The complexity of the basement reflections allows several interpretations. Here we present two possible interpretations, both consistent with all geological and geophysical observations (e.g., Pinheiro et al., this volume).
From the seismic data, we infer at least two phases of rifting (Fig. 10A). First a detachment fault (H) accommodated top-to-the-west motion, accompanied by the development of the small landward-thickening, landward-tilted, wedge-shaped fault blocks riding on H: the wedge-shaped blocks between Site 900 and basement high FB (shots 4025-4425). These wedge-shaped blocks are bound by westward-dipping normal-fault structures that are synthetic to and detach onto H (Fig. 11). We suggest that during progressive extension and exhumation of the lower plate to H, the unloading of the lower plate to H caused this to bow up, as inferred for detachment faults in the western United States (Lister and Davis, 1989). This may have rendered at least part of H inactive.
The detailed seismic image from prestack depth migration (Fig. 12) provides evidence for the presence of lower crustal material and H itself on top of the basement high of Site 900. H can clearly be followed from 9 km depth to the top of the fault block at 6 km depth, approaching the basement surface apparently slightly east of the drilled position, so that we consider it most likely that the lower plate to H was drilled. This is also indicated by the heavily sheared and fractured structure of the analyzed samples (Shipboard Scientific Party, 1994b).
The change of dip direction of structure H from west to east is however remarkable and hard to explain within a single phase of faulting. Instead we infer that H, and the overlying wedge-shaped blocks, were rotated during a later phase of faulting and extension. The most pronounced structure interpreted to have been active during this later phase is the bright L reflection, which cuts down from the easternmost tilted block (Site 901) and probably marks the lowermost boundary of the middle basement high (FB). On the depth migrated section (Fig. 8), L appears approximately planar down to 12 km depth, where it appears to flatten into a deeper level along which the continental block east of Site 900 moved to the west (Fig. 8). This subsequent faulting along L may even be active today, expressed by the 100-m-high fault scarp at Site 901. Other structures active at this phase of faulting may include approximately planar faults bounding the western flanks of basement highs FB and 900, although these structures are less clearly imaged. Together, these faults would have formed an array of oceanward-dipping faults, dissecting the original detachment system (H), and rotating the segments of that system landward.
Thus, we identify a detachment fault, H, accommodating top-to-the-west motion, overlain by two wedge-shaped "horses" (Gibbs, 1987), and itself back-rotated both during the unloading of the foot-wall that accompanied extension, and also by later steeper faulting. This leads us to suggest that various phases of faulting may have controlled continental breakup in the Iberia Abyssal Plain region.
This interpretation is supported by the results of Leg 149 (Sawyer, Whitmarsh, Klaus, et al., 1994; Fig. 2). Whereas Site 897 proved the existence of a peridotite ridge between oceanic crust and the western end of the transition zone between oceanic and continental crust, Site 901 drilled prerift or pretilting sediment of Tithonian age, capping conformably the most westward tilted crustal fault block. This block is interpreted as thinned continental crust, because the overlying sediment is approximately 15 m.y. older than the onset of seafloor spreading in this area (Whitmarsh and Miles, 1995; Miles et al., this volume). The third drilling location where basement was reached, Site 900, sampled mafic rocks (meta gabbros, sensu lato) strongly deformed at about 15 km depth under granulite facies conditions (Cornen et al., this volume) at 136 Ma (Féraud et al., this volume), that is 6 m.y. before final breakup at 130 Ma (Whitmarsh and Miles, 1995). Because of the depth and grade of metamorphism, we interpret these rocks as lower crust exhumed during the rifting process.
Thus, going from east to west, the basement consists of upper crust, lower crust, and upper mantle. Therefore, we conjecture that a cross-section through the upper lithosphere was exposed during the rifting process, probably by top-to-the-west motion along detachment H, and was subsequently dismembered by steeper normal faults.
The complex seismic image of the area west of Site 900 also offers different possibilities for interpretation (Fig. 13). Between lower crust (Site 900) and serpentinized mantle material (found at Site 897, 27 km west of Site 898), the major and controlling structure may be a westward dipping low-angle normal fault corresponding to the reflection F. It cuts down from Site 900 from 6 km depth towards the west below the basement high IAP-7 to 9 km depth (Fig. 13), and extends over at least 15 km along the profile. F may represent on the west flank of Site 900 the same structure as H does on the east flank. If so, the basement high at IAP-7 (see also Fig. 2) would be an upper-plate fault block to this master detachment, which accommodated extension by top-to-the-west motion.
As block-faulting or detachment structures are less pronounced, F may also have been later than H, perhaps coeval to L, in which case IAP-7 would be part of the footwall to H. Within this scope, the Moho may lie nearby, and mantle material could be found at shallow levels. Finally, the IAP-7 basement high could represent a volcanic structure formed by magmatism accompanying breakup on the Iberian continental margin. Nevertheless, this hypothesis seems rather unlikely because of the general absence of volcanism on this passive rifted margin and the local extent of this layer.
