Representative massive chromitite samples were collected from the Gomati and Nea Roda ultramafic bodies. More precisely, Al-rich massive chromitites from the Gomati (G1 chromitites) were collected from the mining sites of Paivouni and Tripes; Cr-rich massive chromitites from the Gomati (G2 chromitites) were collected from the mining sites Agios Georgios and Kroupnos; Massive and banded Cr-rich chromitites from the Nea Roda (NR chromitites) were collected from the sites NR1 and NR2, respectively. Serpentinites from the Gomati region were collected adjacent to the mining sites, whereas diopsidites were found penetrating the G1 chromitites from the Paivouni mining site. During fieldwork, criteria such as alteration degree, mode of occurrence, relationships with other lithologies and stratigraphical distribution were applied. More than fifty samples were collected around the area of seven abandoned mine sites.

Electron microprobe analyses (EPMA) and electron back scattered images (B.S.I.) were carried out at the Eugen F. Stumpfl Laboratory of the Leoben University, Austria, using a Superpobe Jeol JXA 8200 (JEOL, Tokyo, Japan) instrument. Analyses of spinel and silicates were obtained in WDS mode, with an accelerating voltage of 15 kV and a beam current of 10 nA. The Kα lines were used in the analysis of Na, Mg, K, Al, Si, Ca, Ti, V, Cr, Zn, Mn, Fe and Ni, after instrument calibration on natural chromite, rhodonite, ilmenite, albite, pentlandite, wollastonite, kaersutite, sphalerite and metallic vanadium. The used diffracting crystals were: TAP for Na, Mg, Al; PETJ for K, Si, Ca; and LIFH for Ti, V, Cr, Zn, Mn, Fe, Ni. The counting times 20 and 10 s were used for peak and background, respectively, in the analysis of all major elements and were increased up to 60 and 30 s for the trace elements analyses. The following detection limits (in ppm) were automatically calculated by the microprobe software: Ca (50), Mg, K, Mn, Fe and Ni (100), Na, Al, V and Zn (150), Cr (200), Si and Ti (250). The amount of Fe³⁺ of chromite was calculated assuming the spinel stoichiometry R²⁺O R³⁺ ₂O₃.

The PGM and other accessory minerals were located in situ under reflect light microscope on polished chromitite sections, at 250–800 magnification. Grains smaller than 5 microns were only qualitatively analyzed by EDS. The quantitative composition of bigger grains was obtained in WDS mode, at a 20-kV accelerating voltage and 10-nA beam current, with a beam diameter of 1 μm. The counting times of peak and background were 15 and 5 s respectively. The Kα lines were used for S, As, Fe and Ni, Lα for Ir, Ru, Rh, Pd and Pt while Mα were used for Os. Synthetic pure metals, NiS, NiP and natural pyrite and niccolite were used as reference materials for the six PGE (Ru, Rh, Pd, Os, Ir, Pt), Ni, Fe, S, P and As respectively. The following diffracting crystals were selected: PETJ for S and P; PETH for Ru, Os and Rh; LIFH for Fe, Ni, Ir, Pt; and TAP for As. For interferences involving Ru-Rh and Rh-Pd the corrections were automatically performed. The detection limits are (in ppm): S, As and P (100), Fe, Ni and Cu (150), Ru, Rh and Pd (200), Os, Ir and Pt (800). Platinum concentrations were systematically below detection limit and have been removed from the analytical results.

Trace elements and REE contents in clinopyroxene and trace elements in spinel-group minerals were determined on polished thin sections by LA-ICP-MS (laser ablation inductively coupled plasma mass spectrometer) system at the NAWI Graz Central Lab for Water, Minerals and Rocks (University of Graz and Graz University of Technology), with an ESI New Wave 193 Excimer Laser (193 nm wavelength) coupled to a quadrupole Agilent 7500 CX mass spectrometer. A beam size of 50 µm, with a fluence of ∼5 J/cm2, helium flow of 0.8 L/min, 25 s gas blank followed by 50 s of ablation and a dwell time of 20 s for each mass were used for the element analyses. The reference material NIST SRM 612 was used for standardization and Si and Al as internal calibration elements for diopside and spinel-group minerals, respectively. The USGS reference glass BCR-2G was analysed as monitor standard which could be reproduced within errors. For data reduction, the software “GLITTER” was used and the values for NIST SRM 612 were taken from Jochum et al. (2012).

