Apatite has been recognized as a robust tool for the study of magmatic volatiles in terrestrial and extraterrestrial systems due to its ability to incorporate various volatile components and its common occurrence in igneous rocks. Most previous studies have utilized apatite to study individual magmatic systems or regions. However, volatile systematics in terrestrial magmatic apatite formed under different geological environments has been poorly understood. In this study, we filtered a large compilation of data for apatite in terrestrial igneous rocks (n > 20,000), categorized the data according to tectonic settings, rock types, and bulk-rock compositions, and conducted statistical analyses of the F–Cl–OH–S–CO2 contents ( 11,000 data for halogen and less for other volatiles). We find that apatite from volcanic arcs preserves a high Cl signature in comparison to other tectonic settings and the median Cl contents differ between arcs. Apatite in various types and compositions of igneous rocks shows overlapping F–Cl–OH compositions and features in some rock groups. Specifically, apatite in kimberlite is characterized as Cl-poor, whereas apatite in plutonic rocks can contain higher F and lower Cl contents than the volcanic counterparts. Calculation using existing partitioning models indicates that apatite with a high OH (or F) content does not necessarily indicate a H2O-rich (or H2O-poor) liquid because it could be a result of high (or low) magma temperature. Our work may provide a new perspective on the use of apatite to investigate volatile behavior in magma genesis and evolution across tectonic settings, volatile recycling at subduction zones, and the volcanic-plutonic connection.

Understanding the conditions and timescales of storage and remobilization of magma bodies in the upper crust is key to interpret the signals of potential reawakening of the volcanic activity at active volcanic systems. In this study we provide the first volcanological and petrological characterization of the Singkut volcanic system located in northern Sumatra, ~35 km N of the Toba caldera and ~40 km SW of the major city of Medan. Singkut is a ~9 km diameter caldera delimited by ~300 m-high rims where pre-caldera lavas are exposed. The inner part of the collapsed structure is occupied by three post-caldera volcanoes and currently hosts an active geothermal field. We utilize field observations and correlation with a distal marine tephra layer to map the extension and thickness of the tuff erupted during the caldera-forming eruption and use these data to estimate the erupted magma volume. We use major and trace element data of bulk-rock, matrix glasses and minerals to characterize the pre-eruptive conditions of pre- and post-caldera lavas and caldera-forming tuff and 14C and U/Th-He zircon dating to determine the eruption ages. In addition, a combination of U/Th and U/Pb in-situ zircon dating and zircon trace element geochemistry provides insights into the mechanisms and timescales that led to the Singkut caldera-forming eruption and those that controlled the post-caldera activity. Our data show that Singkut caldera formed ~50 ka during a large explosive eruption that deposited ~60 km3 of pyroclastic material. The cataclysmic eruption was preceded by at least 200 ky of mostly effusive pre-caldera activity and followed by effusive and mildly explosive post-caldera activity, with the last eruption reported at 1881 AD. The lavas and pumices have high crystallinity (24-62% crystals) and contain pl+amph+bt+opx+Fe-Ti ox+ap+zr±qtz. Notably, large and strongly resorbed quartz crystals are abundant in the pre-caldera lavas and scarce or absent in the caldera-forming tuff and post-caldera lavas. Bulk-rock composition of pumices and lavas varies from andesitic to dacitic, while the matrix glass in the pumices is rhyolitic. Trace element composition of glass (e.g., positive Eu anomalies) indicate resorption of feldspars. Crystallization ages of the youngest zircons in pre-caldera lavas overlap with eruption ages (~250 ka) while crystallization ages of the youngest zircons in the caldera-forming tuff and post-caldera lavas are significantly older (~100 ka) than the eruption ages (~50 and ~16 ka, respectively). Ti-in-zircon thermometry combined with zircon geochronology show that the Singkut magma body experienced a heating phase which started approximately upon eruption of the pre-caldera lavas and continued at least until the eruption of the post-caldera lavas. Such prolonged heating event determined progressive melting of the least refractory mineral phases (mostly quartz and feldspars) and hampered zircon crystallization for ~50 ky before the caldera-forming eruption and ~80 ky before the effusion of the post-caldera lavas. Heating was likely due to an increase of the recharge flux in the magma reservoir which reduced the crystallinity of the crystal mush and promoted remobilization and eruption of the Singkut magma body.

