Petrological, textural, and volatile constraints on the architecture of Mauna Loa’s magmatic system over 170 years (1852–2022)
Mauna Loa’s reawakening in 2022 highlighted the limited availability of petrological constraints characterizing the geometry and dynamics of pre-eruptive storage at this volcano. In particular, there has been relatively little prior work constraining pre-eruptive volatile budgets, mineral zoning and textures. To address this, we examine olivine- and orthopyroxene-hosted melt inclusions (MIs), their mineral hosts, and matrix glasses from seven Mauna Loa eruptions spanning 170 years (1852–2022). We present the first extensive MI dataset from Mauna Loa with direct Raman measurements of vapor bubble CO 2 . Olivine- and orthopyroxene-hosted MIs contain 0.3–0.5 wt % H 2 O and 100–600 μg/g CO 2 . MI saturation pressures reveal that the dominant reservoir is located at ~2–5 km depth, consistent with fluid inclusion barometry and geophysical constraints. This depth has remained consistent over the past 170 years, although some eruptions appear to have tapped a narrower range of depths than others. Even the high-Mg# orthopyroxene phenocryst-hosted MIs record saturation pressures of <150 MPa (<6 km depth), which is inconsistent with a deep sub-Moho origin proposed by previous studies that rely on thermodynamic modeling. Our Raman measurements also provide an opportunity to assess the accuracy of different methods to reconstruct the amount of CO 2 in vapor bubbles, with applications to MI datasets published without direct Raman measurements. In general, models that partition CO 2 between the melt and bubble using the bubble volume prior to pre-eruptive quenching (e.g. Melt Inclusion Modification Corrections, MIMiC) systematically underestimate CO 2 contents by a median of ~100–200 μg/g (depending on the model and K D choice) and thus underestimate storage depths by ~0.5–1.5 km. While these methods are useful for first-order estimates of bubble CO 2 in published datasets, new studies should measure bubble CO 2 by Raman spectroscopy or perform MI rehomogenization. The zoning and textures of the crystals that host the MIs provide additional constraints on the plumbing system to supplement MI barometry. Orthopyroxene crystals exhibit varied zoning patterns, including oscillatory zoning that records repeated fluctuations in magmatic conditions, and sector zoning indicating rapid, kinetically controlled crystal growth. While faster Fe-Mg diffusion in olivine results in a simpler zoning record, complexly zoned olivines with higher-Fo ‘shoulders’ from 1852, 1868, and 1950 also record the injection of more primitive melts at shallow levels. Electron backscatter diffraction reveals that touching crystals in orthopyroxene glomerocrysts exhibit only very weakly systematic crystallographic orientation relationships, perhaps due to rapid aggregation in a dynamic system. Overall, our data are consistent with a large, shallow (~2–5 km) pre-eruptive reservoir repeatedly recharged by primitive magma injections that homogenized prior to eruption due to their small volume relative to the overall size of this reservoir. This chemical complexity is effectively recorded in mineral and MI compositions (orthopyroxene zoning in particular) but is obscured in glass and whole-rock analyses.