) and crustal thickness are determined.
Seismic and borehole data usually enable three approaches understanding the tectonic history of a given area. The first is a qualitative assessment of seismic sections to determine whether any synrift reflection packages can be identified so that, once the equivalent sediments are sampled for biostratigraphic analysis, the timing of rifting can be established. The other two approaches are quantitative. One uses borehole data to undertake a backstripping analysis to compute basement subsidence: in the Leg 149 area, the incomplete stratigraphic record obtained over basement highs precludes this approach. The other quantitative method uses digitized seismic profiles to estimate total tectonic subsidence along the profiles, from which extension factors () and crustal thickness are determined.
Prosser (1993) proposed that distinct stages of rift evolution can be recognized, each with distinctive expressions on seismic profiles. She recognized four phases of development in rift basins: rift initiation, rift climax, immediate postrift, and late postrift. Her conclusions were based largely on studies of basins developing beneath relatively shallow seas, but included the Armorican Margin on the north side of the Bay of Biscay. She showed that footwall crests and hanging walls may be eroded in subaerial and shallow-marine realms and so contribute sediments to the deeper parts of the basin. In such basins, tectonic movements play a crucial role in creating—or destroying—accommodation space in which sediments may accumulate. However, in rift basins that remain in the submarine realm throughout their history, the crests of tilted fault blocks will not be subjected to subaerial erosion and so will not contribute significant amounts of sediment to adjacent basins, except by mass wasting processes. This means that sediment transport into the basin will be dominated by gravity-driven processes and pelagic/hemipelagic sedimentation. Little sediment will be derived locally by marine or subaerial erosion, but there may be gravity-driven resedimentation from both the hanging wall and footwall as rifting develops and the gradient of the slopes increases.
Rotation of fault blocks during rifting results in three features that may be resolvable on seismic reflection profiles. First, synrift reflection packages should show thickening toward footwalls, and, second, rotation may also result in later packages onlapping earlier ones toward the hanging wall. Some erosional truncation might occur between packages if sediments were transported toward the footwall by slumping or debris flows. Third, the crests of tilted fault blocks will show significant erosional features if they are uplifted above sea level: no such features are visible on seismic profiles in the Leg 149 area. Even the most elevated crest in the study area, the tilted fault block beneath Site 901 (Fig. 3C), which is capped by Tithonian clays and minor sandstones, shows no erosional truncation on a reprocessed version of part of LG12 shown as Figure 9 in Krawczyk et al. (this volume). This suggests that the rift system from Site 901 westward developed entirely in a submarine environment.
No seismic lines in the Leg 149 area show reflection divergence into steep slopes on top of the acoustic basement that are candidate fault scarps. The topography of the acoustic basement is draped by Units 6, 5, and 4, all of which show onlap and thinning toward basement slopes. Even adjacent to the extremely large fault scarp immediately to the west of Site 901 on line LG12 (Fig. 5C; Fig. 1 of Beslier, this volume; Fig. 3 of Krawczyk et al., this volume) there is no clear reflection divergence. The only possible exception occurs on the reprocessed version of LG12 (Krawczyk et al., this volume, their Fig. 3, Fig. 12) between SP 4300 and 4400 at about 8.5 ms TWT, where there is some indication of a reflection package thickening into a low-angle fault plane dipping westward. Beslier et al. (1995) also interpreted this part of the profile as a synrift interval, but we believe it is more likely the result of drape and compaction over an earlier fault-induced topography.
The absence of convincing examples of synrift reflection geometries may be due to very low sedimentation rates during the rift climax in deep-sea settings. The creation of a series of long hanging-wall slopes and footwall scarps will mean that turbidity and debris flows will be ponded back in more landward basins. Therefore, only the more dilute portions of turbidity currents are likely to flow through the maze of ridges and hollows created by rifting and deposit sediment in the deeper parts of deep-sea rift zones. During the rift-climax phase, sedimentation in such areas is most likely dominated by pelagic and hemipelagic settling. In such a situation, rifting and contemporaneous deposition would have to continue for several million years to produce seismically resolvable reflection packages showing divergence.
