A number of major tectonic events indicate that around the late Miocene (5-6 My), the Hellenic Arc accelerated its southward buckling:

• The most intense accretionary activity along the external front of the Aegean Arc, i.e. the Hellenic trench, occurred in the Pliocene, leading to the formation of the so called "Mediterranean ridge" (e.g. Finetti, 1976; Underhill, 1989; Mascle and Chaumillon, 1997).

• The north Aegean and northwestern Anatolian regions were affected by a significant acceleration of transtensional tectonics, associated with a system of SW-NE strike slip faults (Sengor et al., 1985; Hempton, 1987; Yilmaz, 1989; Mercier et al., 1989; Zanchi et al., 1990; Taymaz et al., 1991; Barka, 1992).

• Crustal thinning, with a dominant S-N extensional trend, occurred in the Western Cretan basin (WCB) which almost reached its present configuration around the late Pliocene (Buttner and Kowalczyk, 1978; Angelier et al., 1982; Mercier et al., 1989; Meulenkamp et al., 1994).

The fact that crustal stretching only occurred in such a limited zone, with an almost triangular shape (Figure 3b), and only developed during the Pliocene, imposes important constraints on the driving mechanism of this extensional event.

• The Cyclades arc was affected by intense E-W compressional deformation and by uplift, which generated plurikilometric sized faults and the first deposition of continental facies, after the Miocene marine sedimentation (Buttner and Kowalczyk, 1978; Durr et al., 1978; Mercier et al., 1987, 1989; Avigad et al., 2001).

The deformations listed above may be interpreted, in our opinion, as effects of the continental collision between the Adriatic block and the Tethyan orogenic system at western Greece. After this collision, the convergence between the eastern Anatolia and the Adriatic plate, no longer accommodated by the consumption of the pre-Apulian zone, emphasized E-W stresses on the Aegean arc, accelerating its southward bending/extrusion. We advance the hypothesis that this bending and the different mechanical behavior of the external accretionary belt (Hellenides) with respect to the inner massifs (Cyclades) was responsible for the generation of the WCB, through the mechanism sketched in Figure 4. Due to its high rigidity, the Hellenides belt did not resist the tensional stress induced by bending and broke up in two arcs, the Peloponnesus to the West and the Crete-Rhodes to the East. After this break, the two Hellenic arcs diverged from the Cyclades massifs, causing crustal extension in a roughly triangular zone, the WCB.

Around the late Pliocene–early Pleistocene, the tectonic setting in the southern Aegean region changed considerably. Crustal stretching ceased in the WCB and began in the Eastern Cretan basin (ECB) , with a roughly NW-SE extensional trend. Tectonic activity in the eastern Hellenic Arc (Crete-Rhodes) underwent a significant acceleration, with the formation of several discontinuities and the land interruption between Crete and Rhodes (e.g. Buttner and Kowalczyk, 1978; Armijo et al., 1992). This tectonic change might be a consequence of the incipient collision of African continental domain (Libyan promontory) against the central sector of Hellenic consuming boundary, just southwest of Crete (e.g. Ryan et al., 1982; Lyon-Caen et al., 1988; Armijo et al., 1992; Mascle and Chaumillon, 1997). After this continental collision, the compressional stress field induced by the westward push of southern Anatolia concentrated in the eastern Hellenic Arc, causing its SEward bending/extrusion, accompanied by a considerable fragmentation (Figure 4c). In the wake of the outward migrating eastern Hellenic arc, NW-SE extensional tectonics occurred, with the formation of the ECB. The above mechanism and the related deformation pattern are still going on in the southeastern Aegean area and eastern Hellenic Arc, as indicated by neotectonic, seismological and geodetic data (Mercier et al., 1989; Armijo et al., 1992; Papazachos and Kiratzi, 1996; McClusky et al., 2000; Viti et al., 2001).

The above space-time distribution of tectonic events in the Aegean zone cannot easily be explained by the slab pull model. Here we report some considerations about the major outstanding problems.

• It is widely recognized that the Ionian-Levantine lithosphere has subducted roughly NNE to NEward, along a trench zone extending from the Kefallinia fault system to Crete (Figure 2), while the easternmost sector of the Hellenic trench ( Pliny and Strabo) is instead recognized as a sinistral transpressional border (e.g. McKenzie, 1978; Le Pichon and Angelier, 1979; Mercier et al., 1989; Armijo et al., 1992). Given this geometry of the slab, one could expect that SW-NE back-arc extension induced by its roll back mainly affected the Peloponnesus and the Aegean internal zone (the present Aegean sea). Instead, since the late Miocene the most evident extensional deformation with a roughly S-N trend has occurred in a limited and peculiarly shaped zone of the southern Aegean area, i.e. the WCB, as indicated by geological and geophysical observations (Berckhemer, 1977; Le Pichon and Angelier, 1979; Angelier et al., 1982; Gautier and Brun, 1994) and by the fact that, at present, the crust in the above zone is significantly thinner, roughly 20 km, than that in the surrounding Aegean zones, roughly 30 km (e.g. Makris, 1978; Meissner et al., 1987).

• It has been suggested that Crete during its separation from the central Aegean zone (Cyclades massif) moved roughly southward, (Buttner and Kowalczyk, 1978: Angelier et al., 1982; Mercier et al., 1987, 1989; Armijo et al., 1992). Contemporaneously, the Peloponnesus experienced a significant clockwise rotation, roughly 30° (e.g. Kissel and Laj, 1988). This heterogenous kinematic behavior of the various sectors of the arc, with the consequent separation between the Peloponnesus and Crete (Armijo et al.,1992), cannot easily be reconciled with the presumed cylindrical SWward roll-back of the Hellenic slab.

• Geological evidence indicates that the starting of the more recent extensional phase in the Southern Aegean zone (Pliocene) was more or less coeval with the starting of outward migration of the Hellenic Arc (e.g. Le Pichon and Angelier, 1979; Angelier et al., 1982; Mercier et al., 1989). This would imply that the driving mechanism of this T-A-BA system can not easily be related with the pull of the Pliocenic-Quaternary slab, since in the early Pliocene this slab was not yet sufficiently developed to undergo gravitational instability. Thus, the presumed slab-pull force could only be related to the sinking of a pre-existing subducted lithosphere. In this case, however, one should explain why gravitational instability just occurred in a limited sector of a considerably laterally extended consuming boundary, ranging from the Dinarides to Anatolia (Mercier et al., 1987, 1989).

• Around the late Pliocene-early Pleistocene, the deformation pattern in the internal Aegean area underwent a considerable change. S-N crustal stretching almost ceased in the WCB (e.g. Armijo et al., 1992) and began to develop in the ECB, with a roughly NW-SE extensional trend, and in the western Aegean area between the Peloponnesus and Crete, with a dominant E-W extensional trend (e.g. Mercier et al., 1989; Armijo et al., 1992). If extensional activity in the Aegean back-arc zone was driven by slab-pull forces, one could expect that the observed drastic change of strain pattern around the early Pleistocene was associated with important changes in the Hellenic subduction process. However, there is no clear evidence of any significant change, in subduction rate and geometry, of the above consuming process since the upper Miocene-early Pliocene (e.g. Le Pichon and Angelier, 1979).

• The sinking of the Hellenic slab and the related SWward trench suction can hardly be assumed as the driving mechanism of the Quaternary extensional activity, i.e. the NW-SE extension in the southeastern Aegean zone and of the roughly E-W extension between the Peloponnesus and Crete.