Historical explosive activity of Mount Melbourne Volcanic Field (Antarctica) revealed by englacial tephra deposits

Five tephra layers named BRH1 to 5 were sampled in an ice cliff located on the north-eastern flank of Mount Melbourne (northern Victoria Land, Antarctica). The texture, componentry, mineralogy, and major and trace element compositions of glass shards have been used to characterize these layers. These properties suggest that they are primary fall deposits produced from discrete eruptions that experienced varying degrees of magma/water interaction. The major and trace element glass shard analyses on single glass shards indicate that Mount Melbourne Volcanic Field is the source of these tephra layers and the geochemical diversity highlights that the eruptions were fed by compositionally diverse melts that are interpreted to be from a complex magma system with a mafic melt remobilizing more evolved trachy-andesitic to trachytic magma pockets. Geochemical compositions, along with textural and mineralogical data, have allowed correlations between two of the englacial tephra and distal cryptotephra from Mount Melbourne, recovered within a marine sediment core in the Edisto Inlet (~ 280 km northeast of Mount Melbourne), and constrain the age of these englacial tephra layers to between the third and the fourth century CE. This work provides new evidence of the intense historical explosive activity of the Mount Melbourne Volcanic Field and better constrains the rates of volcanism in northern Victoria Land. These data grant new clues on the eruptive dynamics and tephra dispersal, and considerably expand the geochemical (major and trace elements) dataset available for the Mount Melbourne Volcanic Field. In the future, this will facilitate the precise identification of tephra layers from this volcanic source and will help define the temporal and spatial correlation between Antarctic records using tephra layers. Finally, this work also yields new valuable time-stratigraphic marker horizons for future dating, synchronization, and correlations of different palaeoenvironmental and palaeoclimatic records across large regions of Antarctica.


Introduction
Tephra in Antarctic ice records (ice cores or blue ice) provide important sources of data for evaluating the history and evolution of Antarctic explosive volcanism. The typical absence of pyroclastic deposits at proximal locations on the flanks of the volcanoes, due to high rates of post-depositional erosion and/or remobilization and also (mainly) the extensive cover of snow and ice obscuring deposits on the volcanic centres, limits the reconstruction of past volcanic eruptions, including their timing, scale, and impact on the environment. This gap in knowledge can be overcome by the study of tephra and cryptotephra (not visible, highly disseminated, fine-grained tephra) at medial to distal sites, which are preserved and remain unaltered within the snow, transformed into glacial ice upon burial (Smellie 1999;Harpel et al. 2008;Iverson et al. 2017). In polar regions, and particularly in Antarctica, tephrochronological studies in ice cores and blue ice records have been well developed over recent decades (e.g. Narcisi et al. 2005;Kurbatov et al. 2006;Curzio et al. 2008;Iverson et al. 2014;Wolff et al. 2010;Narcisi et al. 2012;Severi et al. 2012;Kim et al. 2020;Nardin et al. 2021). Along with tephra included in marine sediment sequences (Hillenbrand et al. 2008;Del Carlo et al. 2015;Di Roberto et al. 2019;2021a;, tephra deposits in englacial sequences can help improve near-vent eruption records, in terms of both frequency/chronology of explosive volcanism, and magma chemical evolution through time, whilst facilitating advancements in dating, correlating, and synchronizing high-resolution climate and atmospheric composition records through the Late Quaternary (Narcisi and Petit 2021). In addition, tephrochronological correlations between different archives (e.g. ice and marine records), located at proximal and distal sites from an eruptive source, are greatly enhanced once a numerical age is obtained for tephra. Specifically, the correlation of two tephra layers allows the age-transfer from one site to another by the simple use of stratigraphy, heightening tephrochronology to an age-equivalent dating method (Lowe 2011).
Mount Melbourne is an active volcano in northern Victoria Land, Antarctica, and future activity on the volcano would pose a significant tephra fall risk to nearby scientific stations, e.g. Mario Zucchelli Station (Italian), Jang Bogo (Korea), Gondwana (Germany), and the new China Antarctic research base (Fig. 1a), as well as to Austral hemisphere air traffic. Therefore, a good knowledge of the recent volcanic activity of this volcano is important to better evaluate the possible volcanic hazards (Geyer 2021).
The earliest investigations on Mount Melbourne began at the end of the 1960s whereas the first geophysical observations started in 1988 by Italian National Antarctic Research Program (PNRA); since then no eruptive activity has been observed at this volcano. The age of the most recent eruption is proposed by Lyon (1986) who performed the stable isotope analysis of two snow profiles sampled on Mount Melbourne at ca. 2000 m of altitude and on the Campbell Glacier obtaining snow accumulation rates of 0.5-2.2 m/a. Using these data, the author derived the age of englacial ash layers outcropping in an ice cliff at ca. 1200 m of altitude on the western slope of Mount Melbourne and photographed in 1965 by Adamson and Cavaney (1967). They roughly estimated the depth of snow from the surface to the first major ash layer to be between 29 and 36 m and, considering an accumulation rate of 0.5 ± 0.16 m/a, the last major eruption should have occurred between 1862 and 1922 AD. Although this method has several limitations (e.g. the lack of direct measurement of the snow thickness, or the detailed study of snow profile aimed at identifying possible hiatus or erosion surfaces), it does provide an estimate of the age of the latest eruptive activity of Mount Melbourne. Unfortunately, the englacial tephra layers described in Lyon (1986) and Adamson and Cavaney (1967) were never sampled nor characterized in their textural and geochemical features because the outcrop did not allow safe sampling.
During the 2017 austral summer, the ICE-VOLC project (PNRA) facilitated new geological surveys around Mount Melbourne which revealed the occurrence of multiple englacial volcanic ash layers in an ice cliff located on the northeast flank of the volcano at ca. 800 m of altitude, very close to Baker Rock location (Gambino et al. 2021; Fig. 1). The better accessibility of the outcrop allowed the sampling of tephra and the measurements of the thickness between the different ash layers.
