch et al., 2006). This behavior is characteristic of HP–LT terranes, in which white mica chemistry may preserve peak-pressure conditions, whereas the b0 lattice parameter commonly reflects partial re-equilibration during exhumation, resulting in moderate b0 values that nevertheless remain typical of HP–LT metamorphic belts (Kisch et al., 2006).

Representative geochemical analyses of nine rock samples are provided in Appendix D. The chemical compositions of all analyzed lava and ignimbrite samples fall within similar ranges and will, therefore, be considered as a single group. Although relatively variable, silica concentrations are consistently high (SiO2=69.41 wt %–82.07 wt %). The rocks are characterized by moderately high Al₂O₃ contents (up to 12.51 wt %) and low concentrations of TiO₂ (≤0.29 wt %), Fe₂O₃ (≤4.41 wt %), MgO (up to 1.27 wt %), CaO (≤0.14 wt %), and P2O5 (≤0.06 wt %), while K₂O concentrations display a broad range (1.69 wt %–10.42 wt %). The wide range of K contents partly reflects varying modal amounts of alkali feldspar phenocrysts, consistent with petrographic observations (Fig. 3b, e). In the classification diagram Nb / Y vs. Zr / TiO2×0.0001 (Pearce, 1996), analyzed rocks are plotted in the field of subalkaline rhyolite (Nb / Y <0.5; Fig. 5a), which corresponds to their major mineral assemblage. High-K content (Appendix D) coupled with high values of Ce / Yb and Ta / Yb indicate calc-alkaline affinity (Fig. 5b). Low concentrations of Ni (≤20 ppm) and Cr (≤49 ppm) are in line with the evolved, felsic character of the rocks (Wilson, 1989, chapter 11.5.). Elevated Zr contents, reaching up to 410 ppm, the concomitant decrease in Zr / Hf ratios with falling Zr abundance, and a negative correlation of both Zr and Hf with SiO₂ (not shown) point to zircon fractionation and imply that the melts were zircon-saturated (Linnen and Keppler, 2002). Low- to moderate-level selective mobilization of certain large ion lithophile elements (LILE), such as Cs, Rb, and Ba, likely associated with secondary low-temperature alteration, has been identified (Fig. 6a1). These elements are thus excluded from further petrogenetic interpretation. In contrast, high field strength elements (HFSE), including Th, Nb, Ta, Ti, Hf, Y, as well as the rare earth elements (REE), appear unaffected by post-magmatic overprint and are considered to retain their original magmatic signatures (Appendix E and Fig. 6a1).

In multielement variation patterns normalized to primitive mantle (PMN) analyzed samples show enrichment in LILE (Cs, Rb, K) and Th (up to 1730 times relative to PM), strong negative anomaly of HFSE (Nb–Ta, P, Ti) and Sr [(Sr/Nd)PMN =0.01–0.07] as well as variable positive Pb [(Pb/Ce)PMN =1.64–11.92] and negative Ba [(Ba/Rb)PMN =0.06–0.33] spikes (Fig. 6a1).

Chondrite-normalized (CN) REE patterns display fractionation and enrichment of light rare earth elements (LREE) over middle rare earth elements (MREE), with [(La/Sm)CN =2.47–3.53] and total LREE concentrations ranging from 13 to 113 times chondritic values. This feature is more pronounced in the rhyolitic lava samples (Fig. 6b1). However, all samples exhibit an approximately flat pattern in the heavy rare earth elements (HREE), with [(Tb/Lu)CN =0.86–1.53] and HREE concentrations ranging from 9 to 33 times chondritic values.

Measured ¹⁴³Nd / ¹⁴⁴Nd ratios of four representative samples vary in a narrow range from 0.512250 to 0.512342 (Appendix F). The initial εNd are calculated for 242 and 243 Ma, the crystallization age of the analyzed rhyolitic lava and ignimbrite samples based on the U-Pb zircon dating (see below). The initial εNd varies between -4.42 to -5.13. Very low initial ¹⁴³Nd / ¹⁴⁴Nd and εNd(t) values are close to the subducted continental material field and indicate a very high degree of crustal contamination (Fig. 7a). The influence of sedimentary and/or crustal components is further supported by the well-defined trend in the Th / La vs. Sm / La ratio (Fig. 7b).

