Two contrasting hypotheses for the origin of the ocean/continent transition beneath the southern Iberia Abyssal Plain are proposed in this paper. The first, presented by Sawyer (1994) and as hypothesis 1 in this paper, suggests that the ocean/continent transition is composed largely of very slow-spreading oceanic crust. The second, presented by Whitmarsh and Miles (1994, 1995) and as hypothesis 2 in this paper, suggests that the ocean/continent transition was formed by tectonic and magmatic disruption of continental crust. Two further papers in this volume, which discuss extensional processes during the formation of the ocean/continent transition, are also relevant here. One, presented by Krawczyk and Reston (in Krawczyk et al., this volume), suggests that the ocean/continent transition exposes progressively from east to west, upper continental crust, lower crust, and upper mantle adjacent to a major detachment fault. The other, presented by Beslier et al. (this volume), suggests that the ocean/continent transition was formed by widespread exposure of the upper mantle due to detachment faulting during rifting, followed by mechanical extension of the exposed mantle at the end of rifting.
An important result of the leg, which is not explicitly mentioned in the results of individual sites, is that in-place basement rocks of unequivocal continental crustal character were not recovered during the leg. Whereas there are a few small examples of arkose and mica schist (suggesting a continental crustal source) in debris-flow deposits over basement, the rocks recovered during Leg 149 that bear on the issue of crustal type are almost entirely serpentinized peridotite, gabbro, and bits of basalt. A few of the gabbro and basalt clasts in debris flows at Site 899 are described as having transitional to alkaline affinities, but the vast majority of the mafic rocks recovered show no trace of a continental origin. The simplest interpretation of this observation is that there is little, if any, continental basement around. Starting with this premise, we have tried to develop the hypothesis that all or most of the Leg 149 transect is underlain by oceanic crust. This hypothesis is contrary to the geophysical model upon which Leg 149 was based. We will deal first with the petrological data from basement rocks at Sites 900, 897, and 899. We will then look at the existing geophysical data. We find that all of the existing data, with the possible exception of magnetic anomaly data, are consistent with this hypothesis.
The Site 900 metagabbro seems to have been formed as a cumulate rock from a MORB melt (Seifert et al., this volume). Rocks like these form most commonly as middle to lower oceanic crust at mid- ocean ridges. The Nd isotope evidence for a seafloor spreading origin for the rocks is very strong (Seifert et al., this volume). This makes their emplacement through a screen of extended continental crust or as a continental crust underplate possible, but perhaps less likely. Cornen et al. (this volume) emphasize slight deviations of the rock from that which would be emplaced at a mid-ocean ridge, mentioning slight transitional and alkaline affinities and comparing the rock to those of the Galicia Bank margin that have been interpreted to have been produced by a MORB source interacting with continental mantle. The latter comparison seems at odds with the Nd isotope results from Site 900 (Seifert et al., this volume). The Site 900 rock is basically oceanic in its petrology.
The drilling at Sites 897 and 899 indicates that we are dealing with multiple exposures of peridotite at the seafloor. Although basement at and around Site 899 has not been adequately imaged, the existing seismic data (Sawyer, Whitmarsh, Klaus, et al., 1994) do suggest that Site 899 is located over an isolated basement high. Although we did not drill in-place peridotite basement, we did drill through landslide or mass-flow deposits of virtually 100% peridotite containing large boulders. Gibson et al. (this volume) interpret this to mean an outcrop of peridotite basement must have been nearby. That Site 899 is located at the edge of an isolated positive magnetic anomaly (Miles et al., this volume), probably signifies that the peridotite exposure near Site 899 is local and does not parallel the margin. It does not, however, change the conclusion that, contrary to prior interpretations, peridotite was exposed at the seafloor away from the peridotite ridge, and an adequate model for the formation of this margin must take it into account. It is also significant that the peridotites recovered at both sites are petrologically similar (Cornen et al., this volume), suggesting that they were derived and emplaced at the seafloor from similar sources and by a similar sequence of pressure-temperature conditions.
If the Leg 149 transect is underlain by oceanic crust, then the spreading rate can be constrained to average 6.8 mm/yr half rate. This is based on (1) a seafloor-spreading interpretation of magnetic anomalies that dates the oceanic crust immediately west of the peridotite ridge at about 127 Ma (Whitmarsh et al., this volume), (2) the Tithonian (about 146 Ma) paleontological date for the deepest sediments drilled at Site 901, and (3) the distance between the two locations, 130 km. The rate is an upper bound because the crust at Site 901 had to have been formed prior to the age of the deepest sediment drilled. How much before is unknown because we were unable to core to basement at Site 901. Given the thickness of sediment below the terminal depth of Hole 901 A, we suspect that a rate of about 5 mm/yr is probably more likely. Seafloor spreading at half rates under 10 mm/yr is considered very slow. Seafloor spreading at rates between 10 and about 70 mm/yr is considered slow. Seafloor spreading at rates above about 70 mm/yr is considered fast.
