The MBS-Main core (4.25–294.785 m depth) is supported by three surface firn cores: MBS-Alpha (surface to 20.41 m depth), -Bravo (surface to 20.225 m depth), and -Charlie (surface to 25.89 m depth). Mean accumulation rates derived by annual layer counting throughout the satellite era are 0.309 ± 0.08 myr-1 ice equivalent (IE) for the main core and 0.298 ± 0.07 myr-1 IE for the MBS-Alpha core . This correlates with the ERA5 estimated site annual accumulation of 0.302 ± 0.05 myr-1 IE . The climatology and site conditions of Mount Brown South are discussed in detail in , and .

The MBS-Main and MBS-Alpha cores have been analyzed for both trace impurities and stable water isotopes (δ18O) . Measurements were performed on discrete samples at 3 cm resolution, using ion chromatography chemistry/trace impurities, and cavity ring down spectroscopy stable water isotopes, at the Institute for Marine and Antarctic Studies at the University of Tasmania and Physics of Ice, Climate and Earth, Niels Bohr Institute, at the University of Copenhagen. Additionally, the MBS-Main core was analyzed using continuous flow analysis (CFA) for trace impurities and stable water isotopes , also at the University of Copenhagen.

Limited sample material remains of the MBS-Main core throughout the satellite era depths, after re-sampling was required due to challenges with contamination during discrete chemistry sampling. The outer edge pieces remaining after preparation of samples for CFA remain available to use, but the sample size is small (∼2cm2). The MBS-Alpha core, had a much larger remaining sample area usable for this study (>10cm2). Part of the MBS-Alpha core was damaged in a freezer failure, resulting in some sections with one slightly melted edge along the length of the core. The trace chemistry samples from the damaged cores appear to be reliable see and the damaged edges of the core have been removed. While the core is not suitable for analyses where high temperature or slight meltwater migration would compromise the results, the cores are still able to be used for microparticle analysis, including tephra studies.

The chemistry and stable water isotopes for the MBS-Main and MBS-Alpha cores have been analyzed and compared across cores . The cores align well, and robust chronologies have been produced for both cores . The MBS-Main and surface cores have been dated using independent layer counting and volcanic matching. A thorough description of the process of chronology development and the resulting age scale (MBS2023) is presented in . The MBS2023 chronology provides depths for annual horizons for the main and surface cores . Thanks to the robust chronology development efforts for these cores, we are able to cross-match depths across the MBS ice core array, allowing us to make use of the multiple cores to provide larger volumes of ice core material for sampling.

The Hybrid Single Particle Lagrangian Integrated Trajectory model HYSPLIT; was used in conjunction with the age-at-depth chronology for the Alpha ice core as a one element of our strategy for sample selection as well as to investigate potential sources of volcanic glass shards identified at the MBS site. Specifically, in sample planning, atmospheric trajectories that were deemed favorable to transport of a specific event to the MBS site were identified, and provided an independent timestep to guide initial sampling. While atmospheric trajectory modeling is often used as a tool for validation of ice core cryptotephra correlations and deposition timing , its implementation for targeted sampling is not well documented in the literature. Trajectories were used both to guide the sampling strategy and in the correlation of the cryptotephra horizons found in MBS-Alpha.

Guided by previous studies which have demonstrated the dominance of extreme precipitation events and atmospheric rivers on accumulation at the MBS site , and the important role of meridional transport of maritime air masses in such extreme events , we targeted transport from the Southern Indian Ocean using HYSPLIT trajectory analysis. To this end, daily 120 h forward trajectories were computed, originating from 53.1° S 73.52° E (central Southern Indian Ocean) from 1979–present. Favorable transport conditions were identified by trajectories passing within 0.5° of the MBS site, and were used to assist with development of a directed sampling plan. Time periods with favorable trajectories were compiled together with volcanic signatures in the MBS ice chemistry record and known regional eruption events to prioritize the first pass of low-resolution sampling (six samples per meter) by aligning to the independent age-at-depth chronology developed for MBS-Main.

Trajectory analysis was additionally used in the validation of the tephra horizons identified here. Once potential horizons were identified, six-hourly 10 d back trajectories were generated originating from 1500 meter above ground level (m a.g.l.) at the MBS site (to avoid trajectories hitting the ground; Fig. ). Frequency analysis of these trajectories was used in assessing the proposed tephra horizon correlations.

For ease of trajectory generation and repeatability, we used the PySPLIT package for all trajectory generation. NCEP/NCAR reanalysis meteorology data was used to drive the model. Individual trajectories were clustered using the built-in clustering algorithm in HYSPLIT. The aim of the algorithm is to identify a number of (user-defined) maximally distinct clusters while minimizing variability within each cluster .