Geochemical analyses of four serpentinite samples were performed at Activation Laboratories Ltd. Actlabs according to the packages of 4Lithores. This is a combination of packages 4B (lithium metaborate/tetraborate fusion ICP whole rock) and 4B2 (trace element ICP-MS). Geochemical analyses of PGE concentrations were performed at Activation Laboratories Ltd. Actlabs. Eleven chromitite samples were analyzed; three from G1 (GN1 - GN3), four from G2 (GS1 - GS4) and four from NR (NR1 - NR4) from seven mining sites (Figure 1). Aliquots of 25 g of sample were treated by the NiS fire assay procedure. The NiS beads were dissolved in concentrated HCl, and the PGE + Au residue collected on a filter paper. The residue underwent two irradiations and three separate counting steps for the PGE and Au. The irradiation is induced with neutrons, and the resulted gamma radiation was measured by Instrumental Neutron Activation Analysis (INAA). The detection limits (ppb) were: Os (2), Ir (0.1), Ru (5), Rh (0.2), Pt (5), Pd (2) and Au (0.5).

A Ni fragment enriched in P was extracted and investigated by X-ray diffraction at the university of Florence, Italy, using the procedure described by Zaccarini et al. (2019). Massive chromitites were processed in SGS Mineral Services, Canada, for recovery of heavy minerals, following the procedure described by Sideridis et al. (2018).

PGE concentrations from eleven massive chromitites were examined and the results are presented in Supplementary Table S9. The G1 chromitites display the lowest ΣPGE (45.4–135.6 ppb, avg.= 83.5 ppb), characterized by elevated palladium-PGE (PPGE) concentrations (Avg. PPGE/IPGE = 0.93) and Pd/Ir ratios (1.14–5.53;). Their normalized PGE patterns demonstrate positive Ru anomalies and depleted Ru-Pt profiles, with a slight enrichment in Pd (Figures 10A, B). The G2 chromitites are more enriched in ΣPGE (64.3–264.8 ppb, avg. = 135.7 ppb), described by lower PPGE concentrations (Avg. PPGE/IPGE = 0.31) and low Pd/Ir ratios (0.14–0.22) with the exception of a single sample with Pd/Ir = 7.69. Their normalized PGE patterns are closely comparable to those of G1, albeit exhibiting more prominent positive Ru anomalies and pronounce negative Ru - Pt slopes, with a slight increase in Pd. The normalized PGE patterns of NR chromitites greatly resemble the G2 counterparts. Their ΣPGE (132.0–199.0 ppb, avg.= 169.3 ppb) are slightly higher in concentration than in the G2 samples, displaying enrichment in IPGE concentrations (Avg. PPGE/IPGE = 0.17 and Pd/Ir = 0.20–0.41).

The serpentinites display MgO/SiO₂ and Al₂O₃/SiO₂ ratios ranging between 0.83–0.88 and 0.01–0.06, respectively, exhibiting enrichments in arsenic and antimony when compared to PM values, 16–19 and 2.1–2.8 ppm, respectively (Figure 11). Fluid-mobile-elements (FME) such as Cs, U and Ba display values of <0.5, 0.03–0.94 and 5–12 ppm, respectively. LREE demonstrate enrichment relative to the flat MREE and HREE patterns, whereas Nb-Ta, Zr and Ti are depleted if compared to the PM (0.5*PM). These rocks are rich in Ni (1,620–2,540 ppm) and Cr (2,640–10,000 ppm) consistent with their refractory mantle origin. The whole-rock geochemical analyses are presented in the Supplementary Table S10.

Major and trace elements in spinel-group minerals within chromitites are commonly used to estimate their parental melt composition and classify them according to their normalized patterns, respectively. Both approaches have been implemented upon metamorphosed chromitites (González-Jiménez et al., 2015; Colás et al., 2019). This is achieved by careful examination of chemical changes in core and rims of the analyzed grains. In the present study, individual grains from G1, G2 and NR chromitites have been analyzed via EPMA and LA-ICP-MS to test this theory.

