As said before, this class of interpretations is mainly constituted by extrusion and pull apart mechanisms. The effects of these two kinds of mechanism on the structural and morphological features within and around back arc zones are generally well differentiated. The first type (extrusion) is mainly characterized by an arc undergoing a progressive deformation (buckling) and a back arc basin which progressively opens in the wake of the migrating arc (T-A-BA system). In this context the extensional trend of the basin is roughly parallel to the arc's motion trend, i.e. roughly perpendicular to the trench. The second type of mechanism, instead, requires the presence of a major shear zone between two plates and the direction of extension in the basin is more or less parallel to the relative motion of the two plates.
This mechanism, whose basic features are sketched in Fig. 3, is expected to occur along a consuming border where a sector of the accretionary belt is deformed by an extrusion mechanism and partly separates from the overriding plate. This separation is accommodated by crustal stretching in the back arc zone. Simultaneously, the outward migration of the deforming belt (arc) causes the roll back of the slab lying in front of it (Fig. 3B). The tectonic context which produces the deformation of the arc may be quite variable from case to case. Most often, this phenomenon occurs when a buoyant structure enters a sector of the consuming border, with a direction of motion not perpendicular to the trench. In this oblique constrictional context, the accretionary belt undergoes a longitudinal shortening, which is accommodated by lateral bending/extrusion, at the expense of the adjacent lithosphere (Fig. 3).
The occurrence of this mechanism requires that the buoyancy of the accretionary belt is significantly higher than that of the lithospheric domain lying in front of it. Thus,for instance, the lateral extrusion of the arc is strongly favoured when it faces old oceanic lithosphere, since such kind of lithosphere is presumably characterized by very low, or even negative, buoyancy (Cloos, 1993). This might explain why even the presumably limited loading of a poorly developed arc structure, such as the volcanic arcs of the Mariana and Tonga zones, may have caused the roll back of the adjacent slabs in the related arc-trench systems. Another basic requisite for the occurrence of the proposed mechanism is a rigid behavior and a limited fracturation of the belt, which allows the formation of few relatively large crustal wedges sliding and rotating each other, as shown e.g. in Fig. 3. If this condition is not fulfilled, the highly fragmented extruding material tends to occupy the entire space available and, thus, it does not allow the separation of the arc from the overriding plate and the consequent back arc extension.
To help the description of the proposed model, some examples of T-A-BA systems, which might have formed by the extrusion mechanism, are shown in Fig. 3.
The possible importance of extrusion processes in the generation of back arc basins has been already stressed by a number of authors (e.g. Tapponnier, 1977; Tapponnier et al., 1986; McCabe, 1984; Uyeda, 1986; Lavé et al., 1996; Mantovani et al., 1996, 1997, 2000a, 2001a). The physical plausibility of this kind of mechanism has been demonstrated by analitical computations and by analogue and numerical modelling (Tapponnier et al., 1982; Peltzer and Tapponnier, 1988; Ratschbacher et al., 1991; Faccenna et al., 1996; Mantovani et al., 2000b, 2001b). In the following, we argue that the implications of this model may provide plausible solutions for the outstanding problems of the subduction related interpretations we mentioned before.
The lack of back arc extension in several subduction zones may be barely explained by the fact that the conditions required for the occurrence of the extrusion mechanism were not present at those consuming boundaries. For instance, along the south American subduction zones there is no evidence of lateral extrusion processes along the trench-arc system. This is coherent with the fact that the lithospheric domain entering this trench-zone presents quite a normal oceanic character all along the respective consuming boundaries, unlike what happens in a number of western Pacific subduction zones (see section 6.1.2) and with the fact that plate convergence is perpendicular to the consuming boundary.
The evidence that some areas of former back arc extension are now inactive even though they remain adjacent to active subduction zones, as, e.g. the Japan and Kurile basins, might be due to the fact that at a certain evolutionary stage new boundary conditions, no longer favourable to the occurrence of the extrusion mechanism, began to affect the respective arc-trench systems. This tectonic event may be caused, for instance, by the arrival of a buoyant domain in the sector of the trench zone facing the migrating arc or by the fact that the deforming arc has reached such a configuration, with respect to the dynamic boundary conditions, to inhibit any further deformation. Of course, the effective reliability of these potential explanations must be checked for each T-A-BA system (see the discussions intext).
