The rock magnetic properties of the peridotites are generally characterized by relatively strong NRM intensities and low Koenigsberger ratios. The average intensity of the recovered serpentinized peridotite is 0.29 A/m for samples from Hole 897D, but about 1.28 A/m for samples from Hole 899B. A recent magnetic survey revealed a magnetic anomaly high (about 0.3 A/m) in the vicinity of Site 899, which was previously thought to result from a relatively strongly magnetized nonoceanic crust (Whitmarsh et al., this volume). The much stronger magnetization intensity of Site 899 is apparently in excellent agreement with this observed magnetic anomaly high, suggesting that the serpentinized peridotite body under Site 899 contributes significantly to the magnetic anomaly. Conversely, the fact that Site 897 is located more oceanward, but has a mean NRM intensity less than one-third of the average intensity (~4 A/m) of dredged and drilled oceanic basalts in the Atlantic (Lowrie, 1977), may be used to explain the weakly negative anomaly data observed nearby.
The AMS data suggest that there is no noticeable magnetic anisotropy, and hence, probably no magneto-petrofabrics in these peridotites. This observation is consistent with other types of data, including data obtained by X-ray texture goniometry methods that indicate that there is little variation of fabrics in these peridotites (Morgan, this volume) and by seismic methods that suggest a weak degree of anisotropy (only about 7%) in these rocks (Harry and Batzle, this volume). These consistent observations would attest to the hypothesis that the peridotites were not strongly influenced by the extensional stress field that may prevail in the adjacent continental lithosphere.
Stable components of magnetization are revealed in the results of demagnetization on the peridotites from both holes. In the upper part, the remanence is dominated by a single stable component of normal polarity. Nearly identical directions of this component were found in all samples regardless of lithology and depth. Therefore, the paleomagnetic data of the altered upper part seem best interpreted as indicating that the magnetization is an overprint that was probably imposed during the alteration of these peridotites. The mean inclination of this overprinted magnetization is consistently shallower (~30°) than that of a Holocene field (~60°), suggesting that the overprint was unlikely acquired in the latter. In view of the polarity of magnetization identified from the overlying Cretaceous (see Fig. 11 for an example) and Tertiary sediments at both sites (Zhao et al., this volume), this normal overprint is probably older than the magnetization of the overlying Tertiary sediments.
In the fresher lower part, although the remanence is still dominated by a single component of magnetization in majority cases, several samples display a multicomponent nature, with a characteristic component isolated after removal of a secondary component of opposite polarity (Fig. 5). The characteristic component is carried by magnetite, which is known capable of preserving an early remanent magnetization over a long time. Because of the demagnetization behavior of these samples, I have no reason not to believe that the magnetization directions derived from these fresher peridotites are primary. Moreover, the reversed magnetization zone in the beginning of the fresher lower part is correlative between the two sites, which is perhaps the most compelling argument for the presence of primary magnetization because it would be difficult for a later remagnetization to produce such a preferential polarity pattern at depth. Several lines of overlapping evidence (Table 3) support my contention that the reversed magnetization zone in the beginning of the fresher lower part is correlative:
1. A distinctive increase in susceptibility was observed from shipboard pass-through magnetic susceptibility measurements at both sites (Fig. 12), which corresponds to the boundary between lithostratigraphic Units III and IV (Sawyer, Whitmarsh, Klaus, et al., 1994). The depth difference of this stratigraphic marker between the two sites is about 283 m.
2. Biostratigraphic and lithostratigraphic studies have determined that the unit boundaries between these two sites are also offset consistently by 285.4 m (see Sawyer, Whitmarsh, Klaus, et al., 1994; also see Fig. 2).
3. As mentioned, there is a Koenigsberger ratio peak found in both holes. Although this peak occurs in the more oxidized upper part and the value of the ratio itself does not carry much weight in the overall interpretation, its depth difference (291.7 m) between the two holes is significant, as it roughly coincides with the depth difference (307.4 m) of the reversed magnetic polarity zone found in the fresher lower part.
Therefore, these consistent markers of depth difference between the two sites indicate that the reversely magnetized peridotites in both holes are probably the same unit and recorded the same geomagnetic field during the time of emplacement.
Assuming the origins of the identified magnetic components of the peridotites are correctly established, the next step is to correlate the pattern of polarity changes with the established magnetic reversal sequence. To do so, it may be useful to briefly mention the marine magnetic Anomaly M0 and to review the proposed kinematic evolution and timing of important tectonic events among the major plates (Iberia, North America, Europe, and Africa) bordering the North Atlantic.
