Phenocryst compositional diversity as a consequence of degassing induced crystallization

In volcanic arc lavas, compositional diversity in phenocryst populations has commonly been attributed to magma mingling or mixing. However, the amount of dissolved water in the magma appears to have a significant effect on composition of the phenocrysts that crystallize from the melt. Tens of plagioclase and pyroxene phenocrysts were analyzed from six crystal-poor (<6 vol%) andesite and dacite scoria cones on the flanks of Volcán Tequila in western Mexico. The compositions and phase assemblages in the crystal-poor lavas are remarkably similar to that of the crystal-rich lavas (15-30 vol%) from the main edifice and flank flows of Volcán Tequila. Both lava types have plagioclase phenocrysts that span a wide compositional range, up to 45 mol% anorthite. In the crystal-rich lavas, individual phenocrysts have significant compositional variation, from oscillatory zoning of tens of mol% to relatively homogenous composition cores with a 5-10 um rim of significantly different composition. In contrast, plagioclase in the crystal-poor lavas has compositional variation within the population, but not individual phenocrysts. The plagioclase have little core to rim zoning and remarkable euhedral shapes, irrespective of composition. They are often riddled with melt inclusion channels, which broadly parallel the long axis of the crystal. These textures have been recognized in plagioclase crystallization experiments to be the result of rapid and large degrees of undercooling during crystallization. In the crystal-poor lavas, there is no textural evidence to suggest the phenocrysts were ever out of equilibrium with the host magma, so an alternative to magma mingling/mixing must be considered. The composition of plagioclase is dependent on several parameters, but varies most strongly with H2O content. Because of this relationship, a new plagioclase hygrometer (Lange and Frey, 2006) calibrated on plagioclase compositions from water-saturated experiments in the literature, can be used to determine the dissolved water content in the magma from which the plagioclase crystallized. The range in dissolved water content (~6 to 0.5 wt%) in the crystal-poor lavas recorded by the plagioclase is the result of magma degassing upon ascent through the upper crust to the surface. The rapid loss of volatiles during ascent decreases the dissolved water content in the magma and induces crystallization. This is entirely consistent with the lack of zoning, euhedral shapes, and rapid-growth textures seen in the plagioclase. Degassing-induced crystallization is also supported by the compositions and textures of the pyroxene phenocrysts. In the crystal-poor lavas, the orthopyroxene and clinopyroxene phenocrysts have euhedral shapes, with compositional variation of .15 KD^{Fe-Mg} ((Fe/Mg)pyx/(Fe/Mg)liq) (e.g. 0.30-0.45), with each sample following a continuum in composition. Like the plagioclase, the compositional variation is typically not within individual pyroxene phenocrysts, which display little zoning, but the entire population. Several of the pyroxene crystals have serrated edges, a texture consistent with rapid growth. By looking in detail at the phenocrysts in less complex, crystal-poor lavas, we have recognized a process not previously considered to explain phenocryst compositional diversity in lavas. The results of this work have implications for studies of more crystal-rich lavas, typical of stratovolcanoes. Much of the compositional diversity previously attributed to magma mingling/mixing may simply be a consequence of volatile loss and degassing induced crystallization.