In the following paragraphs we will summarize the characteristics of very slow spreading oceanic crust. This summary is based primarily on Cannat (1993), who studied the emplacement of mantle rocks at the seafloor at mid-ocean ridges globally, and Srivastava and Keen (1995), who studied the extinct mid-ocean ridge in the Labrador Sea. These environments are unambiguously oceanic and are at the slow-spreading end of the spectrum.
Slow-spreading oceanic crust is characterized by exposures of mantle peridotite and gabbro, in addition to basalt, at the seafloor (Cannat, 1993). Such exposures have been identified at slow or very slow spreading ridges in the North, Equatorial, and South Atlantic spreading centers, and at the Southwest Indian Ridge, the Cayman Trough, and the Galapagos rift. Cannat (1993) identifies the critical characteristic to be the lack of adequate magma rising from the asthenosphere to fill the gap between the spreading plates. Bown and White (1994) have shown that when seafloor spreading occurs faster than about 10 mm/yr half rate, there is adequate magma produced by decompression melting of the asthenosphere to form a normal thickness oceanic crust. In this spreading regime, the crustal thickness is remarkably uniform and independent of spreading rate, because the amount of melt is proportional to the space created by spreading. When spreading slows below about 10 mm/yr half rate, significant cooling of the rising magma can occur and the amount of melt becomes less than proportional to the amount of spreading. This creates the magma-starved seafloor spreading environment described by Cannat (1993; Fig. 6). Some of the divergent relative motion between the spreading oceanic plates is accommodated by the intrusion and extrusion of rising melt. The melt cools to form gabbro pods and a thin basalt upper crust. In normal oceanic spreading, the gabbro would be intruded into a screen of older gabbro. In the magma starved ridge model, the gabbro is often intruded into a screen of mantle peridotite. The remaining divergent relative motion between the spreading oceanic plates is accommodated by mechanical extension of the oceanic lithosphere. This means that there will be a regional tendency to passively uplift gabbro and upper mantle rocks. This extension process also faults and refaults the portion of the magma starved lithosphere above the brittle-ductile transition. In Cannat's (1993) model, this produces rotated fault blocks reminiscent of continental rifting. This faulting will often exhume gabbro and upper mantle peridotite exposing it at the seafloor. These rocks will show indications of brittle and/or ductile shearing that occurred during their uplift and faulting induced exhumation. Because of the slow spreading rate, there will be significant conductive cooling of the lithosphere and upwelling asthenosphere. This means that there will be a thicker axial lithosphere than at a faster spreading center and the brittle/ductile transition will be deeper. Thus we would expect to see evidence for brittle shearing at depths normally found more commonly in continental rifts. We also would expect to find that rising melt could cool and solidify at depths greater than typical for normal spreading rate ridges. The mantle rocks exposed at a slow spreading ridge are expected to be relatively undepleted because less melt was extracted during their decompression than at a normal ridge.
Many of the Leg 149 drilling observations are consistent with this model. In this model, it is no surprise that we would observe serpentinized peridotite exposures at Site 899 as well as Site 897. If the transect to the west of Site 901 were underlain by slow spreading oceanic crust, then we might expect a rather random distribution of peridotite, gabbro and basalt exposures in the basement. Some exposures might take the form of margin-parallel ridges, while others might not. In this hypothesis the peridotite ridge has little or no special significance. It no longer necessarily forms a boundary between oceanic crust and extended continental crust, but is just one of perhaps many seafloor peridotite exposures within slow spreading oceanic crust. The peridotite ridge continues to be an interesting feature, however, because no similar features have been observed in slow spreading oceanic crust elsewhere. The peridotites at Sites 897 and 899 were emplaced at shallow levels in a relatively cool thermal field that did not allow significant partial melting (Cornen et al., this volume). This is quite consistent with the slow seafloor spreading hypothesis. The gabbro at Site 900 is also consistent with this model. If formed at a very slow spreading center, the Site 900 rocks could have been emplaced as a gabbro body in a country rock of peridotite and gabbro at a depth of about 13 km. It was subsequently brought upward in a ductile shear zone (Event 1) producing the "flaser" texture and remained at a shallower depth to acquire (Event 2) its lower amphibolite to greenschist facies metamorphic grade, and then brought to the seafloor by block faulting (Event 3) at 136.4 0.3 Ma. It remained exposed at the seafloor, subject to mass wasting (Event 4), for tens of millions of years before it was buried by continent-derived sediment.
