For nomenclatural orientation, ‘Vardar Ocean’ is the name of the western ocean domain of the northern Neotethys (Fig. 2b). We agree with Schmid et al. [3] that ‘Vardar zone’ (Fig. 2b) is not synonymous with ‘Vardar Ocean’. In our opinion, the Vardar zone is not the ‘root’ of Vardar-derived thrust sheets, as has been often suggested [7–9]. Quite the contrary, as will be shown, the ‘Vardar zone’ is where the last slab of the Vardar Ocean was subducted [10,11] and probably corresponds to the ‘Sava suture zone’ [3,12].

The names of geo-tectonic sub-divisions of the Vardar zone used herein are after Kockel [13].

The ‘Vardar zone’ corresponds to the northwest–southeast striking belt (Fig. 1) where remnants of island arc volcanic formations are found [6,14–18,23] and where easternmost Pelagonia is covered by Upper Cretaceous carbonates [3].

We consider it important to use the term ‘ophiolite’, in the strict sense of the ‘Steinmann Trinity’ [19], because there are oceanic formations in the study areas that are composed of basalt +- radiolarite but are devoid of serpentinite and were derived from tectonic environments unrelated to obduction, which will be shown.

Furthermore, the term ‘mélange’, used herein, follows Hsü [20] referring to tectonically produced polymictic fault-zone rocks as opposed to polymictic sedimentary deposits (see also Scherreiks [21]). The mélanges are associated with mylonitic S–C shear fabrics of subduction zones [22] like those found in the Vardar zone [23].

The Vardar Ocean existed during the Middle to Late Triassic, substantiated by radiolarians and pillow basalt found in ophiolite occurrences in our study area in Evvoia [33,36–40] (Table 1a 11.1). Initially, the late Triassic and early Jurassic carbonate platform evolved from a cyclically alternating supratidal to peritidal domain [21,32] and then began sinking, presumably responding with subsidence as Adria converged with the Vardar Oceanic plate [33]. The postulated beginning of intra-oceanic obduction was around the Toarcian to Bajocian time (180–170 Ma), based on the ages of amphibolites found in the ‘metamorphic sole’ of subduction-zone mélanges [41–43]. The platform subsided during the Middle Jurassic, verified by ever deepening carbonate facies [21], and then became emergent during the Callovian time, verified by bauxite deposits (Fig. 4a) [35]. The age of this Callovian upheaval has been verified with Bathonian foraminifera in the limestones below, and Oxfordian foraminifera above the bauxite crusts (Table 1a 5 and 6) [55]. The ‘Callovian event’ has been attributed to plate tectonic stress that affected the entire Mediterranean region [44]. An Oxfordian transgression re-established shallow marine environments which generated a Tethys-wide reef facies that extended from the Alps to Asia and in the Hellenides is characterised by the demosponge, Cladocoropsis mirabilis Felix [21,29] (Table 1a 7 and 8). Rapid platform subsidence and drowning below the carbonate compensation depth (CCD) occurred during the Tithonian–Berriasian time, verified by radiolarian cherts [45]. The final ophiolite emplacement is estimated to have occurred in the Valanginian time, in Evvoia, after flysch-like sedimentation had been shut off by the obduction [21,33,34]. The obduction was followed by a period of ophiolite erosion (post-Eohellenic unconformity: Scherreiks [21]) and a subsequent gradual, widespread, transgression of marine conglomerate in Evvoia and across the Pelagonian zone during the Early Cretaceous time [21,46,47] (Table 1a 9).

Examples of obducted ophiolite s. str. occur in the study areas of northern Evvoia (Fig. 4a) [21,34] and are found throughout the Korab–Pelagonian zone (Fig. 1). They lie, tectonically emplaced, together with mélange on top of Upper Jurassic and Lower Cretaceous carbonate platform rocks [52,53]. The ophiolites are erosional remnants that have been postulated to be parts of a single obducted ophiolite sheet that was emplaced during the Late Jurassic to Early Cretaceous, an age which classifies it as ‘Eohellenic’ after Jacobshagen et al. [52]. The onetime ophiolite sheet is considered to have had a width of at least 200 km – when judged from the width of the ophiolite outcrops on geological maps [3,39] (Fig. 1).

