The geochemistry and petrology of the Rodrigues Ridge (western Indian Ocean)
The Rodrigues Ridge is a linear east-west trending volcanic ridge, located between 18°S and 20°S in the western Indian Ocean. The trend of the Rodrigues Ridge is contrary to the ocean floor fabric of the underlying crust, which formed on the Central Indian Ridge (CIR) between ca. 48 My and 10 My. Dating of dredged basalts from the Rodrigues Ridge showed them to be 8-10 My, with no systematic variations with longitude. All samples recovered from the Rodrigues Ridge were olivine and plagioclase phyric with traces of chrome spinel. Only the most western site contained any phenocryst clinopyroxene. Phenocryst olivine was in the range Fo88-79 and plagioclase was in the range An79-60. Where present the clinopyroxene was titaniferous (2.0-3.7% TiO2) in the range Ca4852Mg30-19Fe 19.33. The Rodrigues Ridge lavas are transitional alkali basalts which display systematic geochemical variations with longitude. Most notably the trace elements Ba, Nb, Rb, Sr, Th, Y, Zr and the LREE increase from east to west, while Sc decreases. These variations may be described as ranging from depleted (in incompatible elements) MORB-like compositions in the east, to enriched OIB-like compositions in the west. This is manifested by at least a three-fold increase in the concentration of the incompatible elements. In accordance with the model presented by Ellam (1992) it is proposed that this behaviour is controlled by the depth to the base of the lithosphere, which acts as an upper limit to melting. In this model the upper limit on the melting is much deeper under the older thicker lithosphere than it is under the young lithosphere. Thus melt composition will increasingly be influenced by the presence of residual garnet under older ocean crust. Even in the absence of residual garnet clinopyroxene in the upper mantle may be capable of retaining the HREE and some trace elements such as Y and Zr. Furthermore shallow mantle is more likely to have been subjected to one or more previous melting episodes beneath spreading ridges, leading to melts depleted in incompatible elements being derived from beneath young ocean crust. Rayleigh-type modelling for most sites along the Rodrigues Ridge produced only a poor correlation with the observed data. At several sites, notably RR3, some incompatible elements (eg. Zr, Y, REE) show buffered or decreasing trends with decreasing MgO. Although the causes of this behaviour remain ambiguous, it is possible that these melts have not experienced significant high-level fractionation, and this unusual behaviour has its origin in melt-rock reactions within the upper mantle. At these depths the partition coefficients for Zr, Y and the mid-HREE in clinopyroxene have been shown to be greater than unity (Blundy et al., 1998 and Vannucci et al., 1998). Thus the reaction between the melt and coexisting clinopyroxene within the upper mantle, in conjunction with the crystallisation of olivine, may explain how these buffered and declining trends (with declining MgO) have developed. To characterise their isotopic signature a subset of samples were analysed for Sr, Nd and Pb. Like the trace elements, the isotopes display clear linear trends with longitude. On paired isotope plots the Rodrigues Ridge lavas form similar linear trends between samples from the CIR (the Marie Celeste Fracture Zone) and the Reunion hotspot (notably Mauritius), suggesting that they are products of mixing between the mantle sources of the CIR and Reunion. It is proposed that the upper mantle has been passively contaminated by the Reunion plume (with circa 87Sr/86Sr = 0.7042, 143 Nd/ 144Nd = 0.7042, 206Pb/204Pb = 18.788, 207Pb/204Pb = 15.585 and 208Pb/204Pb =38.849). At shallow levels, immediately below the lithosphere, the upper mantle is made up predominantly of a depleted MORB-like source, with circa 87Sr/86Sr = 0.7031, 143Nd/144Nd = 0.51305, 206Pb/204Pb = 18.354, 207 Pb/204Pb = 15.517 and 208Pb/204Pb =38.214, while at greater depths more Reunion plume material is available. The observed linear array, on paired isotope plots, reflects the differing contributions made by these deep and shallow sources to the Rodrigues Ridge lavas. It is proposed that the Rodrigues Ridge was formed due to a build up of stress, possibly resulting from its proximity to the Ridge-Ridge-Ridge triple junction (see Patriat &Ségoufin 1988), causing the rigid African plate to rupture parallel to the principal stress direction. This would have resulted in decompression melting in the upper mantle, so facilitating rapid but short-lived volcanism. Although volcanism has ceased, there is still elevated heat-flow within this area, suggesting that the upper mantle is still anomalously hot (von Herzen & Vacquier 1966). Isostatic readjustment could account for the magmatic reactivation at 1.5 My which formed Rodrigues island.