Very slow spreading oceanic basement is typically rougher than normal oceanic crust (Cannat, 1993; Srivastava and Keen, 1995). It is rough because it has been mechanically extended and the extension has been accommodated by normal faulting and block rotation. Srivastava and Keen (1995) interpreted a seismic reflection profile across the now extinct spreading center in the central Labrador Sea. The youngest oceanic crust (Fig. 7) was formed at a minimum half rate of about 3 mm/yr, has a highly faulted basement surface, appears to have been mechanically extended by about 70%, and is about 3 km thick (Srivastava and Roest, 1995; Osier and Louden, 1992, 1995). The older oceanic crust at each end of the line shown in Figure 7 was formed at about 10 mm/yr half rate, has a relatively unfaulted basement surface, appears to have been mechanically extended by only about 15%, and is about 6 km thick. Since the hypothetical spreading rate for the oceanic crust under the Iberia Abyssal Plain transect is about 5 mm/yr, the seismic profile of the 3 mm/yr unequivocal oceanic crust of the central Labrador Sea should give a good idea of the kind of structures that should be observed.
We find that many features of the Iberia Abyssal Plain transect (Fig. 3, Fig. 8; Krawczyk et al., this volume; Whitmarsh et al., this volume), particularly the region between Site 901 and Site 898 along Profile LG-12 (Beslier, this volume), are similar to those seen in Figure 7. The basement in both areas consists of normal-fault-bounded, rotated, blocks of crust. The basement relief in both areas is about 1.0-1.5 s two-way traveltime. The larger blocks in both areas have a typical width of 8-12 km. Most of the faults dip toward the extinct spreading axis in the Labrador Sea and, on Profile LG-12 in the Iberia Abyssal Plain, dip toward the west. We initially argued that the presence of rotated fault blocks was strong evidence for extended continental crust. The Labrador Sea analog draws that argument into question by showing that rotated fault blocks are also a characteristic of very slow spreading oceanic crust. There are also regions of the Iberia ocean/continent transition, particularly to the east of the peridotite ridge, where block faulting is not obvious because the basement is relatively smooth (Fig. 8). In the context of very slow seafloor spreading, these regions are probably areas that have been affected by less mechanical extension and more vigorous magmatism.
Very slow spreading oceanic crust is thinner than normal oceanic crust, may not have normal oceanic crust seismic velocities, and may not have a well-developed Moho (Cannat, 1993). The characteristic seismic velocity structure of oceanic crust is produced by the layering of basalt over gabbro over the upper mantle. At magma-starved ridges, this layering does not develop in the normal way. Because upper mantle rocks are brought to the surface in some places, and form a screen for the basalt and gabbro elsewhere, there are not likely to be clear layers 2 and 3. Seismic velocities in the crust may be laterally variable and are unlikely to match those of normal oceanic crust. There will probably not be a clear Moho because the base of the crust is characterized, not by a fairly sharp boundary between gabbro and peridotite layers, but by a gradual upward increase in the ratio of gabbro to peridotite (Fig. 6). Because the magma supply is low and much divergent motion is accommodated by rifting and thinning the already formed oceanic crust, where you can define a Moho, the crustal thickness is likely to be less than that of normal oceanic crust (Bown and White, 1994).
Osier and Louden (1992) reported seismic refraction results from the region of the reflection data shown in Figure 7. Their refraction data in the region of 10 mm/yr seafloor spreading (the crust just older than that shown in Fig. 7) show crust 4-6 km thick, only slightly thinner than normal oceanic crust (7 km; White et al., 1992). In this crust there are clear layers 2 and 3 with velocities of 5.1-6.2 km/s and 6.9-7.6 km/s respectively. Their refraction data from the region of 3 mm/yr spreading (Fig. 7) suggest the presence of thin (3-4 km), low-velocity crust (3.6-4.5 km/s) and a low-velocity upper mantle (with an average velocity of 7.7 km/s). Srivastava and Keen (1995) noted that the layer thicknesses and velocities are highly variable and that upper mantle velocities as low as 7.1 km/s were observed in the slowest spreading region. This kind of variability is quite consistent with the description of very slow spreading oceanic crust by Cannat (1993).
The seismic refraction data along the Leg 149 transect (Whitmarsh et al., 1990; Fig. 3B) show many of the characteristics of very slow spreading oceanic crust. The crust in Figure 3B identified as transitional crust varies from 2.5 to 4 km in thickness. This is thinner than normal oceanic crust and thinner than most unequivocal extended continental crust. Conversely, it is in the range of thicknesses observed in very slow spreading oceanic crust. There is a layer of low velocity upper mantle (7.4-7.55 km/s along Whitmarsh et al.'s Line 2) under the area of disputed crust type interpreted to be serpentinized peridotite (Whitmarsh et al., 1990; Fig. 3B). A similar layer is seen in the refraction data from the very slow spreading oceanic crust in the Labrador Sea. The refraction data west of the peridotite ridge (Whitmarsh et al., 1990; Fig. 3B) also show a layer of low velocity upper mantle. The crust there is 3 km thick, the crust has velocity 4.0-5.5 km/s, and the upper mantle, also interpreted to be serpentinized peridotite, has velocity 7.0-7.6 km/s. On the basis of these refraction data, the crust in both areas seems to have the thickness predicted for, and observed in, very slow spreading oceanic crust.
