Recently there have been numerous debates on the source of intermediate subduction zone lavas, in particular adakites. These lavas bear a distinctive trace-element signature (Y<18ppm, Sr/Y>40, La/Yb > 20), which, it has been argued, is attributed to the direct melting of the downgoing slab. Adakites from the Mexican Volcanic Belt define a hydrous volcanic front, exemplified by high-MgO (4–9 wt%), intermediate lavas (52–63 wt% SiO₂), with hornblende common in those lavas containing >57 wt% SiO₂. These lavas contained as much as 7 wt% H₂O during phenocryst equilibration (900-1050^oC), representing a minimum in the source with water saturation as the maximum. The presence of hornblende lherzolite nodules in an andesitic lava is indicative of a mantle origin, as are hydrous experiments on partial melting of lherzolite, which match in temperature and composition the andesitic lavas of central Mexico. The association between the adakitic signature and hydrous magmas can be observed in a number of other locations, most notably the Cascades, which also display a hydrous front, and the Aleutians, where the hydrous adakitic lavas form discrete regions.

Ascent velocities calculated for the nodule-bearing lava, coupled with recent results from global U-Th-Ra isotopic surveys of arc lavas, demonstrate that arc magmas are transported at rates ca. 1 – 10 km/yr. At such rates, the extent to which a melt can react with the mantle and lower crust is limited. Thus, adakite genesis by a dacitic slab melt reacting with the mantle to form a high MgO andesite seems unlikely. We propose that the adakite trace-element signature is transported to the mantle by fluids from the breakdown of lawsonite and that the Mexican adakites represent hydrous mantle melts that ascend rapidly to the surface.