The zircon grains from the Margečan rhyolitic ignimbrite (sample MA-11) range in size from 100 to 300 µm and are predominantly elongated, with sharp and straight grain boundaries. A total of 31 zircon grains were separated and analyzed for U-Pb dating from sample MA-11 (Appendix G). Of these, 26 analyses yielded concordant ages, while five were rejected as discordant, likely due to Pb loss or crystal alteration, and were not presented in the Fig. 8. The correction for initial Pb composition was not applied. The weighted mean age of the 26 selected zircon grains is 243.8±0.43 Ma (MSWD =12) (Fig. 8b), although the presence of two slightly distinct Middle Triassic age populations (239.67±0.67 and 246.13±0.55 Ma) cannot be excluded (Appendix H).

A total of 22 analyses were performed on zircon grains from the rhyolitic lava sample GA-1 collected at Kjumberk Hill. The correction for initial Pb composition was not applied, and concordant ages were obtained from 21 out of 22 measurements. The results are presented in Appendix G and illustrated on the concordia diagram (Fig. 8a). The weighted mean age of these 21 selected zircons is 242.2±1.80 Ma (MSWD =0.23). The discordant age that was excluded may reflect Pb loss or crystal alteration. Additionally, three ages classified as concordant exhibit minor deviations from the concordia line but are still considered valid.

The very high SiO₂ contents (>79 wt %) and elevated K₂O concentrations (>4 wt %), coupled with low Na₂O and CaO abundances and slightly higher loss on ignition (LOI =2.02–3.83 wt %), as well as the selective enrichment in Rb, Ba, and Cs (Appendix D; Fig. 6a1), suggest that some of the analyzed calc-alkaline rhyolitic lavas and ignimbrites were affected by alteration processes of variable intensity, likely related to low-temperature hydration, hydrothermal activity, and/or low-grade metamorphism (Riley et al., 2001; Šegvić et al., 2023). The low K₂O+Na₂O (1.69–10.42) with a simultaneously extremely high ratio 100× K₂O/(K₂O+Na₂O) (35.6–100.0), according to the criterion proposed by Hughes (1973), indicates weak to moderate K-metasomatism of the investigated rocks. In light of these characteristics, petrogenetic interpretations of the studied rocks are based on discrimination ratios of immobile trace elements, REE and Nd isotopic data.

The identical chemical and isotopic composition of the investigated Middle Triassic rhyolitic lavas and ignimbrites, along with the coeval acidic Pietra Verde tuffs identified in the broader study area (Slovenec et al., 2023), indicates a shared origin from the same magmatic source (Figs. 6a1–b1, 6a2–b2, 7a–b, 9a–b). The generation of these felsic volcanic rocks most probably involved partial melting followed by fractional crystallization of plagioclase and K-feldspar, as evidenced by the negative anomalies in Sr (Sr depletion, Xu et al., 2010) and Eu ((Eu/Eu* =0.22–0.55); Fig. 6a1–b1), which represent key liquidus phases controlling melt composition (Storey, 1995). Feldspars in this crystallization system acted as stable residual phases, with crystallization taking place within the plagioclase stability field and in the absence of garnet (Wang et al., 2012; Fig. 9b). The magma from which these felsic rocks formed was subduction-related, as indicated by negative anomalies in Nb–Ta (e.g., Pearce et al., 1984; Hofmann, 1997) and probably Ti (Fig. 6a1), contaminated by lithospheric mantle melts, and sourced from a domain enriched in subduction-derived crustal components. The crustal signature, more specifically the influence of subducted and recycled sediments, is reflected in (i) strongly negative initial εNd values (-4.42 to -5.13) accompanied by low ¹⁴⁷Sm / ¹⁴⁴Nd ratios (≤0.134911; Fig. 7a), and (ii) a marked increase in Th / La at nearly constant Sm / La ratios (Fig. 7b). Partial melting of crustal rocks likely contributed to the genesis of these units, as suggested by chondrite-normalized REE patterns resembling those of the average upper continental crust (Fig. 6b1) and the presence of a positive Pb anomaly in the primitive mantle-normalized patterns (Fig. 6a1). The pronounced negative Eu anomaly may also carry a crustal signature (Rollinson, 1993, Sect. 4.3.3.; Rollinson and Pease, 2021; Rudnik and Gao, 2014; Fig. 6b1). This is characteristic of felsic magmas generated through partial melting of subduction-influenced mantle sources, potentially accompanied by partial melting of the upper continental crust (Wilson, 1989, chapter 11.6.).