Lagrangian transport models like HYSPLIT are some of the most commonly used tools for investigating particle atmospheric transport but should be considered with caution. As back trajectory timelines become longer (as in the case of the 10 d trajectories used here), the potential for error due to atmospheric noise introduced into the model increases. Despite the potential for the influence of such noise in the model systems, we find HYSPLIT to be useful as a confirmatory tool, used in parallel with geochemical analysis and existing satellite observations of volcanic ash transport.

For the first low-resolution screening process, sample depths were selected from MBS-Main based on atmospheric circulation modeling, in conjunction with known regional eruption records and volcanic signals in the ice core chemistry, in order to target volcanic material entrained in mid-latitude moisture from the southern Indian Ocean. These moisture-rich air masses were targeted using HYLSPLIT trajectory generation (as described in Sect. ), selecting for meridional transport trajectories, as it has been demonstrated that the meridional transport pathway is a source of significant snowfall accumulation to MBS . We hypothesize that air masses transported along these trajectory lines could entrain material from volcanoes in the Kerguelen Plateau region (Heard and McDonald Islands) or from sources further afield. In order to refine our sampling strategy and to mitigate the need for time-consuming high-resolution sampling of the full length of the core, we targeted our screening sampling towards depth/ages that corresponded with air masses passing over the Kerguelen Plateau region (i.e. meridional transport) prior to arriving at MBS.

As this is the first investigation of cryptotephra in the MBS ice cores, for this study we focus on the satellite era (1979–present). This allows us to utilize atmospheric circulation modeling to assess volcanic ash transport, and to consider satellite volcanic observations to inform our volcanic matching efforts. Additionally, knowledge of the magnitude and timing of eruption events identified in satellite era ice cores can help fine-tune future efforts to reconstruct events identified from volcanic horizons in the more distant past. Furthermore, from a practical perspective, the duplicate cores covering this period allows for larger sample volumes for analysis, increasing the likelihood of capturing sparse tephra in a given sample.

First, a broad low resolution screening of the targeted depths was conducted from the MBS-Main core based on the previously described sampling plan and using samples ranging from 15–18 cm in length (six samples per one-meter-long ice core segment, Fig. b, measured to the nearest half-centimeter), and cut with a band saw from the outer edge of the core (∼2cm2 cross sectional area). With a minimum annual layer thickness throughout the satellite era of 0.256 m firn depth (mean 0.501±0.137m, based on the MBS2023 chronology, ), this sample size represents at least sub-annual sampling resolution. MBS-Main core samples were melted in clean, new polycarbonate bottles before being transferred to clean new 10 mL centrifuge tubes and centrifuged (5 min, 3000 rpm). The resulting concentrate was evaporated on frosted glass slides and the remaining sample material was coated in low-viscosity epoxy resin (Logitech type 301 2-part epoxy resin). For efficiency and because of difficulties resulting from the preparation of the slides, these samples were only investigated using optical microscopy, and samples depths were prioritized based on the presence or absence of potential tephra grains. Of a total of 65 exploratory samples from the MBS-Main core, 22 samples were selected based on visual inspection using optical microscopy for re-sampling of the corresponding depth ranges in the MBS-Alpha core (Fig. a1–2).

As these samples were prepared from a narrow outer edge piece from the ice core, the small cross-sectional area meant that robust sample cleaning/decontamination procedures were not feasible for the MBS-Main samples. We were therefore not able to eliminate the potential for contamination by transfer from one core depth to another. General lab contamination debris also potentially obscured tephra and limited our ability to form robust identifications. Additionally, due to the sample preparation methods used for the MBS-Main samples (uneven surface of the glass slides used), we were not able to reliably fully section the very fine tephra grains for further microanalysis and were thus required to rely only on optical microscopy for these samples. This necessitated secondary sampling from the MBS-Alpha core, of which there was more sample volume remaining (Fig. c). The MBS-Main samples were valuable in guiding our sample selection for MBS-Alpha, allowing very efficient, targeted use of the core.

The second phase of sampling was able to be conducted at a higher resolution due to the available sample volume available from MBS-Alpha. The depths from the 22 selected MBS-Main core samples were transposed to the corresponding MBS-Alpha core depths (based on the annual horizons presented in , Fig. a3), resulting in 70 sample depths ranging from 4–8 cm in length (Fig. a5). Samples were measured to the nearest half-centimeter and cut using a thin bladed pull-saw. Careful decontamination procedures were followed during the cutting and preparation of the MBS-Alpha ice samples to prevent transfer of sample material from one sample to the next. MBS-Alpha samples were cut and decontaminated under a laminar flow hood in the cold lab. The cutting surface and blades were cleaned between each individual ∼5cm sample, and both the ceramic blade used and the cutting surface (a clean, new plastic cutting board) were changed and washed with ultrapure water between samples from each meter of core (e.g. a freshly washed blade and cutting board were used for samples from MBS-Alpha 14, and then washed before decontamination of samples from MBS-Alpha 15).