The Northern Sporades are devoid of serpentinite. The ophiolite sheet, known to have been obducted over Pelagonia, had been eroded from large areas of Pelagonia during the later Lower Cretaceous time (see above). On the Sporades, erosion was extreme; the ophiolite and large parts of the carbonate platform are missing (Fig. 5). The eroded surface of Jurassic and Triassic platform carbonates is covered by a sheet of mélange composed of meta-basalt and radiolarian chert which is chaotically mixed with carbonate breccia and mylonitic phyllonites [10] (Figs 4a and 5). Slices of Cretaceous and Paleocene platform carbonates of reefal origins are tectonically incorporated in the melange (Table 1a 10–10.3). The Cretaceous carbonate platform successions of Alonnisos and Skopelos overlie the mélange. In corroboration with Kelepertsis [54] we suggest that the Cretaceous and Paleocene carbonates of the northern Sporades are of Vardar zone origin, which will be expanded upon in the Discussion and conclusions. The Cretaceous carbonate platform and its mélange substrate, we suggest, correlate with a similar succession in the Almopias sub-zone (Fig. 4a and b).

Sheared Eohellenic ophiolite occurs on top of Pelagonian carbonates in contact with disrupted Cretaceous limestones (Table 1b 1 and 2), along the western border of the Vardar zone, for example, near Panagitsa and Arnissa (Fig. 6) [55] and southwards near Pyrgi–Kato Grammatiko and west of the Vermio mountains [56] (Fig. 6). West verging imbricated thrust faults characterise this western boundary of the Vardar zone, from the Dinarides through the Hellenides (in Jacobshagen [53]). The base of the imbricates is Eohellenic ophiolite and the Triassic–Jurassic carbonate platform of the Pelagonian zone which is covered by disrupted ophiolite followed by schistose pyroclastic units interleaved with slices of radiolarian cherts, volcaniclastic and chloritic marble layers. This tectonic transition between Pelagonia and the western edge of the Vardar zone is shown by Sharp and Robertson [6] in the Arnissa area (Fig. 6): a ~500-m-thick succession of imbricated ophiolite mélange. This succession is topped off by limestone debris with rudists and planktonic foraminifera, Globotruncana [55] (Table 1b 3) (Plate 1). In agreement with these observations, we underscore that the contact between the Vardar and Pelagonian zone is a thrust-fault-zone (see Discussion and conclusions).

Although Cretaceous carbonates have been supposed to transgressively overlie laterite and serpentinite [6,47,55], we are of the opinion that the inferred transgressional conglomerates are cataclasites (Plate 2a,b) and that orthoconglomerates [57] that could substantiate a marine transgression have not been verified (see Discussion and conclusions). Furthermore, the Cretaceous limestones of the Vardar zone are in tectonic contact with the subjacent allochthonous substrate even where post-Eohellenic laterite is found along the contacts. The circumstances here are similar to the Northern Sporades where a sedimentary contact of the Cretaceous Carbonates with their original substrate is nowhere to be found [10].

Serpentinite and ophiolite–carbonate mélange crop out through the Cretaceous limestone cover in tectonic windows along a narrow, elongated zone of north–south striking faults, extending from Kerassia–Karydia–Kedronas [55,58] to Ano Grammatiko [18,56] (Fig. 6). Extensive exposures consist of ‘conglomeratic’ rocks [55], which in our opinion are cataclasites (see Plate 2 and Discussion and conclusions). The ‘conglomeratic’ rocks contain Triassic and Jurassic carbonates as well as limestones ranging in age from Cenomanian to Turonian (Table 1b 2) and overlie Pelagonian serpentinite [55]. Near Nisi and Karydia (Fig. 6) these cataclasites (Plate 2a,b) occur below Campanian limestone (Table 1b 4) (Plate 1). At its base, this succession contains olistolith marbles of Triassic–Jurassic age and overlie white micaceous Triassic marbles in suggested transgressional contact [55]. We dispute a transgressional origin of the Kedronas–Nisi ‘conglomerate’ (see discussion on pseudo-conglomerates). The tectonic windows exposing underthrust Pelagonian ophiolite rocks can be followed in west Almopias from the north near Karydia to the Vermion area [56] (Fig. 6, see section B-B´).