It may be inferred that the origin of the Margečan rhyolitic ignimbrites and Kjumberk Hill rhyolitic lavas, genetically, temporally, and spatially associated with the Pietra Verde tuffs (Slovenec et al., 2023; Figs. 6, 7, 9), likely involved a combination of subduction-related mantle-derived melts and crustal magmas within a tectonic setting conducive to melting–assimilation–storage–homogenization (MASH) processes.

Felsic volcanic rocks may form in a variety of geotectonic settings. Gorton and Schandl (2000) and Pearce et al. (1984) proposed Ta / Yb vs. Th / Yb and Nb vs. Ta diagrams as effective tools for discriminating among them. Trace element ratios in the investigated rhyolitic rocks are analogous to those of felsic extrusive and pyroclastic rocks formed in an Andean-type continental margin volcanic arc (Riley et al., 2001; Garrison et al., 2011), and closely resemble those observed in the Pietra Verde tuffs from the broader study area (Fig. 10a–b). This interpretation is further supported by significant enrichment in LREE, selective depletion in HFSE, and pronounced negative anomalies in the Nb–Ta pair and probably Ti, all of which are characteristic of subduction-related settings (Pearce, 1982; Arculus and Powell, 1986; Hawkesworth et al., 1993; Fig. 6a1–b1). However, this does not necessarily reflect the actual tectonomagmatic setting of formation but is likely inherited from older (ancient) arc-derived lithologies associated with Paleotethyan subduction, which underwent partial melting (Slovenec and Šegvić, 2021; Slovenec et al., 2023). In addition, the consistent presence of a strong positive Pb anomaly and negative Eu anomaly in all samples (Fig. 6a1–b1) suggests that the magma might have formed in areas of mantle upwelling, such as continental rifts, settings where the Pb anomaly tends to be pronounced, in contrast to subduction zones, where it is typically weak or absent (Bachmann and Bergantz, 2008).

Considering the above interpretations, the origin of the studied rhyolitic rocks from the Margečan locality and Kjumberk Hill can be understood within the geodynamic framework proposed by van Hinsbergen et al. (2020) and van Hinsbergen and Schouten (2021). These authors linked the opening of the Neotethys to the southwest-directed subduction of the Balkan Paleotethyan lithosphere beneath the northeastern margin of Greater Adria, followed by slab roll-back of the subducted Paleotethyan plate (Fig. 11a). This model favours passive continental rifting along the Middle Triassic northern margin of Greater Adria (Fig. 11b). Accordingly, the genesis of the investigated rocks involved: (i) partial melting of a heterogeneous subcontinental lithospheric mantle that had been metasomatized during earlier Variscan subduction events in the Late Paleozoic (Saccani et al., 2015), and (ii) to a lesser extent, processes related to crustal melting and fractional crystallization, as suggested by Slovenec et al. (2023). Effusions of rhyolitic lavas and explosive eruptions of pyroclastic (ignimbrite) flows occurred during the Middle Triassic along tectonically weakened zones of the rifted passive margin of Adria, forming basinal areas where such lithologies were alternating with the deposition of deep-marine siliciclastic and carbonate sediments. One such short-lived basin, the Northwestern Croatian Triassic Rift Basin (Kukoč et al., 2023), located along the northeastern Adriatic passive margin (Fig. 11a), served as the site of this felsic magmatic activity during the Anisian to early Ladinian.

