The dating and correlation of an eastern Mediterranean lake sediment sequence: a 46–4 ka tephrostratigraphy for Ioannina (NW Greece)

Terrestrial archives from the Mediterranean have been crucial to expanding our understanding of past environmental variability on a range of timescales. Dating Quaternary sequences in the Mediterranean is, however, often challenging, and age models often have large chronological uncertainties. Tephra deposits can provide crucial age control for detailed environmental reconstructions on sub‐centennial timescales. Here, tephra analysis is undertaken for the first time on a sediment core (I‐08) from Lake Ioannina, northwest Greece, for the interval spanning 46–4 ka bp. Detailed visible and ‘crypto‐’ tephra analysis identifies deposits associated with explosive volcanism at Italian volcanic sources, including Campi Flegrei, Pantelleria, and the Aeolian Islands. We identify two visible tephra layers, the Campanian Ignimbrite (CI/Y‐5; ca. 39.8 ka bp) and Pantelleria Green Tuff (PGT/Y‐6; ca. 45.7 ka), as well as the Holocene Vallone del Gabellotto cryptotephra marker (VG/E‐1; ca. 8.3 ka bp). Evidence for repeated remobilisation and redeposition of CI tephra material is outlined, and the potential mechanisms and effects of sediment reworking in lake environments are examined. Bayesian modelling, which incorporates the new tephra ages with earlier radiocarbon dates, extends the I‐08 core chronology back to ca. 46 ka bp, facilitating direct correlation of the Ioannina sequence to others in the Mediterranean region.


Introduction
Millennial and centennial-scale climate oscillations during the last glacial cycle, such as Dansgaard-Oeschger (D-O) oscillations, were first identified in the Greenland ice core records (e.g. Dansgaard et al., 1982Dansgaard et al., , 1993Johnsen et al., 1992;Andersen et al., 2004) and, subsequently, in terrestrial and marine sequences throughout the North Atlantic realm and into the Mediterranean (e.g. Bond et al., 1993;Allen et al., 1999;Goñi et al., 2000;Roucoux et al., 2001;Tzedakis et al., 2004). Detailed studies of last glacial sequences have provided important insights into the heterogeneity of climate and environmental responses to abrupt climate change. During the last glacial period oscillations between cold D-O stadial and warm D-O interstadial conditions occurred on timescales of decades or shorter, and are argued to reflect the interplay between internal atmospheric, glacial, and ocean dynamics (Li & Born, 2019;Menviel et al., 2020). Detailed proxy studies with secure chronologies are necessary to examine temporal and spatial variation in the expression of D-O cycles in the Mediterranean region (e.g. Allen et al., 1999).
Developing robust and independently derived chronologies to facilitate accurate inter-site comparisons is a necessary, but complex, task (Lowe et al., 2008;Blaauw et al., 2018). Two main dating approaches are typically used to develop chronologies for long terrestrial sedimentary records from the Mediterranean.
Radiocarbon dating is frequently applied to develop chronologies spanning the last 50 ka (e.g. Lawson et al., 2004;Staff et al., 2019). Beyond the 50 ka BP upper dating limit of radiocarbon, pollen records from terrestrial sites can be correlated to their marine equivalents, and, in turn, to the sea surface temperature (SST) and δ O 18 records therein, which provide age constraints based on alignment to the Greenland ice cores and orbital parameter, (e.g. Tzedakis, 2002;Tzedakis et al., 2004;Müller et al., 2011;Roucoux et al., 2011). Refining chronologies through correlation approaches, however, inhibits our ability to interrogate relative leads and lags in environmental responses to climate drivers (Blaauw, 2012). Moreover, the large (often millennial-scale) uncertainties inherent in the chronologies to which the records are tuned, such as the layer-counting uncertainty in the Greenland ice core records (e.g. Rasmussen et al., 2014), are rarely factored into tuned age models.
The central and eastern Mediterranean region is home to numerous active and extant volcanic centres and explosive volcanism has produced widespread tephra layers that comprise a well-dated and interconnected tephrostratigraphic framework (see Blockley et al., 2014;Bronk Ramsey et al., 2015;Lowe & Walker, 2015, and references therein), providing scope to test and improve the existing age-depth models of palaeoenvironmental records in the region (e.g. Giaccio et al., 2017;Leicher et al., 2016). Furthermore, tephra markers form time-parallel event horizons (e.g. Lane et al., 2013;Neugebauer et al., 2017) that allow the direct comparison of sequences at and between precise moments in time. By integrating palaeoclimate records from different sites, tephra studies avoid the often large chronological uncertainties associated with other dating approaches.
Analytical advances over the last two decades have improved our ability to extract, identify, and fingerprint the geochemical composition of glass shards which are not visible within the sedimentary sequence (known as cryptotephra; Blockley et al., 2005;Hayward, 2012). Consequently, tephra deposits can be detected at ever-larger distances from volcanic source regions, increasing the spatial scope of existing tephrostratigraphic frameworks to continental and, in some cases, hemispheric scales (Davies, 2015;van der Bilt et al., 2017).
This paper presents the first tephra study of last glacial and Holocene sediments from the key Mediterranean palaeoecological site of Lake Ioannina, NW Greece. In constructing a last glacial and Holocene tephrostratigraphy for Lake Ioannina we create opportunities for direct correlation of the valuable proxy record of millennial-and centennial-scale change contained therein (see Lawson et al., 2004;Tzedakis et al., 2004;Jones et al., 2013) to other key sites in the central and eastern Mediterranean region. In doing so, we allow for an interrogation of the sequencing of local climate and environmental responses to Dansgaard-Oeschger cycles at the periphery of the North Atlantic climate system.