In this paper, we present data on the texture, mineral phases, and major and trace element geochemical data performed on single glass shards analyzed via Electron Probe Microanalysis (EMPA) and Laser Ablation Inductively Coupled Plasma Mass Spectrometry (LA-ICP-MS) of five englacial tephra layers sampled for the first time from the north-eastern flank of Mount Melbourne. Geochemical data indicate that the tephra layers are derived from Mount Melbourne explosive activity. The texture of the particles and their morphological features were important for clues to the style of the eruptions and fragmentation mechanisms. In addition, geochemical fingerprints of englacial tephra have allowed their correlations and dating with cryptotephra layers recently found in marine sediments in Edisto Inlet near Cape Hallett (Di Roberto et al. 2023), highlighting that Mount Melbourne has been very active during historical times.

Geological setting
Mount Melbourne stands as a 2732-m-high stratovolcano with a basal diameter of about 21-24 km on the coast of northern Victoria Land, between Wood Bay and Terra Nova Bay, in Antarctica (Fig. 1b). The Mount Melbourne volcano also includes several parasitic cones and secondary vents located on the flank of the main edifice (Smellie et al. 2023). The main volcano edifice shows a gentle shape with undissected flanks, apart from a possible slump scar on the east side (Giordano et al. 2012), and a well-formed ice-filled crater ca. 700 m in diameter that Armienti et al. (1991) interpreted as a summit caldera. It is largely covered by snow and ice except for the summit region where rock outcrops extend downslope on the east side to ca. 1800 m. The Mount Melbourne Volcanic Field (MMVF), together with The Pleiades, Mount Overlord, Mount Rittmann, and the Malta Plateau volcanoes, is part of the Melbourne Volcanic Province within the McMurdo Volcanic Group (Smellie and Rocchi 2021). A synthesis of MMVF volcanic history was reported by Giordano et al. (2012) and Smellie et al. (2023) based on stratigraphic and volcanological studies, geochemical data, and age determinations. According to recent updates based on radioisotopic ages, the MMVF volcanism began in the Late Miocene (c. 12.5 Ma for centres in the north by Tinker Glacier) but developed mainly from the Pliocene (c. 4 Ma; Smellie et al. 2023) particularly after c. 3 million years ago (Rocchi and Smellie 2021 and references therein). The activity can be subdivided into different evolution stages: (i) the older Cape Washington shield volcano (Late Miocene-Late Pliocene), (ii) the Random Hills Period (Lower-Middle Pleistocene), (iii) the Shield Nunatak Period (Middle Pleistocene), and (iv) the Mount Melbourne Period (Upper Pleistocene-present; Giordano et al. 2012;Smellie et al. 2023).
The oldest deposit that can be linked directly to an eruption of the Mount Melbourne stratovolcano, re-dated as 115 ka by Smellie et al. (2023) at Edmonson Point, is a trachytic ignimbrite. This was emplaced during a large Plinian eruption and indicates the formation of a crustal magma chamber during the recent Mount Melbourne stage (Giordano et al. 2012). Subsequently, the Adelie Penguin Rookery lava field succession was produced comprising alkali basaltic, hawaiitic, and subordinate benmoreitic lavas, scoria cones, and spatter cones that are dated at 90.7 ± 19.0 ka. (Giordano et al. 2012). The location of the studied outcrop is labelled with a red star whereas the location of the marine sediment core TR17-08 in Edisto Inlet (Di Roberto et al. 2023) and of the deep ice cores at Talos Dome and Styx glacier are annotated with a green star and blue hexagons, respectively. Green dots indicate the location of the permanent scientific bases. The red box outlines the area of the topographic map in (b) that illustrates Mount Melbourne and locations of the BRH sections. (c) Picture of the studied outcrop where the BRH tephra layers are exposed, Mount Melbourne in the background Evidence of more recent activity, based on tephrostratigraphic analysis and 40 Ar-39 Ar dating of proximal pyroclastic sequence exposed on the Mount Melbourne summit, indicates there were at least four Strombolian or Vulcanian style to sub-Plinian/Plinian eruptions during the Late Pleistocene to Holocene . The most intense of these eruptions was a sub-Plinian to Plinian eruption that yielded an age < 17.8 ka (13.5 ± 4.3 ka). Furthermore, three trachytic cryptotephra with glass compositions similar to Mount Melbourne products were found intercalated in marine sediments of Edisto Inlet, near Cape Hallett (Di Roberto et al. 2023). These cryptotephra layers were interpreted as derived from historic explosive eruptions of Mount Melbourne that occurred between 1615 cal. years BP and 1677 cal. years BP, i.e. between the third and fourth century CE (Di Roberto et al. 2023).
Presently, the Mount Melbourne volcano is quiescent with thermal anomalies, including noticeable steaming or fumarolic activity both in the crater and on the north-western side of the volcano. These emissions of steam and volcanic gas have produced several ice towers and a complex network of ice caves, as recently reported after geochemical surveys in the frame of the ICE-VOLC project (Gambino et al. 2021).

Material and methods
During the XXXII Italian Antarctic Expedition in 2017, the area surrounding Mount Melbourne was aerially surveyed by helicopter flights. A series of dark, sub-horizontal englacial tephra layers were found exposed along a ca. 50-m-high ice cliff located on the north-eastern flank of the volcano at about 800 m of altitude close to Baker Rock (74.24096 S, 164.72032 E; red star in Fig. 1). Five sub-horizontal tephra layers, named BRH1 to 5 (from the top to the bottom), were distinguished and sampled by the Italian alpine climbing guides (Fig. 2). Tephra layers are between 10 and 20 cm thick and are each separated by 1.5 to 7.5 m of ice (Fig. 2). Each sample is collected the "bulk" of the tephra layer and represents the whole unit; unfortunately, due to the extreme sampling conditions, it was not possible to sub-sample different stratigraphic heights within each unit. The tephra samples were recovered still embedded in the ice, which was then melted to recover the clastic fraction.