An extensive imbricated belt of ophiolite mélange some 50 km long and 5–10 km wide can be traced from the Lyki–Klissochori area [55,59] to the Naousa and Veria areas (Fig. 6) [18,56,60]. The mélange is interleaved with slices of marble and Jurassic carbonates, which we agree, are of Korab–Pelagonia origin [56,61] (Table 1b 6 and 7–7.2). The carbonates contain an Oxfordian–Kimmeridgian reefal fauna, including Cladocoropsis sp. of Late Jurassic age [59]. As pointed out above, this is a typical Kimmeridgian–Tithonian reef facies of the Pelagonian zone [21] (Table 1a 7–8) that had been overthrust by Eohellenic ophiolite during the Early Cretaceous. In the Vardar zone, the Pelagonian ophiolites are locally interleaved with sericitised basalt schist (Lyki) (see Geochemistry) and are in an underthrust position beneath ‘conglomeratic’, ophiolitic mélange and upper Cretaceous carbonates (east of Margarita, Fig. 6) (Table 1b 7).

In accord with the previously cited researchers and the described geology, we support the opinion that the ophiolites and upper Jurassic carbonates found in the west and central Vardar sub-zones are tectonically inherited from the underthrust Pelagonian plate (Fig. 4b).

A noteworthy difference between the eastern and western units of the Vardar zone is that the eastern Almopias and the Paikon units appear to be devoid of serpentinite which we corroborate from Tranos et al. [62]. Serpentinite, however, probably exists at depth (Fig. 4b), because further north in an area known as Ano Garefi serpentinised peridotite is exposed below basalt [60]. The mélanges of the Nea zoi-Vryssi-Meglenitsa and Krania units (Figs 4b and 6) are composed of dolerite, pillow basalt and tuff and contain upper Jurassic–lower Cretaceous radiolarite [59], with a relict Cretaceous cover (Table 1c 1.–1.2). Slices of Triassic lavas and radiolarites [63] (Table 1c 3 and 4) and upper Cretaceous arenites are also incorporated into the foliated matrix of the mélange of the Krania–Vryssi units [60]. The ‘ophiolite related’ mafic units, ‘ophiolite nappe’ and ‘Meglenitsa Ophiolite’, reported in Sharp and Robertson [6] (from Sharp and Robertson [64] and Sharp and Robertson [65]), in our opinion are not ophiolites s. str. but consist of ocean floor or arc basaltic rocks (see Geochemistry).

The Theodoraki limestone is the youngest formation of the Paikon antiform [23]. This limestone is part of the Cretaceous carbonate platform that covers the entire Vardar zone, and is composed of a wide range of neritic to reefal facies (Table 1b and Table 1c Theodoraki unit). The platform is in tectonic contact with a pile-up of SW dipping slices of Theodoraki limestones and slices of volcano-sedimentary rocks including radiolarites, tuffites and lava, and Triassic-Jurassic Marble and schist of Pelagonian origin [17]. Katrivanos and others [23] corroborate that the tectono-stratigraphic sequence is composed of volcano-clastic rocks together with limestones of Middle to Late Jurassic age, based on micro and macro-faunas including C. mirabilis (Griva–Kastaneri formation Fig. 4b, Fig. 6) (Table 1c Griva–Khromni units). The volcano-sedimentary slices are on top of Triassic–Jurassic Gandatch marbles and schists (Fig. 6). All the volcanic material of this series is strongly mylonitised in discrete, narrow shear zones related to mylonitic foliation [23]. The carbonate rocks are mylonitised, near the contacts with tectonically overlying volcano-sedimentary slices, for example, at Kastaneri [23]. Our investigations corroborate the above observations, which lead us to interpret the volcano-sedimentary formations in the substrate of the Theodoraki limestone as a composite allochthonous mélange complex in which slices of volcanic and sedimentary rock-units can be individually distinguished.

In contrast to the above, the Paikon unit has been depicted [64] to consist of a contiguous sedimentary, stratigraphic, succession extending from the Triassic to Cretaceous time only interrupted by two unconformities, an Oxfordian and a Cenomanian.