The strong curvature of arcs in T-A-BA systems seems to be a plausible consequence of the arc-parallel compression they are supposed to undergo in the extrusion model (Fig. 3). Later in this work it is argued that this kind of kinematically induced dynamic conditions might be recognized in the tectonic contexts where back arc extension occurred.
The extrusion model may explain why back arc basins are systematically associated with subduction, without invoking a causal relationship between the two processes. In fact, this model implies that lithosphere consumption and back arc extension are both side effects of a third process, i.e. the forced outward migration of the arc (Fig. 3). Slab roll back is caused by the push and gravitational load of the advancing arc onto the margin of the subducting lithosphere, while back arc extension is produced by the divergence between the arc and the overriding plate.
Another major feature of back arc basins which may be accounted for by the extrusion model is the fact that crustal stretching and arc buckling do not occur along the entire length of a convergent plate boundary, but only develop along a sector of it (Uyeda and Kanamori, 1979). These processes, in fact, are expected to only occur in the limited zone where the arc has separated from the overriding plate (Fig. 3). The geometry of the back-arc basin is thus controlled by the kinematics and nature of the indenting buoyant block, by the original configuration of the accretionary belt and by the dimensions of the oceanic domain lying in front of the extruding arc. This last factor, for instance, had a crucial influence on the deformation pattern of the Calabrian Arc-south Tyrrhenian and Carpatho-Balkan systems (e.g. Mantovani et al., 1997, 2000a).
A major difference between the Mariana type and the Chilean type subduction zones is the dip of the slab, almost vertical in the first type and nearly horizontal in the second type (e.g. Uyeda and Kanamori, 1979; Taylor and Karner, 1983; Scholz and Campos, 1995). This difference seems to be confirmed by the distribution of subduction zones versus dip angles (Fig. 4b), even though intermediate values of dip angles (40-60°) may characterize both types of subduction. Anyway, to explain the striking difference of slab dips beneath the extreme examples of the two types of subduction, one could consider the rather different rheologies of the mantle zone through which the slab must penetrate and eventually deform in the two cases. Various kinds of geophysical investigations have indicated that the asthenospheric layer is much more developed beneath oceanic domains than under continents (e.g. Pollack and Chapman, 1977; Artemieva and Mooney, 2001 and references therein). This implies that the viscous resistance forces which act on a slab dipping under a continental block (like the South America plate) are much higher than those acting in the 'softer' mantle underlying a T-A-BA system, like the Mariana one. Consequently, the steepening of the slab, driven by its negative buoyancy, is much less resisted beneath a Mariana type than beneath a Chilean type arc. This could explain why the Mariana slab has already reached an almost vertical configuration after a relatively short time life,whereas the Chilean slab is still dipping at a very low angle after a much longer time life.
In this regard, it is useful to point out that the key tectonic process which determines the asthenospheric environment through which the slab penetrates in the Mariana type subduction zones, is the separation of the arc from the continental plate. In fact, due to this separation, subduction must occur some hundreds of km away from the continental margin and, thus, the slab can penetrate and move through the large volume of asthenospheric material which has been attracted by the lithospheric thinning in the back arc zone.