<p>Details on the methods and all fitted diffusion profiles, and a dataset containing the apatite single points and traverses data, water content values, and diffusion modelling parameters. </p>

One of the biggest challenges in volcanology is assessing the role of magma properties (volatile budgets, storage depths, and ascent rates) in controlling eruption explosivity. We use a new approach based on apatite to estimate volatile contents and magma ascent rates from a sequence of sub-Plinian, effusive, and Vulcanian eruption deposits at Rabaul caldera (Papua New Guinea) emplaced in 2006 CE to probe the mechanisms responsible for the sudden transitions in eruption styles. Our findings show that all magmas were originally stored at similar conditions (2–4 km depth and 1.8–2.5 wt% H2O in the melt); only the magma that formed the lava flow stalled and degassed at a shallower level (0.2–1.5 km) for several months. A more energetic batch of magma rose from depth, bypassed the transient reservoir, and ascended within ≤8 h to Earth’s surface (mean velocity ≥0.2 m/s), yielding the initial sub-Plinian phase of the eruption. The shallowly degassed magma was then able to reach the surface as a lava flow, likely through the path opened by the sub-Plinian magma. The magma of the last Vulcanian phase ascended without storage at a shallow depth, albeit more slowly (ascent rate 0.03–0.1 m/s) than the sub-Plinian magma. Our study illustrates how the complexity of plumbing systems may affect eruption styles, including at other volcanic systems, and have implications for interpreting volcano monitoring data.

Unlike their silicic counterparts, mafic eruptions are known for being on the low-end of the explosivity spectrum with eruption styles commonly ranging from effusive to Hawaiian fire fountaining. However, there are increasing discoveries of large mafic Plinian eruptions, sometimes generating ignimbrites, suggesting that this phenomenon might not be so uncommon. So, what processes lead a mafic magma to fragment violently enough to generate extensive ignimbrites?We sampled pumices from ignimbrites and PDCs with a compositional range from basaltic-andesite (Curacautín ignimbrite, Volcàn Llaima, Chile), andesite (Marapi, Indonesia) to trachyte (Gunungkawi ignimbrite, Batur, Indonesia). We use SEM imagery and X-ray Microtomography on pyroclasts from these deposits to characterize phenocryst, microlite and vesicle textures. From vesicle number densities we estimate fragmentation decompression rates in the range of 0.4–1.6 MPa/s for the three deposits. With a combination of EPMA and SIMS analyses we characterise pre-eruptive storage conditions. Based on the bulk and groundmass compositions, the storage temperature (1,050–1,100°C), pressure (50–100 MPa) and phenocryst content (1.0–2.5 vol%), we conclude that the basaltic-andesitic Curacautín magma was at sub-liquidus conditions, which allowed fast and widespread disequilibrium matrix crystallization (0–80 vol%) during ascent to the surface. Combined with the important decompression rate, this intense crystallization led to a magma bulk viscosity jump from 103 up to >107 Pa s and allowed it to fragment brittlely. Conversely, for the Marapi PDC and Gunungkawi ignimbrite, similar decompression rates coupled with larger initial bulk viscosities of 105–106 Pa s were sufficient to fragment the magma brittlely. The fragmentation processes for these latter two deposits were slightly different however, with the Marapi PDC fragmentation being mostly driven by vesicle overpressure, while a combination of bubble overpressure and intense strain-rate were the cause of fragmentation for the Gunungkawi ignimbrite. We conclude that mafic ignimbrites can form due to a combination of peculiar storage conditions that lead to strongly non-linear feedback processes in the conduit, particularly intense microlite crystallization on very short timescales coupled with intense decompression rates. Conversely, the high viscosity determined by pre-eruptive storage conditions, including temperature and volatile-content, are key in controlling the formation of more evolved magmas PDCs'.

Details on the methods and all fitted diffusion profiles, and a dataset containing the apatite single points and traverses data, water content values, and diffusion modelling parameters.