The magnetic observations in the Iberia Abyssal Plain are not as neatly explained by the very slow spreading oceanic crust hypothesis as by the second hypothesis presented below. Effective magnetic stripes are probably not created during very slow (less than 5 mm/yr, see below) seafloor spreading because the addition of magma to the crust is discontinuous and irregular in space. That the anomalies in the region between Sites 901 and 897 are of low amplitude and lineated is generally consistent with the very slow spreading oceanic crust hypothesis. However, Whitmarsh and Miles (1995) were unable to match observations within the ocean/continent transition using a magnetic model using very slow spreading rates. This is a problem for this hypothesis because very slow spreading anomalies have been successfully modeled elsewhere (southern Australia margin at 5 mm/ yr, Mutter et al., 1985; southern Red Sea at 5-11 mm/yr, Roeser, 1975; Miller et al., 1985; southern Aegir Ridge in the Norwegian Sea at 5 mm/yr, Nunns, 1983).
Very slow spreading oceanic crust in the Iberia Abyssal Plain is probably bounded in the east by extended continental crust under Site 901. The Site 901 basement block is constrained to be continental because we do not think that very slow spreading ocean crust would form at 320 m water depth. We do not see equally strong evidence that any basement west of the Site 901 high is continental. We have not yet explored the plate tectonic reconstruction implications of a >130-km-wide region of ultraslow oceanic crust under the Iberia Abyssal Plain.
Very slow spreading oceanic crust in the Iberia Abyssal Plain is probably bounded in the west by a spreading rate change several tens of km west of the J anomaly. The peridotite ridge is often interpreted to be an important crust type boundary and would be a natural place to argue for an increase in oceanic spreading rate. Whitmarsh et al. (this volume) have interpreted magnetic anomaly data between the peridotite ridge and the J anomaly to indicate seafloor spreading at 10 mm/yr half rate. Before accepting this conclusion, we should look carefully at the oceanic crust to the west of the peridotite ridge. Constraint on the structure of the oceanic crust to the west of the peridotite ridge comes from seismic refraction data (Whitmarsh et al., 1990) and seismic reflection profiles. Seismic refraction line L3 was the basis of the portion of Figure 3B just west of the peridotite ridge. Whitmarsh et al. (1990) found anomalously thin oceanic crust with low velocity upper mantle that they interpreted as serpentinized peridotite. This could also be interpreted as very slow spreading oceanic crust (thin crust with a low velocity upper mantle; Srivastava and Keen, 1995). In Figure 3B the thin crust abuts normal thickness crust to the west of the refraction profile, but the location is constrained using only gravity data (Whitmarsh et al. 1993). The seismic reflection profile Sonne-16 extends westward from the peridotite ridge at Site 897. There it crosses two more basement highs with relief of about 1-1.5 s two-way traveltime and spacing of 15-20 km peak to peak (Fig. 9). This is greater relief and a shorter wavelength than observed by Srivastava and Keen (1995) in the Labrador Sea in areas where spreading rate was faster (10 mm/yr half rate). We suggest that there may be a strip of oceanic crust west of the peridotite ridge that was produced by very slow spreading. The rate may have increased gradually to normal Atlantic spreading rates, or there may be an abrupt boundary. A good possibility for the timing of this acceleration is the time at which continental breakup is estimated to have occurred, and hence seafloor spreading to have begun, on the Galicia Bank margin to the north (about 112 Ma).
Following the logic of this hypothesis, we find no equivalently compelling reasons to assert that any crust to the west of the Site 901 basement high is extended continental crust. Krawczyk et al. (this volume) argue that the seismic reflection data show evidence for detachment faulting in the Site 900 high. We believe that this observation is equally consistent with the very slow seafloor-spreading hypothesis in that there was undoubtedly a large amount of mechanical extension of the newly formed oceanic crust. The extension was probably accommodated by the same kinds of faulting/shearing as is found in continental crust.
Beslier et al. (this volume) present a model for the formation of the Iberia Abyssal Plain margin that is somewhat similar to our very slow seafloor-spreading hypothesis. As in this hypothesis, they argue that the Site 901 high is probably the western edge of extended continental crust. In their model, simple shearing late in the rifting process exposed a broad region of ultramafic and mafic rocks at the sea- floor to the west of the Site 901 basement high. The region of denuded upper mantle was then mechanically extended to its current width of about 150 km. They suggest that the denuded mantle was stretched during the last stages of continental rifting offshore Galicia Bank. Our hypothesis differs only (1) in explaining how the mantle is denuded in the first place (we do not invoke regional simple shear) and (2) in the degree of melt generation (we call for some melt to make the Site 900 gabbro and the various gabbros and basalts found in mass-flow deposits). The hypotheses agree as to the location of the edge of extended continental crust, the presence of peridotite in a region rather than confined to a single ridge, the importance of mechanical extension and faulting within the lithosphere to the west of the edge of continental crust, and the possible temporal association of the extension landward of the continental edge in the Iberia Abyssal Plain with the continuing rifting on Galicia Bank.
