The results of recent experimental and geochemical studies demonstrate that typical, low-K, island arc dacites (IAD) and tonalities have liquidus water contents below 6 wt %, while many arc basalts, including evolved, low-Mg high-alumina basalt (HAB; H2O≥2 wt %) are water-rich. If these water contents are typical of arc basalts, fractionation at pressures of 200 MPa or more would produce water-rich magmas with major element chemical characteristics unlike >99% of observed IAD. In light of this, plausible mechanisms for IAD genesis include (1) dehydration melting of amphibolitized arc crust and (2) low-pressure (less than 200 MPa) fractionation or assimilation/fractional crystallization (AFC) (accompanied by devolatilization) of hydrous arc basalts. In general, a partial melting origin is favored for tonalitic plutons emplaced at pressures ≥200 MPa, for bimodal suites where geochemistry rules out a genetic relationship between the mafic and silicic end-members, and for IAD with isotopic characteristics distinct from potential basaltic or andesitic parents. Evidence for partial melting of amphibolite yielding IAD melts has been documented in several ancient island arc complexes. A fractionation origin is favored where there is isotopic homogeneity in a basalt-dacite system and for dacites having very low concentrations of incompatible trace elements. Bimodal magnetism in general and high concentrations of incompatible elements in silicic magmas appear to favor a partial (batch) melting origin over Raleigh fractionation but can also result from convectively driven batch fractionation processes (Brophy, 1991).

Although both fractionation/AFC and partial melting may be legitimately invoked to explain dacitic magmatism in a given situation, a general model must also account for the absence of high-pressure fractionates of hydrous basalts. This observation seems to favor an important role for amphibolite melting in IAD genesis. It is also possible, however, that physical factors (e.g., neutral buoyancy) promote the formation of basaltic magma chambers in the upper crust or that convective processes, enhanced by large temperature gradients in the upper crust, may favor upper crustal over lower crustal fractionation. ¿ American Geophysical Union 1995