Recognising mush disaggregation in basaltic systems: The distribution of olivine compositions in Icelandic basalts and picrites

The importance of magmatic mushes in controlling the evolution of magma has recently been a subject of vigourous investigation, particularly in silicic systems. Constraining the nature of mushes is also a key component of investigations into the plutonic basal wrecks of basaltic volcanoes. However, there has been surprisingly little effort to identify the manifestation of mush zone processes in the products of basaltic eruptions. Here we show that a number of previously unexplained petrological observations of basaltic eruptions can be understood within the framework of disaggregation of magmatic mush. A large compilation of electron microprobe data containing Icelandic olivine and glass compositions was used to examine the distribution of the forsterite content of olivine macrocrysts within individual eruptions, and the relationship of these olivines to their carrier basaltic liquids. The dataset of 7836 olivine and 233 glass point analyses was examined and 11 eruptions were identified where glass data and over 60 crystal core analyses were available. This sampling density permitted statistical investigation of the distribution of olivine compositions. In 10 of the 11 eruptions more than 90% of the observed olivines are too forsteritic to be in equilibrium with their carrier basaltic liquids. The results of both kernel density estimates and Gaussian mixture modelling indicate that each of the 11 eruptions contained at least one robust peak in olivine compositions. Out of these 11 eruptions, 8 show unimodal distributions of macrocryst olivine forsterite content, two are bimodal and one is polymodal. An important feature of the relationship between the carrier glass compositions and the distribution of olivine forsterite contents is that, for 10 of the 11 flows, a strong peak in the olivine compositional distribution occurs at forsterite contents that are 2-3 mol% higher than those expected for olivines in equilibrium with the carrier liquid. This offset peak is not predicted for olivines generated by simple equilibrium or fractional crystallisation models. Instead, the distribution of olivine compositions and its relationship with the carrier liquids can be accounted for using a three-stage model. In the first stage, fractional crystallisation and crystal settling generate a mush pile on the floor of a magma chamber. Compositional stratification is present in this mush, with the olivines at its base being more forsteritic than those at its top, reflecting the evolution of liquid compositions during fractional crystallisation. The olivines in the uppermost part of the mush are close to equilibrium with basaltic liquid in the interior of the chamber. In the second stage, diffusion occurs in the mush pile, reducing the variance in olivine compositions and generating a single peak in olivine compositions close to the mean forsterite content of the olivines in the crystal pile. Finally, the mush is disaggregated throughout the chamber interior shortly before eruption takes place. Quantitative models of this process indicate that the observed offset peak in olivine compositions can be generated after 100-10000 years of diffusion in a mush pile, depending on the mush thickness. The commonly observed lack of equilibrium phenocrysts in basaltic magma can therefore be accounted for by incorporation of magmatic mushes.