The genesis of silicic arc magmas in shallow crustal cold zones

A number of currently popular models for the genesis of evolved arc-magmas (from basaltic andesite to dacite) invoke repeated intrusion, partial-melting and differentiation at the base of the crust. However, several observations suggest that this may be the exception rather than the norm: (1) geobarometry often indicates shallow pressure (0.1-0.3 GPa) evolution; (2) incongruent melting of amphibolite at elevated pressures should yield magmas in equilibrium with high pressure phases like garnet, but rare earth element patterns almost ubiquiously preclude this; (3) compositionally-zoned caldera forming eruptions suggest differentiation at near surface depths; (4) U-series data most commonly indicate differentiation over millennia time-scales. This requires rapid cooling that, in turn, is most easily explained by relatively small magma volumes undergoing crystal fractionation within the shallow (i.e. cool) crust. To further test these ideas, we combined published experimental-data for liquidus equilibria with appropriate silicic arc-magma compositions. On projections of the ternary liquidus system nepheline-silica-olivine, recent data for Tongan silicic lavas plot either on or close to low-pressure (1 atm) cotectics for the rocks' phenocryst phases, suggesting low-pressure differentiation. Using our own and published data from arc volcanoes around the world we find that the majority are consistent with differentiation at shallow depths, regardless of total crustal thickness. Combined with the typical timescales of differentiation, we estimate that the volumes of magma stored during differentiation in shallow crustal zones are usually on the order of only a few km3. There is also a clear role for mixing and recharge that involves magmas that are more deeply-sourced and primitive in character (typically evolved basalts and basaltic andesites). Whether the latter differentiated in the lower-crust or at the crust/mantle boundary has important implications for the constitution and average composition of arc crust. At present, we can only conclude that evidence for more silicic arc-magma generation at these depths is generally lacking.