After ∼1mm of material from each of the outer edges was removed using a ceramic blade, the samples had a cross sectional area of ∼10cm2. Samples were melted at ambient temperature in rinsed sterile Whirl-Pak bags. Melted sample material was transferred to acid-washed 15 mL centrifuge tubes, and Whirl-Pak bags were rinsed at least twice with ultrapure water to minimize the possibility of sample material being left behind in the bags. Samples were then centrifuged (5 min, 3000 rpm) and the resulting concentrate was pipetted into wells created by placing cleaned 25 mm acrylic rings on polyimide (Kapton) tape adhered to flat glass plates, and set to evaporate on a hot plate at 60 °C. The centrifuge tubes were subsequently rinsed at least twice with ultrapure water, and the rinse water was added to the same sample area to evaporate. Each sample well was then backfilled with low-viscosity epoxy (Struers EpoFix) to create round resin mounts.

The sample mount surfaces were polished using 1 µm aluminum oxide polishing compound to remove surface resin and expose any tephra grains present in the mounts, then cleaned in an ultrasonic bath to remove excess polishing compound. Samples were examined using transmitted and reflected light using a petrographic microscope. Mounts containing tephra grains of suitable size and quality e.g. with large enough surface area for the microanalysis beam, accounting for mineral inclusions and/or vesicles, larger than ∼5µm on the shortest axis) were selected for future electron probe microanalysis (EPMA; Fig. a6). Selected samples were carbon-coated in preparation for BSE imaging, scanning electron microscopy by energy dispersive spectroscopy (SEM-EDS, FEI MLA 650 ESEM), and EPMA (Fig. a7–9) at the Central Science Laboratory at the University of Tasmania.

While MBS annual layer thickness allows for sub-annual sampling, the site is characterized highly variable accumulation, both between- and within years . Annuals in MBS-Alpha range 22.6–85.4 cm firn depth, and found that 88 % of inter-annual variability in MBS is associated with extreme precipitation, with on average ∼6% of days accounting for over 50 % of annual accumulation.

No sub-annual chronology has been published for MBS, however characterize the seasonality of impurity flux to the MBS site. Here, we rely on the austral summer peaking sulfate-to-chloride ratio (SO42-/Cl-) and δ18O and austral winter peaking sodium (Na+) to approximate within-year ages. We note, however, that such linear interpolation-derived sub-annual dating does not account for within-year variability, and therefore should be considered with caution.

The mounted volcanic glass shards were analyzed using EPMA at the Central Science Laboratory (CSL) at the University of Tasmania (Fig. ) in 2023 and early 2024. Analyses were conducted on a JEOL JXA-8530F Plus field emission microprobe with five wavelength dispersive spectrometers. Single point analyses used a 2 µm beam diameter with 4 nA beam current, 15 kV accelerating voltage. Analytical conditions were chosen to balance minimizing beam damage, particularly Na ion migration, while maintaining a beam size suitable for our smallest (≤5µm) shards.

EPMA analytical conditions were selected in collaboration with the instrument scientists, with the aim of obtaining as robust analytical totals as possible on our small glass shards while minimizing alkali ion migration and maintaining consistent analytical conditions across all measurements. While larger a beam size could have been used for a few of the larger samples analyzed using the broad-beam overlap method , we chose to prioritize consistent analytical techniques, using the same beam size for all samples analyzed in order to enable better comparison across samples and obtain results from the largest number of glass shards possible in our samples.

Concentrations of 12 major and minor element oxides were measured in the individual shards (SiO2, TiO2, Al2O3, FeO, MnO, MgO, CaO, Na2O, K2O, P2O5, SO3, F). A series of mineral standards were used for instrument calibration, a rhyolitic glass reference standard from the Smithsonian Institution Department of Mineral Sciences VG-568, NMNH 72854; was used to validate accuracy and precision of measurements between measurement sessions. See the Supplement for full details on the EPMA measurements and standards.

The dataset was detection limit filtered (99 % confidence), and erroneous analyses of non-volcanic material (e.g. quartz grains or other mineral particles in the samples) were removed from the dataset. Data were corrected for Cl concentrations and normalized to 100 % anhydrous as recommended by prior to interpretation. Full geochemical results and secondary standards measurements can be found in the Supplement.

Glass shards are most abundant in the 13.28–13.34 and 16.87–16.915 m depth samples. The 13.28–13.34 m sample has the largest glass shards (Fig. ), ranging from 10–20 µm along the longest axis. All shards are angular, while the larger shards in this sample have cuspate, bubble-wall, or y junction morphologies and smaller shards appear more blocky or platy. Glass shards in the 16.87–16.915 m sample are typically smaller, <10µm, with only one shard<10µm (∼15µm) across the longest axis. These shards have simpler angular morphology, and include blocky and platy shapes, some have fluted or cuspate edges. The remaining samples contain glass shards ranging from <5-15µm, with mostly simple angular morphologies.