Marine magnetic anomalies have provided the richest source of information about magnetic reversals. The main reason for the high fidelity of the marine magnetic record is the remarkable continuity of the geologic processes by which new crust is formed along mid-ocean ridges. The polarity chrons of the pre-Aptian sequence are generally described by the M-sequence with designations "M0" through "M29" (Kent and Gradstein, 1985; Channell et al., 1987). There is still no direct isotopic dating of the M-sequence anomalies. The ages assigned by Kent and Gradstein (1985) to the M-sequence anomalies are based on fixing the Barremian/Aptian boundary at 118 Ma and the Oxfordian/Kimmeridgian boundary at 156 Ma. A number of short-period reversed events have been identified within the Cretaceous interval of dominantly normal polarity. The best documented of these is the reversed polarity event, of possible ~1 Ma duration, situated close to the Barremian/Aptian boundary (Hellsey and Steiner, 1969; McElhinny and Burek, 1971; Pechersky and Khramov, 1973; Channell et al., 1987). This reversal has been correlated with marine Anomaly M0. The M0 used in Kent and Gradstein's (1985) time scale is at 118 Ma, which immediately precedes the Cretaceous Long Normal Superchron (84-118 Ma). As most published papers about Iberia use this system, I adhere to the Kent and Gradstein (1985) time scale in this study.
Kinematic models for present-day plate motions (Minster and Jordan, 1978; Argus et al., 1989) and the distribution of earthquakes show that Iberia is now moving as part of Eurasia. However, continental geology strongly suggests that Iberia could not have moved with Eurasia during the formation of the Pyrenees in the Early Cretaceous (Srivastava et al., 1990). Paleomagnetic results on land also suggest that Iberia has rotated counterclockwise about 35° relative to Europe between the Barremian or Aptian and the Maastrichtian (Van der Voo, 1969, 1993) and that most of the rotation (30°) occurred at about Hauterivian to Aptian time (Galdeano et al., 1989). Several plate tectonic reconstructions have been attempted to show the original positions of North America, Iberia, and Europe (Le Pichon et al., 1977; Srivastava et al., 1990; Srivastava and Verhoef, 1992). At present, only models with marine magnetic data have the potential to provide estimates of the age of important events (although these events have to relate strictly to the history of seafloor spreading). By matching synchronous magnetic lineations and fracture zone systems that characterize the seafloor-spreading history between Europe and North America, Srivastava et al. (1990) reconstructed the history of Iberia's motion relative to its neighboring plates from Anomaly M0 (118 Ma on the time scale of Kent and Gradstein, 1985) to the present. They implied that Iberia was alternately attached to Africa or Europe and the opening of the North Atlantic was not an instantaneous process. Several salient features stand out from their kinematic model:
1. For the major part of the Cretaceous magnetic quiet period, Iberia moved as an independent plate.
2. Sometime before Chron 34 (84 Ma), Iberia became attached to the African plate, and the plate boundary between Africa and Eurasia was located at the Bay of Biscay.
3. From Chrons 34 (84 Ma) through 18 (42 Ma), Iberia was part of the African plate.
4. From the late Eocene (Chron 18, 42 Ma) to the early Miocene (Chron 6c, 24 Ma), Iberia again moved as an independent plate. The boundary between Eurasia and Iberia during this time extended west from the Pyrenees to King's Trough. The separation process is further complicated by intraplate deformation. The relative motion between Iberia and North America was transformed into the left-lateral motion of Iberia with respect to Europe (Le Pichon et al., 1977), and evidence of southwesterly motion of Iberia with respect to Europe was found in the fault pattern of the sedimentary basins north of Spain (Lepvrier and Martinez Garcia, 1990).
5. At about early Miocene (Chron 6c, 24 Ma) time, motion along the King's Trough/Azores-Biscay Rise boundary became very small, and Iberia started to move as part of the Eurasian plate (Srivastava et al., 1990).
The general evolution of the North Atlantic is now well documented in the literature (Bullard et al., 1965; Le Pichon et al., 1977; Courtillot, 1982; Srivastava and Tapscott, 1986; Klitgord and Schouten, 1986; Rowley and Lottes, 1988; Srivastava et al., 1990; Srivastava and Verhoef, 1992). Three rift-drift episodes can be recognized. The separation of Africa and North America is the oldest, with a Triassic-Jurassic rifting phase and a final breakup in the Middle Jurassic. The next phase is the separation of Iberia and the Grand Bank, which took place during the Early Cretaceous, with breakup at about 130 Ma. The last phase of seafloor spreading started with the separation of Europe from North America in the middle Cretaceous. Steady-state seafloor spreading between the Grand Bank and Iberia started at M0 time (118 Ma).
