Summary

Mantle replenishment in lithophile elements has been discerned in the patterns of trace elements and isotopes in lavas. One replenishment process is identified as metasomatic replacement, seen in ultramafic xenoliths brought up in high-velocity alkaline eruptions. Thus alkaline magmatism provides the best prima facie evidence of metasomatism and open-system conditions in the upper mantle. The list of added lithophile elements includes the following: H, C, F, Na, Al, P, S, Cl, K, Ca, Ti, Fe, Rb, Y, Zr, Nb, Ba and rare earths. Some metasomatism may be due to wall-rock alteration near magma bodies, but the evidence for metasomatism prior to melting opens the possibility that the process is a precursor to alkaline magmatism, giving the necessary source enrichment in lithophile elements. In some igneous provinces the metasomatism is widespread, intensive and pervasive; in others it appears as veining of variable intensity. Metasomatism as a large-scale process is best indicated by the widespread distribution of alkaline magmatism in space and time: volatile flux through the lithosphere would then be the necessary precursor of metasomatism and magmatism.

Volatile activity, metasomatism and melt enrichment clearly widen the scope for mafic magma generation in the mantle, but some long-standing problems of the alkaline associations (and indeed the calc-alkaline) also call for re-examination in terms of volatile activity in the lithosphere mantle. These include the diversity of magmas, the generation of large felsic volumes, and composition gaps in magma series. Experiments show that felsic minerals are stable to 30 kb, indicating the possibility of felsic-melt generation in the upper part of the mantle. A combination of volatile flux and melt percolation along geotherms that intersect the solidus at depths of less than 80 km would lead to enrichment and metasomatism, providing distinct mantle sources for felsic magmas. Initial (or residual) melts from such a region, as distinct from those from greater depths, would be constrained by equilibria involving felsic minerals. Thus an igneous cycle could generate a bimodal association, with felsic melts forming in the upper regime and the mafic melts originating at depths below the range of felsic-mineral stabilities. Such a magma system is consistent with the observed eruptive characteristics, explains the typical ultramafic nodule and megacryst suites in the alkali olivine-basalt association and is free of difficulties with relative volumes of melts, with eruption timing and with rapid changes in erupted compositions.