Sample preparation was carried out at laboratories of Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Pisa (INGV-Pisa). Samples were washed with deionized water in an ultrasonic bath to remove impurities, dried at 60 °C, mounted with epoxy resin in 1-in. stubs, polished, and prepared for textural and geochemical analyses. Textures, components, and mineral assemblages of each tephra were studied with an optical microscope and a scanning electron microscope (SEM) Zeiss EVO MA and images of 3D silhouettes and 2D cross-sections have been collected in secondary (SE) and back-scattered (BSE) electrons mode, respectively. The major and minor elements glass composition of samples was determined using a JEOL 8600 wavelength-dispersive electron microprobe equipped with four spectrometers at the Research Laboratory for Archaeology and the History of Art, the University of Oxford (operating conditions: 15-kV accelerating voltage, 6-nA beam current, and a beam diameter of 10 µm). The JEOL 8600 electron microprobe was calibrated with a suite of appropriate mineral standards; peak count times were 30 s for all elements except Mn (40 s), Na (12 s), Cl (50 s), and P (60 s). The PAP absorption correction method was used for quantification. Reference glasses from the Max Planck Institute (MPI-DING suite; Jochum et al. 2006) bracketing the possible chemistries were also analyzed. These included felsic [ATHO-G (rhyolite)], through intermediate [StHs6/80-G (andesite)] to mafic [GOR132-G (komatiite)] glasses. All glass data have been normalized to 100% for comparative purposes. Uncertainties are typically < ± 0.8% RSD for Si; ~ ± 5% for most other Fig. 2 Picture of the glacier cliff where BRH tephra layers are exposed, taken during the sampling performed by the Italian alpine guides. On the picture are annotated the five englacial tephra and the ice thickness between them major elements, except for the low abundance elements, for instance Ti (~ ± 7%) and Mn (~ ± 30%).
Trace elements analysis of volcanic glass was performed using an Agilent 8900 triple quadrupole ICP-MS (ICP QQQ) coupled to a Resonetics 193 nm ArF excimer laser-ablation in the Department of Earth Sciences, Royal Holloway, University of London. Full analytical procedures used are reported in Tomlinson et al. (2010). Spot sizes 20 and 25 mm were used depending on the vesicularity, crystal content, and ultimately the size of available glass surfaces. The repetition rate was 5 Hz, with a count time of 40 s on the sample, and 40 s on the gas blank to allow the subtraction of the background signal. Blocks of eight or nine glass shards and one MPI-DING reference glass were bracketed by the NIST612 glass calibration standard (GeoREM 11/2006). In addition, MPI-DING reference glasses were used to monitor analytical accuracy (Jochum et al. 2005). The internal standard applied was 29 Si (determined by the EMPA analysis). Where individual shards were arranged into a grid formation across an epoxy mount, they are given a unique row (A, B, C) and shard number (1, 2, 3). In these instances, the grainspecific 29 Si content was applied as the internal standard. Where tephra samples were mounted in epoxy mounts without shard mapping, average 29 Si contents for the appropriate compositional groupings within the tephra deposit, determined based on EMPA analysis, were applied as the internal standard to the individual glass shards ablated. Internal standard values applied to individual shard ablations are provided in the Supplementary Information. LA-ICP-MS data reduction was performed in Microsoft Excel. Accuracies of LA-ICP-MS analyses of MPI-DING glass standards ATHO-G and StHs6/80-G were typically < 5%. Full glass datasets and MPI-DING standard glass analyses are provided in Supplemental Information.

Texture and components of tephra layers
BRH samples are similar in terms of components, comprising a range of clasts with different textures ranging from dense to highly vesicular and glassy to microlite-rich (Figs. 3 and 4a-e). The morphology of juvenile clasts ranges from equant with rounded vesicles (Fig. 4f) to fluidal with elongated tubular vesicles (Fig. 4g, h). Dense poorly vesicular juveniles display features such as stepped surfaces and hackle lines, quenching cracks, and pitted surfaces (Fig. 4l, m), similar to the glassy shards that range from blocky to platy ( Fig. 4i, n).
Overall, the grain size, scarcity of non-volcanic detrital material, and unabraded pristine shapes of the clasts indicate that these decimetre-thick layers are emplaced as primary fallout deposits. Extremely fragile particles such as glass fibres (Pele's hair) survived, and the external morphology of vesicular particles preserves pristine elements like fragile glass fibres, glass tips, spiny glass edges (Figs. 3 and 4), or glass coatings around magmatic crystals. Particles do not exhibit surficial alteration textures due to weathering, surface abrasion, or rounding. Lithic or detrital fragments are also scarce. All these features along with the geochemical fingerprints (see next paragraph) indicate very minor or no aeolian remobilization, re-sedimentation, or other transport after the deposition from the volcanic plume. Moreover, the layers occur close to the surface of the ice cover and well above the glacier bed, hence are unlikely to be glacier bedload brought to the surface by shearing.
As mentioned in the "Material and methods" section, the studied material comprises "bulk" samples of each tephra layer. Possible internal stratifications resulting from variations in eruptive style, energy, dynamics, magma reflected by changes in the glass composition, grain size, componentry, etc. were lost.