Whitmarsh and Miles (1995) proposed a model to explain geophysical observations over the west Iberia ocean/continent transition that involves a propagating rift which, by means of faulting of the continental crust followed by magmatism (intrusive and possibly extrusive) and eventual breakup, leads to the ocean/continent transition as we find it today. They recognized three zones of magnetic anomalies on a chart of the west Iberia Margin (Fig. 10). West of the peridotite ridge (evidence for the existence and continuity of this ridge is presented below) they recognized, and modeled, M-series seafloor spreading anomalies. East of the ridge they found linear 015°-trending magnetic anomalies in the ocean/continent transition; these anomalies parallel the seafloor-spreading anomalies further west but could not be explained using a constant spreading rate and the Mesozoic geomagnetic reversal timescale. A major premise of their model is that these anomalies are the result of synrift intrusions in the lower continental crust under the same stress regime that later determined the direction and nature of faulting within the oceanic crust. Further east, east of 11°15'W, the more elongate linear anomalies trend roughly east of south (about 165°) but in addition there are other shorter features with a variety of trends. The anomalies in this easternmost zone, which includes Site 901 and other tilted fault blocks that extend landwards beneath the continental slope, and which is assumed to consist of stretched continental crust, are less well organized and the east-of-south trends differ from those in the central zone by about 30°. The Whitmarsh and Miles (1995) model in Figure 11 and Figure 12 is presented in the form of a sketch that stresses the propagation of a continental rift and the implied transition from the almost amagmatic rifting of a nonvolcanic margin to the predominantly magmatic process of seafloor spreading itself at a half-rate of about 10 mm/yr. In the model the semicontinuous peridotite ridge found off the northwest part of the west Iberia Margin is attributed to tectonic exposure, by a so-far unexplained mechanism, of upper mantle material along the line of continental breakup.
We emphasize the evidence for an almost continuous linear peridotite ridge off west Iberia and its relationship to the ocean/continent transition, oceanic crust, and continental crust. North of Gorringe Bank (Fig. 2), serpentinized peridotite has been dredged, sampled by submersible or drilled at seven locations off west Iberia between latitudes 40°46' and 43°N (Boillot et al., 1988a,b; Sawyer, Whitmarsh, Klaus, et al., 1994). Off Galicia Bank, the samples come from the relatively steep west and northwest flanks of the bank and from a narrow (10-20 km wide) ~000°-trending basement ridge traced for over 115 km in a north-south direction on seismic reflection profiles that cross the ridge about every 5 km (Thommeret et al., 1988).
Beslier et al. (1993) used other, less closely spaced, seismic profiles to propose that the above two ridges were part of a 300-km-long feature offset, principally sinistrally and in en echelon fashion, into three or four separate segments. In the southern Iberia Abyssal Plain, the ridge has been traced for over 40 km on seismic reflection profiles (Fig. 9). Here, even though the ridge is directly sampled only at Site 897, its recognition is assisted by its general association with overlying folded sediments; it appears that the peridotite ridge acted as a weak zone during the Miocene northwest-southeast compressional episode (Masson et al., 1994).
Added evidence for the significance of the west Iberia peridotite ridge comes from other geophysical observations and modeling. First, based on modeling deep-tow magnetic profiles a very clear contrast exists between the bulk magnetization of the crust either side of the peridotite ridge both off Galicia Bank (Sibuet et al., 1995) and in the southern Iberia Abyssal Plain, although less abruptly there because the deep-tow profile crosses the strongly magnetic, and possibly unique, Site 899 basement high (Whitmarsh et al., this volume). In both cases the landward crust possesses a significantly weaker magnetization. Second, seismic velocity structures either side of the peridotite ridge also appear to differ significantly (Whitmarsh et al., 1990; Horsefield, 1992; Whitmarsh et al., work in progress) and have been interpreted as thin oceanic crust and ocean/continent transition crust. The thin oceanic crust is explained by a poor magma supply caused by the cooling effect of continental lithosphere (Whitmarsh et al., 1993) or by seafloor spreading that has immediately followed prolonged continental rifting, which has the same effect (Bown and White, 1995). Once continental rifting has ceased, seafloor spreading rapidly (in a few m.y.) reaches a steady state and, provided it exceeds 7.5 mm/yr half-rate (Bown and White, 1994), normal thickness oceanic crust is produced. Last, in the southern Iberia Abyssal Plain there is a distinct change in the form of acoustic basement either side of the peridotite ridge. To the west the basement consists of parallel ridges and troughs trending slightly east of north whereas to the east it consists of more or less isolated highs that have a weak northerly trend and that rarely have a length-to-width ratio of more than 3:1 (Fig. 9 and unpubl. data collected summer, 1995).
