Li, Be, and B abundances in minerals of peridotite xenoliths from Marsabit (Kenya): Disequilibrium processes and implications for subduction zone signatures

Abstract

The light elements Li, Be, and B have been analyzed in situ in minerals from three groups of peridotite xenoliths hosted in Quaternary basanites from the Marsabit volcanic field (northern Kenya). Group I and II are fertile lherzolites that experienced deformation, decompression, and cooling in the context of Mesozoic rifting (Group I), followed by heating, static recrystallization, and associated cryptic metasomatism (Group II) as a result of Tertiary‐Quaternary rifting and magmatism. Group III xenoliths are spinel harzburgites and dunites that experienced strong cryptic and modal metasomatism. The Li‐Be‐B systematics in minerals of Group I and II are similar to unmetasomatized subcontinental lithospheric mantle. In contrast, Group III samples are characterized by significant enrichment in all light elements and disequilibrium partitioning between different phases. Light element concentrations levels are similar to that expected for mantle rocks metasomatized by melts and fluids released from subducting slabs, while light element/rare earth element ratios (especially Li/Yb) approach those of typical Island Arc basalts. However, detailed investigation of textures and chemical zoning shows that at least Li concentrations in primary minerals were modified (i.e., decoupled from Yb) during late‐stage melting and/or fluid percolation related to Tertiary‐Quaternary alkaline magmatism in Marsabit (formation of melt pockets consisting of silicate glass, clinopyroxene, olivine, and chromite), ultimately followed by xenolith entrapment and transport to the surface. Mass balance calculations show that the melt pockets formed at the expense of earlier metasomatic phases. During this process the melt pockets mostly preserved the B, Be, and rare earth element budget of the precursor phase assemblage, whereas Li was added. Elevated B/Be and low Ce/B of metasomatic phases prior to late melting could result from metasomatism by a slab fluid. However, similar characteristics are expected for evolved Si‐ and CO2‐rich fluids derived from basanite melt‐peridotite interaction, not related to any subduction zone process. The results of this study imply that the inference of a “slab signature” exclusively based on trace element data of metasomatized peridotite is ambiguous.