BRH1 tephra is the uppermost tephra exposed at the top of the ice cliff. It is made of coarse to very fine ash ( Fig. 4a) and comprises two discrete particle populations differing in colour, shape, and texture. The first population, which is the more abundant, consists of shiny, black to dark brown, vesicular glassy clasts ( Fig. 3a and g) with spiny to fluidal shapes ( Fig. 3d), volcanic glass fibres, and glass droplets (Pele's hair and tears; Fig. 3c) and minor amounts of grey, dense, and blocky clasts (Fig. 3b). Often, particles with fluidal shapes are bounded by sharp-planar breakage surfaces (Fig. 3g). Fluidal clasts (Fig. 3a) are usually glassy aphyric to microlite-poor, poorly to moderately vesicular with spherical vesicles. Blocky clasts (Fig. 3b) are almost nonvesicular to very poorly vesicular with abundant microlites and microphenocrysts of andesine plagioclase (up to 60 μm), Fe-augite clinopyroxene (up to 30 μm), olivine (~ Fo 45 ; up to 10 μm), Fe-Ti oxides (up to 10 μm), and rare apatite (up to 10 μm). Also, crystals (up to ca. 250 μm) of plagioclase, olivine, and clinopyroxene occur coated in a dark glass.
The second particle population represents approximately 30% of the sample and has a finer grain size (fine ash). Particles comprise white to pale yellow, highly vesicular pumice fragments, and nearly transparent glass shards (Fig. 4a). Pumice fragments often show elongated to tubular vesicles (Fig. 3f), and bear within an aphyric groundmass abundant labradorite plagioclase (up to 80 μm) and olivine (~ Fo 75 ; up to 30 μm) with rare Mg-augite clinopyroxene (up to 20 μm) and Mg-chromite spinel (up to 10 μm). Lithic clasts are scarce in BRH1 tephra and consist of reddish or altered volcanic rocks (Fig. 3e).
BRH2 tephra is a poorly sorted lapilli (up to 4 mm) and ash ( Fig. 4b) layer. It is predominantly composed of black/ dark brown, shiny, vesicular glass fragments with spiny and fluidal shapes again often bounded by sharp-planar breakage surfaces (Fig. 3d), Pele's hair and tears, highly vesicular fragments, and grey dense, blocky clasts. Fluidal clasts ( Fig. 4b) have glassy groundmass with spherical to elliptical vesicles and contain abundant microlites of andesine to oligoclase plagioclase (up to 50 μm), Fe-rich olivine (~ Fo 35 ; up to 15 μm), Fe-augite clinopyroxene (up to 30 μm), and Fe-Ti oxides (~ 10 mm). Grey dense and blocky clasts (Fig. 4b) are porphyritic with microphenocrysts of oligoclase to anorthoclase feldspar, Fe-augite, and Fe-Ti oxides within a glassy to microcrystalline groundmass. Tubular pumice fragments and bubble wall glass shards are abundant in the fine-grained, fine ash-sized portion of the deposit. Pumices have a glassy groundmass sometimes with acicular microlites of anorthitic feldspar. Scarce lithic clasts consist of reddish, altered volcanic rocks and intrusive rock fragments. Loose crystals of anorthoclase, clinopyroxene, and fayalite are abundant.

Major, minor, and trace elements glass geochemistry
Representative major, minor, and trace element glass chemistry from the BRH 1 to 5 tephra layers are provided in Table 1, whilst the full datasets are provided in the Supplementary Information.
The tephra layers analyzed (BRH1 to 5) display more than one compositional component, consistent with their variable textures and componentry; consequently, compositions range from basalts, through basaltic trachy-andesites and trachy-andesites to more evolved trachytes (Fig. 5a). Glasses are all characterized by high analytical totals suggesting a very minor amount of post-depositional alteration and/or reworking.
BRH1 to 5 volcanic glasses predominantly straddle the high-K calc-alkaline and Shoshonitic boundary, with the more evolved trachytic end-members residing more clearly within the shoshonitic field (Fig. 5b). The absence of negative anomalies at Nb and Ta in the mantle-normalized trace element profiles of the BRH1 to 5 volcanic glasses is clearly consistent with Antarctic alkaline regional volcanism within the West Antarctic Rift System (Panter 2021), and the absence of significant crustal involvement in magma genesis (Fig. 6). As is to be expected, overall levels of incompatible trace element enrichment are significantly greater in the trachytic end-member glasses of the tephra units relative to their more mafic components (Fig. 7). Light Rare Earth Elements (LREE) are enriched relative to the Heavy Rare Earth Elements (HREE), and the relative level of enrichment remains fairly constant spanning from the basalts through to the trachytes (La/Yb N ~ 13-14). Trachytic end-member glasses display depletions in Ba, Sr, and Eu consistent with feldspar fractionation, with these anomalies absent in the more primitive (basaltic) glasses. Owing to the significant heterogeneity of the BRH tephra layers, we describe the chemical signature of dominant chemical clusters or components, and therefore rare outlying analyses are not described in detail. BRH1 (n = 18/24), BRH2 (n = 13/33), and BRH5 (n = 15/30) tephra layers all contain a significant mafic glass component associated with the black glassy and fluidal particles, which display a relatively narrow compositional range within the basalt compositional field (44.5 to 46.0 wt% SiO 2 , and 3.9 to 5.3 Na 2 O + K 2 O), with minor overlap into the basanite field (Fig. 5a). More noticeable compositional variability in these basalts is observed in other major oxides, for instance with 7.3-4.7 wt% MgO, 14-12 wt% FeOt, 12.1-9.7 wt% CaO, and 5.0-3.3 wt% TiO 2 , where this variability is consistent between the basalts of all three layers (Fig. 5). Tephra layer BRH3 also contains dark glassy material entirely consistent with the basaltic glasses of the BRH1, 2, and 5 tephra layers; however, these compositions are not as well represented in the chemical analysis of the tephra (n = 10/74), which are dominated by a trachytic end-member (Fig. 5a).