One of the major challenges faced by volcanologists to investigate controls on eruption dynamics is to quantify both pre-eruptive volatile budgets and timescales of magma ascent. Indeed, petrological investigations of the two parameters usually rely on different methods/analytical techniques that are not always applicable/accessible. Recent studies have shown that the abundance and zoning pattern of F, Cl, and OH in apatite can be used to determine both pre-eruptive volatile budget and magma degassing rates that can, under some conditions, be related to magma ascent rates ([1],[2]). Here we apply the two methods to apatite in the Rabaul 2006 eruption deposits (Papua-New-Guinea). This was a VEI-4 eruption and occurred in three main phases: (1) a sub-plinian onset followed 12h after its start by (2) a mixed strombolian-effusive phase, which subsequently evolved into (3) discrete vulcanian explosions. We sampled deposits of the three phases: (1) pumices, (2) fragments of lava flow, and (3) fragments of cow-pad bombs. We calculated pre-eruptive water contents using apatite included in clinopyroxene as they keep a better record of reservoir conditions from the time of entrapment. We found that the magma that fed the sub-plinian phase contained the highest water content of about 2 wt.%, while magmas that fed the lava flow and the vulcanian phase were drier, with 0.2 to 0.5 wt.% less H2O. X-ray maps acquired with an EPMA show that only apatite crystals in the groundmass of the vulcanian and effusive deposits are zoned in F and Cl at the crystal rims, whereas those from the sub-plinian deposits and included in clinopyroxenes are not zoned. This indicates that the zoning is related to syn- or immediately pre-eruptive changes of Cl-F-H2O during magma ascent towards the surface and can thus be modelled as diffusive reequilibration of the crystal and the melt. We obtained maximum diffusion timescales of <8 hours for the unzoned apatite in sub-plinian deposits, timescales of 20–22 hours for apatite in vulcanian deposits, and 600–1500 hours for those in the lava flow. Thus, the time scales increase with decreasing explosivity of the eruptions, as it could be expected if magma ascent rate played the key role of eruption dynamics. However, the degassing timescales of the effusive phase are significantly longer than the eruption duration itself, which can be explained if the magma started rising in the system 1–3 months prior to the onset of the eruption. The volatile-rich, fast-rising magma that fed the initial sub-plinian phase propagated through, disturbed and remobilized the shallower, more degassed batch of magma, which was erupted during the following effusive phase. Deeper, volatile-poor magma that kept moving up the open conduit, was responsible for the late vulcanian explosions. Our results show that apatite is a powerful tool for probing slight changes in magma volatile chemistry and ascent rates that can vary between different phases of the same eruption and produce different eruption styles. [1] Li and Costa, 2020, GCA [2] Li et al. 2020, EPSL

Rabaul in Papua-New-Guinea is an extremely active andesitic caldera complex that displays a large spectrum of eruption styles. Since 1878, four sub-plinian (VEI-4) and ten VEI 1-3 (effusive, strombolian, vulcanian) eruptions occurred from Tavurvur and Vulcan, the two main active vents. We study pumiceous tephra, ballistic bombs and a lava flow from five of these eruptions to investigate (1) magma ascent rates and (2) volatile escapement processes during the ascent. We measured total and connected porosities, permeability, and connectivity and related these results with measurements of crystallinity, Microlite Number Density (MND) and Microlite Size Distribution (MSD) of plagioclases and pyroxenes. From the application of existing percolation models, we find that explosive products yield a percolation threshold comprised between 50 and 60 vol% total porosity, while petrophysical parameters of the lava flow and some of the ballistic bombs are explained by bubble collapse driven by surface tension. Sub-plinian products show low phenocryst contents (5-15 vol%), microlites generated by nucleation-driven crystallization or glassy textures due to kinetic crystallization lags. Vulcanian, strombolian and effusive products on the other hand, show medium to high phenocryst contents (15-40 vol%) and microlites that crystallised by growth-dominated processes. MSDs in sub-plinian pumiceous tephra constitute a partial record of drastic magma acceleration. In comparison, MSDs in vulcanian and strombolian ballistic bombs and lava flow show a classic pattern of crystallization at a steady-state due to gradual magma ascent. We find that magma feeding sub-plinian eruptions ascends 2-3 orders of magnitude faster than magma feeding vulcanian/effusive eruptions (>= 1-100 m/s vs 0.1 m/s). In the case of sub-plinian products, due to the important kinetic crystallization lag, these speeds cannot be estimated using microlite crystallization triggered by decompression as a proxy. Combining petrophysical and textural measurementswe suggest that slight changes in initial conditions in the reservoir such as crystallinity or the presence of exsolved volatiles, can have a profound impact on the ascent rate and generate positive or negative feedback effects leading to powerful sub-plinian activity or intermittent vulcanian/passive effusive activity respectively. (C) 2020 The Authors. Published by Elsevier B.V.