Our scrutiny of published lines from the Leg 103 area (Mauffret and Montadert, 1988; Winterer et al., 1988) and around Site 389 (Groupe Galice, 1979) revealed no convincing reflection divergence into faults indicating significant deposition during rifting. Therefore, we suggest that these earlier studies did not correctly identify the seismostratigraphic expression of the rifting episode. We agree with the view of Prosser (1993, p. 51) that ignoring the importance of infilling remnant topography in the creation of overall wedge-shaped geometries and the possibility of divergent reflector configurations induced by compaction can lead to overestimations in the duration of active faulting. The extent of such an overestimate in the Galicia Margin area is illustrated in Figure 12.
In the light of the discussion above, we conclude that rifting occurred prior to the deposition of sediments equivalent to our Unit 6, which is Hauterivian (and possibly Valanginian) to Aptian in age (Fig. 9, Fig. 10). The faint progradational reflection geometry in Subunit 6C, situated immediately over a possible fault scarp on line SO18 (Fig. 8), may be an immediate postrift feature in terms of Prosser's (1993) rift phases.
At Site 398, ill-defined curved reflection patterns in Unit 4 of Groupe Galice (1979) could indicate drape and compaction over fault-induced topography (Fig. 10). This suggests that this unit is probably postrift in origin, in which case the rifting must predate the Hauterivian age obtained from the oldest sediments drilled at Site 398 and postdate the Tithonian sediments drilled at Site 901. The timing of rifting can be better constrained in the deep Galicia Margin area. Here, Tithonian-Berriasian shelf carbonates drilled during Leg 103 are clearly prerift in origin and are identified as such on published seismic lines (Mauffret and Montadert, 1988; Winterer et al., 1988). However, according to our reinterpretation of the published lines, the reflection package equivalent to the upper Valanginian to Aptian siliciclastic sediments above the carbonate interval is postrift, not synrift, in origin. This suggests that rifting occurred during latest Berriasian and earliest Valanginian times in the Leg 103 area. Given the suggested correlation of seismic units beneath the Iberia Abyssal Plain with those beneath the deep Galicia Margin (Fig. 9, Fig. 10), rifting probably occurred between the late Berriasian and early Valanginian in both areas. This constrains the age of rifting to between about 134 and 140 Ma (using the Mesozoic time scale of Gradstein et al., 1994). The 40Ar/39Ar age of 136.4 ± 0.3 Ma (early Valanginian) related to late low-temperature shear deformation during the last stages of continental rifting (Féraud et al., this volume) is consistent with the timing of rifting suggested in this paper.
Figure 12 summarizes the ages of key events at the eastern margin of the Iberia Abyssal Plain. It suggests that rifting occurred over a relatively short period of time (~5 m.y.) during the late Berriasian and early Valanginian and was followed about 6-8 m.y. later by the onset of seafloor spreading beneath the southeastern part of the Iberia Abyssal Plain. This time gap between the end of rifting could be explained by our conclusion about the age and duration of rifting in this area being incorrect. However, off the Galicia Margin, the commencement of seafloor spreading should have been later if ocean opening propagated northward, but here the age of rifting is much better constrained to be late Berriasian-early Valanginian by correlating seismic and drilling data. The time gap between the cessation of rifting and the onset of seafloor spreading in both areas suggests either that there was a temporary suspension of plate movement (which is unlikely) or that extension was transferred to another location that could have been situated on the Grand Banks conjugate margin.
The conclusion that our Unit 6 is Aptian to possible Valanginian in age only increases the difficulty of interpreting the setting of the Aptian debris flows and rock-fall breccias encountered on the crest of basement highs at Sites 897 and 899. These occurrences are about 1 km higher than Unit 6. Yet there is no evidence for the basement ridge at Site 897 being tectonically uplifted, or rising diapirically, during the deposition of Units 1-4. Elsewhere in areas also presumed underlain by serpentinite ridges (i.e., the area around Site 898), there is no sign (such as significant reflection thinning and rotation toward areas of uplift) of Units 5 and 6 being affected by localized uplifts. This conclusion is at odds with comments made by Comas et al. (this volume) and Gibson, Morgan, and Milliken (this volume) to the effect that the basement topography that existed at the time of formation of the Aptian olistostromes incorporating mantle material drilled at Sites 897 and 899 differed significantly from that observed today. In particular, we see no evidence that Units 5 and 6 were affected by latest Cretaceous postrift tectonics as suggested by Comas et al. (this volume) to account for the occurrence of olistostromes on basement highs. Debris-flow deposits, such as those occurring in lithostratigraphic Unit IV at Site 897 (Shipboard Scientific Party, 1994a) and Subunit IVB at Site 899 (Shipboard Scientific Party, 1994c) require only gentle (<1°) slopes for their formation (Stow, 1994) and so could have been generated on the basement topography observed today. However, the rock-fall origin (Gibson, Morgan, and Milliken, this volume) for the serpentinite breccias of Subunit IVA at Site 899 implies the existence of nearby slopes inclined at 15° or more (Stow, 1994). As this site was surveyed only by low-resolution single-channel JOIDES Resolution data, its topographic setting is not well constrained.
