STRUCTURE AND PETROGENESIS OF PLAGIOCLASE-BEARING MANTLE PERIDOTITES FROM THE ROMANCHE TRANSFORM (EQUATORIAL ATLANTIC OCEAN)

The Romanche Fracture Zone is a lens-shaped, 900 km-long, 100 km-wide multifault deformation zone, the major transform fault that offsets the Equatorial Mid-Atlantic Ridge and one of the longest of the worldwide ridge system. It looks different from “classic” oceanic transform boundaries, which are characterized by a single, few-kilometers-wide, strike-slip deformation zone, so that it has been defined as “megatransform” plate-boundary (Ligi et al., 2002). The Romanche FZ peridotites represent residua of a relatively cold mantle related to a strong cold edge effect, an possibly regional lower mantle temperatures, hence experiencing among the lowest degrees of melting (<10%) of the whole mid-ocean ridge system, in which frequently part of the percolating melts may be retained and crystallize within the lithospheric mantle (Seyler & Bonatti, 1997; Bonatti et al., 2001; Tartarotti et al., 2002). Textural, geochemical and petrological data of abyssal plagioclase-bearing peridotites sampled in the central and western parts of the Romanche FZ during the oceanographic expedition PRIMAR-96 (Russian R/V Gelendzhik) are here reported. Petrography reveals that rocks of Group1 and Group2 represent mantle residua showing protogranular to porphyroclastic textures possibly with relics of equigranular textures, developed during upwelling throughout asthenospheric and lithospheric environments simultaneously to melting and melt percolation. These textures appear to have been diffusively superimposed by further, deep melt percolation at the shallowest levels in mantle, leading to crystallization of large amounts of plagioclase (up to 16%) and new clinopyroxene from the trapped melts. As melting ceases when mantle approaches the conductive layer of the lithosphere, in fact, continuous melt supply from deeper levels by diffuse porous flow results in reactive impregnation, which may start in the shallowest levels of the spinel stability field and then proceeds in the plagioclase stability field, leading to crystallization of new minerals pockets, made of (cpx+ol) or (plag+cpx) depending on pressure conditions. New mineral pockets will exploit the rock porosity, i.e. zone of pervasive grain-size reduction developed in asthenospheric environments, and lithospheric brittle/ductile deformation structures, e.g. fractures, gashes, kink bands, micro-faults. Reactive impregnation may involve total amounts of impregnating melts up to 25%. Major and trace elements of fresh minerals (ol, opx, opx, cpx, sp and pl) have been determined by EMPA and LA-ICP-MS techniques. Geochemistry attests for the rims of the coarser-grained pyroxene to have equilibrated in the plagioclase stability field. Trace elements modeling allows to determine the degree of melting of the peridotites and compositions of the melts which percolated the rock source during its melting and post-melting histories. Percolating melts pertain to two main compositional families, namely: i) enriched melts relatively aggregated with strongly variable garnet fingerprint and ii) markedly depleted melts. Degrees of melting from 7% to 14% have been inferred for rocks of Group1 and Group2; this suggest plagioclase-bearing peridotites of the Romanche FZ to be the result of refertilization of previously depleted peridotites rather than pure residues after low degree of partial melting, therefore the observed scarcity of erupted basalts at the seafloor in this region (Bonatti et al., 2001) may be due to strong melt retention rather than low melt production. This is in contrast with sample of Group3, which attests for significant degrees of melting (18% according to modeling) combined with very low melt retention and weak, nearly negligible post-melting reactive impregnation, suggesting efficient mechanisms of melt production and extraction, thus a scenario more similar to channeled melt transport rather than diffuse porous flow. This type of rock may thus represent an example of efficiently channelized melt extraction that early collects deep melts and re-distributes them partially into the rock, resulting in E-MORB extrusion at the seafloor with significant garnet-fingerprint (Ligi et al., 2005), confirming that channelization of deep melt portions can be strongly efficient from the early melt stages in the deep regions.