A Lithospheric Mantle Source for the Cape Verde Island Magmatism: Trace Element and Isotopic Evidence from the Island of Fogo

Abstract

Young silica undersaturated rocks from the island of Fogo, in the southern Cape Verde islands, show striking correlations between rocks types, incompatible trace element (ITE) ratios and Pb-Sr-Nd isotopic compositions that provide new insights into the generation of the HIMU source. The Cape Verde islands are situated on relatively old and thick oceanic lithosphere that, unlike most hotspots world-wide, has remained relatively steady compared to the underlying mantle plume for a considerable period of time. The magmatism has resulted in highly silica undersaturated rocks, including some of the rare cases, globally, of oceanic carbonatites. The lavas studied are mainly primitive (MgO = 5-12%) nephelinites and tephrites, with less silica undersaturation characterising the historic volcanism. ITE ratios range widely, with Zr/Hf = 37-60, Nb/U = 33-92 and Th/U = 2.7-4.2, whereas Nd and Sr isotopic compositions show a restricted range with 143Nd/144Nd = 0.51272-0.51282 and 87Sr/S68r = 0,7034-0.7038. Pb isotope compositions are within the typical range of 'young HIMU' OIB: 2~176 = 18 .8 -19 .4 , 2~176 = 15 .52 -15 .59 , 2~176 = 38.7-39.1. The isotopic data agree well with previously published results (Gerlach et al., 1988), albeit with tighter arrays. Measured oxidation states of iron vary by a factor of two, and show good correlations with rock types and degrees of silica undersaturation, but also with isotopic and ITE ratios. The Fogo arrays are in accordance with a model of binary mixing between two dominating source components beneath the Cape Verde islands, a HIMU and a EMI-type source. The most distinctive feature of the EMI-type source is its low 143Nd/144Nd ~0.51260 and high Ba/La ratios, and it is interpreted to represent delaminated subcontinental lithosphere. The HIMU component in the Cape Verde islands is characterised by high oxygen fugacity and high ratios of U/Pb, U/Th, Zr/Hf and LREE/HFSE, features that are shared with many carbonatites. It also has high present-day 143Nd/144Nd, -0.51305 and negative A7/ 4 values (down to -5.5), the latter indicating it to be much younger than the 1.5-2 Ga source age, typically inferred for the HIMU source. There are well-defined arrays in diagrams of 2~176 versus ~t (238u/Z~ and of 2~176 versus (232Th/2~ that imply concordant ages o f -120 Ma for the two systems (Fig. 1). No such good correlations exist for the Sm-Nd and Rb-Sr systems, so mixing is restricted to the source regions. For the U-Th-Pb arrays to have preserved significant source age information, provides tight constraints on aspects of the melt generation models. The preferred model includes the derivation of the HIMU component from DMM by mantle degassing (due to the high 143Nd/144Nd in the HIMU endmember), and the best-fit model ages for the HIMU component o f -200 Ma broadly overlap the timing of Pangea break-up at this latitude. In such a model, the source region is situated within the oceanic lithosphere, and the contribution from the plume itself is limited. Geophysically, the model is consistent with (1) the considerable thickness of the Cape Verde plate (100-125 km) which would diminish the probability of any extensive melting in the plume, and (2) the fixed position of the ocean plate, that would allow for heat to be conducted from the plume into the lithosphere. The temporal variations are attributed to exhaustion of the HIMU component, possibly combined with progressive thermal erosion of the lithosphere by the rising plume, causing melting to proceed at higher levels in the oceanic lithosphere. The presented model suggests that 'young HIMU'