BRH1 glasses, consistent with the observed componentry (see above), are chemically bimodal; in addition to the dominant dark glassy basaltic component already described, the tephra layer also contains a homogeneous secondary trachytic component (SiO 2 = 65.2 ± 0.1 wt% and Na 2 O + K 2 O = 10.78 ± 0.1 wt%; n = 5/23). The incompatible trace element contents of these trachytic glasses (n = 4) are relatively homogeneous, for instance 676 ± 40 ppm Zr, 54 ± 4 ppm Y, and 21 ± 2 ppm Th, whilst these levels of enrichment far exceed those of the dominant basaltic end-member (Fig. 7). Ratios of HFSE vs. Th become significantly lower (Zr/Th = 33.0 ± 1.3; Nb/Th = 8.1 ± 0.5; Y/ Th = 2.6 ± 0.1 [1 s.d]) than those of the basaltic component (Fig. 7). These evolved trachytic glasses display particularly strong negative Sr anomalies (Sr/Pr N = 0.09 ± 0.003), but also less pronounced anomalies in Ba and Eu (Fig. 6). Overall, BRH2 glasses cover a wide compositional range, consistent with the mixed componentry (see the "Texture and components of tephra layers" section). As with BRH1, the basaltic glass forms a significant component (1) of BRH2. In addition, at least two other compositional components or clusters are observed in BRH2, which plot along an evolutionary trend that extends towards the highly evolved trachytes, similar to the overlying BRH1 tephra (~ 65 wt% SiO 2 ). BRH2 glass component 2 straddles the boundary between the basaltic trachy-andesite to trachy-andesite compositional fields with between 54.5 and 56.5 wt% SiO 2 and where Na 2 O + K 2 O = 6.9-8.4 wt% (Fig. 5a). BRH2 component 3 glasses are more variable and evolved than component 2 predominantly clustering at the boundary between the trachy-andesite and trachytic field, albeit with a small number of analyses extending towards the more evolved trachytic endmember of BRH1 (Fig. 5a).
Trace element analysis of BRH2 component 2 (basaltic trachy-andesite/trachy-andesite) glasses was unsuccessful due to the presence of microlites, but trace element analyses were acquired for glass component 3. The trachy-andesites through to trachytes are equally heterogeneous at the trace element level, with 388-720 ppm Zr and 10.0-21.9 ppm Th. The glasses show depletions in Sr (Sr/Pr N = 0.30 ± 0.1 [1 s.d]) and Eu; however, these are not as pronounced as those observed in the more evolved trachytes of BRH1 (Fig. 6). HFSE ratios to Th in these glasses remain constant despite the variability in absolute concentrations (Zr/Th 33.6 ± 1.6; Nb/Th = 8.4 ± 0.9; Y/Th = 3.1 ± 0.2 [1 s.d]). The Zr/Th and Nb/Th ratios are broadly consistent with the more evolved (trachytic) glasses of the overlying BRH1, but again are significantly lower than those of the basaltic glasses in this tephra (and BRH1; Fig. 7).
In addition to a basaltic component, the BRH3 tephra has a minor population of intermediate glasses which straddle the basaltic trachy-andesite to trachy-andesite compositional fields, similar to component 2 of the overlying tephra BRH2 (Fig. 5a). A significant component of BRH3 analyses (n = 51/73) lies firmly in the trachytic compositional field, with 62.1-67.2 wt% SiO 2 and 8.8-10.6 wt% Na 2 O + K 2 O and relate to the pumiceous material within this tephra layer. Within these variable trachytic compositions, a dominant cluster of shoshonitic glasses is clearly recognized at ~ 66 wt% SiO 2 , and ~ 4.9 K 2 O (Fig. 5b), where Na 2 O is > K 2 O (K 2 O/Na 2 O = 0.9). These most evolved trachytes (n = 16) are extremely enriched in incompatible trace elements, for instance 613-728 ppm Zr and 18.3-22.1 ppm Th. HFSE ratios vs. Th in these trachytic glasses remain constant where Zr/Th = 33.0 ± 0.8, Nb/Th = 7.5 ± 0.3, and Y/ Th = 2.5 ± 0.1 (Fig. 7). They also display depletions in Ba, Sr (Sr/Pr N = 0.10 ± 0.01), and Eu (Fig. 6). The glass compositions of BRH4 are heterogeneous, largely bimodal, with some glasses extending between the two dominant end-members. The least evolved glass component plots in the basaltic trachy-andesite compositional field (SiO 2 = 53.8-55.7 wt% SiO 2 , Na 2 O + K 2 O = 7.0-8.3), whilst straddling the boundary with trachy-andesites (Fig. 5a). The more evolved end-member is characterized by slightly more variable trachytic glass compositions with 62.6-66.5 wt% SiO 2 and 9.1-10.5 wt% Na 2 O + K 2 O (Fig. 5a) and these relate to the pumice component of the deposit. The basaltic trachy-andesitic glasses (n = 5) have some incompatible trace element variability, for instance 400-449 ppm Zr and 11.3-12.3 ppm Th (Fig. 7). The HFSE ratios in Fig. 6 Primitive mantle normalized (McDonough and Sun 1995) trace elements concentrations of volcanic glass (component averages) for the BRH1 to 5 tephra layers. The data presented clearly illustrate a chemical link between the trachytic end-member glasses of the BRH tephra layers and those found at the summit of Mount Melbourne ). Bas-basalt; Tra-Bas-trachy-basalt; Bas Tra-And-basaltic trachy-andesite; Tra-And-trachy-andesite; Tra-trachyte Fig. 7 Bivariate trace elements plots, specifically (A) Th vs Zr, (B) Th vs Nb, (C) Th vs Y, and (D) Th vs Hf, depicting the significant heterogeneity of the BRH1 to 5 tephra layers consistent with their major elements variability ranging from basalts to trachytes. The plots illustrate the significant chemical overlap of these successive tephra deposits, which clearly lie upon an evolutionary trend indistinguishable from the products of Mount Melbourne (mainly trachytic), and offset from those of the nearby Mount Rittmann (after Del . The grey compositional field is for Edisto Inlet tephra TR17-08, whereas the black symbols indicate the average composition of Brimstone Peak tephras (+ BIT121, × BIT122;  these glasses are constant (36.7 ± 0.2 Zr/Th, 9.6 ± 0.1 Nb/ Th, and 4.0 ± 0.02 Y/Th [1 s.d]), whilst they show a small anomaly in Sr (Sr/Pr N = 0.4 ± 0.1 [1 s.d]). The trachytic endmember glasses (n = 5) are more enriched in incompatible elements than the basaltic trachy-andesites (Fig. 7), and show considerable variability, for instance 472-604 ppm Zr and 13.8-17.2 ppm Th. These trachytic glasses display constant HFSE ratios (e.g. 34.7 ± 1.0 Zr/Th, 8.2 ± 0.4 Nb/ Th, and 2.8 ± 0.3 Y/Th), but are slightly lower than those of the basaltic trachy-andesites. The trachytic glasses display a greater depletion in Sr (Sr/Pr N = 0.3 ± 0.1), along with additional depletions in Ba and Eu in contrast to the basaltic trachy-andesites (Fig. 6). However, the removal of Sr from the melt composition of BRH4 is not as significant as for some of the more evolved trachytic glasses observed in the overlying tephra deposits (e.g. BRH1, BRH3; Fig. 6).