Major element oxides of all tephra identified in the MBS-Alpha core are provided in Table 2. The MBS-Alpha glass shards have SiO2 values ranging from 51.70 wt %–77.31 wt %, with 1.38 wt %–9.12 wt % Na2O, and 0.28 wt %–6.0 wt % K2O (Fig.  and Table 2). When plotted on the TAS diagram , compositions for the glass shards include basaltic andesite, trachyandesite, andesite, dacite, trachyte/trachydacite, rhyolite, and phonolite. The most abundant compositions are dacite (15 shards), phonolite (11 shards), and rhyolite (10 shards). Remaining compositions have fewer shards (one basaltic andesite, two trachyandesite, one andesite, and two trachyte/trachydacite shards).

The glass shards in the 13.28–13.34 m sample (Table ) are mainly dacite in composition, with one andesite shard (Fig. ). The analyses from this depth have SiO2 ranging from 61.87 wt %–66.23 wt %, with total alkalis (Na2O + K2O) ranging from 3.16 wt %–7.27 wt %. Bivariate diagrams demonstrate there is little variation in the abundance of other major element oxides when plotted against SiO2 (Fig. ).

All glass shards analyzed in the 16.87–16.915 m sample (Table  have a homogeneous phonolitic composition. Compared to the 13.28–13.34 m horizon, these glass shards have a somewhat lower SiO2 concentration, ranging from 56.31 wt %–58.76 wt % and are strongly alkaline (Na2O+K2O), ranging from 12.80 wt %–14.23 wt % (Fig. ). As a group, these shards have among the highest Na2O and K2O contents of all MBS-Alpha glass analyzed here (7.04 wt %–9.12 wt % and 4.78 wt %–5.76 wt %, respectively).

EPMA analytical conditions were selected and analyses completed prior to the publication of the recommendations for best practices for EPMA analysis of fine grained tephra . Our analytical conditions produced meaningful results, however some limiting aspects require consideration.

Many of the analyses have analytical total oxide values of >90%, due to the very small size of the shards (<5µm). However, approximately half of the sample analyses resulted in lower analytical totals. Of the 42 EPMA data measurements presented here, five have analytical totals below 60 % (indicated in Table ). We attribute low analytical totals to the following scenarios (1) glass shards that are very thin (either naturally, or due to the amount of polishing required), resulting in an interaction volume depth that includes the resin substrate below the shards and/or (2) larger shards with complex morphology causing the resin to be not uniformly removed from the surface of the shards, resulting in some resin being included in the beam area. This could have been resolved by additional polishing, however this risked fully polishing away some of the smaller shards in the sample. As a result, we chose to accept slightly lower totals in order to include analyses of as many shards as possible.

show with their “broad beam overlap” method that analytical totals as low as 67 % are statistically similar to the same analyses with higher analytical totals, with only a slight decrease in precision. show reliable results with analytical totals as low as 60 %. report good accuracy despite somewhat lower precision for EPMA analyses of very fine grained tephra shards resulting in analytical totals ranging from as low as 35 % up to 101 wt %. Following this approach, analyses with analytical totals of at least 50 % are presented here. Only one sample with <60% total (shard 17-9_009) is included in the horizons that we pursue source correlations for. As the major element compositions of this shard fall within the range of the other glass shards in this sample, we have chosen to include it in our analyses.

The probe beam used in EPMA has the potential to damage the sample by inducing migration of alkali ions typically Na2O;. Analytical conditions for EPMA were selected to minimize potential migration of alkali ions for small glass shards and repeated analysis of secondary standards did not show a loss of sodium over time (see the Supplement). This loss of sodium is due to probe beam damage in cases of repeated analysis in a small area or single glass shard. While most of the glass shards in our sample were too small to fit multiple analyses, some of the glass shards in the 13.28–13.34 m sample were large enough for multiple analyses. Data from these analyses are presented in Table 2 as the average of multiple analyses (see the Supplement for individual point measurements). Some of these glass shards did show decreases in Na2O over multiple analyses (see Fig. S1 in the Supplement) and consequently we have not relied exclusively on Na2O values for our tephra source identification.

The 13.28–13.34 m sample falls between the 1991 and 1992 annual horizons 13.761 and 13.099 m respectively, MBS2023;. In Fig. , sample depth ranges are shown alongside seasonally varying isotope and chemistry species measured in the MBS-Alpha core. It is notable that during the development of the MBS2023 chronology, it was determined that trace chemistry was as reliable as δ18O, and at times more so, in determining annual horizon positions in the MBS records see Sect. 4.3 in. It can be seen in Fig.  that the 13.28–13.34 m sample falls towards the end of the austral winter sodium peak, but just at the start of the austral summer peak in sulfate-on-chloride, indicating that the glass shards were likely deposited with snow that fell in mid to late 1991. It can also be seen that the sample falls just at the beginning of a significant peak in non-sea-salt sulfate, likely volcanic in origin.

