Evaluation of crystal mush extraction models to explain crystal-poor rhyolites

• Mush extraction models are evaluated using rhyolites from Oregon and Colorado.
• Intermediate crystal mushes can't generate observed A-type rhyolites from Oregon.
• Subalkaline magmas with titanite or abundant amphibole can't make typical rhyolites.
• Repeated granitic mush extraction can't explain zoned high-silica rhyolite bodies.

Mush models have become the new paradigm for explaining crystal-poor rhyolites in a variety of settings. Despite this general acceptance, there are cases where this model is problematic. Rhyolites from two specific areas are used to highlight examples where mush extraction models are inconsistent with erupted compositions. Rhyolites from eastern Oregon are used to address a mush origin of hot and dry (or A-type) rhyolites from bimodal volcanic suites and rocks from the San Luis Caldera Complex of the San Juan Volcanic Field in Colorado are used to address crystal-poor rhyolites of calc-alkaline suites.Crystal-poor A-type rhyolites from Oregon resemble those from the neighboring Snake River Plain–Yellowstone centers. They are Fe-rich and high-field-strength-element enriched in comparison to regional calc-alkaline rhyolites and they vary widely in their degree of fractionation. A compositional assessment between least fractionated A-type rhyolites and a variety of intermediate magmas, including co-genetic intermediate magmas that erupted along with rhyolites during ignimbrite eruptions, indicates that intermediate, calc-alkaline and alkaline crystal mushes are unlikely to be able to generate interstitials melts after > 50% crystallization that match observed rhyolites with high Ba/Rb and Ba/Sr as long as alkali-feldspar, low An plagioclase (~< An40) or biotite are part of the crystallizing magma mush assemblage before extraction. Arguments specifically against granodioritic mush as rhyolitic nursery for Oregon hot & dry, Fe-rich, A-type rhyolites are multifold and strong.The San Luis Caldera Complex consists of crystal-rich intermediate magmas as well as crystal-poor rhyolites that erupted over a narrow time window. This association allows us to directly apply the mush model by comparing silicic interstitial melts of crystal-rich magmas with erupted rhyolites. REE contents of rhyolitic interstitial melt have MREE depleted patterns relative to bulk rock in all cases where titanite and/or abundant amphibole are observed. On the contrary, erupted San Luis Caldera Complex rhyolites do not show a depleted middle REE pattern but rather have patterns common to typical rhyolite. Consequently, whenever titanite and/or abundant amphibole is part of a mush mineral assemblage of calc-alkaline intermediate magma prior to melt extraction, the interstitial rhyolitic melt is unlikely to represent commonly observed rhyolitic lava or tuff compositions.Granite bodies have also been invoked as crystal mush sources to produce compositionally zoned high-silica rhyolites upon repeated extraction. This model is tested on two voluminous high-silica rhyolite bodies with strong compositional zoning.Considering Nb and Rb contents of most enriched rhyolite of both tuffs as the first extracted interstitial melt of the granitic mush constrains the bulk composition of the granite and in turn the second extracted rhyolite using batch melting models. Results indicate the second extracted rhyolite at 50% crystallinity is much more depleted than any observed rhyolite of either of the zoned tuffs. Therefore, a repeated granitic mush extraction scenario as explanation of chemically zoned high-silica rhyolite magmas is doubtful.While physically attractive, an origin of crystal-poor rhyolites from crystal-rich mushes requires careful testing to determine instances where a mush model is applicable and where it is not.