Various hypotheses for the origin of alkaline sodic mafic magmas have been proposed. This diversity of models is mainly related to the various constraints used to develop them. The goal of this paper is to test these different models using petrological and geochemical constraints in an attempt to understand why alkaline sodic rocks are so similar even while their environment of formation varies from oceanic to continental rift. Incompatible trace-element contents of alkaline basalts from ocean islands and continents show that the sources of these rocks are more enriched than primitive mantle. A fundamental question then is how the sources of alkaline rocks acquire these trace-element enrichments.

Recycled oceanic crust, with or without sediment, is often invoked as a source component of alkaline magmas to account for their trace-element and isotopic characteristics. However, the fact that melting of oceanic crust produces silica-rich liquids seems to exclude the direct melting of eclogite derived from mid-ocean-ridge basalt to produce alkaline lavas. Recycling oceanic crust in the source of alkaline magma requires either (1) that the mantle “digests” this component producing metasomatized CO₂-rich peridotitic sources or (2) that low-degree melt from recycled oceanic crust reacts with peridotite in the presence of CO₂, producing low-silica alkaline melt by olivine dissolution and orthopyroxene precipitation. These two hypotheses are plausible in terms of major elements. However, they have specific implications about the type and proportion of recycled lithologies present in the asthenosphere to explain the specific trace-element pattern of intraplate alkaline lavas. A third hypothesis for the formation of alkaline magmas is the melting of metasomatized lithosphere. In this model, the major- and trace-element signature of alkaline magma is not controlled by the asthenospheric source (i.e., the amount of oceanic crust or CO₂ present in the asthenosphere), but by the petrological process that controls the percolation and differentiation of low-degree asthenospheric melts across the lithosphere. This process forms amphibole-bearing metasomatic veins that are a candidate source of alkaline rocks. This hypothesis offers an explanation for the generation of the Na-alkaline lavas with similar major- and trace-element composition that are observed worldwide and for the generation of K- and Na-alkaline magma observed in continental settings. This hypothesis requires the formation of significant amounts of metasomatic veins within the lithosphere.

Qualitative analyses of the thermal implication of the potential models for the generation of alkaline rocks demonstrate that such magma requires low potential temperature (Tp: 1320 °C to 1350 °C). If temperatures are higher, melting of the convecting mantle will erase any signature of low-degree melts produced from fertile mantle lithologies. This analysis suggests that a role for hot thermal plumes in the generation of intraplate volcanoes dominated by alkaline magmas is unrealistic.