Compositionally, the BRH5 tephra layer is largely bimodal, with some very minor evidence of intermediate glass compositions spanning an evolutionary trend between the two end-members, which are characterized by basaltic and trachytic glass populations (Fig. 5a). The basaltic component is entirely consistent with those observed and described (above) from the BRH1, BRH2, and BRH3 tephra layers (Figs. 5, 6, and 7). Whilst the trachytic end-member glasses are variable, they remain consistent with both dominant trachytic components of BRH1 and BRH3, where SiO 2 contents of ~ 64.9 wt% and ~ 66.8 wt% respectively are observed (Fig. 5a). The incompatible trace element contents of the BRH5 trachytic endmember (n = 5) glasses are homogeneous with 737 ± 30 ppm Zr and 23 ± 1 ppm Th. The levels of enrichment in some of these trachytic glasses extend to the highest observed throughout this englacial succession (Fig. 7). Furthermore, strong depletions in Ba, Sr (Sr/Pr N = 0.09 ± 0.01 [1 s.d]), and Eu are observed in these glasses (Fig. 6). HFSE ratios to Th are constant (32.0 ± 0.3 Zr/Th, 7.3 ± 0.2 Nb/Th, 2.5 ± 0.1 Y/Th [1 s.d]) and are entirely consistent with the most evolved trachytic products seen in the BRH1 and BRH3 layers.

Volcanic source and proximal-distal correlation
Major and trace elements glass composition data show a strong geochemical affinity between the BRH tephra and the products of MMVF, plotting along the same evolutionary trend. This compositional similarity and the proximity of these tephra layers to MMVF suggest they were erupted from its vents (Figs. 5 and 7). BRH major and trace elements compositions plot along the evolutionary trend defined by volcanic glasses of MMVF (Figs. 5 and 7), which is distinct from other volcanoes in the region. However, there is no direct match between the composition of BRH tephra and that of late Pleistocene to Holocene tephra layers exposed in the summit of Mount Melbourne, recently characterized compositionally by Del  are characterized by generally higher alkali and lower silica contents. In addition, Mount Melbourne summit proximal deposits lack the basaltic end-member compositions which instead are common in the BRH products except for the BRH4 tephra. Thus, the BRH tephra layers are likely to result from eruptions at separate vent areas, possibly from parasitic cones located on the flank of the volcano or at the periphery of the Mount Melbourne volcanic complex. On the eastern flank of Mount Melbourne, a few kilometres NNW of Edmonson Point ropey basaltic lava and a hawaiite scoria cone are observed, known respectively as the ROL and SCC outcrops of Giordano et al. (2012). These sources are close enough and their deposits are compatible with the characteristics and the eruption styles of BRH tephra. Also, the parasitic cones NNW of the summit of Mount Melbourne, which have trachytic compositions, and possibly scoria cones on the south side of Random Hills further to the north (Smellie et al. 2023) could be considered possible sources. Unfortunately, these only the eruptive centres close to Edmonson Point are dated, and not precisely. They are only constrained to the last 90 ka (Giordano et al. 2012;Smellie et al. 2023), hampering direct correlations.
We also compared the composition of the BRH tephra with visible tephra and cryptotephra layers identified in the marine record of the Ross Sea and in the glacial records of northern Victoria Land.
Interestingly, the composition of trachytic particle population in BRH3 and BRH5, the deepest and middle tephra respectively among those studied, correlates very well with the composition of three cryptotephra recently identified in the TR17-08 piston sediment core recovered in Edisto Inlet, near Cape Hallett, more than 280 km from Melbourne volcano (Di Roberto et al. 2023; Fig. 1). These cryptotephra layers, namely TR17-08-512, -518, and -524, have a trachytic composition (Figs. 5 and 7) and comprise colourless to light-green glass shards and pumice fragments, characterized by pristine textures among blocky, y-shaped, and bubble wall morphology whilst preserving fragile glass tips. Within core TR17-08, tephra age was defined using radiocarbon dating of carbonate material, intercalated with tephra, and was constrained between 1615 cal. years BP and 1677 cal. years BP, i.e. between the third and fourth century AD (Di Roberto et al. 2023). These TR17-08 tephra layers have been interpreted as produced by a series of explosive eruptions from the Mount Melbourne volcanic complex that occurred closely spaced in time, with a maximum interval of c. 60 years between events (Di Roberto et al. 2023).
The geochemical correlation defined between BRH proximal tephra and distal marine tephra allows the former to be indirectly dated and consequently defines the onset (and an overall age) of this eruptive period of Mount Melbourne between the third and fourth century AD. The results confirm that Mount Melbourne has been very active in historical times, with more explosive eruptive events than previously thought.