Study site
Lake Ioannina is located in the interior of the Epirus region of northwest Greece, in the western foothills of the Pindus mountain range (Fig. 1a). The Ioannina basin, ca. 470 m a.s.l., is situated within a tectonic depression bounded by the gently sloping Tomarochoria mountains to the west and the steepsided Mitsikeli mountain to the east. The growth of the basin has been attributed to karst solution and subsidence (Lawson, 2001, and references therein). The lake itself (also known as Lake Pamvotis) has undergone extensive artificial drainage, culminating in 1959 when the northerly Lapsista sub-basin (previous water depth 1-3 m) was converted to agricultural land (Romero et al., 2002). Whilst the 'natural' (pre-drainage) scale of the lake basin is not known, it may have been as large as 20 km (Conispoliatis et al., 1986). The longest retrieved sediment record is core I-284, recovered by the Greek Institute of Geology and Mineral Exploration (IGME), which records continuous sedimentation back >250 ka BP (Tzedakis, 2002;Lawson et al., 2004;Roucoux et al., 2008Roucoux et al., , 2011. The present-day lake (Fig. 1b), located near the foot of Mitsikeli mountain in the southeast of the original basin, is 11 km on its longest axis, with a surface area of ca. 23 km 2 . The modern lake is shallow, with a maximum water depth of 10 m, and is a closed system with no major fluvial inputs, although ephemeral streams have been identified within the basin (Lawson, 2001). The lake is primarily fed by springs, most notably at the foot of the Mitsikeli Ridge (Fig. 1b;Higgs et al., 1967). Drainage occurs through sinkholes, termed katavothrai, located throughout the basin (Higgs et al., 1967).
Previous work at Ioannina has identified millennial-scale expansions and contractions of tree populations which have been linked to climate oscillations recorded in Greenland and throughout the North Atlantic, however, tuned chronologies inhibit interrogation of the timing of these responses relative to other sites in the Central and Eastern Mediterranean (Tzedakis et al., 2004). More recent studies have sought to generate independent, well-resolved age models for the site (e.g. Jones et al., 2013), however, chronological limitations continue to complicate inter-site comparisons. In particular, the hardwater effect, whereby C 14 ages are artificially inflated by unquantifiable amounts of old, inert carbon derived from the karstic bedrock (Shotton, 1972), coupled with the absence of terrestrial macrofossils and macrocharcoal, limit the utility of radiocarbon dating at the site. Through integrating the Ioannina site into the wider Mediterranean tephra framework, we seek to open up opportunities for direct correlation, providing new insights into how the relative timing of proxy responses to climate forcing vary spatially.
There is significant potential for locating tephra layers in the Ioannina sediment record. Ioannina lies to the east of several Italian volcanoes which have been active throughout the Quaternary (Fig. 1a), and thus the site is located downwind of these volcanoes assuming the present-day prevailing westerlies were dominant during the last glacial. Widespread tephra layers, found in sedimentary sequences across the Central and Eastern Mediterranean basin, have been generated by eruptions from the Campanian and Roman Volcanic Zones Marra et al., 2020), the Aeolian Island volcanoes (Albert et al., 2017), and Pantelleria Island (Jordan et al., 2018). The Aegean Arc volcanoes, e.g., Santorini, Nisyros, Yali and Kos, lie<1, 000 km southeast of Ioannina and were also active during the last glacial and tephra layers have been found in both terrestrial and marine sedimentary archives (e.g. Eastwood et al., 1999;Karkanas et al., 2014;Wulf et al., 2020

Core recovery
The I-08 core site ( ∘ 39 39.0350'N, ∘ 20 54.8990'E; Fig. 1b) lies near the earlier I-249 core site (Tzedakis, 1994), towards what would have been the centre of the lake prior to drainage. Core I-08 was recovered in 2008 from a single borehole using a truck-mounted drill, which maintained the vertical integrity of samples, with a core recovery >90%. Cores were stored in steel tubing at temperatures ∘ <6 C. Coarse sand dominates between 21.00 and 19.71 m, and the remainder of the sequence consists mainly of carbonate clays and silts. Jones et al. (2013) developed a chronology for the upper 21 m of the I-08 sequence through radiocarbon dating using a novel approach which combined radiocarbon dates from microcharcoal concentrates with dates determined through compound-specific radiocarbon analysis (CSRA). More specifically, Jones et al. (2013) measured of the C 14 content of long-chain, odd-numbered n-alkanes which originate from epicuticular waxes of terrestrial higher plants, thus circumventing the impact of local lake reservoir effects. Here we build on this chronology through tephra study from 12 m depth (early Holocene) to the base of the core at 37 m, which covers part of the last glacial (coeval with Marine Isotope Stages 3 and 2).

Sediment analyses
Sediment logging was undertaken throughout the I-08 core sequence to detect any visible changes in sedimentology which may reflect changes in the sedimentary regime at the core site. Visible tephra layers were identified on visual inspection of the core, and confirmed using low-powered microscopy to identify volcanic glass shards. Visible tephra deposits were wet-sieved to remove < μ 25 m particles, allowing the characterisation of glass shard morphology and subsequent geochemical analysis.
Particle size analysis was undertaken to supplement sediment logging on a complex unit between 28 and 32 m depth where multiple changes in grain size were observed. 323 contiguous 1 cm (ca. 1 g) sediment samples were treated with sodium pyrophosphate to deflocculate clays and analysed using a Malvern Mastersizer in the Department of Geography, University of Cambridge.
X-ray fluorescence scanning was used to analyse the chemical composition of I-08 core sediments. Measurements span the entire 38 m sequence, undertaken at 2 mm intervals using the Avaatech core scanner in the Godwin Laboratory for Palaeoclimate Research, Department of Earth Sciences, University of Cambridge.

Cryptotephra extraction and identification
Cryptotephra investigations were carried out in two stages. An initial range-finding investigation analysed contiguous 10 cm long (ca. 5 g) samples of sediment, and was followed by contiguous 1 cm analyses through sections of the core that were identified as containing peaks in tephra glass shard concentrations.
Samples were dried, weighed, and soaked in 3% hydrochloric acid to remove carbonates. For sections of the core where the initial 10 cm samples had glass shard concentrations of >10 000 shards g −1 , Lycopodium tablets of known concentration were added to subsequent 1 cm (ca. 1 g) samples (Gehrels et al., 2006). Samples were wet sieved at 25 µm, with the >25 µm remainder density separated using a heavy liquid (sodium polytungstate) solution at 1.95 and 2.55 g − cm 3 (Blockley et al., 2005).
After extraction the 1.95-2.55 g − cm 3 fraction was sieved again at 25 µm to remove any residues and mounted onto microscope slides using Canada Balsam.
Glass shards were identified at ×400 using a high-powered, polarising optical microscope. In samples where marker spores were added glass shard concentrations were calculated in shards − g 1 dry weight calculated by multiplying the number of Lycopodium spores in the tablet by the ratio of glass shards counted to Lycopodium marker spores counted. Peaks in glass shard concentration were identified as potential ash fall layers and assigned depth codes based on the lower bound of the sediment unit containing the peak.

Geochemical analysis of tephra samples
Intervals with distinct peaks in glass shard concentration at 1 cm resolution, identified as containing potential primary airfall deposits, were re-extracted using the above protocols and prepared for geochemical analysis. Low concentration tephra horizons (<500 shards − g 1 ) were concentrated by picking out individual glass shards using a gas chromatography syringe mounted on a micromanipulator . Both visible and cryptotephra samples were mounted in epoxy resin, then ground and polished to expose flat internal glass shard surfaces for electron microprobe analysis.
Samples from 12 to 20 m depth were analysed for major and minor element concentrations using the JEOL, Freising, Germany-8600 wavelength-dispersive electron microprobe (WDS-EPMA) at the Research Laboratory for Archaeology and History of Art, University of Oxford. A 15 keV accelerating voltage and 6nA beam current were used, along with a defocused (10 µm) beam. Secondary glass standards Atho-G and StHs6/80-G were used as a check on accuracy and precision of the EPMA data. All other samples were analysed using a Cameca, Gennevilliers, France SX100 WDS EPMA at the Department of Earth Sciences, University of Cambridge. Analyses utilised a 15 keV accelerating voltage, a 10 nA beam current, with a 10 µm diameter unfocused beam. Secondary standards KL2-G, T1-G, GOR 128-G, ATHO-G and STHS6/180G (Jochum & Willbold, 2006) were analysed before, between, and after batches of analyses to ensure consistency between sessions.
Trace element analyses used the Agilent 7500es ICP-MS coupled to a Resonetics 193 nm ArF excimer laser-ablation in the Department of Earth Sciences, Royal Holloway, University of London following analytical procedures outlined in Tomlinson et al. (2010). The repetition rate was 5 Hz and the count time 40 s on the sample and 40 s on the gas blank to determine the background signal. Blocks of eight sample/shards of glass and one MPI-DING reference glass were bracketed by NIST612 glass calibration standard (GeoREM 11/2006). The internal standard applied was Si 29 , as determined by grain-specific EPMA analysis.