The 16.87–16.915 m sample falls between the 1985 and 1986 annual horizons (17.282 and 16.679 m respectively). This sample depth coincides with the end of the austral winter sodium peak and the start of the the sulfate-on-chloride peak, indicating that this sample likely represents snow from the middle of the year (late austral winter to early spring) 1985.

For clarity, we will hereafter refer to these two samples by their year of origin, according to the MBS2023 chronology; 1991 for the 13.28–13.34 m sample, and 1985 for the 16.87–16.915 m sample.

A small number of shards originating from a range of sample depths (from 6.48–17.27 m depth), form a compositionally similar group of rhyolites. Despite originating from a broad range of depths (and therefore ages), if treated as a group, the composition of these shards cluster together across most major element oxides (Fig. ). The composition of this cluster is similar to compositions determined for volcanic glass shards produced in the 1991 eruption of Pinatubo . However none of the depths where glass shards of this composition were found correspond within reasonable dating error to the 1991 eruption of Pinatubo, where a sulfate peak is often used as a dating horizon for 1991/1992 in other ice cores .

The rhyolitic glass shard compositions fall within the range of compositions of sparse mid- to late-Holocene rhyolite shards reported in the B53 and B54 ice cores, which correlate with shards identified in the Talos Dome ice core. attribute these Talos Dome rhyolite shards to various extra-Antarctic sources (including Andean and New Zealand sources). have correlated similar high-K rhyolites with either “extremely evolved” products of West Antarctic intraplate volcanoes (including McMurdo group and Marie Byrd Land volcanoes, not active during the satellite era), but also show similarities to Antarctic samples correlated with South American (Aguilera) as well as South Shetland Islands volcanoes. Additionally, these rhyolites also show some compositional similarity to the rhyolitic component of a sample of ash collected from the crater rim of Mt. Erebus in 2000 . Due to the multiplicity of sources attributed to tephra with similar compositions, as well as their sparseness in the MBS-Alpha core, we cannot confidently attribute a single source to this compositional group of rhyolites.

An alternative interpretation of such sparse and heterogeneous samples is to suppose they represent a fraction of the background “dust” signal often seen in ice core analyses . This background signal could include re-mobilized volcanic ash from a variety of sources, transported from nearby ice-free areas or other extra-Antarctic sources . Typically volcanic glass shards transported by wind display evidence of reworking e.g. abraded margins . Textural alteration depends on the duration and conditions of erosion and weathering glass shards are exposed to. While all glass shards chosen for analysis have angular shard-like morphologies that we interpret to be primary textures (see Fig. S2 for an example BSE image of the 8.015 m sample), we cannot rule out the possibility that these glass shards come from re-mobilized material.

While robust contamination preventative precautions were taken during sampling, there is some possibility that some of these unidentified rhyolites (as well as the other unidentified sparse glass shards) could have been the result of contamination from unknown lab sources or between samples. We cannot definitively rule out this as a possibility, however we consider it unlikely, as the sparse shards have been characterized as tephra both based on visual inspection and geochemical composition, and all sample-contact surfaces were rigorously cleaned between preparation of each sample.

The presence of a substantial spike in nss-SO42- Fig. ,, likely of volcanic origin, co-occurring with the 1991 sample bearing multiple glass shards suggests that the tephra found could be associated with the volcanic event that produced the sulfate signal in the ice core. Based on the age of the sample, we consider the two primary candidates for the source of the 1991 tephra horizon to be the eruptions of Pinatubo (Philippines, volcanic explosivity index (VEI) 6) and Cerro Hudson (Chile, VEI 5), both occuring in mid-1991 . However we also investigate all other significant (VEI 3 or greater) known tropical and Southern Hemisphere eruptions from 1991 (Nyamulagira, Democratic Republic of Congo and Lokon-Empung, Indonesia). Additionally, we include other Antarctic and sub-Antarctic volcanic sources (Gaussberg, South Sandwich Islands, and South Shetland Islands), as although there are no known eruptions of these volcanoes during 1991, they are known or speculated possible sources of tephra (primary or reworked) identified in other Antarctic ice cores .

It is clear from the TAS diagram (Fig. ) that products of Nyamulagira and Lokon-Empung do not correlate with glass from the MBS-Alpha 1991 horizon, and neither do the products of Gaussberg or the South Shetland Islands. We therefore rule these out as potential sources. While the South Sandwich Islands do appear to correlate somewhat better on the TAS diagram, when compared with literature values Fig. ;, the 1991 tephra is characterized by significantly higher K2O values (1.99 wt %–3.09 wt %) than South Sandwich Islands (< 1.5 wt % K2O) for similar SiO2 ranges. This combined with the lack of recorded eruptions of the South Sandwich Island volcanoes during this time period, leads us to rule out South Sandwich Islands as a potential source.

