Composition and processes of the mantle lithosphere in northeastern Brazil and Fernando de Noronha: evidence from mantle xenoliths

Spinel–peridotite facies mantle xenoliths in Cenozoic alkali basalts of the Pico Cabuji volcano (Rio Grande do Norte State, Northeast Brazil) and the adjacent South Atlantic oceanic island of Fernando de Noronha are studied for: (1) the information they provide on the composition of the lithospheric component in the erupted basalt geochemistry, and (2) to check the effects of the Fernando de Noronha plume track on the mantle lithosphere. Xenoliths from Pico Cabuji are protogranular lherzolites and porphyroclastic harzburgites recording average equilibrium temperatures of 825 ± 116 and 1248 ± 19 °C, respectively. Pressure in the porphyroclastic xenoliths ranges from 1.9 to 2.7 GPa (Ca-in-olivine geobarometer). Both groups show major element chemical variation trends in whole-rock and Ti and HREE (Er, Yb) variations in clinopyroxene consistent with fractional melting and basalt extraction. REE (rare earth element) profiles of clinopyroxenes vary from LREE (La, Ce) enriched (spoon shaped) to LREE depleted in the protogranular group, whereas they are slightly convex upward in most porphyroclastic clinopyroxenes. HFSE (Ti and Zr) negative anomalies are in general modest in the clinopyroxenes of both groups. Xenoliths from Fernando de Noronha have textural variations similar to those of Pico Cabuji. Protogranular and porphyroclastic samples have similar temperature (1035 ± 80 °C) and the pressure is 1–1.9 and 2.3 GPa, respectively. Whole-rock chemical variation trends overlap and extend further than those of Pico Cabuji. The trace element profiles of the clinopyroxenes of the porphyroclastic xenoliths are enriched in La up to 30 × PM and are smoothly fractionated from LREE to HREE, with deep, negative, Zr and Ti anomalies. The geochemical heterogeneities of the xenoliths from both localities are interpreted in terms of reactive porous percolation. The porphyroclastic xenoliths from Pico Cabuji represent the lower part of a mantle column (the head of a mantle diapir, at the transition conductive–adiabatic mantle), where OIB infiltration triggers melting, and the protogranular xenoliths the top of the mantle column, chromatographically enriched by percolation at a low melt/rock ratio. This interpretation may also apply for Fernando de Noronha, but the different geochemical signature recorded by the clinopyroxenes requires a different composition of the infiltrated melt. Nd and Sr isotopes of the Pico Cabuji porphyroclastic clinopyroxenes (¹⁴³Nd/¹⁴⁴Nd= 0.51339–0.51255, ⁸⁷Sr/⁸⁶Sr=0.70275–0.70319) and of Fernando de Noronha (¹⁴³Nd/¹⁴⁴Nd=0.51323–0.51285, ⁸⁷Sr/⁸⁶Sr=0.70323–0.70465) plot on distinct arrays originating from a similar, isotopically depleted composition and trending to low Nd–low Sr (EMI) and low Nd–high Sr (EMII), respectively. Correlation of the isotope variation with geochemical parameters indicates that the isotopic variation was induced by the metasomatic component, of EMI type at Pico Cabuji and of EMII type at Fernando de Noronha. These different components enriched a lithosphere isotopically similar to DMM (depleted MORB mantle) at both localities. At Fernando de Noronha, the isotopic signature of the metasomatic component is similar to that of the ∼ 8 Ma old lavas of the Remedios Formation, suggesting that this is the age of metasomatism. At Pico Cabuji, the mantle xenoliths do not record the high ⁸⁷Sr/⁸⁶Sr component present in the basalts. We speculate that the EMII component derives from a lithospheric reservoir, which was not thermally affected during mantle metasomatism at Pico Cabuji, but was mobilized by the hotspot thermal influence at Fernando de Noronha. This interpretation provides a plausible explanation for the presence of distinct metasomatic components at the two localities, which would be difficult to reconcile with their genetic relationship with the same plume. Received: 12 June 1999 / Accepted: 13 December 1999