Age-depth model
The new I-08 age-depth model incorporates new dates imported through the application of tephrochronology alongside previously-published radiocarbon dates (Jones et al., 2013). An updated I-08 age-depth model was constructed using a Bayesian approach implemented in OxCal v4.4 (Bronk Ramsey, 2020), using the IntCal20 calibration curve (Reimer, 2020). The P_Sequence deposition model was used, with a low rigidity ( = k 10) applied, allowing for increased uncertainty ranges in the sections of core between dates (Ramsey, 2008;Ramsey & Lee, 2013

Results
In this section we present the results of both visible and cryptotephra study of the I-08 core, as well as our correlations to published proximal and distal glass shard geochemical data from sites in the Mediterranean region.

Visible tephra deposits
The I-08 core contains two visible tephra deposits. The lower tephra deposit, from 31.93 to 31.92 m depth, henceforth I08T_31.93, forms a clearly defined 1 cm layer in the stratigraphy with an abrupt contact at the base. The much larger upper tephra deposit 29.91 to 30.14 m depth, I08T_30.14, shows some fining towards the surface, however, the base is less clearly delineated as it falls at the end of a core section. Microscopic inspection of sediments within these lighter horizons revealed that they contained a high concentration of volcanic glass shards.
The major and minor element composition of I08T_31.93 (Table 1) corresponds in full to the proximal glass compositions of the Pantelleria Green Tuff (PGT; Civetta et al., 1984;Tomlinson et al., 2015), correlated to the Y-6 tephra marker identified in marine sequences from the Ionian Sea (Keller et al., 1978,  content . Whilst other widespread Pantellerian tephra markers have been identified in the Mediterranean, the Green Tuff is easily discriminated on the basis of FeO t concentrations (Hardiman, 2012). I08T_31.93 FeO t concentrations (Fig. S1B) closely match both proximal PGT  and distal Y-6 Tamburrino et al., 2012).
The Pantelleria Green Tuff was produced during a calderaforming eruption of the Island of Pantelleria in the Sicily channel, dated to ± 45.7 1.0 ka using the / Ar Ar  Table S7). We incorporate this best age for the Pantelleria Green Tuff eruption into the I-08 age-depth model at 31.93m (Fig. 2).
I08T_30.14 -Y-5/Campanian Ignimbrite The I08T_30.14 tephra unit (Fig. 2) appears ca. 18 cm thick, however, as 7 cm of material is missing between 30.21 and 30.14 m depth, due to the core extraction process, the true thickness of this tephra deposit is uncertain. As the underlying core section, with its surface at 30.21 m, contains a ca. 5 cm drop-stone of agglomerated ash it is likely that the missing section of the lake sequence was predominantly tephra. Overlying and underlying sediments are generally well compacted clays and silts, less likely to be lost during core extraction than disaggregated, coarse grained tephra. Therefore, it is possible that the thickness of the I08T_30.14 deposit is as much as 25 cm, indicative of an eruption that generated a significant amount of ash-fall and potential secondary thickening from re-deposition at the upper contact.
The I08T_30.14 tephra deposit primarily consists of colourless glass shards with varied morphologies, however, the majority of shards are either platy or fluted. Particle size distributions of tephra glass shards are unimodal, with some evidence of fining upwards through the deposit (Fig. S2).
I08T_30.14 glass compositions are consistent with glass compositions of the Campanian Ignimbrite (CI) generated from a caldera forming eruption of the Campi Flegrei volcano, Italy, ca. 40 ka BP and correlated to the widely-recognised Y-5 marine tephra in the Mediterranean tephrostratigraphy (Keller et al., 1978). The Campanian Ignimbrite eruption is one of the largest known Quaternary eruptions, with tephra deposits  correlated to the eruption found thousands of kilometres from source (Pyle et al., 2006). On the basis of the correlation of I08T_30.14 to the CI (Y-5) tephra isochron (Table S7), the widely used / Ar Ar 40 3 9 and C 14 age of the CI ± 39.85 0.14 ka BP (Giaccio et al., 2017) is imported at the upper bound of the Campanian Ignimbrite visible tephra deposit, at 30.04 m depth in the I-08 core chronology (Fig. 2).

