Tracing garnet clinopyroxenite formation and recycling using arc magma geochemistry: an example from the modern Andean continental arc

Foundering of lower crustal material in arcs is a key mechanism controlling the composition and evolution of Earths continental crust, and igneous geochemistry is an important tool used to elucidate this process. Studies have used first row transition element (FRTE) ratios for fingerprinting pyroxenite melting in arcs induced by garnet pyroxenite foundering. However, interpreting the geochemical linkages between lava FRTE geochemistry and lower crust processes in subduction zones is challenging because lower arc crust assemblages are typically not accessible, and direct petrologic observations remain scarce. The Granatifera Tuff, a Quaternary eruptive succession in the Andean Northern Volcanic Zone (NVZ), Colombia, hosts basaltic bombs and (ultra)mafic xenoliths from the density-unstable section of the modern arc crust. Thus, this site provides a unique opportunity to explore connections between the modern lower crustal dynamics and magma geochemistry in an active continental arc. We report major and trace element geochemistry for garnet clinopyroxenite cumulates and near-primitive arc magmas from the Granatifera Tuff and use nonmodal batch melting models to predict the FRTE compositions of melts that would be produced by partially melting the cumulates. Foundering of the pyroxenites may produce melts with lower Mn/ FeT*100 and Mn/Zn, and higher Zn/FeT(*104) than peridotite-derived melts, and the Granatifera magmas may record a minor pyroxenite-melt contribution. However, our results indicate that FRTEs ratios cannot unequivocally distinguish pyroxenite-melting from the effects of intra-crustal differentiation. Furthermore, garnet/clinopyroxene partition coefficients for many FRTEs measured from our unretrogressed samples do not agree with experiments, indicating that the existing partition coefficients may not be applicable to arc-related igneous pyroxenites. In addition, we show that garnet pyroxenite crystallization in the lower crust can produce the adakite-like trace element signatures observed in NVZ andesites and dacites. Altogether, this study highlights the challenges of using FRTEs to trace pyroxenite melting in arcs and demonstrates that adakitic magmas in the North Andes are a consequence of garnet pyroxenite crystallization and not slab melting.