We are left with Pinatubo (15 June 1991), and Cerro Hudson (8–14 August) as potential sources for the glass in the MBS-Alpha 1991 horizon. While we are able to approximate general sub-annual ages of the individual samples, due to the intra-annual variability seen in MBS accumulation (see Sect. ), the 2-month difference between the Pinatubo and Cerro Hudson eruptions is below our ability to resolve based on dating uncertainty and the size of the sample (∼6cm). Since we cannot rely on sample age, we focus on geochemistry and atmospheric transport modeling to guide our source identification. The products of the 1991 eruption of Pinatubo are primarily rhyolitic (Fig. ), and have much lower TiO2 (< 0.52 wt %) and FeO (< 3.01 wt %; ) than both the MBS-Alpha 1991 glass and the majority of the Cerro Hudson products. Based on the difference in TiO2 and FeO, we rule out Pinatubo as a potential source candidate for the dacite cluster in the 1991 tephra. The geochemical compositions of the glass shards in the 1991 sample show a strong similarity to the Holocene volcanic products of the Cerro Hudson volcano (Fig. ).

The 1991 eruption of Cerro Hudson was one of the 20th century's largest explosive eruptions, producing 43 km3 (bulk volume) of tephra . Beginning with a phreatomagmatic event on 8 August 1991, with a VEI of 3. A second phase of the eruption began on 12 August with a Plinian eruption with VEI of 5, which produced the majority of the tephra ejected during the eruption . The eruption ejected ash as high as >18km into the stratosphere, and ash dispersal was well documented spanning thousands of kilometers to the southeast .

The major elements of the 1991 dacite tephra match well with Holocene volcanic products of Cerro Hudson (Fig. ; variation diagrams against MgO included in the Supplement, Fig. S3). This is seen particularly with the later products from the 1991 eruption, as the composition shifted from basaltic-andesite towards large volumes of trachyandesite to rhyo-dacite products (Fig. , ). The dacite glass from the 1991 tephra horizon show a slightly less alkaline composition than some Cerro Hudson products, including glass shards from the 1991 eruption recovered from visible tephra deposits in lake and terrestrial cores from southern Chile (, Fig. ). This lower alkalinity is driven largely by a lower concentration of Na2O, while the remaining major element oxides show similar concentrations to the reported literature values for Cerro Hudson eruption products Fig. ;. The lower Na2O value could be in part due to the minimal migration of sodium during EPMA analysis described in Sect.  above. Despite the potential Na2O loss, the majority of our analyses are within the range of concentrations of major element oxide concentrations seen in the literature for Cerro Hudson (Fig. ). Based the geochemistry and chronology of this cryptotephra horizon, we propose that the MBS-Alpha 1991 tephra originate from the 1991 eruption of Cerro Hudson.

There are no globally significant volcanic eruptions on the order of magnitude of Cerro Hudson or Pinatubo recorded in 1985 to help narrow the search for the source of the glass shards identified in the 1985 horizon. The phonolitic composition of the glass shards is uncommon enough to narrow down the candidates for volcanic source matching. To our knowledge, there are only two phonolitic volcanoes active during the last 1000 years . These two volcanoes are the McDonald Islands volcano, located near Heard Island on the Kerguelen Plateau, ∼2000km NNW of MBS, and Mount Erebus, Ross Island, Antarctica, ∼2500km SSE of MBS.

There are no known observations of volcanic activity from the McDonald Islands from their discovery in 1874 until 1997. Since 1997, only two eruptive events have been described, with ongoing hydrothermal activity . However, the eruptive history of the islands is poorly known due to their extreme remoteness, and near-constant cloud cover in the region making satellite observation difficult for most of the year .

Mount Erebus has been continuously active since at least 1972 with a maximum VEI of 2, and periods of increased activity from 1984 through 1985 . Reports from nearby Scott Base describe frequent activity, including the detection of strombolian eruptions, audible explosions, and incandescent ejecta seen from McMurdo Sound and as far away as Butter Point (70 km from the volcano) .

The phonolite tephra identified in the 1985 horizon show similarities to the composition of volcanic glasses from both Erebus and McDonald Islands (Fig. a).

Few analyses of McDonald Islands volcanic material exist in the published literature . investigated an obsidian-rich phonolitic pumice washed up on a beach in the Chatham Islands, Aotearoa, and characterized it as a product of McDonald Islands, likely from the 1997 eruption. Due to the rigorous matching efforts of , and to expand the available match dataset to include recent eruptive products, we include the Chatham Island sample as McDonald for comparison. Analyses from MBS-Alpha 1985 show a less alkaline composition than that of McDonald Islands, largely driven by a lower concentration of Na2O (Fig. ). Additionally, MBS-Alpha 1985 has higher concentrations of TiO2, FeO, and MnO, and generally appears distinct from the literature values for McDonald Islands .