Irrespective of the origin of the ocean/continent transition crust, there is compelling evidence from independent observations of its magnetization, seismic velocity structure, and upper surface that it differs significantly from the thin oceanic crust further west and that, wherever measurements have been made off west Iberia, the peridotite ridge itself is closely associated with the boundary between the two crustal types.
Although there is evidence for a more or less continuous margin-parallel linear peridotite ridge from about 40° to 43°N, we note that the west Iberia Margin is possibly unique in possessing such a ridge of peridotite; no such ridge has yet been convincingly demonstrated on the conjugate Newfoundland margin. Peridotite has been sampled off other rifted margins in the approximate vicinity of the ocean/continent transition (Nicholls et al., 1981; Bonatti et al., 1986) but only rarely in circumstances that could be described as similar to west Iberia (Site 651 in the Tyrrhenian Sea, Kastens et al., 1988). This may simply be because on no other rifted margin is the ridge so accessible and/or the margin so thoroughly surveyed.
In spite of the strong evidence for the peridotite ridge, it is difficult to explain how it was emplaced. An explanation in terms of a major low-angle detachment fault, such as has been postulated beneath Galicia Bank (Boillot et al., 1987), has yet to receive support from reflection profiles across Site 897. Whitmarsh and Miles (1995) invoked tectonic emplacement at continental breakup by an "as yet unknown mechanism." S. Pickup (pers. comm., 1995) has recognized a 30° landward-dipping reflector beneath the southern Iberia Abyssal Plain, south of the Leg 149 transect, which intersects the east side of the peridotite ridge and can be traced to a depth of at least 12 km. It is possible that the peridotite ridge is associated with the final breakup of continental crust immediately preceding the onset of seafloor spreading but we cannot at present explain its continuity or long narrow ridge-like character. The evidence of limited melting of the peridotite implies that emplacement was slow (to inhibit significant adiabatic melting) and/or affected by the adjacent cold continental lithosphere.
We now consider how the results of Leg 149 fit hypothesis 2. Site 897 was drilled on the crest of the peridotite ridge and provided the first direct evidence that this ridge exists within the southern Iberia Abyssal Plain. The peridotite cores from this site, and the cores from Sites 637 (drilled on the peridotite ridge off Galicia Bank), all indicate a very similar early pressure-temperature history (crystallization at up to 30 km depth and 1200°C and ductile deformation at around 1000°C) and hence, presumably, point to a similar mode of emplacement of these rocks. Beslier et al. (this volume) describe the emplacement as happening during lithospheric stretching as adiabatic uplift of a mantle dome which was subsequently tectonically exposed at the seafloor. This view is entirely consistent with the Whitmarsh and Miles (1995) model (Fig. 12). The greater mylonitization of the Site 637 cores may simply be the chance result of the borehole intersecting a zone which had experienced extensive shearing while Site 897 did not.
Site 899 was unusual in that it penetrated a subcircular basement high that, on the basis of a surface magnetic anomaly chart of the west Iberia Margin (Miles et al., this volume, Fig. 1), may be in a location that is unique within the Iberia Abyssal Plain ocean/continent transition; the chart shows the site to lie on the west flank of a north-northwestsouth-southeast isolated steep-sided positive magnetic anomaly of over 125 nT amplitude, quite different from the generally lower amplitude and more gently sloping anomalies elsewhere in the ocean/continent transition. A similar conclusion emerges when a deep-towed magnetometer profile across the site is studied; this shows an 8-km-wide ~600 nT positive anomaly over Site 899, which is far larger than the weaker anomalies immediately to the east (Whitmarsh et al., this volume). Modeling of the deep-tow profile indicates that the bulk magnetization of the crust is unusually high, higher than oceanic crust, in the vicinity of Site 899 and this is supported by measurements on the cores (Zhao, this volume; Whitmarsh et al., this volume) which indicate that the Site 899 peridotites are on average five times more strongly magnetized than the Site 897 peridotites. Since the amplitudes of magnetic anomalies are strongly correlated with the iron oxide mineralogy of the (near surface?) crustal rocks, which in this case appear to be ultramafic, we caution against extrapolating the results from this site to the ocean/continent transition in general and, in particular, against assuming that there is a continuous peridotite basement between Sites 897 and 899, which are 20 km apart. The question whether in fact the Site 899 cores are representative of the underlying basement, or of more distant location(s), is addressed later. If the underlying basement is ultramafic, one explanation could be that the site is located at the northern end of a peridotite ridge segment that overlaps, en echelon fashion, the segment on which Site 897 is situated (Beslier et al., 1993). However, we are unable to explain why the Site 899 basement should be more strongly magnetized and the strong magnetization may have an independent origin. For example, the greater lherzolite content implies more Fe and hence, possibly, more magnetic minerals. Nevertheless, a similar mode of peridotite emplacement to Sites 637 and 897 is indicated by a very similar early pressure-temperature history (crystallization at up to 30 km depth and 1200°C and ductile deformation at around 1000°C) and hence, presumably, to a similar mode of emplacement of these rocks.
