Melting, Melt Reservoirs and Magma Transport in Earth's Continental Crust - How Does the Top-down View from the Igneous World Compare With the Bottom-up View From the Metamorphic World?

At the volcanic end of the spectrum, the evolution of large, typically caldera-forming, rhyolitic magma systems in believed to occur in shallow-crustal magma chambers. Magma chambers are viewed either as bodies of fractionating liquid, or as bodies of crystal mush that may be activated by input of new melt. However, the extent to which any pluton represents some integrated expression of either end member is unclear. Furthermore, in the deeper crust, where evidence of melting is preserved in anatectic migmatites and residual granulites, magma chambers of any kind appear to be absent. The main magma ascent mechanisms are: low-inertia flow or diapirism; fracture-driven flow or dyking; and, pervasive viscous (visco-elastic) flow in self-generated melt-induced shear and fracture systems. The first of these involves melt plus residue, implies partial convective overturn and is limited by the brittle-viscous transition zone; a second mechanism, such as fracture-driven flow, might enable draining of melt into an upper crustal chamber. Fracture-driven flow or dyking has not been demonstrated as a viable mechanism for extracting melt from melt-bearing crustal sources. Viscous (or visco-elastic) flow in self-generated melt-induced shear and fracture systems appears to conform to observations from (some) residual high-grade metamorphic complexes, and appears to be consistent with observations of (some) mid-crustal plutons in continental arc environments. Orogens are complex systems that are self-organized at all scales of observation. The strain field, which emerges under subsolidus conditions, controls the initial distribution of melt. In the transition from grain-scale to channelized flow in veins the direction of melt flow will be controlled by elements of the metamorphic fabric, particularly the foliation and lineation. In strike-slip shear zones, and in zones of highly oblique transpressive deformation, melt flow is expected to be sub-horizontal in the plane of the foliation and along the lineation. Thus, melt flow is likely to be along the layering parallel to a stretching lineation down gradients in melt pressure to dilatant sites such as interboudin partitions, shear bands and sheets. With increasing melt volume a mesoscale shear and fracture system develops, in which the direction of greatest mesoscale permeability will be vertical. The system evolves to achieve the optimum dynamic structure according to the rate of melt production; in this manner, the melt flow network evolves to a critical point enabling transfer of melt from lower to upper crust. Initially the system may be conditionally open, but with increasing melt production it transitions to a fully open system at the critical point (self-organized criticality) enabling the source to be rapidly drained of melt. There is a tendency to separate the processes of melt generation and segregation from melt extraction, ascent and emplacement. However, this separation is an artificial one, and there is not a simple succession of events without feedback. Melt distribution at depth controls strain distribution in the overlying crust, although strain distribution, in turn, controls melt ascent and emplacement. There is a coupled thermal weakening associated with melt segregation, ascent and emplacement, and, therefore, a feedback relation between these processes and deformation. Segregation, ascent and emplacement of leucogranite magma is intimately linked to regional-scale orogenic deformation.