The subsidence experienced by any point on a rifted continental margin includes thermal subsidence with a time constant of approximately 63 Ma (Parsons and Sclater, 1977), which begins at about the end of the rifting process, and subsidence in response to sediment loading, which depends on the sedimentation history of the margin. To begin to understand the tectonic history of the Iberia Abyssal Plain Margin, we intended to investigate the subsidence of the three Leg 149 sites that had the greatest penetration of the sedimentary column (Sites 897, 899, and 900; Sawyer, Whitmarsh, Klaus, et al., 1994). At these three sites we had hoped to obtain the information required to conduct a backstripping analysis (paleodepth estimates and a knowledge of the lithology, porosity, and biostratigraphic ages of the sediments). In principle this should have enabled us to compute the basement subsidence by taking account of sediment loading, water depth at the time of deposition, and sea-level changes (Steckler and Watts, 1978). A similar approach was used successfully at the Galicia Bank Margin by Moullade et al. (1988). However, given the above time constant, in theory most subsidence occurs within the first 50 Ma after rifting ceases. The latest onset of thermal subsidence along the Leg 149 transect was at about 130 Ma, which is our best estimate of the beginning of seafloor spreading there (see Whitmarsh et al., this volume). Thus, to constrain the subsidence, we need paleodepth estimates from sediments laid down in the Aptian to latest Cretaceous age interval. Unfortunately, almost no in situ sediments of this age were recovered from the Leg 149 sites. Consequently, only the most limited estimates have been made of paleodepth (Table 3). We are almost sure that the Tithonian sediments at Site 901 were deposited in neritic depths (approximately 200 m) but all other cores containing benthic foraminifers that were used to estimate paleodepth are Paleocene or younger in age and indicate depths in excess of 4200 m (Collins, Scott, and Zhang, this volume; Collins, Kuhnt, and Scott, this volume; Kuhnt and Collins, this volume). While these depths are broadly consistent with the onset of subsidence about 130 Ma, they give virtually no constraints on the history of the critical first 50 Ma of subsidence.
We used the bathymetry and basement depth along the digitized seismic profiles to estimate total tectonic subsidence, the sediment-unloaded depth to basement (Sawyer, 1985). We included the effects of sediment compaction and local isostatic compensation in our calculation. We computed the average sediment density (ρs) between the seafloor and basement assuming an exponential decrease in porosity (
) with depth z, sediment grain density (ρsg) of 2650 kg/m3, porosity at the seafloor (0) of 0.55, an exponential porosity decrement c of 4.5 × 10-4/m, and sediment thickness (S). The result is not very sensitive to the sediment compaction parameters:
Using this sediment density model, water depth (W), water density (ρw) = 1030 kg/m3, and mantle density (ρm) = 3300 kg/m3, we computed TTS (Y) on a point-by-point basis along the profiles assuming local isostatic compensation (Steckler and Watts, 1978):
To eliminate the roughness of the TTS owing to local basement topography (Sawyer, 1986), the TTS was smoothed for subsequent calculations using a 50-km running-average filter. Extension during rifting may be estimated using TTS if one assumes isostasy and that the current basement surface was at sea level prior to rifting. In this case, the present depth to basement, after sediment unloading, is the total amount of tectonic subsidence of the crust during and since rifting. Within extended continental crust, the TTS is related to the crustal extension ()
(Fig. 13) (Le Pichon and Sibuet, 1981; Dunbar, 1988) by
The constant of proportionality (C) is a function of the age of continental breakup (t). The thermal time constant (
) is 62.8
m.y. (Sclater et al., 1980). For old rifted margins, with ages greater than one thermal time constant, the relationship between TTS and (
is relatively insensitive to the duration of the breakup process, variations in the initial crustal thickness, or whether extension is accomplished by stretching or dike intrusion (Dunbar and Sawyer, 1989). We have
estimated
in this study for rifting at 135 Ma (Fig.