Returning now to the age assignment of the polarity patterns identified from this study, I first want to point out that Leg 149 drilling successfully confirmed the timing of the second phase of the above-mentioned rift-drift episodes, originally proposed by Whitmarsh et al. (1990), in which Iberia started to drift in late Hauterivian time (M4, ~130 Ma), about 12 Ma earlier than the adjacent Galicia Bank. It is generally believed that continental breakups initiating ocean basins are preceded by a period of lithospheric stretching lasting about 20-50 Ma (Cochran, 1983), and the onset of oceanic spreading is assumed to be contemporaneous with the end of rifting and extensional tectonics on the margins (Malod and Mauffret, 1990). Therefore, it seems that the emplacement of Leg 149 peridotites probably occurred during the continental rifting and breakup of Iberia from the Grand Bank in late Hauterivian time (M4), and the observed reversed magnetization zone in the fresher lower part of the peridotite section most probably coincides with the onset of seafloor spreading between the Grand Bank and Iberia at M0 time (118 Ma). Because Anomaly M0 lasted only about 1 Ma, the overprinted normal magnetization signal observed in the more oxidized upper part of the peridotite section could be imposed any time within the Cretaceous Long Normal Superchron (84-118 Ma), most likely soon after the M0 time. These interpretations are fully consistent with the time constraints from the magnetic signatures of the overlying Cretaceous sediments, as mentioned above. The most conspicuous feature of the polarity records for the Cretaceous red and brown clays is a short reversed polarity zone, which may suggest that the deposition of these cores occurred during the M0 subchron.
Several lines of geological observations agree with the above age assignments to the magnetic polarity zones:
1. Biostratigraphic arguments suggest that the oldest units of these peridotites are Early Cretaceous in age (Hauterivian, about 135-132 Ma), and the first sediment deposited on top of the peridotite section was found to be Late Cretaceous in age. Therefore, the entire peridotite section should have been emplaced during the Cretaceous. Recent micropaleonotologic dating of cores from Hole 899B revealed that the emplacement of these peridotites probably lasted only about 1-2 Ma (de Kaenel and Bergen, this volume).
2. Magnetic Anomaly M0 is identified from only the east side of the peridotite sites (less than 150 km), suggesting the peridotites are of approximately the same age as Anomaly M0 (118 Ma).
3. Numerous major unconformities (at about 118 Ma) in many sedimentary basins surrounding the peridotite sites are concurrent with seafloor spreading at M0 time. On the Iberian margin side, these tectonostratigraphic basins include the Inner Galicia Basin (Murillas et al., 1990) and Lusitanian Basin (Wilson et al., 1989), and on the Newfoundland side, the Whale Basin (Balkwill and Legall, 1989) and Jeanne d'Arc Basin (Tankard et al., 1989). These unconformities not only provide a direct linkage between the evolution of these basins and separation of Iberia from North America, but also suggest that the emplacement and alteration events in these peridotites should have taken place at about M0 time.
4. Shipboard observations suggest that the alteration of the peridotites recovered during Leg 149 took place soon after the peridotites were emplaced at or near the seafloor surface and that the fluid responsible for serpentinization was probably seawater. Therefore, both the paleomagnetic and geologic data are compatible with the suggestion that alteration by fluid circulation was associated with the first stages of accretion of the oceanic crust, during the middle Cretaceous.
It is beyond the scope of this paper to propose a detailed and quantitative model for the emplacement process of the peridotites recovered during Leg 149. However, it is appropriate to discuss qualitatively some of the constraints from paleomagnetic data and other relevant geologic data that can be placed on the potential models for the mode of emplacement and the late-stage alteration of these peridotites.
The sites at which peridotite was recovered during Leg 149 are regions where the crust is missing and where the mantle rocks crop out directly on the seafloor. This would imply that progressive uplift and final emplacement of peridotites at the Earth's surface seem to be related to the thinning of the continental crust beneath the rift. As mentioned in the "Introduction" section, several distinct models have been proposed to account for the uplift of mantle rocks on the seafloor in relation to continental rifting. These models include diapiric emplacement (Nicolas et al., 1987; Bonatti, 1987) and tectonic denudation by detachment faulting (Boillot et al., 1987). Although both models can be used to explain the exposure of upper mantle rocks at the end of the rifting, they both pose certain difficulties when applied to the peridotites at Iberia margin. Specifically, the vertical asthenosphere diapiric model predicts basalt flow should exist on top of the "diapirs," and the detachment faulting model requires that the mantle rocks experience ductile deformation at great depth and brittle deformation near the surface. Shipboard petrologic observations found a complete absence of basaltic clasts within the serpentinized peridotites at both Holes 897D and 899B. Therefore, the vertical diapirism model cannot completely account for the emplacement process at these sites. On the other hand, the current lack of evidence for the presence of anisotropy in these peridotites, as indicated by the AMS data from this study, would pose some difficulties for a pure detachment faulting model. Because uplift of the peridotite by some kind of diapiric mechanism is needed and detachment faulting accounts nicely for the crustal thinning and stretching, I propose a combination of these two models for the process of emplacement of the Iberian peridotites. Uplift by diapirism could occur prior to the main stage of stretching. At the end of the stretching phase, when the crust is particularly thinned, these peridotites would be serpentinized by seawater circulation (which is responsible for remagnetizing the upper part of the peridotites) and hence become less dense than the surrounding crustal material. Finally, the serpentinized peridotites rose to the seafloor and cropped out at the ocean/continent boundary just before "true" seafloor spreading started between the Iberian and Newfoundland margins (i.e., the opening of North Atlantic).