The basement we drilled at Site 900 is massive flasered (sheared) cumulate gabbro with a primary mineralogy indicating metamorphism at at least 0.4 GPa (~13 km depth) and REE patterns characteristic of a transitional MORB parent magma (Cornen et al., this volume; Seifert et al., this volume) and Nd and Pb isotope ratios indicative of a MORB parent magma (Seifert et al., this volume). The gabbro has many compositional and textural similarities to oceanic cumulate gabbros (Seifert et al., this volume). While the Site 900 Nd isotope data are consistent with formation of the gabbro from such a parent magma, on the other hand, rocks generated from a single source generally have a narrow range, say 0.5, of epsilon Nd; the larger range found here (6.3-10.3) may be indicative of some mixing of sources (e.g., MORB and continental crust). Further, it is interesting that Schärer et al. (1995) studied a gabbro and an associated chlorite rock (dated as synrift) from the western margin of Galicia Bank that had small positive epsilon Nd values in the range 3.6 to 5.6. They concluded that the gabbro crystallized beneath stretched continental crust prior to continental breakup between Iberia and Newfoundland and was emplaced as synrift underplated material. Thus, a range of epsilon Nd values from 3.6 to 10.3 has been observed off west Iberia, suggesting perhaps a varying degree of continental contamination of the synrift melt along the margin. An absolute age of 134.6 Ma for the last thermal event experienced by the Site 900 gabbro indicates its tectonic deformation, and probable crystallization, during synrift time (Féraud et al., this volume). We therefore conclude that the Site 900 gabbro may also have been the product of synrift underplating.
Given that Site 897 was drilled on the peridotite ridge, and that Site 899, for reasons already described, appears to lie over crust that is probably atypical of the ocean/continent transition, only Site 900 can be considered to have sampled the basement of the ocean/continent transition itself. Therefore we have to look to Site 900 cores and to the allochthonous clasts found in debris-flow deposits at Site 897 (Subunit IIIB) and at Site 899 (Unit IV) to make some estimate of the nature of the crust exposed within the ocean/continent transition towards the end of rifting.
At Site 897 the conglomerates of Subunit IIIB yield rounded clasts of different lithologies, ages, and degrees of lithification, suggesting that many different lithologies cropped out on the seabed. The lithologies include claystone and chalk/limestone with minor dolomite, basalt, arkosic to lithic sandstones, and a sandstone with fragments of shallow-water fossils, mica schist, basalt, serpentine and mica (Shipboard Scientific Party, 1994a; Comas et al., this volume). At Site 899, Unit IV contains clasts of mafic rocks (submarine basalts, microgabbros, mafic clasts and amphibolite) which display transitional MORB to alkaline features.
Thus, quite a wide variety of igneous, and some metamorphic, rocks can be found as clasts in the coarser fraction of sedimentary units at Sites 897 and 899. Except for the Site 900 gabbro, the igneous rocks have transitional MORB to alkaline affinities and do not strongly support the idea of a widespread MORB-like oceanic crust in the ocean/continent transition. REE patterns of some basalt clasts are close to CFB, relatively free of contamination by continental lithosphere. Other clasts suggest the existence of a variety of prerift sedimentary rocks but it is impossible to say whether these cropped out on the continental shelf and slope or closer to the sites.
The lithologies present from the southern Iberia Abyssal Plain are rather different from the metamorphosed sediments and granitic rocks that have been sampled around Galicia Bank and the granitic rocks that crop out in the Berlenga-Farilhoes Islands near the head of the Nazaré Canyon (Fig. 1). Black et al. (1964) reported dredged limestones from Vigo Seamount and Galicia Bank but, on the basis of their shape and smooth striated surfaces, attributed many plutonic and metamorphic rocks to glacial erratics. Capdevila and Mougenot (1988) report dredged samples from around Galicia Bank of predominantly metamorphosed sediment (mica schist, gneiss, granulite, Paleozoic sediments(?), phyllite, and metaarkose) and a variety of granitic rocks (granodiorite, granite, tonalite, granophyre), but it is interesting that they do not mention a single sample of gabbro. Although it is not clear what criteria were used by Capdevila and Mougenot to confirm that these samples were not glacial erratics (Davies and Laughton, 1972 show that erratics occur as far south as 30°N in the northeast Atlantic) granite and granodiorite were also sampled, presumably in situ, by submersible at about 12°30'W (Boillot et al., 1988a). Therefore, it seems that the difference between the southern Iberia Abyssal Plain and Galicia Bank (principally a lack of granitic rocks) is real and is probably attributable to the different outcrop geology of the adjacent onshore regions where granite outcrops only north of about 41°N. We therefore conclude that the lack of granitic rocks in cores from the southern Iberia Abyssal Plain part of the west Iberia Margin does not necessarily signify a lack of continental crust there.