Further insights into this problem may be gained by considering the features of the Calabrian and Hellenic subduction zones. The slabs under these arcs present quite different dip angles (65 and 30° respectively), in spite that the subducting lithosphere (the Ionian-Levantine oceanic domain) is the same in the two zones (e.g., Dercourt et al., 1986; Finetti and Del Ben, 1986). The difference of dip angles could be explained, for instance, by the different rheology of the mantle through which the respective slabs must penetrate and move. Beneath the Calabrian arc and the southern Tyrrhenian basin one can reasonably expect the presence of a soft mantle, due to the relatively large extension of the stretched zone, whose formation has certainly induced an abundant flow of asthenospheric material from the surrounding mantle. This hypothesis is strongly supported by geophysical investigations in the Tyrrhenian area, which indicate an oceanic-like lithosphere and a well developed asthenospheric mantle (Mele, 1998; Morelli and Piromallo, 2000; Martinez et al., 2000). Beneath the Aegean zone, instead, crustal stretching is very limited, being confined to the small Cretan sea, where a crustal thickness of roughly 20-25 km is estimated. In the remaining Aegean area, a continental character of the lithosphere is widely recognized (e.g. Makris, 1978; Meissner et al., 1987; Papazachos and Nolet, 1997).
The tectonic pattern implied by the extrusion model could also provide the ground for explaining another basic feature of T-A-BA systems, i.e. the weak seismicity observed at this kind of trench zone, like the Mariana one, with respect to the very strong earthquakes recorded at the Chilean type subduction zones (e.g., Uyeda and Kanamori,1979; Scholz and Campos,1995). To this purpose, one must consider that the dimensions of the subduction fault (controlling the magnitude of decoupling earthquakes) implied by the extrusion mechanism are expected to be considerably smaller than those of Chilean type subduction zones. In fact, in T-A-BA systems the overriding arc is constituted by relatively thin crustal wedges (Fig. 3) and, thus, the frictional interface with the descending lithosphere cannot exceed few tens of km. In the Chilean type boundaries, instead, the subduction fault may involve the entire lithosphere and, thus, the magnitudes of the decoupling earthquakes may be much larger (e.g. Uyeda and Kanamori, 1979; Carlowiczs, 1995; Wang, 2000). Another factor which could contribute to mitigate seismic activity at T-A-BA systems with respect to Chilean type subduction zones is the increased role of slab pull among the forces acting on the slab. This, in fact, could reduce the coupling between the subducting and the overriding lithosphere, with significant effects on the amount of seismic energy release (Scholz and Campos, 1995).
The uprise of hot asthenospheric material up to crustal levels beneath the back arc zone, implied by the extrusion model (Fig. 3), may explain the high heat flow observed in this kind of regions (e.g. Uyeda, 1986).
In this section we have discussed on how the implications of the extrusion model might account for the major features of T-A-BA systems. However, to understand the plausibility of this mechanism it would be necessary to try to recognize if the boundary conditions and structural-rheological properties required for its occurrence were present in the tectonic contexts within which back arc basins opened up. This problem is discussed in the next sections.
We start the discussion from this region since its evolution has involved a number of T-A-BA systems and also because the analysis of the deformation pattern of this area led us to believe that the extrusion model is the most plausible driving mechanism of back arc opening (Mantovani et al., 1997, 2000a, 2001a, 2002). For a detailed discussion about the large amount of evidence and arguments which may support this conviction we make reference to the above papers. Here we only point out, by the help of the proposed evolutionary reconstruction (Figs.5 and 6), that the boundary conditions and structural features implied by the extrusion mechanism might be recognized in the tectonic contexts which led to the formation of the major back arc basins in the Mediterranean area.
The formation of this T-A-BA system was triggered, around the upper Oligocene, by the oblique continental collision between Africa and Western Europe (Fig. 5a). After this contact, plate convergence was accommodated by the East to SEward extrusion of crustal wedges of the Alpine belt and of a fragment of the European foreland (Corsica-Sardinia microplate), at the expense of the old oceanic lithosphere in the Western Apulian zone (Fig. 5b,c). In the wake of the migrating arc, extensional tectonics developed, generating the Balearic basin. The migration of the southern branch of this arc underwent a progressive stop, as it collided with more eastern sectors (Algeria-Tunisia) of the northern African continental margin (Fig. 5c). The opening of the Balearic basin definitively ceased around the upper Miocene when the Corsica-Sardinia microplate reached its present position (e.g. Rehault et al., 1984; Dercourt et al., 1986). It is interesting to note that the morphology of the arc, suggested by geophysical investigations (e.g. Rehault et al., 1984), is constituted by a number of wedges decoupled by strike-slip faults (Fig. 5b,c), with a pattern very similar to that shown in Fig. 3b.