Given their closely matching ages and the clear geochemical similarity between the investigated rhyolitic lavas and the Pietra Verde tuffs from the broader study area in northern Croatia (Strahinjčica and Ivanščica Mts., Slovenec et al., 2023; Figs. 6–7, 9–10; Žumberak–Samoborska Gora Mts., Goričan et al., 2005; Fig. 1a), as well as with those from the External Dinarides (e.g., Donje Pazarište, Smirčić et al., 2018; vicinity of the town Gračac, Sokač et al., 1974; vicinity of the town Knin, Kuljak, 2004; Zelovo–Suvaja locality, Šćavničar et al., 1984), it can be inferred that the rhyolitic lavas and ignimbrites investigated in this study represent potential source, or one of the sources, of the Pietra Verde tuffs distributed along the northeastern passive margin of the Adria Plate (Fig. 11).

Felsic rock suites from both investigated localities, as well as the Middle Triassic Pietra Verde tuffs from the adjacent Ivanščica and Strahinjčica Mountains, share the same paleogeographic provenance, i.e., the northern to northeastern Adriatic passive margin. Consequently, their lithological characteristics, along with those of spatially and temporally associated units, are broadly similar and do not allow for clear discrimination into distinct tectonic units based solely on lithology. However, the three major tectonic domains that converge in the broader study area are characterized by diachronous orogenic histories, and the pressure–temperature paths of the rocks composing these units differ significantly. Therefore, the distinct p-T evolution of each domain may serve as a key discriminant for distinguishing between rock units in the transitional zone between the Southern Alps, Dinarides and the Austroalpine units (Fig. 1b).