The variation diagrams for K2O, TiO2, Al2O3, Fe2O3, MgO, and CaO show that the MBS-Alpha 1985 glass correlates strongly with literature values for Mt. Erebus (Fig. ). Comparison glass data for Mt. Erebus are sourced from englacial (blue ice and ice core) tephra correlated with Erebus , lava bombs ejected from 1972–2004 collected near the flanks of Erebus , and ash sampled near the crater rim in 2000 . While some of the literature shows very stable, homogeneous populations for the products of Mt. Erebus , analysis of the 2000 eruption crater rim ash has more highly variable compositions, ranging from phonolitic to trachytic to rhyolitic. Additionally, when whole rock samples are considered , the 1985 tephra fall well within the range of compositions expected from Erebus. MgO bivariate diagrams were also considered (Fig. S4), and the MBS-Alpha 1985 tephra cluster more strongly with Mt. Erebus products than with those of the McDonald Islands.

Here we have identified and described two cryptotephra horizons in the MBS-Alpha ice core which we interpret to be the products of eruptions of Mt. Erebus (1985) and Cerro Hudson (1991). In total, we identified geochemically verifiable tephra in 17 % of the samples collected from MBS-Alpha (12 out of 70 samples). Volcanic sulfate and elevated aerosol particle counts have been previously identified in Antarctic ice congruent with the timing of the 1991 eruption of Cerro Hudson . The MBS 1991 horizon is, to our knowledge, the first geochemically linked identification of Cerro Hudson 1991 glass tephra shards identified in Antarctic ice. Our identification of these cryptotephra horizons highlights both the importance of developing new sampling strategies for locating cryptotephra in ice cores and the potential of the coastal East Antarctic site as a millennial-length archive for future tephrochronology studies.

Tephra and cryptotephra have not been reported for many coastal East Antarctica ice cores. This is because visible tephra layers are uncommon; cryptotephra in these cores tend to be small and sparse, and traditional techniques for identification of cryptotephra require time-consuming, comprehensive sampling of tens to hundreds of meters of ice core material . Cryptotephra sampling is most often a “needle in a haystack” type of effort, and even in the relatively tephra-rich ice cores from Greenland that are in close proximity to active volcanic centers in Iceland, success rates for tephra sampling efforts are low, especially for broad, comprehensive sampling efforts. It is not uncommon to collect hundreds of samples from a single ice core and identify tephra in as few as 2 % of samples . Targeted sampling plans based on ice core chemistry and known eruptions are highly effective in identifying tephra , and we demonstrate here that incorporating atmospheric transport into sampling strategies can further improve the likelihood of identifying cryptotephra in ice cores distal to major volcanic sources.

Here, we have incorporated atmospheric modeling tools that enable the hindcasting of atmospheric circulation processes potentially relevant to the cryptotephra transport (over the satellite era). This allowed the development of a sampling strategy tailored to the MBS ice core, by targeting depths where we expected cryptotephra to be found by using the atmospheric transport characteristics of the site to guide sampling together with existing volcanic records and ice core chemistry. The efficiency is gained by the initial atmospheric analysis targeting circulation patterns that would be beneficial to transport to the ice core site as a complementary factor (rather than bulk sampling at low resolution the entire length of the Alpha ice core). Using this initial guidance of episodes of favorable meridional transport from the southern Indian Ocean region to the MBS site, our method demonstrates the potential of this type of sampling strategy in successfully and more efficiently locating sparse cryptotephra that are not necessarily associated with sulfate or insoluble particle anomalies. We suggest this method could be adapted to other coastal Antarctic ice core locations such as Roosevelt Island, Fig. ; where sampling based on ice core chemistry alone is made challenging by intermittent precipitation or high background levels of marine aerosols.

The presence of volcanic material from Mt. Erebus and Cerro Hudson in the MBS-Alpha ice cores indicates the potential for the preservation of broader Antarctic and ex-Antarctic sources for cryptotephra in East Antarctic ice cores and the MBS ice core array in particular. We expect that other cryptotephra layers exist in the MBS-Alpha ice core, transported by atmospheric pathways not used to guide the sampling performed in this study. Cryptotephra from large, satellite era eruptions outside of our targeted source area (e.g. Puyehue-Cordón Caulle, El Chichón) may exist in MBS-Alpha but were not captured by our method. This is the first tephrochronlogical investigation of MBS ice cores, and provides insight into a region underrepresented in the Antarctic ice core array. Future tephra studies, either from the remaining MBS-Main archive, or from a new core drilled in a similar location, would greatly increase our understanding of the volcanic archive potential of marginal East Antarctic sites. While the available remaining volume of MBS-Main archive is small (2 cm2 cross section, a volume often used in tephrochronology studies; ), we propose that our findings here justify using a atmospheric-transport focused sampling to develop a tephrochronology of the deeper section of MBS-Main.