Cryptotephra
Rangefinder (10 cm contiguous sampling) investigations revealed 11 intervals of increased volcanic glass concentration which warranted further 1 cm resolution study (Fig. 2). Sections of the core were selected for high resolution study by identifying significant increases in the tephra glass shard concentration profile.
The depths selected for high resolution study in the I-08 core were: Three further small peaks at ca. 26.3, 25.5 and 24.9m depth were also studied at 1 cm resolution, however, the results of these investigations are not included here as the shape of a shard concentration profiles and inconsistent morphologies of glass shards within these sections suggest post-depositional reworking rather than primary airfall of tephra. Due to a gap in the core between 32 and 33 m, associated with core extraction, the increase in glass shard concentration at ca. 33.1 m was not interrogated.
Pre-Pantelleria Green Tuff (>30.93 m depth; > ca. 46 ka BP) Tephra glass shards were found in all but seven of the 1 cm samples studied between 34.11 and 33.81 m depth, albeit at low concentrations (<1, 50<1500 shards − g 1 ). Throughout this interval, glass shards were generally light olive in colour and of varying relief, with both fluted and cuspate forms. Shards were generally small, <40 µm on the longest axis, however, at 34.03 m depth larger shards (ca. 80-100 µm) accounted for ca. 5% of the total shards counted. In the sample at 33.85 m depth, a few shards containing microcrysts were identified. The highest concentration of tephra glass shards in this section of the core, with ca. 1600 shards − g 1 , was found at 33.97 m depth.
Considering the distribution, appearance and size of the glass shards in each sample, five peaks in cryptotephra concentration were targeted for geochemical analysis (Fig. 4a) at 34.07, 34.03, 33.99, 33.97 and 33.85 m depth. WDS-EPMA analysis of shards in this section of core was challenging due to their small size, however, at least one analysis was possible for shards in three of samples, hereon I08T_34.02, I08T_33.97, and I08T_33.85.
There are three known widespread Mediterranean tephra isochrons of Pantellerian origin (Table S7), the ca. 46ka BP Pantelleria Green Tuff (Y-6; Scaillet et al., 2013), the ca. 77 ka BP P-10 (Paterne et al., 1988) and the older, ca. 131 ka BP, P-11 tephra marker (Paterne et al., 2008). The P-10 tephra marker identified by Paterne et al. (1988), subsequently identified in Lago Grande di Monticchio (TM-22; Wulf et al., 2004), Adriatic Sea core PRAD 1-2 (PRAD-2375; Bourne et al., 2015) and correlated to the ca. 85 ka BP Proximal Ignimbrite Z of Pantelleria , is typically associated with a NNW dispersal axis. Correlation to the P-10 tephra marker can be ruled out as these deposits are, at all sites, associated with higher Al O 2 3 and lower FeO t values ( Fig. 4) than the Ioannina tephra layers. The P-11 tephra marker has been identified in Greece, at Theopetra Cave (Karkanas et al., 2014) and, putatively, at Megali Limni (Lesvos; Margari et al., 2007;Vogel et al., 2010. Correlation of the ML-5 tephra layer at Megali Limni to the P-11 tephra marker is uncertain as, whilst geochemically similar, the age of the P-11 eruption is difficult to reconcile with its stratigraphic position within the Megali Limni pollen record. Both the P-11 and ML-5 tephra deposits can be ruled out as correlatives as the FeO t concentrations found in I08T_34.02, I08T_33.97 and I08T_33.85 fall considerably above the range of FeO t in glass shards from the P-11 and ML-5 deposits (Fig. 4).
The geochemical affinity of glass shards in layers I08T_34.02, I08T_33.97, and I08T_33.85 leaves open the possibility that these layers are stratigraphically displaced contamination from the visible I08T_31.93 tephra layer either through post-depositional reworking, or coring-related displacement. The absence of tephra glass shards from a number of samples, for example, between 33.96 and 33.86 m depth, as well as the absence of tephra from the blank samples run during the extraction process rules out laboratory contamination during cryptotephra extraction and analysis.
Alternatively, tephra layers I08T_34.02, I08T_33.97, and I08T_33.85 may be coring artefacts, given that the borehole from which cores were retrieved would still have contained tephra from the I08T_31.93 visible deposit which may have been retrieved in subsequent core drives. Similar stratigraphic displacement of tephra glass shards has been identified in the TP-2005 core from Tenaghi Philippon, with glass shards associated with the Campanian Ignimbrite identified 22.78 m below the visible CI deposit, attributed by Wulf et al. (2018) to coring-related contamination. Particularly, following retrieval of a core drive the load of the water column in the borehole is suggested to have forced glass shards to depths of 20 cm or more into the upper part of the subsequent drive (Wulf et al., 2018). In contrast to the peat-dominated Tenaghi Philippon sequence, the sediment matrix of the Ioannina I-08 core primarily consists of silts and clays, and thus the denser, more cohesive Ioannina cores should have reduced porewater, making the sediments less susceptible to such displacement. Furthermore, the glass shard distribution profiles (Fig. 4) do not support the coring-associated displacement hypothesis. If tephra glass shards were coring artefacts, it would be expected that from the top of the 70 cm core segment at 33.81 m depth water load should disperse shards throughout the stratigraphy. Instead, distinct peaks in glass shard concentration are observed at 21, 16 and 4 cm below the top of the core drive (I08T_34.02, I08T_33.97 and I08T_33.85 respectively).
The I08T_34.02, I08T_33.97 and I08T_33.85 cryptotephra peaks may represent smaller, previously unknown eruptions of Pantelleria. Much of our understanding of Pantellerian volcanism is based on the proximal stratigraphy and smaller eruptions from Pantelleria predating the large P-11 and Green Tuff (Y-6) deposits may not have been observed due to subsequent burial by more voluminous deposits. Whilst distal tephra study has been undertaken on visible units identified in marine sequences from the nearby Sicily Channel, very few detailed cryptotephra studies have been undertaken in the southern Mediterranean. Correlation of these tephra deposits to known volcanic events or tephra isochrons identified in other sequences is not possible at present. Therefore, we suggest that tephra layers between 34.11 and 33.81 m depth in the I-08 core to genuine but previously unrecognised eruptions of Pantelleria, however, confirmation is needed by replication at another site.
Post-Pantelleria Green Tuff and Pre-Campanian Ignimbrite (31.93-30.14 m depth; ca. 46-40 ka BP) Two intervals of increased glass shard concentration were identified for further study in the section of core bracketed by the visible I-08 tephra layers, I08T_31.93 and I08T_30.14, which are correlated to the ca. 46 ka Pantelleria Green Tuff and 40 ka BP Campanian Ignimbrite respectively. Three depths within which glass shard concentrations peaked well above background, were identified and selected for geochemical analysis: I08T_31.19, I08T_31.17 and I08T_30.74 (Fig. 5a) associated with distal deposits from Pantelleria (Fig. 5b). Within this grouping there is some variation in Na O 2 with values between 3.2 and 6.8 wt.%, however, values for FeO t and K O 2 are well clustered (Fig. 5c-e). Whilst evidence exists for post-PGT eruptive activity at Pantelleria, a period of quiescence has been suggested following the PGT, ending at 35 ka BP with the eruption of the Serra di Ghirlanda tephra (Civetta et al., 1984). In contrast, Mahood and Hildreth (1986) suggest that volcanism at Pantelleria resumes rapidly after the PGT eruption, identifying trachyte lavas from a vent at Monte Gibele which they date using K/Ar to ca. 41-27 ka BP, which would indicate that some eruptive activity occurred between the PGT and CI eruptions. However, no distal studies have yet identified tephra deposits associated with post-PGT Pantellerian volcanism that could provide a correlative for I08T_31.19.
Alternatively, I08T_31.19 may be the result of remobilisation of tephra glass shards associated with the I08T_31.93 tephra marker some 70 cm deeper in the core. Mechanisms that may have caused post-depositional reworking in the Ioannina cores will be discussed in greater detail in 'Discussion' section.
Multiple pre-CI eruptions have been recorded in both proximal and distal archives, including marine cores from the Adriatic (S7; Tomlinson et al., 2012;Bourne et al., 2010;Matthews et al., 2015). Most notably, the tephra record from Lago Grande di Monticchio identifies at least four smallerscale tephra markers associated with pre-CI volcanism from Campi Flegrei which occur ca. 600 varve years prior to the CI (Wulf et al., 2004Wutke et al., 2015). The pre-CI LGdM tephra markers, termed the TM-18 tephras, are geochemically difficult to distinguish from the CI without high-precision trace element glass data (Wutke et al., 2015). Of these markers Wulf et al. (2018) identify medial distal tephras TM-18-1d, TM-18-4, TM-18-9e as the most likely correlatives for pre-CI Campi Flegrei tephra horizons identified in distal settings. Tephra deposits associated with the TM-18 pre-CI sequence have been identified in a limited number of distal archives. TM-18-1d, for example, has been identified in the Tenaghi Philippon sequence (Wulf et al., 2018). The geochemical fingerprint of the Ioannina tephra deposits can, however, be separated from many of these pre-CI deposits on the basis of As with the ca. 33 m depth tephra markers, it is also possible that I08T_31.17 and I08T_30.74 are related to downcore displacement of tephra shards associated with the coring process. All the tephra deposits identified fall within the major element composition of the I08T_30.14 tephra deposit (Fig. S4) High resolution (1 cm) cryptotephra investigations were undertaken between 29.88 and 29.51 m depth and 29.01-28.51 m depth. The section of core between 29.51 and 29.01 m depth, where glass shard concentrations fell below 100 shards − g 1 in rangefinder investigations, was not studied further (Fig. 2). The hiatus in tephra deposition between 29.51 and 29.01 m, observed via cryptotephra analyses, is verified by the XRF scanning data. XRF-derived elemental ratios for bulk sediment, most notably the K/Ti ratio, closely tracks tephra glass shard concentration throughout this section of the core (Fig. 6b).
Three abrupt increases in glass shard concentration relative to the underlying sample were observed at 29.72, 29.65 and 29.62 m depth, which is above the I08T_30.14 CI tephra marker but below the hiatus in shard deposition at 29.51 m depth (Fig. 6a). Whilst peaks in glass shard concentration at 29.65 and 29.62 m depth are lower than peaks at 29.83 and 29.72 m depth, tephra glass shard concentrations at 29.65 and 29.62 m depth (ca. × 5 10 5 and × 11 10 5 shards − g 1 respectively) represent large increases relative to the underlying samples (Fig. 6a). Glass shards at all depths generally measured between 60 and 100 µm on the longest axis. At 29.72 m depth, however, shards were much larger, typically ca. 100 µm on the longest axis (Fig. 6c).
A fourth sample at 29.83 m depth, was investigated as a potential primary ash-fall horizon on the basis of shard morphology as, similarly to the sample from 29.72 m depth, shards in this sample were far larger than in surrounding tephra samples, typically >100 µm on the longest axis (Fig. 6c). Glass shards throughout this section of core were generally colourless, of low relief, and largely consisted of platy forms, however, shards with a range of other morphologies, including fluted and cuspate shards, were also present.
A distinct increase in tephra glass shard concentration is observed at 28.98 m depth and is also identifiable through a peak in the K/Ti XRF ratio as well as a coarsening in the particle size data. Above 28.98 m there is a return to glass shard concentrations ca. × 1 10 5 shards − g 1 until 28.83 m depth, at which point concentrations begin to increase, albeit with a number of sharp jumps and oscillations in glass shard concentration. The increase in tephra glass shard concentration throughout this section of the core correlates with an increase in K/Ti ratio (Fig. 6b), reflected in the particle size data as an increase in the frequency of particles in the 100-200 µm fraction (Fig. 6c).
Above the peak at 28.98 m depth, a further six samples were investigated since they show a large increase in glass shard concentration relative to their underlying sample (Fig. 6a). Observed glass shard morphologies are broadly consistent in samples between 29.01 and 28.51 m depth, with shards largely low relief, colourless, and dominated by platy, curvilinear, and cuspate forms. A change in glass shard morphology is observed in at 28.66 m depth, where shards display an increase in ventricular features, microcryst inclusions with a limited number of shards (n = 2) with pumicious forms. Due to a return to platy, low relief shards in samples above 28. depth, this influx of new morphologies may reflect primary airfall, henceforth the I08T_28.66 tephra deposit. Most shards observed in the 29.88-28.51 m interval were between 30 and 50 µm on their longest axis, however, occasional larger shards (up to 200 µm on the longest axis) were noted throughout the section. Geochemical analysis of the 11 potential primary airfall tephra layers between 29.99 and 28.51 m depth characterises glass shards as phono-trachytic, with limited variability in SiO 2 (between 60 and 63 wt. %). CaO (1.1-2.8 wt. %) and MgO (0.2-0.9 wt.%) values are consistent (Fig. 6b) Glass shards from all 11 depths are phono-trachytic (Fig. 6a). Post-CI (<40 ka BP) Mediterranean tephra deposits with phono-trachytic glass compositions may correlate to eruptions from the Campanian Volcanic Zone (trachytic and phonolitic; Tomlinson et al., 2015), Gölcük (trachytic and phonolitic; Tomlinson et al., 2015) and Mt. Etna (trachytic; Albert et al., 2013). The MgO and CaO concentrations of all 11 Ioannina tephra deposits between 29.99 and 28.51 m depth indicate that they are all associated with eruptive activity from Campi Flegrei (Fig. 6e).
Two possible explanations for the unusual stratigraphic distribution of tephra glass shards through this section of the core are suggested. The first is that the peaks, which are often characterised by abrupt increases in glass shard concentration relative to the underlying deposits, are primary air-fall deposits associated with post-CI eruptions in the Mediterranean. The second is that these peaks represent a complex process of postdepositional reworking of the Campanian Ignimbrite tephra deposit in the Ioannina basin. It should be noted that the two explanations presented here are not mutually exclusive and it is possible that both reworking and the complex eruptive history of Campi Flegrei act in concert to produce the complex tephra record in this section of the I-08 core. Both possible interpretations of the I-08 tephra record are further discussed in 'Discussion' section.
Last Glacial to Interglacial Transition (16.20-14.20 m depth; ca. 25-7 ka BP) Three discrete peaks in cryptotephra glass shard concentration (I08T_16.07, I08T_15.23 and I08T_14.39) are located between 16.20 and 14.20 m depth in the I-08 core. This section overlies the coarse sand deposits associated with low lake levels during the last full glacial (MIS2, ca. 22 ka BP) and shown by Jones et al. (2013) to cover the Last Glacial-Interglacial Transition (LGIT). In contrast to tephra peaks between 29.88 and 28.51 m depth peaks I08T_16.07, I08T_15.23 and I08T_14.39 are isolated, with no tephra glass shards found in the overlying and underlying sediments. We can rule out coring-related displacement for I08T_16.07 and I08T_14.39 as these isochrons are 17 and 59 cm from the top of their core sections respectively. We are also confident that the isolated peak at I08T_15.23 does not represent coring-related stratigraphic displacement as, whilst glass shards are identified in the 1 cm samples above and below the core, no shards are present in the samples which bound these, suggesting cryptotephra glass shards form a distinct layer within the stratigraphy. Therefore, we are confident that these are independent isochrons.
Shards from I08T_16.07, I08T_15.23 and I08T_14.39 are highly evolved rhyolites (Fig. 7b)  and Acigöl (Tryon et al., 2009). Reference compositional data from these volcanic centres are compared to the Ioannina LGIT tephra compositions in Fig. 7. The Ioannina LGIT tephra correlate well to high-K calc-alkaline activity at Aeolian Island volcano Lipari. Multiple eruption phases have been identified in the proximal Lipari record duing the last glacial cycle, some of which are indistinguishable based on their major element geochemistry (Albert et al., 2017). However, the Vallone del Gabellotto eruptive cycle (sensu. Albert et al., 2017), which is dated to ca. 8 ka BP (Caron et al., 2012;Siani et al., 2004), provides the only likely LGIT correlative for the Ioannina tephra markers (Fig. 7). Of particular note is the widespread Vallone del Gabellotto (VG/E-1) tephra dated to 8630-8430 cal years BP using the IntCal20 calibration curve (Fig. S7). The VG/E-1 tephra has been correlated to horizons present in Tyrrhenian, Ionian, and Adriatic marine cores (Albert et al., 2017) as well as Tenaghi Philippon in NE Greece (Wulf et al., 2018). Of the three Ioannina LGIT tephra deposits the most likely correlative for this widespread Holocene tephra marker is the uppermost I-08 tephra layer, I08T_14.39, the only tephra deposit that sits within the Holocene section of the core (Fig. 7b). The correlation of I08T_14.39 with the Vallone del Gabellotto (E-1) isochron is further supported by new trace element analysis, Supplementary Information 1. Fig. 8 shows good correspondence between I08T_14.39, three proximal deposits of the Vallone del Gabellotto, and its distal correlative M25/4-12-28cm tephra from the Ionian Sea (Albert et al., 2017).
The proximal Lipari record does contain eruptive deposits of LGIT age, however, these are 'localised' and not traced widely across the island (Albert et al., 2017). In the distal realm an older layer with a Gabellotto-like Lipari composition has been recorded at 44 cm depth in the Ionian Sea core M25/4-12 (Albert et al., 2013). The lowermost Ioannina LGIT tephra layer, I08T_16.07, has a trace element geochemistry that corresponds well to the distal Ionian Sea tephra marker M25/4-12-44 cm, which has yet to be linked to a proximal deposit. Therefore, we propose a tentative correlation of I08T_16.07 to Ionian Sea tephra marker M25/4-12-44 cm, which does not yet have a correlative in the proximal tephrostratigraphy. The identification of this tephra marker ca. 500 km from source is currently the most distal deposit and the first outside of Italy and the surrounding seas.
The intermediate Ioannina LGIT tephra,I08T_15.23, has yet to be correlated to a proximal or distal isochron, however, has distinctly lower Th, La, Eu and Ce concentrations (Fig. 8a) than the other tephra layers and is thus likely to be the product of a separate eruption. Therefore, I08T_15.23 has the potential to provide an additional stratigraphic marker in the future.
On the basis of the correlation of I08T_14.39 to the E-1 tephra isochron, the uncalibrated C 14 age of this marker, ± 7.77 0.04ka BP (Caron et al., 2012;Albert et al., 2017), is imported to this depth in the I-08 core chronology and recalibrated as part of the OxCal P_Sequence.

