Serpentinized peridotite cores were obtained from the acoustic basement at Sites 897, 899, 1068, and 1070. Comparable rocks have been dredged, sampled by submersible, and drilled both off Galicia Bank and from Gorringe Bank (e.g., Cornen et al., 1999; Girardeau et al., 1998). Sites 897 and 899 lie within the region of uplifted basement (Region B in Fig. F3) of the OCT, which has been suggested to consist of exhumed upper mantle (Brun and Beslier, 1996; Dean et al., 2000; Discovery 215 Working Group, 1998; Krawczyk et al., 1996; Pickup et al., 1996). Site 1068, on the west flank of Hobby High, lies between two continental basement ridges (Fig. F3) and adjacent to the deep part of the OCT (Region C in Fig. F3), of which it may be representative. Site 1070 lies in a very different geophysical situation within Region A (Fig. F3) of oceanic crust. Questions of the subcontinental or suboceanic origin of the mantle rocks, the degree of depletion (partial melting) that they have undergone, and the nature and origin of the 300-km-long, margin-parallel peridotite ridge are very relevant to models of the development of this nonvolcanic margin. The serpentinization history is treated later.
Serpentinized peridotites that were recovered at Sites 897 and 899 have been described by Cornen et al. (1996a), Sawyer, Whitmarsh, Klaus, et al. (1994), and Seifert and Brunotte (1996). The peridotites are spinel- and plagioclase-bearing harzburgite and lherzolite with minor pyroxenite and dunite. No in situ serpentinized peridotite was cored at Site 899, only a serpentinite breccia and serpentinized peridotite unit resulting from mass wasting that also includes clasts of high K2O basalt (altered or non-MORB), diabase, and microgabbro. Cornen et al. (1996a) describe the cores as coarse-grained websterites (Site 897 only) and depleted peridotites with minor plagioclase-rich lherzolites. Variations in the modal abundance and composition of primary phases provide evidence of heterogeneous partial melting of the peridotites. However, the extent of partial melting was probably relatively low (<10%) based on the modal composition of the peridotites and the abundance of lherzolite (Site 899) and websterite (Site 897) relative to more depleted harzburgite and dunite (Cornen et al., 1996a).
Some of the peridotites contain locally pervasive patches or veinlets of plagioclase, suggesting that the rocks underwent impregnation by mafic melts at pressures below 1 GPa (i.e., in the plagioclase stability field). This impregnation probably occurred at the end of the high-temperature deformation stage, because, in many samples, the plagioclase framework parallels the main porphyroclastic foliation observed in adjacent rocks (Cornen et al., 1996a). These authors further suggest that impregnation was caused by percolation of undersaturated alkaline melts based on local enrichments of Fe, Ti, and Na in olivine and pyroxene and, in one peridotite, the presence of Na-Ti-rich phases such as kaersutite, phlogopite, rutile, and ilmenite. However, Seifert and Brunotte (1996) showed that some lherzolites from Site 897 have flat REE patterns with near chondritic abundances, whereas others have light REE-depleted patterns similar to normal mid-ocean ridge basalts (N-MORB). They interpreted the latter as evidence that melt impregnation of the lherzolites involved melts with N-MORB-like composition. Thus, there is no evidence from the REE data for pervasive infiltration of undersaturated alkaline melts, which would be much more highly enriched in incompatible trace elements and probably have light REE-enriched patterns.
As mentioned already, serpentinized peridotite was cored at Site 1068 beneath a seaward-dipping normal fault zone represented by breccias, flanking the west side of Hobby High. Site 1070 lay on the crest of a north-south basement high 20 km west of the peridotite ridge and 30 km east of the crest of magnetic anomaly J, which, together with magnetic anomaly modeling, a probably normal oceanic velocity structure, and the presence of elongate, isochron-parallel basement highs and lows, is a definitive indicator of oceanic crust. Basement cores from Site 1070 consist of matrix-supported serpentinized peridotite breccias separated by a tectonic contact from an underlying pegmatitic gabbro, which in turn has an intrusive contact with the underlying peridotite.
Hébert et al. (in press) studied the peridotites from Sites 1068 and 1070. The peridotites differ in that those of Site 1068 are fine grained with a well defined high-temperature foliation, whereas those of Site 1070 are coarse grained with little evidence of high-temperature foliation. From visual descriptions, the Site 1068 peridotites were originally mostly plagioclase bearing, whereas at Site 1070 plagioclase was absent. However, in terms of mineral chemistry, peridotites from both sites show many common features. In particular, the wide range in pyroxene composition at both sites suggests a range of primary peridotite compositions from aluminous lherzolite to more aluminum-poor harzburgite. Gabbroic intrusions within Site 1070 peridotites contain primary kaersutite and biotite, providing evidence for relatively K- and H2O-rich magmas. Because the most Ti-rich spinels in peridotite at Site 1070 are near gabbroic veins, Hébert et al. (in press) suggest that much of the variation in spinel compositions at both Site 1070 and nearby Site 1068 may be due to intrusion and percolation of such incompatible element-enriched melts into mantle peridotite. The abundance of platinum group elements is very low in peridotites from both Sites 1068 and 1070, on the basis of which Hébert et al. (in press) suggest that the peridotites were derived from subcontinental mantle.
Abe (in press) presents major and trace element data for primary mantle minerals in the Site 1068 and Site 1070 peridotites. These data show that the trace element concentrations in clinopyroxene are intermediate between values typical of abyssal (oceanic) peridotites and peridotites from continental regions, and are most closely similar to mantle peridotite xenoliths from suprasubduction zone volcanic arcs. It should be noted, however, that there is significant overlap between the different fields on most trace element discrimination diagrams, especially between the arc and subcontinental mantle fields. Abe also found that the REE patterns for clinopyroxenes are light-REE depleted, similar to the pattern observed for whole-rock samples of some lherzolites from Site 897 mentioned above. Given the compositional similarities of pyroxenes to those in mantle xenoliths from arcs, it is possible that Site 1068 and Site 1070 sampled Proterozoic Ossa-Morena Zone mantle that experienced suprasubduction zone processes within the ancient Precambrian Ibero-Armorican Arc, which exists adjacent to the Iberia margin (Abalos and Cusi, 1995, fig. 12A; Silva et al., 2000).
It is too early to draw detailed petrogenetic comparisons between all the peridotites cored in the southern Iberia Abyssal Plain. However, Abe's results clearly indicate that Site 1068 and 1070 peridotites are not as depleted as typical abyssal peridotites and are more likely to be derived from suprasubduction zone or subcontinental mantle. Whitmarsh, Beslier, Wallace, et al. (1998) have noted that although the Site 1070 peridotites differ from those at Sites 897 and 1068 in the lower initial plagioclase mode, in the smaller proportion of coarser grained spinel, in the coarser grain size, and in the near lack of high-temperature foliation, preliminary shipboard geochemical analyses indicate that these peridotites are compositionally close to the less plagioclase-rich peridotites of Sites 897 and 1068. At present, it is thus not possible to find any petrological or geochemical difference between peridotites drawn from sites that it is suspected lie in at least two geophysically different domains.