No abstract

Three major lithospheric plates-Antarctic, South American, and African-meet in the South Atlantic near Bouvet Island where the Mid-Atlantic Ridge (MAR), the Southwest Indian Ridge (SWIR), and the American Antarctic Ridge converge toward a fast evolving triple junction. A major magmatic pulse has recently built a new, swollen segment of the SWIR (Spiess Ridge) that is propagating toward the MAR at a rate of 4 to 5 centimeters per year, disrupting a former ridge-ridge-ridge (RRR) triple junction. A new triple junction will be established about 70 kilometers to the north when the propagating SWIR/Spiess segment will impact with the MAR, probably within the next 1 million years. The American Antarctic Ridge will take advantage of the MAR/SWIR duel by capturing an approximately 70-kilometer stretch of MAR, whereas the Antarctic plate will increase its size.

The Romanche is a long offset (-950 km), slow slip (~ 1.7 cm/yr) transform; thus a hot ridge axis should meet a ~50-m.y.-old, thick and cold lithosphere at the ridge-transform intersection (RTI). A strong thermal/topographic "transform cold edge effect" is therefore predicted. A morphobathymetric, seismic reflection and petrologic study of the eastern Romanche RTI shows that as the Mid-Atlantic Ridge approaches the transform, a well-formed axial rift valley disappears about 80 km from the RTI and is substituted by short en echelon, poorly developed axial ridge segments; they too disappear about 30 km from the edge of the transform valley. The predicted gradual deepening of the ridge axis toward the transform was not observed. An active nodal deep and an "inside comer high" are also absent. These observations, and the distribution of earthquake epicenters, suggest a poorly developed, diffuse RTI. An inactive rift valley ~80 km west of the present RTI suggests ridge jumping within the last ~4 m.y. The present poorly developed RTI may reflect the attempts of an embryonic spreading axis to become established and to propagate toward the transform. We infer from bottom rock sampling that the basaltic crust is patchy or absent and mantle-derived serpentinized peridotites outcrop ubiquitously on the seafloor starting ~30 km from the edge of the transform valley. The unusually deep (~4 km below sea level) axial ridge segments, the lack of crust, and the chemistry of the peridotites suggest a prevalently amagmatic regime due to an ultracold upper mantle in this region. Absence of basaltic crust would favor massive serpentinization of a several kilometers thick peridotite column. Mass balance modeling suggests that the decrease of density and volume expansion resulting from serpentinization could explain the absence of the predicted deepening of the seafloor as it approaches the transform. These results suggest that the topographic effect of the transform edge thermal contrast may disappear at ultracold RTIs and that ultracold RTIs are magma starved, short lived, and unstable in time and space. following generalizations [Fox and Gallo, 1984; Bender et al., 1984; Karson and Dick, 1983; Phipps Morgan and Forsyth, 1988; Blackman and Forsyth, 1989]: (1) the floor of the axial rift valley deepens and widens approaching a transform boundary; (2) a characteristic, areally limited topographic Paper number 95JB02249. 0148-0227/95/95JB-02249505.00 depression ("nodal basin") occurs as the rift valley enters the transform valley; (3) a large topographic high ("inside corner high") is observed commonly at the transform side of the rift valley as it approaches the transform valley; (4) morphostructural lineations oriented obliquely to both ridge axis and transform occur in the transform side corner but not in the non transform side corner; and (5) a systematic change in the geochemistry of axial rift basalts occurs approaching a transform. These features, and particularly the deepening of the axial valley as it approaches a transform, an...