Concerning the G1 chromitite bodies, two cases were encountered and examined: 1) chromitites comprised of zoned spinel grains with porous rims demonstrating significant trace element variations from core to rim with increase of Zn-Co-Mn, in short ZCM, (on average, from core 1,636 to rim 2,633 ppm) and Ti coupled with a decrease in Sc (Figure 5A) and 2) chromitites comprised of mosaic-like texture spinel grains with homogenous normalized patterns characterized by positive Ti and negative Sc anomalies (Figure 5B). Based on these, it is considered that increase in ZCM + Ti and decrease in Sc (coupled with decrease in Al₂O₃ and increase in Fe³⁺# = Fe³⁺/ΣFe) are traits related to alteration (González-Jiménez et al., 2015; Colás et al., 2019). Hence the pristine patterns of G1 ought to be flat (MORB-level) with negative Sc and weak negative Ni anomalies. Such patterns are typical of high-Al (Zhou et al., 2014) and more precisely, of back-arc Al-chromitites of Coto, Tehuitzingo, Moa-Baracoa, Sagua de Tanamo and Dobromitsi (Yao, 1999; González-Jiménez et al., 2011, González-Jiménez et al., 2015; Colás et al., 2014, Colás et al., 2019 and references therein; Figure 5E). Concerning the mosaic textured case, most likely, represents a product of complete re-equilibration as its annealing texture suggests. The pristine G1 spinel normalized patterns are also comparable with fore-arc chromitites (Figure 5F) such as: a) the case of Al-rich, subduction initiation Moa-Baracoa chromitites reported by Rui et al. (2022) though in this occurrence prominent negative Ti anomalies are noted and b) Ouen Island (González-Jiménez et al., 2011).

Regarding the G2 grains, two cases were encountered and examined: a) grains with low degree of alteration and chlorite intergrowths as confined altered zones, with an increase in ZCM being noted in the latter (on average, from core 2,500 to rim 4,175 ppm; Figure 5C), and b) grains with low-alteration degrees and little chemical variation in trace elements and a small increase around the crystal rims. Since the core analyses have low ZCM concentrations and they are free of chlorite inclusions, they represent pristine composition. The similarities of the normalized trace element patterns with other pristine chromite grains from Cr-rich fore-arc chromitites of Antalya and Thetford mines (Pagé and Barnes, 2009; Zhou et al., 2014; González-Jiménez et al., 2015; Akbulut et al., 2016; Figure 5F), are also in favor of the core analyses representing pristine compositions. The normalized patterns are characterized by mildly negative V-Ga patterns, positive Ti and negative Ni anomalies and almost flat Zn-Mn patterns. However, similar patterns are also consistent with back-arc Al- and Cr-chromitites of Dobromirtsi and Al-chromitites of Moa-Baracoa (Colás et al., 2014).

The seemingly unaltered NR grains present no trace element chemical variation from core to rim (Figure 5D). Since these chromitites were most-likely re-equilibrated (Bussolesi et al., 2022a), it can be assumed that their patterns depict this effect and therefore not primary compositions. Despite the above, it is generally considered that chromitites with high spinel/silicate ratios are only slightly affected by post-magmatic processes (e.g. González-Jiménez et al., 2015; Akbulut et al., 2022), this is why although NR are re-equilibrated their chromite major composition must have not been significantly altered. The normalized patterns of the studied grains demonstrate characteristics that are intermediate between the G1 and G2, with weak negative Sc, positive Ti and pronounced negative Ni anomaly. These patterns are also reminiscent of Cr-rich chromitites and are almost identical to those of Bulgarian Al-rich chromitites reported by Colás et al. (2014).