For the sake of completeness, it should be said that some differences exist between the BRH and TR17-08 cryptotephra. The most notable regards the lack of basaltic glass populations in TR17-08 samples. The lower viscosity of the denser melts typically results in less fragmentation and lower plume heights of the basaltic phases that limit the tephra dispersal. This is consistent with the significantly larger grain size characterizing the basaltic ash (and lapilli) in the studied tephra. The higher abundance of fine ash in the trachytic end-member testifies to more efficient magmatic fragmentation, and these finer particles are transported further, especially those associated with the intense phases of the eruption with high plumes. Moreover, whilst three cryptotephra have been identified within the TR17-08 marine core, only two tephra with similar trachytic compositions were recovered at the BRH site. Another layer may be preserved deeper in the ice sequence, below the BRH5 tephra, or was eroded at the BRH site.
Furthermore, we explored correlations with tephra from the ice core records of Styx Glacier (SG), Talos Dome (TD), and the blue ice areas of Brimstone Peak (75.888S 158.55E; BIT Dunbar et al. 2003), located at ca. 35 km, 230 km, and 250 km ESE from the sampling site, respectively. Unfortunately, most of the geochemical data for SG, TD, and BIT tephra layers comprise only major element glass composition for which the accuracy is not clear, and lack trace element glass data, which hamper reliable correlations with the BRH tephra layers.
In spite of this, we found a major similarity between the trachytic compositions of BRH4 tephra and the TD85 tephra layer in the Talos Dome record dated at c. 670 yrs BP (1280 CE; Narcisi et al. 2012;Severi et al. 2012), but it is not a viable correlation based on other tephrostratigraphic evidence. In the Talos Dome record, the TD87a tephra correlates to the 1254 CE eruption of Mount Rittmann, and is located c. 2 m below the TD85 (Severi et al. 2012). Given that the 1254 CE tephra is ubiquitous in the marine and glacial records of the Ross Sea and northern Victoria Land, a significant and widespread tephra marker (Di Roberto et al. 2019 and references therein), and it is not between BRH4 and BRH5 in the proximal exposure, we conclude that BRH4 does not correlate to TD85.
A tephrostratigraphy similar to that of Talos Dome has been developed for the Styx glacier ice core (which is the ice record closer to the studied BRH site), where the 1254 CE tephra was detected at a depth of 99.18 m and is only overlaid by another tephra at 97.01 m (Han et al. 2015), instead of five ash layers.
Vice versa, our data confirm that TD85 likely derives from Mount Melbourne volcanic activity, as previously suggested by Narcisi et al. (2012), based on the major element data. If this correlation would be verified using trace element glass compositions, it would affirm that Mount Melbourne was active in even more recent times, less than 700 years ago. This further shows that incomplete geochemical databases strongly limit correlation efficacy, and highquality geochemical data are necessary to make reliable correlations.
A general geochemical affinity also exists between BRH1, BRH2, and BRH5 tephra and the composition of BIT volcanic ash . In detail, there is a strong similarity in major element compositions between the BRH1, BRH2, and BRH5 basalt to basanite glass population and the BIT121 and BIT122 ash layers (Fig. 5). Trace element compositions of BIT are not entirely determined, and the same similarity can be only documented in the Th vs Hf plot (Fig. 7d). BIT121 and BIT122 layers are respectively a thick dark brown unit made of clear, aphyric, and dense blocky shards (BIT121) and a grey diffuse layer with clear to olive green glass, blocky shards with spherical vesicles, plus droplets and glass hairs. In particular, the latter has strikingly similar textural characteristics to the BRH tephra. Unfortunately, BIT volcanic ash layers lack chronological and appropriate trace element information, and thus cannot be definitively correlated with BRH.

Eruption dynamics
Componentry, textures, and geochemical fingerprints of the studied tephra indicate that these layers have been emplaced as primary fallout deposits. As mentioned above, the external morphology of particles that preserves pristine features like fragile glass tips, Pele's hair, or spiny glass edges (Figs. 3 and 4) and glass coatings around magmatic crystals indicate minor or no aeolian remobilization, resedimentation, or other transport after deposition from the volcanic plume.
The volcanic glass composition of each sample, which is strongly bimodal with other analyses along geochemical trends, always lies on the same evolutionary compositional lineage indicating the same volcanic source (see Del Carlo et al. 2018). The proximity of these decimetre-thick units to the MMVF, and their shared geochemical affinity, is consistently being from Mt. Melbourne or other vents around the summit. The absence of large amounts of detrital material, including glass particles from other volcanic sources, is evident and indicates negligible reworking, re-sedimentation, and redistribution of pyroclastic products.
Glass particles forming the studied deposits comprise a variety of fragments differing in external shape, vesicularity, mineral content, and composition. Internal and external textural inhomogeneity is widely documented in the products of explosive eruptions from many volcanoes and has been ascribed to changes in the ascent and/or flow conditions during the eruption and to diverse mechanisms of particle fragmentation (i.e. hydromagmatic vs magmatic), transport, and the eruptive environment (see, e.g. D'Oriano et al. 2022 and references therein).
