Search

Explosive silicic eruptions pose a significant threat to society, yet the development and destabilization of the underlying silicic magmatic systems are still controversial. Zircons provide simultaneous information on the trace element composition and age of silicic magmatic systems, while melt inclusions in quartz and plagioclase yield important constraints on their volatile content as well as magma storage depth. Melt inclusions in zircons (MIZs) combine these data from a single mineral grain, recording the age, storage depth, temperature, and composition of magmas, and thus provide unique constraints on the structure and evolution of silicic magmatic systems. We studied MIZs from the Laguna del Maule (LdM) volcanic field in the southern Andes that is among the most active Pleistocene-Holocene rhyolitic volcanic centers worldwide and a potentially hazardous system displaying inflation rates in excess of 25 cm/yr. The host zircon ages suggest that the LdM MIZ record extends to ~ 30 kyr before eruption, in contrast to the melt inclusions in LdM plagioclase and quartz crystals that formed only decades to centuries before eruption. The major element compositions of MIZs are minimally affected by post-entrapment crystallization, and agree well with the LdM rhyolitic whole rock data. The MIZs record long-term differences in zircon-saturated melt composition between two eruptive units (rdm: Rhyolite of the Laguna del Maule vs. rle: Rhyolite of Los Espejos). The more evolved major element composition of rle MIZs than rdm MIZs, suggests a long-term deeper connection of the rdm crystal mush to a more primitive magma body than that of the rle. The evidence of slow H diffusion observed in MIZs suggest that their H2O contents are not significantly affected by diffusion of H through the host zircon. The magma storage pressures of 1.1 to 2.8 kbars recorded by the H2O contents of rdm and rle MIZs are consistent with the optimal emplacement window (2.0 ± 0.5 kbar) of silicic magma reservoir growth, storage, and eruptibility based on thermomechanical modeling (Huber et al. 2019). [ABSTRACT FROM AUTHOR]