Discussion
The Campanian Ignimbrite tephra marker The I08T_30.14 tephra deposit, which we correlate to the 39.85 ka BP campanian Ignimbrite eruption of Campi Flegrei caldera, Italy, is interesting as it contains the full major and minor element variation observed in the CI proximal deposit ( Fig. S3; Table S7). Two closely-related compositional groups can be recognised within the I08T_30.14 tephra deposit, illustrated by differences in CaO, MgO and K O 2 at overlapping SiO 2 concentrations (Fig. S3). The two composition groups in I08T_30.14 closely correspond to the end member associated with the lower and intermediate fall deposits of the CI eruption in the proximal deposits (CaO <2; , as well as as the higher CaO and K O 2 and lower Na O 2 end member associated with the upper flow . Of the distal occurrences of the CI/Y-5 tephra this upper flow end member is only identified in Megali Limni, Tenaghi Philippon, and Kalodiki (Greece; Pyle et al., 2006;Margari et al., 2007;Wulf et al., 2018) and the archaeological site of Crvena Stijena (Montenegro; Morley & Woodward, 2011). Elsewhere tephra deposits associated with the CI eruption, for example, in PRAD 1-2 (Adriatic Sea; Bourne et al., 2010) and Lake Ohrid (Albania/Montenegro; Vogel et al., 2010), have chemical compositions which more closely match the lower and intermediate fall deposits. The offset in glass geochemical composition of the CI/Y-5 tephra layer between these different sites may arise from a variety of factors, including the height of the volcanic plume at different stages of the eruption, the prevailing wind direction, and the varying atmospheric transport and deposition of tephra glass shards due to differences in the eruption dynamics (e.g. injection height, wind speed, volume) and physical properties between glass shards generated in different phases of the eruption.
With regards to the thickness of the Ioannina I08T_30.14 (CI/Y-5) deposit, estimates from the Costa et al. (2012) ash fall out model for the Campanian Ignimbrite suggest that a ca. 10 cm deposit is expected at the site. That the deposit in the I-08 sequence, at even the lowest estimate for layer thickness, exceeds the model estimate is likely related to site-specific factors. The increased thickness of the I-08_30.14 layer is not without precedent. In their work mapping the thickness of the CI/Y-5 tephra marker Engwell et al. (2014) argue that in distal subaerial environments deposits are typically thicker than for equivalent deep sea deposits. Primarily, the large scale of the Ioannina catchment means large volumes of tephra would have been deposited into the lake environment through subsequent in-wash processes. Post-depositional reworking of unconsolidated tephra deposits from the surrounding landscape, facilitated by the steep topography, seems likely in the Ioannina basin and may have led to the redeposition of sediments in the lake.