A final new piece of evidence is afforded by seismic reflection images of the crust and upper mantle of the west Iberia ocean/continent transition (Krawczyk et al., this volume). Prestack depth migration of the east-west LG-12 profile, which passes over or close to almost all the Leg 149 drill sites, enabled the identification of steepand low-angle normal faults (labeled H and L in Fig. 8) between Sites 900 and 901. Such signs of extensional tectonics are consistent with, but do not prove, tectonic disruption of continental crust. Of particular significance is the apparent location of Site 900 on the lower plate to an important detachment fault; this provides a means to explain the close to synrift exhumation and retrograde metamorphism of the Site 900 gabbro from around 13 km depth, where it underwent dynamic recrystallization, to the seafloor. Further west, as far as the peridotite ridge., the acoustic basement has a much smoother appearance, on this and other reflection profiles, with relief of under 1 s two-way traveltime. Here we take the smoother basement to be evidence of the growing oceanward importance of extrusive magmatism which lead to the eventual onset of 10-mm/yr seafloor spreading west of the peridotite ridge.
Hypothesis 2 can now be refined in the light of the new ODP data. The most striking new result is the discovery of an apparently synrift MORB-like cumulate gabbro within the ocean/continent transition in a location where continental crust was predicted. Although the original model predicted widespread intrusive and even extrusive material, in or over the continental crust, respectively, it now seems that underplating at the base of the continental crust, or even the emplacement of the gabbro as a layer trapped within the uppermost mantle, is also possible (Beslier et al., this volume). We do not know whether underplating is widespread within the ocean/continent transition but the sort of gabbro drilled at Site 900 could form in local pockets, akin to an aborted form of the punctiform initiation of seafloor spreading proposed by Bonatti (1985) for the northern Red Sea, just as well as in a widespread sheet. In other words, the first synrift magmatism may begin as a series of isolated patches of MORB-like products at the base of the continental crust. Eventually such patches may merge to form a continuous belt which is the forerunner of oceanic crust produced by seafloor spreading. The exhumation of cumulate gabbro from 13 km depth to the top of basement is simply explained by late-stage low-angle detachment faulting (implying significant horizontal extension), which acts at a smaller scale than envisaged in the original model. Why has underplating not been detected by seismic refraction lines? First, it may be discontinuous on the scale of ~80-km-long refraction lines, as suggested above, and therefore hard to detect. Second, even if the underplated gabbros should form a continuous sheet, the seismic refraction method is unlikely to be able to resolve layers at the base of the crust, or within the uppermost mantle, that are thinner than a few hundred meters, and we have no evidence at present that the gabbro is thicker than this.
The presence, within mass-wasting deposits, of basaltic and microgabbroic clasts with transitional to alkaline affinities is to be expected from melts that have passed through, or been contaminated by, continental lithosphere. The minor amounts of mica schist and arkose in the same deposits is evidence of continentally derived material. The absence of granitic rocks in the cores is inconclusive and may have a simple geological explanation linked to the distribution of onshore outcrops.
Whitmarsh and Miles (1995) supported their case for this type of model by reference to analogous, but better exposed, nonvolcanic rifted margin locations. The best active analogue is probably the northern Red Sea which is at least 1500 km from the Afar hotspot. Here Bonatti (1985) envisages a thinned and stretched continental crust injected by diffuse basaltic intrusions and he quotes examples of samples of "...basaltic/gabbro rocks...showing transitional or alkaline affinities" from the Brothers Islands at 26°15'N. Bonatti and Seyler (1987) report that on Zabargad Island at 23°40'N there is a peridotite-silicic gneiss-gabbroic association which is a sample of continental upper mantle/lower crust. The pressure-temperature history of the Zabargad gabbros and gneisses is very similar to that deduced from the Site 900 gabbros. Similarly Pautot et al. (1984) report a basalt from Charcot Deep at 25°15'N with alkaline/transitional affinities. The comparison with the results of Leg 149 is striking and suggests a fundamental similarity between the evolution of the northern Red Sea and the west Iberia Margins.
Finally, a principal factor which led to the original model was the apparent absence of seafloor-spreading anomalies within the west Iberia ocean/continent transition and the presence there of relatively low-amplitude and long wavelength isochron-parallel anomalies. The origin of these anomalies is still debatable but cumulate gabbros, if distributed with a predominant margin-parallel trend, could make an important contribution to such anomalies. The discovery of cumulate gabbro within the ocean/continent transition does not preclude the original model but has allowed for its refinement; the model will receive strongest support, however, should future basement drilling discover evidence of unequivocal in situ continental rocks within the ocean/continent transition.