The formation of these extensional zones (almost coeval with the Balearic basin) was a side effect of the northward indentation of the Arabian promontory against the orogenic zone (Tethyan belt) created by the consumption of the Tethyan ocean (Fig. 5b,c). This belt (Fig. 5a) was constituted, in the northern part, by an accretionary chain of European affinity (Carpathians, Balkanides and Pontides), by oceanic remnants, crystallin massifs and metamorphic units in the inner part (Pelagonian, Aegean and Anatolian massifs), and by an accretionary chain of African affinity (Dinarides, Hellenides and Taurides) in the southern part.
Under the push of the Arabian indenter, the Anatolian sector of the Tethyan belt decoupled from the Iranian sector, through a system of right lateral faults (Hempton, 1987) and moved roughly NWward. This displacement was accommodated by a complex deformation of the eastern Tethyan belt, which involved a differentiated behavior of the northern chain (Carpatho-Balkan) with respect to the inner massifs and southern belt (Dinarides-Hellenides). In the Balkanides and Carpathian belt the longitudinal shortening was accommodated by the lateral extrusion of crustal wedges, at the expense of consumable zones in the southern Moesian margin and, more evidently, in the European Carpathian foreland, respectively (Fig. 5b,c). In the wake of the migrating Carpathian arc transtensional tectonics occurred in the Pannonian area. The progressive collision of the Carpatho-Balkan arc with the continental European domain caused the end of that extrusion process and of the consequent back arc extension in the Pannonian region (Fig. 5c). The inner massifs and the Dinarides belt underwent a more gentle deformation, in terms of a southward buckling, at the expense of the Ionian-Levantine old oceanic zone. The divergence between the Balkanides and the Aegean inner massifs caused the extension in the North Aegean and Northwestern Anatolian area documented by geological and vulcanological evidence (e.g. Papazachos, 1989).
The evolutionary phase which led to the present configuration of the Hellenic Arc (Fig. 6) started around the late Miocene, after two major tectonic events: the continental collision between the Adriatic plate and the Aegean zone, at the outer Hellenides (e.g. Mercier et al., 1989), and the activation of the western segment of the North Anatolian fault system (Barka, 1992). These events determined a strengthening of the E-W compression on the Hellenic Arc, which accelerated its southward extrusion/buckling, at the expense of the Ionian-Levantine zone. The higher rigidity of the external belt (Hellenides) with respect to the inner massifs (Cyclades) led to the separation and opposite rotations of the eastern (Crete-Rhodes) and western (Peloponnesus) segments of the Hellenic Arc, with the consequent formation of the Cretan basin (Mantovani et al., 1997,2000a,2001a).
The Plio-Quaternary East to SEward migration of the Alpine-Apenninic orogenic belt, which lay east of Sardinia after the opening of the Balearic basin (Fig. 6a), was produced by an important change of the kinematics of the Adriatic plate and by its lateral effects. This change resulted from the continental collision between the Anatolian-Aegean system and the Adriatic block, occurred around the late Miocene (Mercier et al., 1989). After this contact, the Adriatic plate began a clockwise rotation, which caused the lateral expulsion (NWward) of an African fragment, the Iblean wedge (Fig. 6b). Then, in the constrictional regime induced by the convergence between the African and Adriatic blocks and the Iblean microplate, the intervening orogenic material was expulsed laterally, at the expense of the remnant part of the western Apulian zone and of the Ionian area (Mantovani et al., 1997,2000a,2001a). During the first stage (late Miocene to late Pliocene), the lateral escape of wedges was directed mainly eastward, as indicated by the extensional trend in the central Tyrrhenian basin and the features of accretionary activity in the Southern Apennines (e.g. Ortolani et al., 1979, 1992; Sartori, 1990). In the second stage (Fig. 6c) after the suture of the Southern Apennines consuming boundary (e.g. Patacca et al., 1990), the extrusion mainly involved the SEward escape of the Calabrian wedge, as indicated by the extensional trend in the southernmost Tyrrhenian basin and the features of accretionary activity in the external Calabrian Arc (e.g. Finetti and Del Ben, 1986).