Interpreting the post-CI tephra record
Here we consider two interpretations of the tephra distributions outlined in 'Post-Campanian Ignimbrite (29.88-28.51 m depth; ca. 40-30 ka BP)' section, in sediments from 29.88 to 28.51 m depth which postdate the 39.85 ka Campanian Ignimbrite eruption. First we consider the evidence that peaks in tephra glass shards represent primary air-fall of post-CI eruptive activity at Campi Flegrei. Second, we argue that they may represent reworking of tephra within the Ioannina catchment as well as within the lake basin itself.

Primary air-fall
Multiple post-CI last glacial tephra layers have been identified in the proximal Campi Flegrei stratigraphy, often termed the 'Tufi Biancastri' sequence (Table S7). Discriminating these eruptions in distal archives is challenging, however, given their close, often overlapping, glass shard geochemical composi-tions (e.g. Paterne et al., 1988). Within the caldera, Tufi Biancastri tephra deposits have been identified at Verdolino (the ca. 28 ka BP VRb and ca. 30 ka BP VRa; Pappalardo et al., 1999), and Trefola (the <38 ka BP TLo; Tomlinson et al., 2012). At Ponti Rossi, to the northeast of the caldera, the Masseria del Monte Tuff (MdM), associated with a large VEI 6 eruption ca. 30 ka BP, has recently been correlated with the widespread Y-3 marine tephra (Albert et al., 2019). The MdM/ Y-3 tephra marker has been identified at a range of sites throughout the Mediterranean, including at Tenaghi Philippon (Wulf et al., 2018), to the east of the Ioannina site.
Additionally, a number of extra-caldera outcrops linked to post-CI activity at Campi Flegrei have also been identified; most notably the SMP1-e and CE-1 deposits (Sulpizio et al., 2003;Albert et al., 2015).
In the distal realm, a range of post-CI tephra markers associated with eruptions of Campi Flegrei have been described that have not been correlated to an eruption identified within the proximal stratigraphy (Table S7). At the nearby Lago Grande di Monticchio the TM-17-2 tephra, associated with post-CI volcanism at Campi Flegrei, is dated at ± 35530 1780 varve years BP. Further afield, sediment cores from the Adriatic Sea Matthews et al. (2015) highlight a layer consisting of both trachytic and rhyolitic shards, termed SA03-11-T1535 (Table S7). The trachytic component of SA03-11-T1535 is chemically trachytic component is chemically identical to the Campanian Ignimbrite, but the stratigraphic position and association to the rhyolitic component of the marker lead  to argue a link to the CI eruption is implausible. The SA03-11-T1535 tephra marker correlates with a post-CI tephra marker reported in the Ionian Sea and the replication of the bimodal rhyolitic and trachytic chemical signature indicate it to be a distinctive layer from a single eruption event. The absence of any rhyolitic component in the post-CI cryptotephra peaks observed in the I-08 record mean that we can not identify a correlative with SA03-11-T1535.
The I08T_30.14 visible tephra marker has a broad geochemical composition, containing not only the low CaO, MgO and FeO t and high Na O 2 end members of the CI fall and the lower and intermediate flow units, but also the raised K O 2 , lower Na O 2 end member associated with the upper flow in the proximal stratigraphy (Table S7; Tomlinson et al., 2012). All of the analysed tephra deposits which postdate the I08T_30.14 tephra marker have major and minor element geochemical compositions which fall within the envelope of the I08T_30.14 ( Fig. S5) and, therefore, may represent reworked material from this deposit. Perhaps the most diagnostic component of the Campanian Ignimbrite tephra marker, which helps to discriminate it from other post-CI eruptions, is the upper flow end member . Thus, I-08 tephra layers which contain glass shards with upper flow-like compositions are likely to be reworked CI from the catchment (Fig. S5).
As a result of the broad geochemical composition of the I08T_30.14 tephra marker, identifying subsequent eruptions is challenging. For example, second to the CI, the most widespread last glacial tephra marker from Campi Flegrei is the ca. 30 ka BP Y-3 stratigraphic marker correlated by Albert et al. (2019) to the proximal Masseria del Monte Tuff. Across the full range of major and minor elements the geochemical fingerprint of the Y-3 tephra marker falls within the envelope of the I08T_30.14 ( Fig. S6). It is possible to differentiate the markers, however, through FeO t wt.% and the bimodal K O 2 distribution associated with the MdM/Y-3 deposit (Fig. S6).
The geochemical composition of the I08T_28.98 tephra horizon does not match the MdM/Y-3 across the full major element geochemical assemblage of the Y-3 (Fig. S6) yet nor does the I08T_28.98 perfectly match the geochemistry of I08T_30.14. Given the difference in composition between the I08T_28.98 cryptotephra layer and the I08T_30.14 (CI/Y-5) deposit, the I08T_28.98 is suggested to represent a post-CI primary air-fall deposit, although correlation is not possible at this stage. Future trace element analysis may provide a means of discriminating primary tephra inputs associated with post-CI volcanism from reworked material associated with the Campanian Ignimbrite and I08T_30.14 tephra markers.

