Abstract In recent studies of crustal growth using global zircon Hf–O isotopic datasets, high-δ18O zircons are typically attributed to intra-crustal reworking during which very little juvenile mantle-derived magmas were added to the crust. Although arc magmas may originate from a high-δ18O mantle wedge, it has been difficult to decipher the contribution of high-δ18O mantle to zircon-saturated felsic magma due to superimposed intra-crustal processes. We address this issue by combining the data from high-δ18O zircon-bearing ultramafic cumulates and coeval lavas from a Cretaceous magmatic arc in southern Tibet. The cumulates mainly consist of different proportions of cumulus olivine and intercumulus amphibole. Amphibole analyses show a transition from increasing to decreasing Zr with increasing SiO2 (50–74 wt.%) contents in the intercumulus melts, indicating zircon saturation in late-stage interstitial melts. The εNd(t) values (2.4 ± 1.4) of the apatite grains crystallized before and after zircon remain almost constant. Interstitial zircons have δ18O (6.1–7.2‰) values similar to the earliest crystallized olivine (δ18O = 6.3–7.1‰) in the cumulates. The coeval lavas may represent the intercumulus melts extracted from amphibole-rich cumulates at different depths. Both the lavas and cumulates were ultimately derived from high-δ18O arc mantle modified by small amounts (<12%) of subducted sediments, and crystallized zircon during intra-crustal magma evolution without involving crustal contamination or melting. These high-δ18O zircons therefore are not products of crustal reworking, but record crustal growth during their crystallization (110 ± 2 Ma). Our study shows that the combination of zircon and olivine oxygen isotopes for ultramafic to felsic rocks is more effective than zircon data alone in evaluating the role of crustal growth vs. reworking in an arc system. The implication is that global zircon-based crustal evolution models that attribute all high-δ18O zircons to crustal reworking may conceal recent crustal growth.
Read full abstract