How important is the role of crystal fractionation in making intermediate magmas? Insights from Zr and P systematics

Most magmatism on Earth forms by direct melting of the mantle, generating basalts at the low silica end of the terrestrial compositional spectrum. However, most subduction zone magmas erupted or sampled at the surface are basalt-andesitic to andesitic and hence have higher Si contents. Endmember hypotheses for the origin of andesites are: (1) direct melting of the mantle at water-saturated conditions, (2) partial re-melting of altered basaltic crust, (3) crystal fractionation of arc basalts in crustal magma chambers, and (4) mixing of mafic magmas with high Si crust or magmas, e.g., dacite–rhyolite. Here, we explore the possibility of using Zr and P systematics to evaluate the importance of some of these processes. Direct melting of the mantle generates magmas with low Zr (0.2 wt.%) in island- and continental-arc magmas initially increase to levels higher than what can be achieved if andesites form by direct mantle melting. As Si increases, both Zr and P decrease with Zr decreasing at higher Si, and hence lagging the decrease in P. These systematics, particularly the decoupled decrease in Zr and P, cannot be explained by mixing, and instead, are more easily explained if andesites are dominantly formed by crystal–liquid segregation from moderately hydrous basalt, wherein P and Zr are controlled, respectively, by early and later saturation in apatite and zircon. Although there is clear isotopic and outcrop (enclaves) evidence for mixing in magmatic systems, crystal–liquid segregation appears to be the dominant process in generating intermediate magmas, with mixing playing a secondary role. Finally, recent studies have suggested that the abundance of certain magma compositions in a given volcanic setting may be dictated by the optimal crystallinity window for efficient crystal–liquid separation (50–70 vol%). We show that the SiO2 content of the residual