13). Modeling magnetic anomaly data from the western part of the study area indicates that seafloor spreading commenced at 129-127 Ma
(Fig. 12). As discussed earlier, we believe rifting occurred between 134 and 140 Ma.
Given the variation of
across a margin, one can estimate the variation in crustal thickness. This estimate is valid to the extent that the original crustal thickness was constant in the rifted region and can be inferred from the present crustal thickness in the adjacent unextended region and to the extent that no new material was added to the crust by magmatic intrusion or
underplating. The estimate of extended crustal thickness is the original crust thickness divided by
. In this study we estimated crustal thickness using values of
derived for a rifting age of 135 Ma. We estimated the depth of the crust/mantle boundary by adding the estimated crustal thickness to the smoothed (50-km running average) basement depth.
Our TTS method estimations of
and crustal thickness do not include the affects of dike intrusion. Dike intrusion during rifting should be accounted for in the estimated
in that it is just one of several possible mechanisms of crustal extension. Dike intrusion probably increases the density of the extended continental crust, which will cause the basement to subside more than predicted for a given
. Thus, it may lead to a small overestimate of
. In addition, the intrusion of new material into the extended continental crust will make the actual crust, as measured using seismic refraction, thicker than we estimate using the TTS method.
The TTS estimation of crustal thickness does not include the affects of serpentinization of the upper mantle. Serpentinization of a layer of upper mantle, because it involves a density reduction, would cause the TTS to be less than it would be otherwise. This would work its way through the calculations to reduce the estimated
and increase the estimated crustal thickness. If the serpentinized layer was correctly identified using seismic refraction and not included as part of the crust, then the TTS crustal thickness estimate would be greater than the actual crustal thickness.
The smoothed TTS for the western (seaward) end of profile Lusigal 12
(Fig. 14; distance 0-140 km) is between 6.0 and 6.3 km. This is consistent with the subsidence of normal oceanic crust formed ≥100 Ma or continental crust thinned by
of between 5 and 7 at 135 Ma. There is an inflection at about 140 km with decreasing TTS and
to the east.
This is most simply interpreted as extended continental crust with decreasing extension to the east. The Moho depths predicted using the TTS data are somewhat deeper than those observed using seismic refraction (Whitmarsh et al., 1990a). The TTS estimates should be compared to the depth of the base of the crust (Fig. 14; squares for L2 and L3 and circle for L1). The material below that boundary for lines L2 and L3 is interpreted to be serpentinized peridotite formed by hydration of the upper mantle. For the purposes of this analysis, this serpentine is part of the upper mantle, rather than the crust. The TTS-predicted Moho is about 2-4 km deeper than the seismic refraction base of the crust. We see three possibilities for explaining this discrepancy. First, it may indicate that the prerift crustal thickness we chose, 40 km, was too large. Using 30 km would remove about one-half of the discrepancy. Second, as discussed above, the process of serpentinization of the upper mantle will bias the TTS estimate of crustal thickness upward and therefore Moho depth downward. The third possibility is that we are incorrectly interpreting this crust as extended continental crust. The method used to estimate β and crust thickness is based on that assumption. If this were not extended continental crust, then the top and bottom panels are meaningless.
The remaining profiles (Figs.
14B-14L) show TTS consistent with
of between 5 and 7 with sporadic excursions as high as 9 and as low as 4. These may be interpreted as the same as the west end of profile Lusigal 12
(Fig. 14). That is, they are consistent with the subsidence of normal oceanic crust formed ≥100 Ma or continental crust thinned by the indicated
β with rifting occurring at 135 Ma. Using these data alone, there is no way to select from these interpretations.
The second part of this study documents the following key points concerning the tectonic history of the Iberia Abyssal Plain.
1. Rifting. No direct evidence for the accumulation of synrift sediments was observed on seismic sections. Therefore, the main rifting event must predate our Unit 6, which is Aptian to Hauterivian (and possibly Valanginian in age), and postdate Tithonian siliciclastics drilled at Site 901 and Tithonian-Berriasian carbonates penetrated in the Leg 103 area. Rifting probably occurred between 140 and 134 Ma.
2. Basement topography. There is no evidence to suggest that basement topography established during rifting was modified by later deformation.
3. Total tectonic subsidence calculated along the interpreted seismic profiles indicates continental crust thinned by
