A crystal window into the crustal arc magma plumbing system

Abstract

Open magmatic processes, including magma mingling, fractionation and assimilation, are responsible for the diversity of magma and crystal compositions common in subduction zone settings. This thesis focuses on investigating pre-eruptive open-system processes using crystal-specific studies. Particularly, we develop calcic amphibole as a robust recorder of magmatic processes. Amphibole is a common mineral phase in water-bearing arc magmas. Its stability is a complex function of temperature, pressure, oxygen fugacity, and melt and volatile compositions. We have developed a new multiple regression analysis of published trace element partitioning data between calcic amphibole and melt. We are able to retrieve statistically significant relationships for REE, Y, Sr, Pb, Ti, Zr and Nb. We also present new pressure-independent and temperature-independent empirical chemometric equations to predict melt major element chemistry from amphibole crystal compositions. This enables us to reconstruct melt chemistry from in situ analyses of amphibole in magmas and plutonic xenoliths. Linking these inverted melt compositions to the observed crystal textures allows us to make robust interpretations of magmatic processes throughout the magma plumbing system.
We take the 1951 eruption products of Mt. Lamington volcano as a primary case study. Mt. Lamington is a composite volcano sitting on the Papuan Ultramafic Belt (PUB) ophiolite. The 1951 eruption produced andesitic dome lavas with numerous basaltic-andesitic enclaves and a few PUB ultramafic xenoliths. The mineralogy of the enclaves is dominated by amph+plag, similar to the andesitic lava hosts. The textures of the enclaves vary from fine-grained diktytaxitic to coarser-grained plutonic textured. We interpret this variation to result from variable cooling rates in the enclave-forming magma body when it invades the overlying andesite. The diktytaxitic enclaves contain variable proportions of host-derived amph+plag antecrysts and xenocrysts of ol+sp±cpx±amph with disequilibrium textures, indicating interaction with host lava and assimilation of foreign materials, respectively. A previous study argued that the olivine xenocrysts with chromian spinel inclusions are derived from the PUB, and thus that the PUB contaminated the Mt. Lamington magmas. We demonstrate that this is highly unlikely on the basis of morphological and compositional discrepancies between PUB ol+sp and the xenocrysts. The olivines are considered to represent crystal mush fractionated from precursor(s) of andesitic and/or pre-1951 shoshonitic lavas. Their presence in enclaves represents recycling of earlier-fractionated components through magma recharge.
We also revisit and bring new insights on magmatic processes shaping the Ongatiti eruption of Mangakino volcano, amphibole-bearing plutonic nodules exposed in the lavas in Grenada, and the 1991 eruption of Mt. Pinatubo. We demonstrate that reconstructed melt compositions inferred from the rims of amphiboles in pumice clasts of the Ongatiti ignimbrite are in good agreement with the matrix glass compositions. This suggests that equilibrium between the amphibole rims and melts of matrix glass compositions is achieved. The cores of amphiboles from the Ongatiti ignimbrite show large compositional variation as well as disequilibrium textures (e.g. patchy zoning, resorption/dissolution texture), and the predicted melt compositions also display large variations. We interpret that these variations may be due to different degrees of equilibration of the amphiboles derived from crystal mushes with evolved melts over a range of timescales. This interpretation is an alternative to the model provided by an earlier study which instead suggests that source heterogeneity is a major contribution to the crystal chemical diversity in the Ongatiti ignimbrite.
For amphiboles in plutonic nodules in Grenada lavas, we are able to predict the melts compositional variations from amphiboles in clinopyroxenite, hornblendite and hornblende gabbro xenoliths, in consistency with melt inclusions hosted in those cumulates. We interpret that the variations may be a result of in situ melt evolution due to extensive crystallization of the cumulate mineral phases, or equilibration of cumulate fragments with later evolved melts.
Bimodal amphibole populations and hence inferred melt compositions in Mt. Pinatubo indicate magma mingling process, in consistency with conclusions of earlier studies. The inferred melt trace element compositions from bimodal groups also record co-crystallization of plagioclase, ilmenite, zircon and apatite together with amphiboles to varying extent, and the interpretation can be generally supported by petrography evidences.