The discussion on how the above geodynamic interpretations may provide plausible and coherent explanations for the complex space-time distribution of post-Eocenic tectonic events occurred in the Mediterranean area is reported by Mantovani et al., 1997, 2000a, 2001a, 2002. Numerical modelling experiments (Mantovani et al., 2000b, 2001b) have shown that a satisfactory match of the strain pattern in the central-eastern Mediterranean area, deduced by a large amount of geological and geophysical information, can be obtained by adopting the convergence of the confining blocks (Africa, Arabia and Eurasia) as the only driving mechanism of tectonic activity in this region.
In this section, we describe how the tectonic conditions required for the occurrence (and the stop) of the extrusion mechanism might be recognized in the zones where T-A-BA systems developed (Figs. 7,8).
The formation of the Japan basin has tentatively been explained as an effect of the extrusion of the Japan arc, due to the transpressional collision between the Okhotsk block and Eurasia (Dickinson, 1978; Tapponier et al., 1982; Kimura et al., 1983; Seno et al., 1996; Kusunoki and Kimura, 1998; Altis, 1999). In this view, crustal extension occurred in the wake of the Japan arc, which was forced to bend by the above mentioned compressional boundary conditions. Altis (1999) argued that this interpretation can plausibly account for the major features of the Early-Middle Miocene deformation in Japan and surrounding zones, and that the application of an indentation-extrusion model to the Okhotsk-Eurasia collision zone allows a simpler and more coherent interpretation of the origin and development of the T-A-BA system in the Japan area, compared with the achievements of previous models. The same author also suggested that when the Japan arc began overriding the young and hot Shikoku basin (see below), in the Middle Miocene, its extrusion (and thus back arc opening) came to an end, due to the resistance that this last basin opposed to further subduction.
Other non subduction-related interpretations of back arc extension in the Japan sea have been proposed by a number of authors (e.g. Otofuji and Matsuda, 1983; Lallemand and Jolivet, 1985; Hayashida et al., 1991; Jolivet et al., 1994, 1995; Lee et al., 1999; Itoh, 2001). A discussion about the difficulties that subduction-related mechanisms may encounter in explaining the opening of the Japan sea is reported by Tatsumi et al. (1990).
The fact that the opening of the Kurile basin was more or less coeval with the one of the Japan basin (see the Table) and that the Kurile arc lay in between the Japan system and the Okhotsk block could suggest that the constrictional tectonic context which determined the formation of the Japan T-A-BA system was also responsible for the outward buckling of the Kurile arc and for the extension in the related back arc zone.
The geodynamic context which led to the coeval generation of the first two basins (e.g., Honza, 1995) might have been characterized by conditions very similar to those implied by the extrusion model:
* An arc (Izu-Bonin and proto Mariana ridges), stressed more or less parallelly to its main trend (N-S) by the convergence of the confining blocks, which were constituted, on one side, by the Australian plate moving roughly northward, and, on the other side, by the Japan Arc, extruding roughly southward (e.g., Altis, 1999).
* The presence, on the outer side of the migrating arc (Izu Bonin and proto Mariana), of a Mesozoic oceanic lithosphere (Pacific domain), playing the role of a weak lateral boundary.
Under such N-S compression, the buoyant arc was forced to migrate eastward, overriding the adjacent Pacific oceanic domain. In the wake of the migrating arc, extensional tectonics developed in the back arc zone, forming the Shikoku and Parece Vela basins (Fig. 7). This interpretation is consistent with the evolutionary reconstruction of the western Pacific area proposed by Lee and Lawver (1995). In particular, one could note that the start of extension in the above two basins coincided with the onset of bending in the Izu Bonin and Mariana arc-trench systems.