We share the opinion that the Paikon is an antiform and a Pelagonian tectonic window [23], and that the Paikon unit of the Vardar zone was most probably part of a volcanic island arc complex [3,8,15,17,60]. Our mutually envisioned island arc scenario evolved as the eastern Vardar Ocean subducted north-eastwards towards the margin of the European continent, which initiated supra-subduction arc volcanism [8,17,60]. This was accompanied by back-arc spreading (3,5), represented by the Guevgueli ophiolite complex (Fig. 4b) [61,66–68].

It should be taken into consideration that some remnants of the well-documented Cretaceous Pelagonian carbonate platform (Fig. 8a), may have been subducted (‘piggy-backed’) beneath the Cretaceous carbonate platform of the west Almopias, at the latest during the Paleocene time, and thus possibly inherited Pelagonian-orthoconglomerates could occur in the mélanges beneath the Vardar zone (e.g., Vermion: Photiades et al. [47]).

From the evidence presented here and from seismic tomography it is postulated that the Vardar Ocean subducted along two subduction zones (Fig. 9a). The western intra-oceanic subduction zone evolved about the Toarcian to Aalenian time, based on radiometric ages of amphibolites in sub-ophiolite mélanges, and continued to subduct through the Middle Jurassic verified by late Middle Jurassic radiolarians in the sub-ophiolite mélange in Evvoia [34,36,38] (Table 1a 11.2 and 12). A supra-subduction volcanic arc evolved during the Middle Jurassic, documented by the Elias complex of northern Evvoia (Fig. 4a) which presumably was part of a more extensive supra-subduction ‘Eohellenic arc’ (Fig. 9a) [34]. The beginning of the Eohellenic obduction, is suggested to have begun during the Bathonian time together with the Callovian upheaval [44] and the eastward subduction of the eastern Vardar Ocean (Fig. 9b 3). The Vardar, supra-subduction, volcanic island arc and the spreading Guevgueli back arc ophiolite complex evolved during the (Middle?) Late Jurassic and Cretaceous times. We envisage a Paikon forearc basin, rimmed by an accretionary wedge like that shown in Saccani et al. [67] in which the basin floor was initially covered by (volcanoclastic) basalt without carbonates during the lower Middle Jurassic. To the best of our knowledge, a Jurassic carbonate platform did not evolve in the Vardar zone. We suggest that Jurassic–early Cretaceous volcanoclastic deposits accumulated on the flanks of the Vardar volcanic arc and became the substrate of carbonate accumulation beginning in Aptian time. Investigations of the Guevgueli back arc basin have not disclosed relicts of a carbonate platform [67].

We have interpreted the perturbations beneath Hellenides as sunken Vardar Ocean lithosphere which indicate two episodes of subduction [10] (Fig. 3c).

The vertical section (Fig. 3c) shows that the leading edges of each slab has sunk to a depth approaching 2000 km. Presently, the trailing edge of the western slab (x in Fig. 3c) is about 900 km below the Earth’s surface and the trailing edge of slab (y) is about 400 km below the surface. These are the depths to which the slabs have sunk since their breakoffs. In estimating the width of a slab, however, one must consider that a subsiding lithospheric plate certainly undergoes compression and deformation which can make width-estimates inaccurate (Fig. 3e). The seismic tomographic images are, nevertheless, presently the best possible way to estimate the onetime width of the subducted oceanic lithosphere, which we estimate to have been about 3000 km [determined by adding together the lengths of the slabs (x + y) ~1200 + ~1700 and adding, to that sum, the width of the obducted Eohellenic ophiolite sheet, which has been assumed to be about ~200 km (Fig. 8c)]. However, 3100 km is the composite width, not necessarily the surface width that the Vardar Ocean had at any one time. We do not know when the ocean ridge stopped spreading: subduction and ocean spreading at the ocean ridge could have taken place simultaneously.

The western slab (x) is supposed to have broken off and began sinking after the Eohellenic ophiolite had been emplaced during Valanginian time. The eastern Almopias slab (y) is supposed to have broken off after Pelagonia crashed and underthrust the Vardar zone-carbonate platform and volcanic arc complex.