Discussion

Non-steady state ridge/trasform geometry - A first analysis of the results obtained to date indicates that the MAR/transform geometry in the Romanche area might have changed considerably through time. Two types of changes are suggested by the data: (a) migration and reorientation of the transform boundary and (b) migration or jump of the MAR axis segment impacting against the transform.

The migration and reorientation of the transform boundary is suggested by the presence of the aseismic valley running subparallel to and on the northern side ofthe presently active transform valley. We have mentioned in a previous section of this paper that the inactive valley might have represented the active transform boundary prior to about 8-10 my ago. A jump to the east of the eastern ridge segment, suggested by the presence of an aseismic former rift valley $\sim$ 80 Km west of the present RTI, might have taken place about $<$ 5 My ago, thus increasing the offset length significantly. This migth be the last of a series of events that have nearly doubled the length of offset of thc Romanche transform since the opening of the Atlantic, when the proto Romanche probably acted as a continent/Continent transform.

Figure: Portion of seismic reflection profile ROM-2 over relief D, A: migrated time section; B, line drawing. Location of this profile is shown in Fig. 7.

Non straight transform boundary - If we assume as a first order approssimation that the Romanche transform boundary presently runs along the deeper part of the seismically active transform valley, the boundary appears not to be straight. It shows a number of kinks and bends that change the overall orientation of the transform from about E- W in its western part to WNW-ENE in its eastern part. This change in orientation is confirmed if we consider the trace of the transform boundary identified by BELDERSON1984, SEARLE1994 based on GLORIA data. A non straight transform boundary may determine zones of transpression and of transtension along the transform boundary, as indicated schematically in Fig. 13.

Figure 13: Schematic and simplified model indication that transpression and transtension can develop due to a bend in the transform boundary.

Vertical tectonics and the origin of the transverse ridge - An analysis of possible factors responsible for the development of the large topographic anomalies observed along the Romanche transform zone has been carried out elsewhere BONATTI1994.

Briefly, one possibility is that the transverse ridge is a constructional, volcanic structure. However its elongated, narrow morphology is quite untypical of volcanic structures. Only at one site on the southern slope of the transverse ridge across the RTI, fragments of fresh-looking basalts were sampled, raising the possibility that some volcanism might have spilled over on the transverse ridge from the MAR axis across the RTI. However. basalts recovered elsewhere from several sites on the upper slopes are strongly altered, consistently with their being part of $\sim$ 50 My old crust originally formed at the western MAR axis. The recovery of lower crustal gabbros and upper mantle peridotites on the slopes of the transverse ridge as well as its morphology suggest that this feature is not a constructional volcanic structure but an uplifted sliver of oceanic lithosphere.

A number of processes related directly or indirectly to the thermal structure of the lithosphere adjacent to the transform can contribute to the vertical motions. Horizontal conduction of heat from the hot ridge axis to the old cold plate across the RTI can cause swelling and uplift of the old lithosphere LANGSETH1976, LOUDEN1976, CHEN1988. It has been estimated, however, that for a long-offset, slow slip transform such as the Romanche this effect may cause a few hundred meters uplift at the most CHEN1988, BONATTI1994. Shear frictional heating occur in a long offset transform but cannot contribute significantly to the required uplift CHEN1988. Lithospheric flexure due to the thermal contraction of the aging and cooling plate may cause uplift adjacent to the transform that will not be over $\sim$ 200 m PARMENTIER1986. Hydration of the mantle ultramafic rocks may take place below 500C  due to sea water penetration into crust that is highly permeable and of reduced crustal thickness near the RTI. The consequent decrease of density of hydrated upper mantle may cause uplift in the first few million years of spreading away from the RTI. However, dehydration and subsidence will occur when the old lithosphere is reheated upon transiting across the other RTI. Thus, dehydration-induced subsidence works against reheating-induced uplift when the old lithosphere transits opposite the RTI BONATTI1994.

All these factors taken together can account for no more than about 20% of the up to $\sim$ 4 km uplift inferred for portions of the Romanche transverse ridge.

It has been recognised for a longtime that adjacent blocks in continental strike-slip ter- rains can either converge or diverge, with important structural consequences WILCOX1973, CHRISTIE-BLICK1985. Anomalous topography can result from vertical tectonic movements of upper lithospheric slivers, due to transpressional or transtensional events related either to non straight transform boundaries, or to changes in the direction of spreading and consequent reorientation of transforms, as well as to transform migration connected with longitudinal propagation of ridge segments MENARD1968, BONATTI1978, BONATTI1982, KARSON1986. Transpression due to clockwise changes in the direction of motion of the plates adjacent to right-lateral transforms such as the Romanche, Vema and Kane is probably particularly effeetive in inducing vertical tectonics BONATTI1978, TUCHOLKE1988.

In addition, based on GLORIA data, showing a non-straight Romanche transform boundary, portions of the lithospheric plate on the northern side of the transform are likely to impact against the boundary determining a transpressional regime (Fig. 13).Conversely, estension should prevail in some areas along the southern margin of the transform boundary (Fig. 13), possibly with a tendency for pull-apart basins to form.

We suggest that compression due to the oblique impact of the direction of spreading against the eastern branch of the Romanche transform boundary might have been, in a general way, the main cause of the anomalous uplifted lithosphere observed on the northeastern side of the Romanche. Compression was documented on the easternmost relief (relief D) of the transverse ridge, where the thick sedimentary deposits are affected by folding and overthrusts.

The geometry of Fig. 13 implies areas of transtension on the southern side of the transform close to bends in strike of the transform boundary. We can further speculate that in the area where the suspended valley (i.e., the old transform) merges with the presently active transform, the tectonic regime changed to transtensional about 5 mybp. The ``Vema Depth'' HEEZEN1964, an ultradeep ($>$ 7.5 km below sea level) sediment free basin at about 18:30W along the transform valley, is probably a pull apart basin that formed as a result (Fig. 13). At the same time reliefs A, B and C, that 5 my bp were located close to the confluence of the old with the new transform, started to subside. Another event that probably occurred around 5 my bp is the eastward jumping of the MAR axis at the eastern RTI by about 80 km to its present position.