The tectonic and geologic evolution of these two regions prior to the late Miocene was essentially the same (Taylor and Karner, 1983), while in the successive evolution the Shikoku basin underwent a 25° CCW rotation, probably related to a N-S squeezing of the region (Altis, 1999) and extension has only continued behind the Mariana arc-trench system. This last change was probably connected with the collision between the Caroline ridge and the proto Mariana-Yap trench, which caused the sharp bend of the southern Mariana arc and the consequent back arc extension in the Mariana trough (McCabe and Uyeda, 1983; McCabe, 1984; Eguchi,1984). This explanation is supported by paleomagnetic, geological and geophysical observations (McCabe, 1984) and is also consistent with the evolutionary reconstruction proposed by Lee and Lawver (1995).
The formation of this extensional feature (Fig. 7) has been explained as an effect of the convergence between the Luzon arc and the East Asia continental margin at Taiwan (Letouzey and Kimura, 1985; Lee and Lawver, 1995; Huang et al., 1997). This convergence would have caused the outward migration of the Ryukyu Arc at the expense of the Pacific oceanic domain, with the consequent occurrence of crustal extension in the internal part of the arc (Okinawa trough). The evidence that the uplift of the Ryukyu Arc preceded the formation of the Okinawa trough (Sibuet et al., 1987) is consistent with the hypothesis that the above arc underwent a longitudinal shortening, in line with the extrusion model.
The formation of this extensional zone has been interpreted as an effect of the divergence between the Caroline block (Fig. 7), rotating anticlockwise with respect to the surrounding zones and the northwestward moving Halmahera region. This kinematic pattern has started around the middle Miocene, when the above regions collided with the New Guinea promontory, moving northward (Lee and Lawver, 1995; Hall, 2001). A subduction-related explanation for this trough would be extremely problematic, since it can not easily be associated with any subduction zone.
Crustal extension in this zone has been interpreted as a consequence of the collision between the Ontong-Java plateau of the Pacific plate and the New Guinea continental promontory of the Australian plate (Ripper, 1982; Weissel et al., 1982; Taylor and Karner, 1983). This constrictional context was accommodated by a complex deformation pattern of the intervening zones. In particular, the most buoyant structures, i.e. the New Britain and Woodlark ridges (Figs.7 and 10), underwent bending and rotations, with the generation of extensional zones (as the Woodlark basin) in the wake of the rotating ridges and the activation of a new consuming boundary (New Britain trench) in front of the advancing arc (Lee and Lawver, 1995; Hall, 2001).
The present tectonic setting and kinematic pattern of the interaction zone between the Australian and Pacific plates is illustrated in Fig. 8. The genetic mechanism of the North Fiji and Lau-Havre basins, and of the younger Taupo rift, is still matter of debate (Uyeda and Kanamori, 1979; Taylor and Karner, 1983; Uyeda, 1986; Hall, 2001). In literature, the North Fiji zone has been classified as a plateau, probably due to its not very deep bathymetry (e.g. Taylor and Karner, 1983), but geophysical and geological data have pointed out the recent (Plio-Quaternary) thinning of this basin (e.g., Hamburger and Isacks, 1987; Auzende et al., 1988, 1995; Honza,1995 and references therein).
\fFigure 8 full
Tectonic setting and kinematic pattern of the Tonga-Kermadec-Fiji zone
Present tectonic setting and kinematic pattern of the Tonga-Kermadec-Fiji zone (Taylor and Karner, 1983; Pelletier et al., 1998; Hall, 2001). Symbols as in Fig. 7. The numbers close to converging and diverging little arrows indicate relative plate motion rates (cm/yr) at consuming boundaries and extensional troughs, deduced by geological, seismological and geodetic data (Pelletier et al., 1998) NFFZ = North Fiji fault zone. (For enlargement)
A number of authors (e.g. Taylor and Karner 1983; Hall, 2001) pointed out the difficulty of explaining the time-space distribution of the observed deformation in the above basins and surrounding zones as an effect of subduction-related processes. In particular, it is not clear why around the upper Miocene the New Hebrides arc started a clockwise rotation (which separated it from the Melanesian consuming boundary, also known as Vitiaz trench) up to reach its present position (Hall, 2001). This arc migration cannot be certainly associated with a subduction process, since no lithosphere consumption beneath the New Hebrides arc occurred before the onset of its rotation.