In all studied cases, the non-primary spinel-group analyses are enriched in ZCM + Ti and this has been previously reported for metamorphic spinels (González-Jiménez et al., 2015 and references therein). PGM enclosed in metamorphic chromite of this study are also affected; laurite enclosed in chromite of the NR chromitite appears compositionally modified whereas laurite in pristine chromite of the G2 chromitite represents primary compositions (see section 5.4). Hence, whereas Ti contents in NR are not considered primary, Ti in unaltered cores of Gomati is taken to depict primary compositions and this also applies to other metamorphosed chromitites (Colás et al., 2019). Non-stoichiometric spinel has been previously reported by researchers (Rollinson and Adetunji, 2013; Lenaz et al., 2014). Despite the chemical changes reported in this study and based on the methodology used, the spinel-group minerals are stoichiometric, and this is the case for other metamorphosed occurrences (Lenaz et al., 2018).

Many studies argue for the connection of chromite composition, within massive chromitites, and the parental melts thereof. Equations to calculate the Al₂O₃ and TiO₂ of the parental melts were proposed by Rollinson (2008) based on previous work done by Kamenetsky et al. (2001) (Eqs. 1–4).

Equations for high-Cr chromitite: A l 2 O 3 m e l t = 5.2182 × ln A l 2 O 3 c h r o m i t e − 1.0505 (1) T i O 2 m e l t = 1.0963 × T i O 2 c h r o m i t e 0.7863 (2)

Equations for high-Al chromitite: A l 2 O 3 m e l t = 7.1518 × A l 2 O 3 c h r o m i t e 0.2387 (3) T i O 2 m e l t = 1.5907 × T i O 2 c h r o m i t e 0.6322 (4)

Maurel (1984) on the other hand, proposed Eq. 5 in order to determine FeO/MgO_melt ratios, where Al#_chromite = Al/(Al+Cr +Fe³⁺) and Fe³⁺#_chromite=Fe³⁺/(Al+Cr+Fe³⁺). ln F e O / M g O c h r o m i t e = 0.47 − 1.07 A 1 # c h r o m i t e + 0.64 F e 3 + # c h r o m i t e + ln F e O / M g O m e l t (5)

The most primitive compositions from the cores of spinel-group minerals of massive chromitites were used for the calculations. The parental melts of G1 are richer in Al₂O₃ (14.34–14.51 wt%), followed by NR (13.06–13.35 wt%) and lastly G2 (10.49–10.94 wt%) (Figure 12A). The concentration of TiO_2melt is comparably low in the cases of G2 with values lower than 0.26 wt%. The G1 records a slight enrichment in TiO_2melt (0.61–0.81 wt%) (Figure 12A), the same concentrations, although calculated for the NR, should be taken with caution, since Ti was re-mobilized (see section 5.1). The FeO/MgO_melt ratios for the G1, NR and G2 chromitites range between 0.73–0.76, 0.77–0.79 and 0.79–0.83 wt% respectively (Figure 12B). These ratios are generally consistent for all chromitite types. The composition of the G1 parental melts is akin to a MORB source (Melcher et al., 1997; Rollinson, 2008; Zaccarini et al., 2011; Chen et al., 2019), while the G2 and NR parental melts demonstrate boninitic affinities (Falloon et al., 2008; Koutsovitis and Magganas, 2016; Chen et al., 2019; Huang et al., 2021).

G1 chromitites consist of spinels resembling those in equilibrium with MORB lavas (Figures 4A, B). On the contrary, the G2 and NR chromite analyses compare well with spinel hosted within boninitic lavas (Figures 4A, B). Coexistence of Al- and Cr-rich chromitites in such proximity and without any fault bringing together two separate sections has been interpreted in the following ways: 1) evolution of a single initially Cr-rich batch magma of boninitic affinity into Al-enriched magma residue (Zaccarini et al., 2011; Uysal et al., 2016), 2) subduction initiation starting with formation of Al-rich chromitites after MORB-like melts produced in fore-arc spreading followed by Cr-rich varieties with progressive arc maturation and production of transitional and boninitic melts (Rollinson and Adetunji, 2013; Xiong et al., 2017; Rui et al., 2022) and 3) generation of Al-rich chromitites in either MOR or back-arc settings and subsequent transition into a SSZ (Rollinson, 2008; Uysal et al., 2009; González-Jiménez et al., 2011). Generation of the bimodal chromitites of the RM in a back-arc has been proposed (González-Jiménez et al., 2015).