The maximum diagenetic temperature experienced by Middle Triassic rocks of the northern Dinarides and eastern Southern Alps in Slovenia reached up to 300 °C, interpreted as the result of sedimentary burial, with peak thermal conditions attained during the Middle Eocene to Early Oligocene (Rainer et al., 2016). Assuming an average geothermal gradient of 25–30 °C km⁻¹, this corresponds to lithostatic pressures of approximately 0.3 GPa. Similar temperature conditions were recorded on Mt. Ivanščica, located 5 km to the south of the Margečan locality, where Middle Triassic rocks also reached ∼300 °C (Kukoč et al., 2023). However, in this case, the observed increase in thermal maturity was attributed to Early Cretaceous overthrusting and orogenic loading (Vukovski et al., 2024). Further westward, in the central Southern Alps, peak temperatures in Middle Triassic strata did not exceed 150 °C, corresponding to a maximum lithostatic pressure of ∼0.15 GPa, related to Late Cretaceous to Paleogene sedimentary burial rather than tectonic loading (Grobe et al., 2015). In this context, the studied fresh, unmetamorphosed rhyolitic lavas from Kjumberk Hill are fully consistent with their current placement within the Southern Alps tectonic unit (Schmid et al., 2020). In contrast, the presence of phengitic muscovite in the rhyolitic ignimbrite from Margečan area (Appendices B and C) points to a significant overprint by HP–LT metamorphism, at conditions of 1.1–1.2 GPa and ∼300 °C. These conditions sharply contradict the maximum p-T parameters recorded in the eastern Southern Alps and clearly exceed the pressure range required for the growth of phengitic muscovite under greenschist or burial metamorphic regimes. Instead, the observed assemblage corresponds to blueschist facies conditions, typical of subduction zones, and cannot be produced by any of the thermal regimes previously described for the Southern Alps. Although the timing of this metamorphic overprint remains unconstrained, the only regional geodynamic event postdating the Middle Triassic that could have generated blueschist facies metamorphism is the subduction associated with the closure of the Neotethys. Although the Dinarides dominantly preserve moderate- to high-pressure metamorphic assemblages related to high-temperature subduction-obduction processes (Šegvić et al., 2019, 2020; Putiš et al., 2025), rare occurrences of HP–LT blueschist facies rocks have been reported from the distal Adriatic passive margin units, such as the Mt. Medvednica (Belak and Tibljaš, 1998; Belak et al., 2026) or the oceanic Jaklovce Unit (Putiš et al., 2019 and references therein). Nevertheless, the Margečan ignimbrites cannot be correlated with these two occurrences because metamorphic rocks of Mt. Medvednica initially occupied a much more distal position on the Adriatic passive margin during the pre-subduction stage compared to Margečan ignimbrites. This difference is even more pronounced in the case of the Jaklovce Unit, which is of oceanic rather than continental provenance. Alternatively, the Austroalpine units of the Eastern Alps underwent intracontinental subduction and nappe stacking during the Cretaceous, culminating in high- to ultrahigh-pressure metamorphism around 95 Ma (Hoinkes, 1981; Faryad and Hoinkes, 2003; Thöni, 2006; Schmid et al., 2008; Janák et al., 2015). Considering the present-day regional tectonic position of the Margečan locality, which was so far considered as the northernmost Southern Alpine domain close to the border with Austroalpine units, it is the most plausible explanation that the rhyolitic ignimbrites are part of the northern Adriatic passive margin involved in the Cretaceous Eoalpine subduction system and consequently should be regarded as part of the Eoalpine high-pressure belt (Koralpe-Wölz) of the Austroalpine units (sensu Schmid et al., 2008) of the Eastern Alps. In this context, we propose relocating the southern boundary of the Austroalpine units farther south, up to the Labot and Šoštanj faults (Fig. 12). Under this interpretation, the non-metamorphic Permo–Mesozoic succession exposed in Boč and Ravna Gora mountains represent the sedimentary cover of the Austroalpine rather than the Southern Alpine units, correlative with Mesozoic successions exposed in the North Karawanken Mts. (Fig. 12). Southward extension of the Austroalpine units also provides a simpler explanation for the borehole samples found about a hundred kilometres eastward along MHZ in the Igal unit where sericite K-Ar dating of slate and metarhyolite tuff yielded typical Eoalpine ages (Árkai et al., 1991).

During the Miocene, these segments of the Eoalpine high-pressure belt were laterally extruded eastward into the Carpathian embayment along major dextral strike-slip faults, most notably the Labot, Donat and Šoštanj faults, eventually reaching their current position in what is today northern Croatia. This is further supported by the recent discovery that the Oligocene to Lower Miocene sedimentary succession of the Hrvatsko Zagorje Basin (sensu Avanić, 2012; Avanić et al., 2021), exposed north of the Šoštanj Fault and representing sedimentary cover of Margečan ignimbrites, differs drastically from the succession exposed south of the same structure, and that it represents allochthonous units displaced to the present position from the west, where correlative units are exposed in Slovenia (Vukovski, 2024; Vukovski et al., 2024). Considering available kinematic constraints and estimated offsets along these strike-slip faults (Tari, 1994; Fodor et al., 1998; Sachsenhofer et al., 2001; Vukovski et al., 2024), a first-order reconstruction suggests that the pre-extrusion (i.e., pre-Neogene) position of the Margečan ignimbrite might have been located as much as 70 km to the W–NW along the fault (Fig. 12). This location is near the present-day position of the southern slopes of Pohorje Massif, which represents the closest outcrops of the Eoalpine high-pressure belt rocks. This new finding indicates that the Austroalpine units were extruded much farther to the southeast than previously thought (Fig. 12), which has a direct impact on the amount of crustal stretching in the southernmost border of the extruding wedge, where during lateral extrusion, extension was predominantly accommodated by high-offset strike-slip faults.