Around the late Eocene-early Oligocene the border zone between the western European foreland and the western Apulian region was constituted by an orogenic/metamorphic belt (Iberian, in Figure 2a) built up by the consumption and closure of the Tethyan oceanic domain (Cohen, 1980; Dercourt et al., 1986; Finetti et al., 2001). Around the early Oligocene, the central sector of this belt, along with a fragment of the European foreland (the Corsica-Sardinia microplate), detached from Western Europe, and extensional tectonics began in the Balearic basin (Figure 2b). After this detachment, the Balearic arc underwent a long east to SEward migration and a considerable bending, up to reach the final configuration shown in Figure 2c. The major features of the kinematic reconstruction of this T-A-BA system shown in Figure 2 are widely recognized (e.g. Cohen, 1980; Rehault et al., 1984; Dercourt et al., 1986).

The kinematic patterns of Africa and Arabia adopted in the maps are based on the evidence and arguments given by Dercourt et al. (1986), Hempton (1987), Albarello et al. (1995), Mantovani et al. (1997, 2000a,b, 2001b). Motion rates are only indicative. Present geographical contours and the paleoposition of the African, Sardinia and Corsica coastal line are reported for reference in each evolutionary phase. (Click for enlargement)

Here we hypothesize that the detachment of the Iberian belt from Western Europe was driven by the SW-NE Africa-Europe convergence, after the collision of the southwestern edge of this belt with the African continental domain (Figure 2a), roughly connected with Morocco (Le Pichon et al., 1988; Sengor, 1993). The arc parallel compression induced by this oblique plate convergence implied a longitudinal shortening of the belt, which was accommodated by its outward bending/extrusion, at the expense of the Western Apulian domain, recognized as an oceanic low buoyancy zone (Beccaluva et al., 1994; Doglioni et al., 1999). Extensional tectonics developed in the wake of the extruding arc, with the formation of the Balearic back arc basin. This process progressively slowed down and ceased as more and more eastern sectors of the southern migrating arc (Iberian-Maghrebian belt) collided with the African continent, along the Algero-Tunisian sector (Figures 2b and 2c). Extension in the Balearic basin finally ceased around the middle Miocene (12-13 My), when the Corsica-Sardinia microplate underwent a complete stop. At the end of this evolutionary phase (Figure 2c), the western Mediterranean region was characterized by a structural/tectonic setting not much different from the present one (e.g. Rehault et al., 1987; Dercourt et al., 1986; Vigliotti and Langenhein, 1995).

To evaluate the reliability of the driving mechanism here proposed and to understand if it can offer better chances to explain the observed deformation with respect to the slab pull model, it may be useful to make some considerations about the expected differences between the deformation patterns implied by these two types of interpretation. The most evident differences are expected in the migrating arc, since the extrusion model (Figure 1) predicts a shortening of the arc in the direction of the driving stress, whereas the slab pull model implies the lengthening and disruption of the original belt. Another important difference is expected in the back arc basin, since the slab pull model predicts a pure extensional, trenchward oriented, strain regime, whereas the extrusion model involves a transtensional tectonics due to the simultaneous action of dominant trenchward extension and perpendicular compression (e.g. Philip, 1987). Minor differences are expected, instead, in the external part of the arc, where both models predict slab roll back under the advancing arc and consequent accretionary activity along the trench.

Here, we argue that the major features of the arc deformation pattern observed in the western Mediterranean area (Figures 2b and 2c) can more easily be reconciled with the implications of the extrusion model on the basis of the following evidence and arguments:

• The tectonic structural setting of the Balearic basin, reconstructed by seismic surveys (e.g., Rehault et al. 1984, Figure 3) is characterized by a series of crustal wedges, decoupled by strike slip faults, as shown in Figure 2c, with an overall geometry of the arc very similar to that predicted by the extrusion model (Figure 1). It is hard to believe that the migration of such a fractured arc, while maintaining the ordered distribution and close contact of wedges shown by the present configuration (Figure 2c), may have been produced by a driving mechanism not involving any longitudinal compression of the belt, as the one implied by the roughly eastward roll-back of the Apulian slab.

• The strong bending that the Balearic arc underwent during its migration, changing from a more or less straight configuration (Figure 2a) to the final shape characterized by two almost perpendicular sectors (Figure 2c), could be seen as a shortening process able to accommodate the SW-NE convergence between Africa and Europe. Instead, to interpret such bending as an effect of slab pull forces, one should explain why the rate of trench retreat was higher in the central part of the arc with respect to its peripheral segments. This explanation must necessarily invoke a peculiar distribution of densities in the roll backing slab or other ‘ad hoc’ structural/geometrical conditions of the subduction process, which should be supported by observational evidence. Other considerations about this last problem are given by Mantovani et al. (2001d).

• A sinistral transpressional deformation is recognized in the Iberian belt during the collisional phase which preceded the detachment of this belt from Western Europe (Marroni and Treves, 1998; Finetti et al, 2001). This kind of strain regime is consistent with the oblique Africa-Europe convergence suggested by our interpretation. This agreement between predicted and observed features is also corroborated by the fact that the successive extensional phase along this plate border was characterized by a sinistral transtensional regime (Finetti et al, 2001). This kind of deformation is not predicted by the slab pull driving mechanism, which would only involve a trenchward extensional regime.

A severe difficulty for the slab pull model, in this T-A-BA system, is the fact that the development of the Apulian slab (i.e. the one which formed during the opening of the Balearic basin) began simultaneously with the onset of arc migration in this area. In fact, on the basis of geological and petrological evidence, a number of authors suggested that the previously active consuming process at this plate border involved the subduction of the northern European domain under the southern Apulian one (Cohen, 1980; Rehault et al., 1984; Doglioni et al., 1999). This would mean that the starting of back arc extension in the Balearic basin coincided with an inversion of the subduction vergence at that plate border and, that, consequently, the new embrional NWward dipping slab could not certainly induce back arc extension by a slab pull mechanism.

Another significant feature of the Balearic arc is the strong curvature that it shows in the segment comprised between the western Alps and Corsica (Figure 2c). To explain such strong distortion, it seems necessary to assume a roughly northward displacement of the belt, in line with the evidence mentioned in the previous point and with the proposed arc parallel compression.