Reworking
The presence of glass shards with geochemical compositions unique to the CI is the primary indicator that peaks in glass shard concentration overlying the I08T_30.14 (CI/Y-5) are most likely the products of the remobilisation of this tephra marker (Fig. S5). Specifically, glass shard compositions that match the products of the lower and intermediate flow (sensu. Tomlinson et al., 2012), characterised by low CaO, MgO, and FeO t and high Na O 2 concentrations are unique identifiers of the reworked material.
The profile of glass shard concentrations against depth further suggests reworking, particularly in the upper layers (<28.8 m depth). In low energy lake sediment sequences primary air-fall tephra inputs are typically associated with a distinct peak followed by a decrease in the overlying sediments (Davies, 2015). The gradual increase in glass shard concentration between 28.97 and 28.66 m depth in the I-08 core does not suggest a primary input and could instead be interpreted as a gradual change in the sedimentary regime in the catchment.
A particular feature of this section of the core is the presence of shell layers absent from the underlying sediments. The presence of intervals within the Ioannina sediment sequence that contain molloscan faunal remains are argued to reflect lake-level variation (Frogley et al., 2009). Molluscan shells are preserved in the sediment above ca. 29.9 m, suggesting that water levels were decreasing at the time of deposition. Tephra material at the lake margins would have been subaerially exposed following lake level lowering, then remobilised and deposited further into the lake basin, including to the I-08 core site. Such a process is consistent with the trend of gradual increase in tephra glass shard concentrations between 29.88 and 28.51 m depth (Fig. 9).
Alternatively, the recurrent nature of tephra remobilisation suggested by the stratigraphy of glass shard distributions (Fig. 6a), which shows pulses of increased glass shard concentration, may reflect surface run-off from a continuously eroding catchment. Lake Ioannina is primarily groundwaterfed and located in a basin largely bounded by steep slopes; thus it seems unlikely that processes of alluvial erosion and deposition are responsible for the recurrent inputs of reworked tephra.
A final mechanism which may explain the remobilisation of tephra deposits associated with the Campanian Ignimbrite eruption in the I-08 core may be that the redeposited tephra layers reflect a regional, extra-basin, erosional signal. Reworking of the Campanian Ignimbrite tephra layer has been identified in both the Tenaghi Phillipon (Wulf et al., 2018) and Kopais (Hardiman, 2012) Greek cryptotephra records. Remobilisation of tephra deposits is becoming increasingly recognised as an important stage in the taphonomy of volcanic ash (Dominguez et al., 2020;Buckland et al., 2020) and the Campanian Ignimbrite deposit, widespread and deposited contemporaneously to the dry stadial conditions associated with Heinrich stadial 4, would likely have been frequently eroded and remobilised from the land surface. Indeed, tephra associated with the Campanian Ignimbrite eruption is found in aeolian sequences as far afield as Ukraine (Melekestsev et al., 1984) and Romania where deposits of the tephra >1 m thick have been identified (Fitzsimmons et al., 2013). It is therefore possible that aeolian remobilisation of tephra material may also be reflected in the Ioannina record, however, as noted, little evidence exists of geochemical alteration of tephra which would be expected if tephra has been remobilised from exposed environments.
To summarise, the consistent geochemical signal of glass shards deposited within the I-08 core between 29.88 and 28.51 m depth suggests erosion and redeposition of Campanian Ignimbrite tephra into the lake from the wider Ioannina catchment and the lake basin edges. Remobilisation appears to take place in two phases representing at least two separate processes. The shape of the shard concentration profile, which gradually decreases, suggests that immediately above the visible I08T_30.14 (CI) deposit, from 29.92 m through to 29.70 m depth tephra input is related to erosion of primary tephra fall from the Campanian Ignimbrite eruption within the Ioannina catchment. Subsequently, a hiatus in tephra deposition is interpreted as reflecting landscape stabilisation, with tephra deposits are no longer exposed to processes of surface run-off. The hiatus is clear in both the XRF data as well as the tephra glass shard concentration (Fig. 6a-c). We suggest the I08T_28.98 tephra marker, which is constrained to a 2 cm depth interval, is a primary air-fall deposit associated with Tufi Biancastri volcanism at Campi Flegrei caldera, however, we cannot yet correlate this tephra deposit to a specific eruption. From ca. 28 m depth upwards, there is a change in the sedimentary regime at the I-08 core site, which we interpret as a reduction in lake level resulting in the remobilisation of tephra material from the exposed lake margins. Extra-basin inputs from either aeolian remobilisation of exposed tephra surfaces or subsequent primary air-fall tephra inputs cannot be ruled out, however, given hiatuses in deposition of tephra glass shards, are unlikely to act as the dominant processes driving tephra deposition.

Age-depth model
The identification of the Pantelleria Green Tuff (Y-6) and Campanian Ignimbrite (Y-5) tephra markers within I-08 provides an opportunity to extend the chronology of Jones et al. (2013) back to ca. 45 ka BP. Furthermore, the identification of the Vallone del Gabellotto (E-1) provides an opportunity to refine the Holocene chronology of the I-08 core. Here we present a revised I-08 age-depth model which incorporates the three new tephra ages (Table 2; Fig. 10). The new full corelength age model incorporates earlier radiocarbon dates from macrocharcoal samples and compound-specific radiocarbon analysis (CSRA) for the upper section of the core (Jones et al., 2013). The uncalibrated radiocarbon age estimate for the E-1 marker (Caron et al., 2012) is recalibrated using IntCal20 as part of the OxCal model. Visible tephra deposits are most likely deposited over the course of days and weeks as opposed to years, therefore the 18 cm thick I08T_30.14 deposit is treated as an event horizon in the age model. The ± 39.85 0.14ka age for the CI eruption is input at the lower (30.14 m) and upper (29.96 m) bounds of the I08T_30.14 tephra.
The new age model provides age estimates for the uncorrelated Aeolian Island tephra markers, I08T_16.07 and I08T_15.23, dated here to 12.46-10.21 ka BP and 10.65-8.52 ka BP, respectively, Fig. 10b,c. The ca. 11 ka BP age for I08T_16.07 further supports the proposed correlation of that layer to the Ionian Sea tephra marker M25/4-12-44cm Ionian Sea tephra marker, which occurs in sediments of Late Glacial age within the M25/4-12 core oxygen isotope stratigraphy (Negri et al., 1999;Albert et al., 2017).

