LAUSR

Granites are generally the final products of crustal anatexis. The composition of the initial melts may be changed by fractional crystallization during magma evolution. Thus, it is crucial to retrieve the temperature and pressure conditions of crustal anatexis on the basis of the composition of the initial melts rather than the evolved melts. Here we use a suite of ∼46–41 Ma granites from the Himalayan orogen to address this issue. These rocks can be divided into two groups in terms of their petrological and geochemical features. One group has high maficity (MgO + FeOt = 2–4 wt%) and mainly consists of two-mica granites, and is characterized by apparent adakite geochemical signatures, including high Sr concentrations and Sr/Y and La/Yb ratios, and low concentrations of heavy rare earth elements and Y. The other group has low maficity (MgO + FeOt <1 wt%) and consists of subvolcanic porphyritic granites and garnet/tourmaline-bearing leucogranites. This group does not possess apparent adakite signatures. The low-maficity group (LMG) has lower MgO + FeOt contents and the high-maficity group (HMG) has higher Mg# compared with initial anatectic melts determined by experimental petrology and melt inclusions studies. Petrological observations indicate that the HMG and the LMG can be explained as a crystal-rich cumulate and its fractionated melt, respectively, such that the initial anatectic melt is best represented by an intermediate composition. Such a cogenetic relationship is supported by the comparable Sr–Nd isotopic compositions of the two coeval groups. However, these compositions are also highly variable, pointing to a mixed source that was composed of amphibolite and metapelite with contrasting isotope compositions. We model the major and trace element compositions of anatectic melts generated by partial melting of a mixed source at four apparent thermobarometric ratios of 600, 800, 1000 and 1200 °C GPa–1. Modeling results indicate that melt produced at 1000 °C GPa–1 best matches the major and trace element compositions of the inferred initial melt compositions. In particular, a binary mixture generated from 10 vol% partial melting of amphibolite and 30 vol% melting of metapelite at 850 ± 50 °C and 8·5 ± 0·5 kbar gives the best match. Therefore, this study highlights that high thermobarometric ratios and subsequent fractional crystallization are responsible for the generation of the apparent adakitic geochemical signatures, rather than melting at the base of the thickened crust as previously proposed. The thermal anomaly responsible for the Eocene magmatism in the Himalayan orogen was probably related to asthenosphere upwelling in response to rollback of the subducting Neo-Tethyan oceanic slab at the terminal stage of continental collision between India and Asia. As such, a transition in dynamic regime from compression to extension is necessary for the generation of high thermobarometric ratios in continental collision zone. Therefore, after correcting for potential effects of fractional crystallization and crystal accumulation on melt composition, granite geochemistry coupled with thermodynamic modeling can better elucidate the petrogenesis of granites and the geodynamic mechanisms associated with anatexis at convergent plate boundaries.