The tentative evolutionary reconstruction illustrated in Fig. 9 suggests that the generation of both the North Fiji and the Lau-Havre extensional zones might be connected with the deformation of arcs driven by plate convergence, in line with the extrusion mechanism. Fig. 9a shows the presumed tectonic setting of the zone considered in the middle-upper Miocene, mostly taken from the paleogeographic reconstructions proposed by Hamburger and Isacks (1987), Little and Roberts (1997) and Hall (2001). At this evolutionary stage, the subduction of the Pacific lithosphere at the Melanesian consuming boundary was building up an arc formed by accretionary material and volcanic products , which may be actually recognized in the Solomon-New Hebrides-Fiji-Tonga-Kermadec ridges, as suggested by Hall (2001). During this phase, the convergence between the Lord Howe and Chatham plateaux, two buoyant zones of the Australian and Pacific domains respectively, was accommodated by the consumption of the low buoyancy lithosphere comprised between them (Little and Roberts, 1997).
Around the upper Miocene (Fig. 9b), the incipient collision between the above domains induced an acceleration of the northward motion of the Australian plate, as suggested by the kinematic reconstruction of Gordon and Jurdy (1986). This event determined the onset of a strong deformation and disruption of the New Hebrides-Fiji-Tonga-Kermadec arc.
In the first stage (Fig. 9b), the central part of this arc (Fiji-New Hebrides) underwent a lateral extrusion (southwestward), at the expense of the low buoyancy lithosphere lying in front of it (New Hebrides basin), which caused the separation of the arc from the trench zone (Vitiaz) and the consequent formation of the North Fiji basin. The proposed deformation pattern of the Fiji-New Hebrides arc involved a double bending, which determined its horizontal delamination and the flexural separation of its inner sector (Fiji-Lau ridge) from two lateral slats (southernmost part of the New Hebrides ridge, on one side, and Tonga ridge, on the other side). Extensional deformation occurred in the zones of separation between the inner Fiji-Lau ridge and the two lateral slats (Fig. 9b), triggering the generation of the Lau sphenocasm. Evidence of Late Miocene extension in the other sphenocasm, which opened up between the southernmost New Hebrides arc and the Fiji ridge, is provided by magnetic lineations (Honza, 1995).
In the second stage (Fig. 9c), the Fiji-Tonga-Kermadec arc moved roughly eastward (under the push of the Australian block) and detached from the New Hebrides arc. This divergence was responsible for the opening of troughs in the southern part of the North Fiji basin and for the evident counterclockwise torsion of the Fiji segment of the arc. The separation between the Lau-Fiji and Kermadec ridges has then continued, with the consequent opening of the Havre trough (Fig. 9d). This divergence might also be a consequence of the fast lateral escape of the Tonga ridge, guided by the North Fiji fracture zone. A significant role in this extrusion mechanism might be played by the presence of a subducted lithospheric body under the Vitiaz trench. In particular, the shallowest part of this slab could represent an obstacle against the northward motion of the Tonga wedge, which, consequently, could prefer to extrude eastward ,at the expense of the thin Pacific lithosphere. In this regard, it could be noted that seismological investigations (Fisher et al., 1991) indicate a severe slab contortion beneath the northernmost Tonga trench.
Of course, one must be aware that the proposed evolutionary pattern only represents a working hypothesis. However, it must be pointed out that it provides a possible coherent interpretation for the very complex space distribution and time succession of tectonic events in this zone.
The fact that this extensional zone has developed along the transpressional boundary between the South American and Antarctica plates, could induce to interpret this event as an effect of a pull apart mechanism. However, such hypothesis cannot easily account for the fact that the extensional rate observed in this basin (70 mm/y, Carlson and Melia, 1984; Barker, 1984) is higher than the relative transcurrent motion between the two confining plates (about 30 mm/y). This would suggest the presence of an additional driving force. Such force could be connected with the strong buckling of the South Sandwich arc (Barker, 1984; Royden, 1993b), which emphasized extensional activity in the Scotia basin, in agreement with the main concepts of the extrusion mechanism (Fig. 3).
