Revisiting the discrimination and distribution of S-type granites from zircon trace element composition

Trace element compositions of zircon can be used to estimate the chemistry of their host magmas; as such they provide a useful tool in zircon provenance, and in the assessment of changing magma chemistries in time and space. Granites derived from the melting of sedimentary protoliths (S-types) have previously been discriminated by their P contents and P vs. REE+Y correlations, largely based on data from the Lachlan Fold Belt. Using a range of magmatic suites from different locations, we show that this discrimination commonly fails to discriminate S-type granite from others. We propose an alternative discrimination tool, based on a plot of Ce/U vs. Th/U, which makes use of low LREE/U and Th/U in metapelite-derived melts. Through coupled thermodynamic and accessory mineral saturation modelling, we demonstrate that these low ratios can be explained by monazite co-crystallisation. We demonstrate that Himalayan S-types, which are inferred to have formed from partial melting of metapelite, and thus can be classified as pure S-types, exhibit the lowest Ce/U and Th/U ratios, and overlap those of metapelitic zircon. Granites formed in oceanic arcs (I-types) and mantle-derived suites both have the highest Ce/U and Th/U ratios. Other S-types, such as those known to have mixed sedimentary and igneous protoliths, which we term Hybrid S-types, form a field overlapping pure I- and S-types. We use Ce/U versus Th/U to demonstrate the dominant I-type origin to early Earth (>3.6 Ga) zircon, and using a large detrital zircon database we assess the proportion of S-type zircon through Earth history. In contrast to previous findings, we find that the supercontinent Rodinia had a normal abundance of S-type zircon, as with other supercontinents, and that instead the period 1.7–1.2 Ga exhibits a marked low in S-type zircon, corresponding to fewer continental collisions.