Our results are consistent with a simple model for the evolution of the Iberia Abyssal Plain (Fig. 14). Although an oversimplification and although other models (e.g., Masson and Miles, 1994; Whitmarsh and Miles, 1995; Miles et al., this volume) may be equally valid, this model provides the simplest explanation for the apparent deepening of lithospheric level found going from east to west across the drilled transect. We thus advance this model as a basis for further work and investigations of this margin. As lithospheric extension started, Newfoundland and Iberia moved apart along a large detachment, accommodating top-to-the-west simple-shear motion in the upper lithosphere. Asthenospheric upwelling and consequent decompression melting may have led to the local intrusion of melt into the lower crust. If such melting occurred beneath highly thinned lithosphere, it may be indistinguishable from that occurring at mid-ocean ridges.
Extension of the upper lithosphere along a top-to-the-west detachment would have brought lower crustal and mantle rocks close to the surface to the west (Wernicke, 1981). After the detachment had become inactive, block-faulting may have accommodated continued extension. This faulting would have dissected the lower plate and the detachment itself, and rotation of the blocks would have tilted this back towards the continent. As the block-faulting appears to have dissected the entire detachment system, we infer that mantle thinning (and hence lithospheric weakening) may have taken place over a broad zone during detachment faulting (Fig. 14), rather than being localized down-dip of the detachment. This in turn might be taken as an indication that no single lithosphere penetrating detachment developed, but rather that extension in the lower lithosphere may have been accommodated ductilely (Fig. 14) or along a system of extensional structures.
The model in Figure 14 is consistent with the drilling results (Sawyer, Whitmarsh, Klaus, et al., 1994). The transect between Sites 901, 900, and 898 represents a transition from upper continental crust to lower crust (perhaps synrift intrusion) to mantle, as predicted by the model. It predicts additionally a significant amount of mantle material between oceanic and continental crust, locally exposed at the seafloor during the last rift stage prior to continental breakup. Site 901 sampled continental material; Site 900 drilled a lower crustal tilted block, identified by its lithology (amphibolite facies gabbro, probably synrift); Sites 897 and 899 sampled mantle rocks.
The model (Fig. 14) is also broadly consistent with the structure of the Newfoundland Basin margin. Although Reid and Keen (1990) describe large east-cutting normal faults on this conjugate margin, these do not appear to cut down into the mantle, but rather detach at lower crustal levels. Furthermore, we note that the reported (Reid, 1994) crustal structure of the conjugate Newfoundland Basin margin (reconstruction of Malod and Mauffret, 1990) is consistent with a west-cutting detachment: landward of the Newfoundland Basin, the lower crust appears to be truncated by a west-dipping Moho beneath the top of the continental slope, and is absent beneath most of the continental slope and rise, as in an upper plate margin (Fig. 14). Thus the structure of the conjugate margin is also compatible with a west-cutting master detachment fault that exposed mantle rocks at the seafloor on the Iberian side.
As in the previous interpretation, H and L are considered to be features similar to the S reflector imaged on the Galicia Margin, i.e., as the seismic signature of tectonic contacts formed during extension. However, the seismic data do not fully constrain the relationship between H and L, thus allowing the two interpretations presented in Figure 10B-C. It is however clear that H cuts up to top basement just east of Site 900, implying that the whole area located between Site 900 and the oceanic crust (accreted more than 100 km to the west) is a tectonic window opened on deep lithospheric levels (mantle rocks now partially serpentinized and gabbros probably underplated during continental rifting). This interpretation is in good agreement with the petrostructural evolution of serpentinized peridotites and gabbros, both of which underwent a ductile shear deformation at high and decreasing temperature followed by a complex deformation in subsurface conditions (Beslier et al., 1994; this volume). This evolution likely occurred in extensional lithospheric shear zones which led to tectonic denudation and exposure of deep lithospheric levels at the seafloor.
However, in contrast to interpretation 1 (Fig. 10A), this second interpretation (Fig. 10B-C) postulates that the Iberia Abyssal Plain margin belongs entirely to the upper plate, the main detachment being rooted beneath Iberia (Fig. 10D; for more detailed discussion see Beslier and Brun, 1991; Boillot et al., in press; Beslier et al., in press). However, although the structure of the ocean/continent transition of the two margin segments (Galicia Bank and Iberia Abyssal Plain) is comparable, the ocean/continent transition is much wider beneath the Iberia Abyssal Plain (150 km) than on the western Galicia Margin (30 km). The complex late deformation and structure of the ocean/continent transition beneath the Iberia Abyssal Plain suggest that the mantle and associated mafic rocks can be stretched over a wide region after the breakup of continental crust is completed and before oceanic accretion has started.