Global systematics of formation conditions of subduction zone magmas and their tectonic implications

Subduction zone magmatism plays an important role in material recycling of the earth’s interior, and it is imperative to understand melt generation mechanisms in the wedge mantle. The subduction zone is more complicated and diverse as a magma generation site than that of mid-ocean ridge or hot spot mostly because of the more significant and variable contribution of H2O-rich fluid expelled from the subducting slab, and because of the complexity of thermal and flow structures in the wedge mantle. The spatial variations of compositions of primary magmas in subduction zones have been recognized (e.g., Kuno, 1966; Sakuyama and Nesbitt, 1986), and are attributed to the diversity of melt segregation depth, degree of melting, or involvement of hydrous fluids with peculiar geochemical signature. It is particularly critical if melting at subduction zone is controlled mostly by the addition of H2O-rich fluid (Tatsumi, 1986; Iwamori, 1998) or by thermal and flow structure in the wedge mantle (Plank and Langmuir, 1993; Schmidt and Poli, 1998). In order to address this issue, quantitative estimation of melting conditions with clarification of the critical tectonic factors controlling magma generation is requisite and has been attempted in many studies with limited success. In this study, melting conditions of frontal volcanoes of world subduction zones are quantitatively estimated on the basis of major element composition of volcanic rocks. For quantitative estimation of melting conditions, we adopted appropriate models not only for mantle melting but also for fractional crystallization in the crust, simultaneous determination of both of which ensures the consistency of estimation procedure (Ogitsu and Ozawa, in prep.). Unknown parameters are optimized by least-squares method with input from a major element data of volcanic rocks. This approach is applicable to more differentiated rocks, which are not suitable for olivine addition methods widely used in estimation of primary magmas and melting conditions. We have applied this method to basalt-basaltic andesite from more than 30 frontal volcanoes of 13 subduction zones. From the estimated melting conditions, we have the following important results. (1) Volcanoes from a specific subduction zone show consistent melting conditions particular for the arc. (2) The degree of melting and melting pressure show a good positive global correlation. (3) The melting temperature and melting pressure also show a good positive global correlation. (4) The degree of melting and H2O content in the mantle show a positive global correlation at low melting degree, but significant scatter at higher melting degree. The positive correlation between degree of melting and melting pressure suggest that the melting in the wedge mantle is primarily controlled by decompressional melting of mantle with various potential temperatures as in the case of MORB generation suggested by the similar global systematics. Comparison of the results with tectonic parameters suggests that subduction zones where a hot mantle upwells easily show higher degree of melting. It is concluded that the return flow induced by the slab subduction might be the most critical factor that controls subduction zone magmatism.