LAUSR

Abstract It is popular in geochemical models to assume that chemical differentiation occurs by crystal fractionation from high-melt-fraction magma reservoirs. However, a “crystal mush” reservoir characterized by low-melt-fraction has been gradually accepted in the past decade. Magma differentiation in this scenario remains poorly understand. Located in the East Kunlun orogenic belt, northern Tibet Plateau, the Baishiya granodiorite is an ideal example to investigate magma reservoir processes. Although major elements of the Baishiya granodiorite are relatively homogeneous, there is a distinct variation of magmatic texture, trace elements concentrations, and isotopic ratios from top to bottom. The textural and geochemical signatures above cannot be explained by assimilation-fractionation crystallization, magma mixing or convective self-mixing alone. Within the concept model of crystal mush, we propose a multistage formation process of the studied granodiorite. First, a rhyolitic mushy reservoir once existed. It was replenished by episodic basaltic melts from below, resulting in the textural coarsening of K-feldspar. When the heat and/or melts injection was high enough, the crystal mush could be gradually rejuvenated and subsequently stirred. Some K-feldspar antecrysts would be partly recycled to form mantled plagioclase through convective self-mixing. Thus, the evolved and buoyancy melts exhibited an equilibrated geochemical signature with plagioclase, and ultimately formed the porphyry phase on the top. Moreover, recycling of K-feldspar antecrysts contributed to a depletion of heavy rare earth elements and Y contents, as well as an elevation of Cr concentrations and Eu/Eu* ratios in the residual crystal mush, and finally formed the porphyritic granodiorite. Due to a continuous injection and convective self-mixing, the overall magma reservoir became more mafic gradually until it reached a granodioritic composition. Within a waning intensity of magma flux, the magma reservoir would be locked again. The injected melts would go through a rapid crystallization, and subsequently form microlites, such as the dioritic enclaves and the ground mass of the studied granodiorite. Consequently, traditional mechanisms that are based on low-crystallinity magmas may be a specific part of the protracted history rather than the whole story of the magma reservoir. Geochemical index of granitoid rocks should be used with caution due to this inherited affinity. It should be prerequisite to investigate crystal genesis carefully before utilizing geochemical diagrams.