Successful preservation of tephra in an ice core relies on an explosive eruption producing an ash plume coinciding with favorable atmospheric circulation patterns to transport the ash to the ice core site. Multiple volcanic factors influence this process including the height of the ash plume, the volume of ash produced, duration of eruption and the location of the volcano relative to the deposition site . In the case of known eruptions, the size of the eruption and the location of the volcano are often the criteria guiding the sampling of Antarctic ice core for tephra and crypotephra, while unknown eruptions are often identified through the presence of associated sulfate and/or insoluble particle anomalies. It has often been expected, however, that only volcanoes erupting in Antarctica or large, globally significant eruptions outside of Antarctica (VEI>5) are likely to be preserved in Antarcic ice cores . Here we have expanded our sampling strategy to include atmospheric trajectory analysis as a method to successfully sample cryptotephra in ice core. MBS is well suited to test our strategy because we have a strong understanding of the synoptic setups that govern atmospheric transport to the ice core site .

The atmospheric linkage of coastal East Antarctica and the MBS region with the southern mid-latitudes is well documented . When the MBS ice core site was selected, one of the selection criteria was the prevalence of teleconnections to lower latitudes, especially capturing interconnectedness with large scale modes of climate variability including the Southern Annular Mode and the Indian Ocean Dipole;. The existing body of knowledge about the role of meridional transport and the observed coastal easterlies at the MBS site support MBS as a potential cryptotephra archive.

The back trajectory analysis performed in the planning of our study provides evidence that climatological transport pathways linking MBS go beyond its connection with the southern Indian Ocean, and connect the MBS site with a broad range of high latitude air masses. This is consistent with recent investigations that demonstrate that atmospheric rivers and extreme precipitation across the Antarctic continent bring significant heat and precipitation from the sub-tropics and mid-latitudes to polar sites like MBS over short timescales . The characteristics that made MBS so well suited as a high resolution record of climate variability also increase its chance of preserving tephra from lower latitudes when eruptions coincide with these kinds of atmospheric events.

The identification of the 1991 Cerro Hudson eruption in the MBS-Alpha ice core has potential implications for ice core dating of both the MBS ice core array and other East Antarctic ice cores. Volcanic events are commonly recognized in ice cores via nss-SO4 (volcanic sulphate) anomalies. Deposition of volcanic sulfate from stratospheric eruptions is relatively spatially homogeneous, and measurable in ice cores using standard ice core chemistry analyses .These volcanic sulfate records have long been used in the development of robust ice core chronologies, allowing for the synchronization of ice core records across hemispheres .

The use of volcanic sulfate alone in identifying volcanic tie points at annual resolution, however, is not without challenges. Without sulfur isotope analysis, it can be difficult to determine the specific source of nss-SO4 deposition between multiple potential source volcanoes . Additionally, volcanic sulfate in ice cores can be altered through diffusion within the ice . Sulfate aerosols from distal sources can take months to years to be deposited, with elevated sulfate levels in ice core chemistry often apparent for multiple years following an eruption . For example, elevated sulfate concentrations caused by the 1991 eruptions of Cerro Hudson and Pinatubo were found in aerosol measurements at coastal Antarctic stations and in snowfall across the Antarctic Plateau with peaks measured until at least the end of 1993 . These factors can make defining the precise timing of a volcanic event challenging . In the dating of a snowpit record from the Dome Fuji site in East Antarctica were unable to precisely define the Cerro Hudson/Pinatubo volcanic horizon as a dating tie-point due to the presence of three nss-SO4 peaks within a ∼0.5m interval, likely due to seasonal variability in volcanic sulfate deposition, but possibly also from multiple volcanic eruptions.

Because individual tephra shards can often be geochemically linked to a specific eruption, tephrochronology can help with the refinement of ice core chronologies. Rapid tropospheric transport can carry tephra to an ice core site on the order of days to weeks . If tephra can be identified, it can be used to add precision to sulfate-based chronologies . Geochemical confirmation of Cerro Hudson as the source of the 1991 horizon confirms the accuracy of the dating of the MBS-Alpha ice core, as the 1991 sample is appropriately linked to mid-1991 in the current MBS chronology MBS2023,. As the time period around 1991 has previously provided challenges with ice core dating , the ability to assign a specific date to the 1991 cryptotephra horizon identified here could prove instrumental in verifying future ice core dating efforts, and enable the more accurate characterization of climate proxy-signal links in ice core research.

Several globally significant volcanic eruptions have been identified in the sulfate and/or conductivity record of the MBS-Main . These eruptions have informed the chronology and synchronization of MBS with other Antarctic ice cores WAIS, Law Dome, and Roosevelt Island,. The identification here of extra-Antarctic (Cerro Hudson) cryptotephra in MBS confirms the potential to locate and identify (crypto)tephra from other globally significant eruptions in the deeper ice of MBS. Identification of any of the volcanic events proposed by as volcanic tie-points would have important implications for refinement of the chronology of MBS and other Antarctic ice cores spanning similar time periods.