The formation of high-Sr/Y plutons in cordilleran-arc crust by crystal accumulation and melt loss

Abstract

Bulk-rock data are commonly used in geochemical studies as a proxy for melt compositions in order to understand the evolution of crustal melts. However, processes of crystal accumulation and melt migration out of deep-crustal, crystal-rich mush zones to shallower storage regions raise questions about how faithfully bulk-rock compositions in plutons approximate melt compositions. This problem is particularly acute in the lower crust of arcs, where melt reservoirs are subject to periodic melt extraction that leaves behind a cumulate residue. Here, we examine bulk-rock data from the perspective of high-Sr/Y plutonic rocks in the lower crust of a well-exposed Early Cretaceous cordilleran-arc system in Fiordland, New Zealand. We test the validity of using high-Sr/Y bulk-rock compositions as proxies for melts by comparing bulk-rock compositions to melts modeled from >100 major- and trace-element analyses of 23 magmatic clinopyroxene grains from the same samples. The sampling locations of the igneous clinopyroxenes and encompassing bulk rocks are distributed across ~550 km2 of exhumed lower crust and are representative of Mesozoic lower-crustal arc rocks in the Median batholith. We confirm that bulk-rock data have characteristics of high-Sr/Y plutons (Sr/Y >50, Na2O >3.5 wt%, Sr >1000 ppm, and Y <20 ppm), features that have been previously interpreted to indicate the presence of garnet as a residual or fractionating phase. In contrast to bulk rocks, igneous clinopyroxenes have low Sr (<100 ppm), high Y (25–100 ppm), and low molar Mg# [100 × Mg/(Mg + Fe)] values (60–70), which are consistent with derivation from fractionated, low-Sr/Y melts. Chondrite-normalized rare-earth-element patterns and Sm/Yb values in clinopyroxenes also show little to no evidence for involvement of garnet in the source or in differentiation processes. Fe-Mg partitioning relationships indicate that clinopyroxenes are not in equilibrium with their encompassing bulk rocks but could have been in equilibrium with melt compositions determined from chemometry of coexisting igneous hornblendes. Moho-depth calculations based on bulk-rock Sr/Y values also yield Moho depths (average = 69 km) that are inconsistent with Moho depths based on bulk-rock Ce/Y, contact aureole studies, Al-in-horn- blende crystallization pressures, and our modeled clinopyroxene crystallization pressures. These data indicate that most Mesozoic high-Sr/Y bulk rocks in the lower crust of Fiordland are cumulates formed by plagioclase + amphibole + clinopyroxene accumulation and interstitial melt loss from crystal-rich mush zones. Our data do not support widespread fractionation of igneous garnet nor partial melting of a garnet-bearing source in the petrogenesis of these melts. We speculate that melt extraction and the production of voluminous cumulates in the lower crust were aided by unusually high heat flow and high magma addition rates associated with an Early Cretaceous arc flareup. We conclude that bulk-rock compositions are poor proxies for melt compositions in the lower crust of the Median batholith, and geochemical modeling of these high-Sr/Y bulk rocks would overemphasize the role of garnet in their petrogenesis.