Geodynamic model
The convergence between Africa and Greece can be hypothesized back to the Eocene at a conservative mean speed of 3 cm/yr. With this velocity, a 1200 km long slab should have formed. The present African slab depicted by seismicity has a low dip (15-20°) and is shallow (no deeper than 200-250 km), becoming more horizontal northeastward, and assuming a spoon shape (Papazachos and Comninakis, 1977; Christova and Nikolova, 1993). Most of the seismicity in the Aegean Sea is rather superficial and the high geothermic gradient is probably responsible for the seismic disappearance of the slab underneath the basin. The opening of the Aegean-western Anatolian rift can kinematically be interpreted as due to the faster southwestward advancement of Greece over Africa, with respect to Cyprus-Anatolia over Africa (Figure 10). The rift generated a mantle uplift to compensate the lithospheric thinning. Therefore the underlying slab itself should have been involved and folded by the mantle uprise beneath the rift (Figure 11). The stretching between Greece and Anatolia, and the differential velocity of convergence with the underlying slabs should have generated a sort of “horizontal windows” both in the hangingwall and in the footwall of the subduction, allowing melting of mantle, and generating the OIB magmatism after regular subduction/collision evolution.
Figure 10. Reconstruction
Considering fixed Africa (A), Greece (B) is overriding Africa faster than Cyprus and Anatolia (C). This implies extension between Greece and Anatolia. Differences in thickness and composition of the subducting Africa lithosphere may determine the faster subduction of the Ionian oceanic lithosphere with respect to the Levantine Sea. The margin between Greece and Turkey is a diffused margin in the entire broad area of extension, and the straight line of the figure is positioned only for distinguishing the two plates. Larger rollback of the Hellenic-Greek slab should determine stretching in the downgoing Africa lithosphere with respect to the Cyprus-Anatolia slab.
No relative motion occurs between the central and eastern Mediterranean, since both sides belong to the same African plate. However, the plate is subducting both below Greece and Cyprus, but at different velocities, or in another reference frame, the hangingwall plate is overriding at different velocities. The Cyprus subduction is in fact slower as indicated by the minor shortening of the Quaternary sediments, by the lower seismicity with respect to the Hellenic subduction, and by geodetic data (Figure 4). In fact thinned continental lithosphere (Makris and Stobbe, 1984) occurs in the footwall of the arc of the eastern Mediterranean, whereas oceanic lithosphere characterizes the Ionian Sea to the west (de Voogd et al., 1992) beneath the Hellenic arc. If no relative motion between central and eastern Mediterranean occurs, the different convergent rates at the two subduction zones has to be related to differential velocities between the hangingwall plates, enabling a faster motion of Greece southwestward over the Ionian relative to Turkey and Cyprus over the eastern Mediterranean, responsible for the extension in between. In other words Turkey is relatively moving apart from Greece toward the northeast in the absolute reference frame, and not converging. This extension may or may not be coeval with compression elsewhere, and the related normal faults and shear zones should flat out in the decollement planes at base of the lithosphere. Similar “backarc” extension could be classified the Andaman Sea rift, in the hangingwall of the western Indonesia subduction zone, in the hinge where the arc advances southwestward faster than Asia over the Indian plate.
Figure 11. Cross-section
Cartoon showing that Greece lithosphere (B) is overriding Africa fixed (A) faster than Anatolia (C), generating extension between B and C. See a reference trace of the section in the previous figure. The extension in the ‘backarc’ is due to differential velocity of the hangingwall lithosphere. The African slab should be folded by the isostatic uplift of the mantle in the rift. This should results in a sort of window in the hangingwall lithosphere that is splitting apart into two independent plates, i.e., Greece and Anatolia, and it would ipothetically be coupled with a sort of horizontal window in the underlying stretched slab, allowing melting and uprise of OIB basalts. These kinematics should be envisaged in a 3D view, with the previous figure, where the faster rollback of the Hellenic trench with respect to Cyprus-Anatolia is determining a stretching of the slab also in a map view. The subduction zone migrated southwestward, and it was replaced by the extension. This is coherent with first the emplacement of the calc-alkaline rocks, and later the OIB basalts in the western Anatolia and the central-northern Aegean Sea.
Since the Aegean backarc basin developed in the hangingwall of a northeast directed subduction, it has been used as a case where the theory that backarc basins form only in the hangingwall of west-directed subduction zones (Doglioni et al., 1999) fails. It has rather been considered as an example of slab retreat faster than the convergence rate, determined by the slab-pull (Royden, 1993). However, from the aforementioned issues, it can be argued that the Aegean rift is not a classic backarc basin. In fact, one of the most debated type of rift are the so-called "backarc basins" related to E-NE-NNE-dipping subductions (e.g. the Aegean Sea or the Indonesia backarc). In this paper they are considered to have a different origin from backarc basins due to west-directed subduction zones because of the following points: a) They have lower rates of opening and subsidence with respect to backarcs related to west-directed subduction zones; b) They are often characterized by thick continental crust in spite of longstanding subduction, while in the opposite subduction, backarcs more frequently experienced fast generation of new oceanic crust; c) They may be inactive during contemporaneous subduction, while this does not occur for west-directed subduction zones; d) They have opposite polarity of opening toward WSW or SSW; e) They often are inverted in compressional regime.
Unlike the Apennines or other accretionary prisms related to west-directed subduction zones, the Hellenic arc and the related Mediterranean Ridge (the accretionary prism, e.g., Camerlenghi et al., 1995) have a low dip foreland monocline (almost flat, Clément et al., 2000), deep metamorphic rocks involved in the orogen, and higher structural and a morphologic elevation. Moreover, the slab has regularly low dip and it is shallow.
In the west-Pacific backarc basins (e.g., Honza, 1995), or in the Apennines, Carpathians, Barbados, Sandwich and Banda arcs subduction zones (Doglioni et al., 1999), the asthenosphere replaced the retreated subvertical slab (type 1 of extension in convergent settings, Doglioni, 1995). Due to the low dip of the Hellenic slab, in the Aegean Sea, the hangingwall has not enough space for a thick asthenospheric wedge like in W-Pacific or Apennines subduction zones. The hangingwall and footwall lithospheres are almost stacked one on top of the other, with thin (if any) sandwiched asthenosphere in between. Therefore the Aegean Sea represents a different type of extension associated to a subduction zone, where the hangingwall plate overrode the slab at different velocities, implying internal deformation (type 6 of extension), as shown in Figure 10 and Figure 11.
Considering the "east-northeastward" mantle flow indicated by the hot spot reference frame (Ricard et al., 1991), the origin of backarc basins due to E-NE-directed subduction should not be due to lithospheric disappearance as found in west-directed subduction zones. It could rather be related to differential drag of the lithosphere in the hangingwall of the subduction (Figure 11), due to different viscosity contrasts generated by lateral heterogeneities that control the amount of decoupling at the interface between lithosphere and asthenosphere. The system is composed by three plates (A-B-C, Figure 11), in contrast with back-arc basins due to W-directed subductions where the backarc may develop with two converging plates or even within one single plate.