The almost ubiquitous presence of fluidal fragments including Pele's hair and tears (i.e. fibres and droplets of volcanic glass; Figs. 3 and 4) indicates specific eruptive dynamics and magma properties typical of magmatic eruptions. Fluidal fragments and Pele's hairs and tears formed by the stretching, deformation, and final magma breakup of very hot, low-viscosity magma due to interfacial shear between gas and melt (see Lin and Reitz 1998;Eggers and Villermaux 2008). Pele's hairs and tears form during subaerial Hawaiian-style eruptions fed by low-viscosity basaltic melts (Heiken and Wohletz 1985) and are also common, but minor, in basaltic eruptions occurring in sub-aqueous environments (Clague et al. 2000(Clague et al. , 2003. Similarly, the presence of abundant moderately to highly vesicular basaltic particles with cuspate to spongy external surfaces is indicative of pure magmatic fragmentation of gas-rich melts dominated by exsolution and expansion of magmatic gas, for example in a Strombolian column. More equant and blocky particles have irregular contours, are bounded by sharp planar edges, and are made of dense to poorly vesicular glass, bearing features like stepped surfaces and quenching cracks, as well as hackle lines and surface pitting ( Fig. 4i-l), which are distinctive of phreatomagmatic fragmentation (Fisher and Schmincke 1984;Heiken and Wohletz 1985;Morrissey et al. 2000;Dürig et al. 2012;Zimanowski et al. 2015;Ross et al. 2022). Stepped features are a consequence of the extreme and intensive brittle fracturing of the melt (Zimanowski et al. 2015). The water, in direct contact with the hot melt, expands and exerts a massive hydraulic pressure onto the melt itself, which behaves in a brittle way, so that cracks are occurring, driven by the applied stress; still, liquid water is pushed into the propagating crack, thus increasing the interface area between magma and water and therefore increasing the heat flux from magma to water, generating an accelerating thermohydraulic feedback loop Dürig et al. 2012). Hackle lines indicate the direction of propagation of a crack and, since these can form during slow or fast cracking, are not strictly diagnostic of melt-water interaction, although some fragmentation experiments with water produce increased proportions of particles with hackle lines relative to others without water (Ross et al. 2022). Quenching cracks form immediately after fragmentation due to the sudden quenching and consequent contraction of still-hot particles due to the fast contact with liquid water (Zimanowski et al. 2015). Pitted surfaces indicate incipient alteration resulting from the interaction of glass with hydrothermal fluids (or excess water) in the eruptive column (Zimanowski et al. 2015;Ross et al. 2022).
The simultaneous presence of particle morphologies with features indicative of magmatic and hydromagmatic fragmentation in BRH tephra layers is evidence that various degrees of magma-water interaction occurred during the forming eruptions. For example, texturally mixed deposits could derive from the transformation of an initially subglacial, hydromagmatic eruption fuelled by the interaction between magma and meltwater, into a relatively dry magmatic subaerial Hawaiian-to Strombolian-style eruption, once the water source was exhausted and drained or the magma no longer came into contact with the meltwater (vent completely subaerial). As mentioned above, the major and trace element glass compositions of the heterogeneous BRH1 to 5 tephra deposits are largely overlapping, and clearly reside on the same evolutionary trends indicating they all likely derive from the same magmatic system. The overall chemical variability (ranging from basalts through to trachytes) and in particular the observed clustering of the erupted volcanic glass compositions along the overall evolutionary trend might indicate that the successive eruptions were fed by a complex and vertically extensive magma storage region beneath the source volcano (Cashman et al. 2017;Giordano and Caricchi 2022). This is quite common as reported for several eruptions (see Shane and Hoverd 2002;Shane et al. 2007Shane et al. , 2008. Interestingly, apart from BRH4, all the remaining BRH tephra deposits comprise a basaltic glass component, and the interaction of the mafic melt with more evolved trachy-andesitic to trachytic magma pockets, or mush, is the likely trigger of these eruptions. In this regard, it should be highlighted that at least during some of the eruptive phases emplacing BRH tephra, a trachytic magma was fragmented efficiently enough to disperse the ash several hundred kilometres away. This is possible only during high-intensity eruptions able to produce eruptive columns of significant heights, as hypothesized by Di Roberto et al. (2023).

Conclusions
BRH1 to 5 tephra are primary fallout layers derived from five explosive eruptions that occurred in a short time from within the Mount Melbourne Volcanic Field. The nature and texture of particles forming the tephra layers indicate that the eruptions were mostly Hawaiian to Strombolian in style and characterized by magmatic and possibly hydromagmatic fragmentation occurring over the time of the eruption. The distal dispersal of trachytic ashes to some hundreds of kilometres from the source (c. 280 km) also demonstrates that the BRH eruptions were at some time characterized by a strong explosivity and very efficient magma fragmentation, and possibly produced several kilometre-high eruptive columns.
The magma compositions feeding the eruptions range from basalts to trachytes and the clustering of the erupted product compositions indicates that the eruptions were fed by a complex and vertically extensive magma system region beneath the source volcano with the mafic melt remobilizing more evolved trachy-andesitic to trachytic magma pockets.
Although the geochemical data confirm that the volcanic source of BRH1 to 5 tephra is the Mount Melbourne volcano, there is no direct match between the composition of tephra studied here and Pleistocene to Holocene tephra layers exposed in the summit part of Mount Melbourne characterized by Del MEL samples). This potentially indicates that BRH tephra originated from vents located away from the summit area.
The geochemical correlations assessed between the BRH tephra deposits and three marine cryptotephra intercalated in sediments of the Edisto Inlet (TR17-08-512, -518, and -524), which were dated by the radiocarbon method between 1615 cal. years BP and 1677 cal. years BP, allowed to indirectly constrain the ages of BRH eruptions between the third and fourth century CE. This finding improves our knowledge of the eruptive history of Mount Melbourne volcano, which has produced at least five eruptions during historical time, increasing the awareness of potential hazards to the several permanent scientific bases close to Mount Melbourne.
Moreover, the attribution of a numerical age to the BRH tephra elevates them to new regional isochron markers that will possibly enhance the correlation and synchronization of climatic records of the northern Victoria Land and Ross Sea areas.
This study demonstrates the robustness of the tephrochronological method in terms of correlation and dating of glacial, terrestrial, and marine records, if carried out with a modern approach and up-to-date techniques. The need for high-quality textural, mineralogical, and compositional data (major and trace element glass compositions) on tephra is even more evident for Antarctica, where outcrops are scarce. Only with an exhaustive fingerprinting of tephra layers is possible to establish effective correlations between different records enhancing synchronization and correlations for palaeoenvironmental and palaeoclimatic reconstructions.