Silicic magmas, crystal mushes, granitic plutons and rhyolitic eruptions

A crystallising magma must necessarily pass through a mushy (crystal-dominated) state before it fully solidifies. Similarly, partially melted magma source regions start out as solid-liquid mixtures, albeit with different initial conditions and physical behaviours to melt crystallisation. Thus, localised crystal mushes must be geologically commonplace. Here we examine the pervasive paradigm in which crystal mushes are thought of as the main sources of erupted silicic magmas. Combining geophysical, petrological, chemical and isotopic evidence, as well as theoretical considerations, we emphasise the following points.

• Models of plutonic rocks as mush cumulates left in magma reservoirs after extraction of fractionated and eruptible rhyolitic magmas are untenable. Compositions of rhyolites and granitic rocks show that, in general and even in well-constrained crustal sections, these are neither compositional equivalents nor compositional complements.
• In rhyolitic rocks, apparent resorption textures in autocrysts and antecrysts should not necessarily be ascribed to mush heating events or percolative reactive flow of melt through mushes. Embayed and partially resorbed crystals in volcanic rocks, can also reflect rapid disequilibrium crystallisation and growth, partial resorption on near-adiabatic magma ascent or pre-eruption magma mingling/mixing. Likewise, the commonly monomineralic character of glomerocrysts shows that these cannot generally represent disaggregated crystal mushes.
• The concept of rheologically locked crystal mush is not soundly based in mechanics. Under shear stress, a magma reservoir with 20 vol.% (and less) silicic melt can undergo rapid flow and melt segregation, and potentially collapse any mush column that might exist.
• Geological mapping demonstrates that mafic floors to silicic plutons are uncommon, calling into question the idea of mush reactivation through heating by mafic magma influx.
• An implication of the mush model is that most magma, mush and rock that is generated cannot be erupted, and the plutonic:volcanic ratio is probably rather greater than the 10:1 that is generally supposed. Thus, very large silicic eruptions (>1000 to 10,000 km³ in erupted volume) pose problems for mush models, in terms of the complementary mush volume that would be required.

We recommend that mush-based petrogenetic models be seriously reconsidered. Crystal mushes play a role, but this model should not be invoked to explain all volcanism. We present an internally consistent vision for silicic magma systems, underpinned by fundamental geological, petrological and mechanical observations and principles. This model obviates the need for ubiquitous, expedient mush zones or columns, and allows the crust to remain mechanically stable.