Chasing the mantle: Deciphering cryptic mantle signals through Earth's thickest continental magmatic arc

•Mafic mineral assemblage records high-Mg basalt beneath CVZ.•High-Mg basalt consistent with partial melting of wet, oxidized mantle peridotite.•Modification at the crust-mantle boundary dominates early evolution of CVZ arc magmas.•Primitive CVZ arc magmas are more oxidized and hydrous than CVZ back-arc magmas. Understanding the magmatic processes occurring in the upper mantle and lowermost crust above subduction zones is paramount if we hope to understand the primary mechanisms driving arc magmatism. In the Central Volcanic Zone of the Andes (CVZ), extreme crustal thicknesses and the prominence of silicic magmatism in the Neogene have prevented a clear understanding of the primitive magma input into what is arguably the Earth's archetypal active continental margin. However, we propose that phase chemistry and equilibria from mafic mineral assemblages in basaltic andesite enclaves from the Purico-Chascon Volcanic Complex reveal rare constraints on the character of primitive mantle-derived magmas feeding the CVZ arc. We interpret olivine, Cr-spinel, and olivine-hosted melt inclusion compositions from these mafic assemblages to record the partial melting of wet, oxidized mantle peridotite and the production of high-Mg basalt in the upper mantle. In turn, clinopyroxene and spinel-free olivine are interpreted to reflect magma modification in the lower crust and the adiabatic ascent of differentiated basaltic magma up to ∼20 km. Interestingly, olivine-hosted Cr-spinel compositions suggest that magmas in the Purico-Chascon system ascend from the upper mantle to surface with no significant change in Fe-oxidation state despite obvious differentiation and crustal interactions. Finally, comparison of high Mg-basalt in melt inclusions from Purico-Chascon (CVZ arc) and CVZ back-arc lavas show that the magmas may share a common peridotite source. However, the peridotite beneath the arc appears to be more hydrous and oxidized than that beneath the back-arc, confirming that back-arc lava compositions may not adequately represent mafic magma input into the arc. We posit that the relationships between H2O-content, oxidation conditions, and mafic phase equilibria discussed in this study are directly linked to subduction zone geometry, and thus may be applicable to arcs globally.