No abstract

Potassium-rich adakitic rocks have been used to infer high-pressure (HP) melting of thickened or foundered mafic lower crust in post-collisional settings. However, their origin remains debated because of their potassic rather than sodic nature. We address this debate by investigating the source of post-collisional Oligocene high-K adakitic granites in southern Tibet, which are widely attributed to melting of overthickened mafic lower crust. Our new data provide evidence for the generation of these high-K adakitic granites by anatexis of exposed migmatitic intermediate to felsic arc orthogneisses. These granites contain high-Th/U Oligocene magmatic zircons (30–22 Ma), and inherited zircons (66–48 Ma) with the same εHf(t) (0 to +12) as the coeval migmatites. The migmatites have in turn low-Th/U metamorphic Oligocene zircon rims around 66–48 Ma magmatic zircon cores recording the anatectic event at 29–25 Ma. Phase equilibrium modeling yields an anatectic temperature of ~740 °C and pressure (P) of ~0.9 GPa and shows that garnet is stable at P > 0.7 GPa in the melt-present field. The results indicate that high-K adakitic magmas are derived from melting of older intermediate to felsic arc rocks at intermediate pressures, in the garnet stability field, without any involvement of HP melting of metabasaltic rocks. We propose that hybridization between such purely crustal magmas and subcontinental lithospheric mantle–derived shoshonitic magmas results in the hybridized post-collisional Oligocene–Miocene mafic microgranular enclave–bearing potassic adakitic granitic rocks in southern Tibet.

Thermo-chemical modelling and chemical data suggest that the genesis of arc melts commonly involves re-melting of older intrusions, triggered by the injection of mantle-derived melts, followed by magma mixing. Remelting and mixing may lead to complex zircon populations, which can be used to gain insights into the conditions of mixing. This paper investigates a case where such processes can be studied through the compositional and thermal record provided by zircons preserved in a diorite dyke swarm that remelted host rock tonalites in the Gangdese Batholith in Tibet. Autocrystic zircons from the diorite yield consistent ages of 46-47 Ma even though they can be altered, having highly enriched trace elements, reaching ~1 wt% in Fe, Ca, Y, U, Th, and anomalously high values of LREE and Ti. Granitic magmas derived from the tonalite, back-veined the dykes and mixed with the dioritic mush, transferring small quantities of 77-79 Ma zircon xenocrysts. The xenocrysts are euhedral with little evidence for resorption, indicating they were apparently stable during the process of tonalite anatexis and transfer to the diorite magma. This requires that: (i) tonalite melting occurred at low temperatures with minimal zircon dissolution, and (ii) the diorite either cooled rapidly before significant resorption of the grains or was already saturated in zircon when mixing occurred. Zircon saturation temperatures of the diorite are relatively low, indicating that xenocrystic zircons were transferred to a highly crystalline dioritic mush. This requires either liquefaction by injection of the back-veining felsic magma to allow for mixing or pervasive throughflow of the diorite mush by the felsic magma leaving behind zircon xenocrysts. The findings suggest that the dykes triggered low-T, water-fluxed melting of the host tonalite, and that these anatectic melts invaded the diorite mush initially through the interstices leading to mixing and this may have caused the breakdown of the crystalline framework leading to liquefaction and renewed magma flow. Dyking and assimilation of wall-rock through back-veining as recorded in these rocks could be common in transcrustal arcs. However, this process could be hidden due to a combination of the similarity in the isotopic and chemical nature of arc rocks, and resorption of zircon xenocrysts during mixing. This process may explain some complex chemistry of arc magmatic rocks and their minerals that are not easily explained by endmember models, such as pure re-melting of older arc rocks or fractional crystallization of mantle-derived melts.