This kind of mechanism has been recognized as responsible for crustal stretching in some circum-Pacific basins.
Crustal stretching in this basin (Fig. 7) has been interpreted as an effect of a pull apart mechanism developed along the transcurrent decoupling zone between the Pacific and North American Plates (e.g. Yogodzinski et al., 1993 and references therein). The boundary conditions which caused the beginning of the above strike-slip motion were created by an important change of the Pacific motion trend (Gripp and Gordon, 1990). After this change, the Pacific plate started an highly oblique convergence with the accretionary belt (Western Aleutians) lying along the North American margin. This induced a differentiated motion between the external (trenchward) and internal sectors of the western Aleutians, causing the opening of the Komandorsky basin. This mechanism stopped around 15 My, when the migrating arc collided with the North American continental domain at Kamchatka (Yogodzinski et al., 1993).
A number of authors suggested that the formation of this basin (Fig. 7) was a side effect of the collision of India against Eurasia and of the consequent lateral escape of Indochina (Tapponnier et al., 1982; Letouzey and Kimura, 1985; Kimura and Tamaki, 1986; Briais et al., 1993; Jolivet et al., 1994; Lee and Lawver, 1995; Chung et al., 1997). In particular, crustal extension developed in the above basin by a sort of pull apart mechanism along the Red River fault, in response to the forced separation between the Sunda block and China (Briais et al., 1993; Lee and Lawver, 1995). The end of crustal extension in the South China basin was caused by the continental collision between the Australian plate and the Sunda block, which stopped the separation of this latter microplate from China (Briais et al., 1993). During the opening of the South China basin, other troughs (as the Sulu basin) might have opened up inside the Sunda block, in response to the relative motion between microplates (e.g. Lee and Lawver, 1995).
Most authors (e.g. Uyeda and Kanamori, 1979; Taylor and Karner, 1983; Lee and Lawver, 1995) recognize that this extensional feature (Fig. 7) has developed along a leaky transform segment of the megashear zone (Andaman fault) between the Indo-Australian domain and the Sunda-Indochina block. This old shear zone acted as a western strike slip guide for the extrusion of the Indochina block (50-20 My, Tapponnier et al., 1986) in response to the indentation of the Indian plate. Then, the collision of Indochina with the Sunda land and Australian blocks caused the stop of the above extrusion process. After this event, the Andaman fault system, recently prolonged through the Sumatra zone (Sumatra fault), reactivated, due to the lateral escape of the Sumatra forearc sliver plate (Fig. 7), as an effect of the oblique convergence with the Indo-Australian plate (Lee and Lawver, 1995).
Several authors (e.g. Uyeda and Kanamori, 1979; Taylor and Karner, 1983; Eguchi et al., 1989; Taylor et al., 1991; Lee and Lawver, 1995; Tregoning et al., 1998) suggested that the extensional tectonics observed in the Manus basin, inside the Bismarck sea, has developed by a pull apart mechanism, along the left lateral mega-shear zone between the New Guinea (Australian plate) and the Ontong-Java and Caroline ridges of the Pacific domain (Fig. 7). This interpretation seems to be reasonable, since extension occurs along a releasing sector of a well recognized strike-slip fault system. However, it cannot account for the fact that the extensional rates inferred for the Manus basin from magnetic anomaly data (130 mm/yr, Taylor, 1979) and geodetic data (141 mm/yr, Tregoning et al., 1998) are much higher than the relative motion rate (70-100 mm/yr) between the confining plates (Figs.7 and 10). To overcome this difficulty, one could consider that the New Britain arc, being squeezed between the Ontong-Java and the Australia (New Guinea) blocks, is undergoing a southward extrusion accompanied by clockwise rotation, as indicated by geodetic data (Fig. 10). This process might contribute to emphasize the extensional deformation at the releasing sectors of the Bismarck fault system, providing, thus, a possible explanation of the observed extensional rates.