ABSTRACT A compilation of 561 recent basalt analyses including major/trace elements, Nd‐, Pb‐ and Sr‐isotopes from the main volcanic zones of Iceland is benchmarked in a novel way with reference to average normal mid‐ocean ridge (N‐MORB) and average global intraplate basalt (WLD) compositions. Ten active‐rift, off‐rift and Tertiary zones are compared with surrounding sea floor basalts. The zones are effectively distinguished by magnesium number (Mg#) compared with various incompatible trace element ratios (e.g., Ti/K and Zr/Y). The ternary Th–Hf/3‐Ta plot distinguishes the rift zones, each plotting closer to N‐MORB than WLD, and the off‐rift zones plotting closer to WLD and other Atlantic Ocean islands. Three of the rift zones include high‐Mg# basalts that are unusually enriched in Ni and Cr, displaying very high Ni/Ta and Ni/Th ratios. The average compositions of the zones normalised to WLD reveal distinctive incompatible trace element and isotope ratio profiles. The active rift zones display profiles close to N‐MORB that contrast with the off‐rift zones. The rift zones and Reykjanes Ridge basalts are variably depleted in radiogenic Pb isotopes relative to WLD, with the average Pb‐isotope ratios for the northern rift zone being more depleted than N‐MORB. A petrogenetic model involving mantle sources that have been variably veined with small fractions of low‐solidus veins derived from melts of subducted crustal materials over the past 2 Ga is consistent with the observed trace element and isotope distributions. Combining such a mantle with sequential dynamic partial melting (with some melt retention) in a progressively rising mantle plume can conceptually explain the observed local and regional basalt geochemical variations observed in Iceland. In conjunction with mantle heterogeneities and dynamic partial melting, lower crust magma mixing and fractionation are also required to explain details of the observed local compositional distributions in many of Iceland's volcanic systems.
David A. Wood (Fri,) studied this question.