Identifying the crystal graveyards remaining after large silicic eruptions

The formation of crystal-poor high-silica rhyolite via extraction of interstitial melt from an upper crustal mush predicts the complementary formation of large amounts of (typically unerupted) silicic cumulates. However, identification of these cumulates remains controversial. One hindrance to our ability to identify them is a lack of clear predictions for complementary chemical signatures between extracted melts and their residues. To address this discrepancy, we present a generalized geochemical model tracking the evolution of trace elements in a magma reservoir concurrently experiencing crystallization and extraction of interstitial melt. Our method uses a numerical solution rather than analytical, thereby allowing for various dependencies between crystallinity, partition coefficients for variably compatible and/or incompatible elements, and melt extraction efficiency. Results reveal unambiguous fractionation signatures for the extracted melts, while those signatures are muted for their cumulate counterparts. Our model is first applied to a well-constrained example (Searchlight pluton, USA), and provides a good fit to geochemical data. We then extrapolate our results to understanding the relationship between volcanic and plutonic silicic suites on a global scale. Utilizing the NAVDAT database to identify crystal accumulation or depletion signatures for each suite, we suggest that many large granitoids are indeed silicic cumulates, although their crystal accumulation signature is expected to be subtle. •New trace element models realistically capture the processes in magma mushes.•Silicic cumulates are predicted to have subtle geochemical fingerprints.•Melts extracted from a silicic mush have more obvious trace element signatures.•Rhyolites may be the extracted melts from some granitic ‘cumulates’.