Four empirical chlorite geothermometers were applied in the studied chromitites (Cathelineau and Nieva, 1985; Kranidiotis and MacLean, 1987; Cathelineau, 1988; Zang and Fyfe, 1995). The temperature dependance in chlorite is controlled by the Al substitution over Si in the tetrahedral site (Cathelineau and Nieva, 1985). The temperatures (°C) calculated for the chlorites in the G1, G2 and NR chromitites are presented in Table 1.

These temperature results are generally consistent between the three domains. Consequently, the studied ultramafic bodies must have undergone similar post-magmatic effects that point to a lower-greenschist facies overprint; similar conditions have been reported from metamorphic rocks within the SMM and the RM (Kydonakis et al., 2015). Moreover, chlorite in contact with PGM grains or included in spinel grain displays insignificant temperature differences compared to chlorite hosted within the chromitite interstices. Grossular–andradite–uvarovite garnet is a mineral phase that can form under a variety of conditions, ranging from low-grade metasomatic processes to local regional metamorphism. Temperature estimations of the two garnet groups yielded values ≤400°C for the G1 and <500°C for the G2 counterparts, which typically falls under the temperature range of greenschist facies.

Colás et al. (2019) provided insights regarding the thermal control of the ZCM enrichments in metamorphosed porous chromite. Based upon temperature modelling, the same authors concluded that qualitative assessment of temperature can be applied in metamorphosed spinel grains. In the present study the preserved cores record on average 1,636 ppm of ZCM, whereas porous rims concentrate 2,633 ppm. These contents are comparable with those in the case of East Rhodope (2,772 ppm; Gervilla et al., 2012; Colás et al., 2014), with temperatures ranging between 450–700°C, pointing to amphibolite facies. These metamorphic conditions were also reported for the case 1) of re-equilibrated Nea Roda chromitites (Bussolesi et al., 2022a), and b) the Askos meta-ultramafics (adjacent to the Volvi occurrence) at 0.4 GPa and 450–550°C (Michailidis, 1991). Therefore, the SMM-hosted occurrences of Gomati and Nea Roda have undergone amphibolite facies metamorphism coinciding with the RM meta-ophiolites.

In ophiolites, arc-related melts produced in SSZ settings require high degrees of partial melting of a mantle source to develop chromitites enriched in PGE, especially for refractory IPGE, up to a few thousand ppb (Economou-Eliopoulos, 1996; González-Jiménez et al., 2011; Zaccarini et al., 2011). Lower degrees of partial melting account for only a small portion of melt, leading to PGE-poor chromitites yet relatively enriched in the less refractory PPGE (Economou-Eliopoulos, 1996; González-Jiménez et al., 2011 and references therein). PPGE/IPGE ratios are consistently low for the G2 and NR Cr-rich chromitites with an average of 0.24 unlike the G1 chromitite with average ratios of 0.93. Previous studies refer to ratios of 0.07 and 0.23 for Cr- and Al-rich chromitites of Gomati, respectively (Economou-Eliopoulos, 1996; Bussolesi et al., 2022b). Yet, the similar PGE patterns of the compositionally different chromitites studied may be interpreted as products of continuous melt production from a common source (Rollinson, 2005). This would have led to gradual depletion of the initially fertile mantle and its modification into a harzburgite residue that became enriched in PGE-bearing sulfides, dominated by IPGE, following the initial removal of PPGE (Büchl et al., 2002). Indeed, the studied chromitites display a gradual enrichment in IPGE and ΣPGE from the G1 towards NR and G2 occurrences. The average ΣPGE contents demonstrate a decrease from G2-245.0 to NR-154.6 and to G1-98.8 (collective data: this study; Economou-Eliopoulos, 1996; Bussolesi et al., 2022b). This trend has been previously described by Uysal et al. (2016) referring to chromitites produced by a successively fractionating melt with deeper-crystallized Cr-rich chromitites being more enriched when compared to the more shallow “intermediate” counterparts.

The higher Pd/Ir ratios (>1) of the G1 and one sample of G2 chromitites can be interpreted via two scenarios: 1) enrichment of PPGE due to metasomatic processes or 2) due to fractional crystallization (Garuti et al., 1997 and references therein). The first scenario, although plausible in principle, is supported by only one piece of evidence, which is the presence of a single secondary Pd-Sb PGM within the G1 chromitite. The analyzed primary PGM from the Gomati chromitites (subhedral laurite included in chromite) contain no appreciable PPGE and this agrees with findings by Bussolesi et al. (2022b). Irarsite was encountered in all chromitites. In the NR chromitites, primary laurite appears modified (Tarkian and Prichard, 1987; Zaccarini et al., 2005) with high arsenic contents and oscillatory zoning (euhedral to subhedral included in chromite) along with Pt-As-bearing secondary platarsite and sperrylite (crystallized in chlorite matrix) which do not greatly affect the low PPGE/IPGE ratios of NR chromitites. Therefore, we attribute the Pd/Ir enrichments of G1 and G2 to parental melt fractionation and not to alteration processes. Based on asthenospheric mantle-normalized PGE concentrations, low Pt/Pt* = Pt_N/(Rh_N × Pd_N)^1/2 ratios (<1) (Figure 10B) are consistent with increasing degrees of partial melting of the mantle source, which seems to be the main mechanism that control the PGE distribution since the elevated Pd/Ir ratios were only encountered in four out of twenty nine samples from chromitites of SMM and RM. The generally low Pt/Pt* ratios point to a residual mantle source (Garuti et al., 1997).

Taking into consideration the secondary and modified primary PGM, as well as other base metals that were encountered (Ni-Sb) as also reported by Bussolesi et al. (2020, 2022b), it is evident that there is an enrichment in As + Sb during a post-magmatic event. Bussolesi et al. (2020, 2022b) assigned this metasomatism to the 53–55 Ma Ierissos pluton intrusion or porphyry copper related mineralization. However, no alteration effects related to the metallogenesis of Kassandra district have been reported for the Ierissos pluton as well as along the 12 km Gomati Fault (Siron et al., 2018 and references therein). The Kassandra metallogenesis is linked to two temporally distinct intrusions ranging between 25–27 and 19–20 Ma, situated more than 5 km NW from Gomati ultramafic bodies (Siron et al., 2018 and references therein). Therefore, the aforementioned enrichments should be attributed to a more widespread event. The best fitting scenario is the effect of pervasive serpentinization, evident in all the ultramafic lithologies. Geochemical analysis of the serpentinites with respect to the Gomati Ni-Sb-As-bearing mineralization, demonstrate enrichments in those elements, typically originating from contamination of sedimentary origins (Deschamps et al., 2011; Koutsovitis, 2017). These characteristics, combined with the elevated contents of Cs, U and Ba point to serpentinites related to a subduction zone (Deschamps et al., 2013). According to Deschamps et al. (2013 and references therein) the protolith of such serpentinites may be: 1) subduction-related oceanic peridotites or 2) OCT-related continental peridotites that were exhumed and altered during rifting. The LREE enrichments in serpentinites are considered to have derived from a process of fluid circulation within SSZ environment (Deschamps et al., 2013; Xie et al., 2021), although arguably an already metasomatized protolith could also display the same enrichments (Deschamps et al., 2013).

LREE enrichments are recorded in the diopside hosted within G1-chromitites and diopsidites. In general, primary diopside related to such rocks is expected to demonstrate LREE depletion. G1 diopsidite veins are metasomatic products based on titanite chemistry (Supplementary Figure S3; metamorphic and not magmatic affinity: Ling et al., 2015; Kiss and Zaccarini, 2020), and diopside chemistry is similar to the Omani metasomatic diopsidites (major elements: Figure 6A and HREE Figure 6C; Python et al., 2011; Arai et al., 2020). Diopside within G1 chromitite is comparable with diopside hosted within the Al-rich subduction initiation-related chromitite of Moa-Baracoa (major elements: Figure 6A and HREE Figure 6C). However, Rui et al. (2022) argue that these diopsides are not syngenetic with the chromitites. Moreover, Zhang et al. (2016) attribute LREE enrichments in clinopyroxene hosted in subduction-initiation related dunite and Cr-chromitites (demonstrating concave spoon-shaped patterns) to their post-magmatic equilibration with metasomatic fluids derive from the subducted slab. Therefore, the diopside hosted in G1 chromitites is also a metasomatic affected phase.

LREE, for both diopside types as reported in this study, are generally enriched and their convex upward patterns are often interpreted as due to interaction with alkaline melts (Matusiak-Małek et al., 2021); yet, no alkali basalts occurrences were located in the field nor K-bearing minerals have crystallized. However, similar patterns could be due to fluid metasomatism; negative Nb, P and Ti anomalies are noted in SSZ-related rocks, which are usually ascribed to reaction with hydrous fluids within the mantle (Walker and Cameron, 1983; Tatsumi and Eggins, 1995; Horodyskyj et al., 2009; Guice et al., 2018). In such cases, H₂O and CO₂-rich hydrothermal fluids can favor LREE mobilization relatively to HFSE within subduction zone settings (Walker and Cameron, 1983; Guice et al., 2018). Amphibolitization (Guice et al., 2018) and serpentine-eclogite interaction (Horodyskyj et al., 2009) within serpentinite channels, can control such remobilization. Another important aspect of the studied diopside is the positive Pb anomalies, which are a feature of serpentinites and subducted oceanic crusts (Horodyskyj et al., 2009; Deschamps et al., 2011; Cruciani et al., 2017). Subduction-related fluid metasomatism likely influenced significantly the diopsidites since their diopside grains accumulate higher concentrations of FME and LILE. The low Zr/Hf ratios of the clinopyroxene hosted within the titanite bearing diopsidite is likely attributed to the preference of Zr entering the titanite rather than clinopyroxene (Bea et al., 2006). From the above it is evident that post-magmatic subduction related fluids have altered the primary trace element composition of the studied mineral phases.

The TVG complex, along with other meta-ophiolitic occurrences within RM, has a rather complex geological history. The podiform chromitites display bimodal character with coexistence of Cr- and Al-rich chromitites within the same section. To our knowledge, no tectonic structure exists that could juxtapose those different sections. Hence, the proposed geotectonic model should account for the genesis of those chromitites within the same geotectonic context. Subduction initiation is considered by many researchers as a key process in the evolution of plate tectonics that can also adequately explain the genesis of Al-chromitites with mixed MORB and arc characteristics. However, the mineral products (spinel, diopside) of this environment appear typically Ti-poor (Zhang et al., 2016; Rui et al., 2022) when compared to the mineral components of the G1 chromitite. Therefore, it is suggested that the G1 chromitites were formed from a MORB-like melt produced in extensional settings, as has also been proposed for the RM-hosted meta-chromitites of Dobromirtsi (González-Jiménez et al., 2015). The extensional phase is followed by oceanic closure, whereas the pre-existing mantle section is introduced into SSZ settings (Uysal et al., 2009), translating into higher degrees of partial melting and subsequent depletion of the mantle source. The ensuing chromitites are thus Cr-rich, represented by the G2 and NR chromitite end members. In this light, the SSZ NR chromitites could represent a product of fractionation of the previously Cr-rich G2 since they are Al-richer (Figure 4B) (Zaccarini et al., 2011; Uysal et al., 2016).

Along with the gradual maturation of the subduction zone, fluids liberated from the subduction channel brought about serpentinization and interaction with diopside resulting into LREE enrichments, accompanied by depletion in Nb-Ta, Zr, Ti and increase in Pb. The peak of metamorphism reached amphibolite facies while porous chromite was formed (450°C–700°C). During the regional extensional deformation, the chromitites experienced a greenschist facies overprint (Siron et al., 2018 and references therein) and garnet was crystallized within shear and brecciated zones at temperatures <400 C. Solutions enriched in Fe³⁺ interacted with the porous chromite creating Fe³⁺ enriched chromite and magnetite. Chlorite was re-equilibrated at temperatures <300 C. Secondary PPGE-rich PGM and As-Sb bearing minerals were observed within the chlorite matrix, representing a low-T stage (greenschist facies) whereby PPGE mobile (Tsikouras et al., 2006; Farré-de-Pablo et al., 2021) and As-Sb were incorporated during serpentinization due to input from a sedimentary source (Deschamps et al., 2011; Koutsovitis, 2017).