Regional tephra correlations
Three widespread tephra horizons have been identified in the Ioannina sequence: the Pantelleria Green Tuff (PGT/Y-6), the Campanian Ignimbrite (CI/Y-5), and the Vallone del Gabellotto (VG/E-1), facilitating direct correlation of the Ioannina I-08 sequence with regional palaeoecological and palaeoclimatic reconstructions from different environmental contexts (Fig. 11).
The ca. 40 ka BP CI/Y-5 tephra layer in the Ioannina record permits correlation to key southern European terrestrial palaeoenvironmental records, such as Lago Grande di Monticchio, Italy (Allen et al., 1999;Wulf et al., 2004), Lake Ohrid, Albania/ Montenegro Leicher et al., 2016) and Tenaghi Philippon, Greece (Müller et al., 2011;Wulf et al., 2018), as well as central and eastern Mediterranean marine proxy records including cores from the Ionian (Keller et al., 1978;Albert et al., 2012), Tyrrhenian (Paterne et al., 1988), Adriatic (Bourne et al., 2010), Aegean (Keller et al., 1978;Satow et al., 2015) and Black Seas . The CI/Y-5 marker can be used Scaillet et al. . High resolution analysis of the I-08 vegetation record in sediments surrounding the I08T_30.14 tephra marker would allow the pattern and timing of the ecosystem response to HS4 at the site to be precisely evaluated and compared to other sites in the eastern Mediterranean. The ca. 46 ka BP PGT/Y-6 tephra marker facilitates direct correlation to terrestrial records with palaeoenvironmental sequences including Lake Ohrid, and to marine cores from the Sicily channel (Tamburrino et al., 2012) and Ionian Sea (Keller et al., 1978). Together the PGT/Y-6 and CI/Y-5 tephra layers bracket an interval of ca. 6 ka in the Ioannina, Lake Ohrid and Ionian Sea records associated with four Dansgaard-Oeschger events observed in the Greenland ice-core records (Rasmussen et al., 2014). Comparison of palaeoenvironmental proxies within these archives could allow both local to regional and proxy-specific responses to these climate oscillations to be explored with high precision.
The presence of the Campanian Ignimbrite and Pantelleria Green Tuff tephra layers also provides an exciting opportunity to link the valuable palaeoenvironmental record contained within the Ioannina sequence to regional archaeological sites. For example, CI/Y-5 deposits have been identified in the Klissoura and Franchthi sequences (Lowe et al., 2012), and the PGT/Y-6 marker has been identified at Theopetra (Karkanas et al., 2014). These tephra linkages establish a chronological framework which, going forward, we hope will facilitate more thorough investigation of the environmental niches occupied by early Europeans in the region following their expansion out of Africa.
The identification of three cryptotephra layers correlated to Lipari between ca. 16 and 14 m depth in the I-08 sequence provides the first securely dated evidence that Lipari erupted at least three times during the LGIT and Early Holocene. The Vallone del Gabellotto (E-1) isochron connects the Ioannina sequence to archives in the Ionian, Adriatic and Tyrrhenian seas (Albert et al., 2017), and the terrestrial Tenaghi Philippon peat sequence (Wulf et al., 2018). Preliminary dates for the two earlier Lipari cryptotephra layers suggests two more tie points may exist between Ioannina and the Ionian Sea, however, further geochemical data is needed to confirm these findings.
Challenges in cryptotephra analysis of the I-08 core The frequency of tephra dispersal from explosive eruptions in the Mediterranean region during the last glacial period and the geographical location of Lake Ioannina down-wind from the productive Italian Arc volcanoes, suggested that many cryptotephra layers would be found in the I-08 core sediments. However, cryptotephra analyses did not identify horizons of primary ash fall in the section of the I-08 core between 29 and 18 m.
The identification of discrete tephra layers from Lipari in the Late Glacial and Early Holocene sediments at Ioannina highlight the suitability of the Lake Ioannina sequence for capturing cryptotephra layers. However, during the last glacial when high concentrations of tephra were deposited in the Ioannina catchment from the voluminous Campanian Ignimbrite eruption, the swamping effect of the reworked CI/Y-5 tephra through processes of intra-basin focussing and redistribution of tephra glass shards made the detection of cryptotephra from primary air-fall impossible. Thus, even if primary air-fall tephra layers were present in the I-08 core sequence as cryptotephra horizons they would be difficult to identify against the background of reworked tephra. The challenges associated with identifying primary airfall from closely-spaced eruptions of similar geochemical composition is not a new one in tephrostratigraphic research (e.g. Pyne-O'Donnell et al., 2008). In an attempt to overcome these challenges glass shard colour, grain size, and morphology were carefully considered during microscopic analysis of samples with the aim of detecting any changes in glass shard characteristics that might indicate input of tephra from different sources (McLean et al., 2018). However, as has been Figure 11. Simplified summary diagram of key tephra horizons within the Mediterranean tephrostratigraphic framework for the past 50 ka. Archives included here are: Tyrrhenian Sea (Paterne et al., 1988;Albert et al., 2012), Lago Grande di Monticchio (Wulf et al., 2004, Adriatic Sea (Bourne et al., 2010;Matthews et al., 2015), Ionian Sea (Keller et al., 1978;Albert et al., 2012;Insinga et al., 2014), Lake Ohrid (Wagner et al., 2008;Vogel et al., 2010;Sulpizio et al., 2010;Leicher et al., 2016), Lake Ioannina (this study), Tenaghi Philippon (Wulf et al., 2018), the Aegean Sea (Satow et al., 2015), and Megali Limni (Margari et al., 2007). Eruption ages from Bronk Ramsey et al. (2015) other than Agnano Mt Spina (Smith et al., 2011), Vallone del Gabellotto (E-1) (Albert et al., 2019), Mercato (Santacroce et al., 2008) and Pantelleria Green Tuff (Y-6) . Site locations are shown in Fig. 1. [Color figure can be viewed at wileyonlinelibrary.com] noted from many distal tephra archives, silicic tephra from Mediterranean sources have quite similar morphologies, so few eruptions are distinguishable on the basis of morphology alone (Bourne et al., 2010;Matthews et al., 2015). Thus, it is possible that cryptotephra layers were not found during this study because the shards from different eruptions in the Mediterranean tend to be of similar morphology and colour.
Unexpectedly, this study has demonstrated that the identification of reworked tephra horizons can, in supplement to other proxy data such as grain size and geochemistry, provide an insight into changing processes of erosion, sediment inwash, and reworking within a lake catchment. The value of detailed sedimentary analyses to understand tephra taphonomy has been demonstrated in this study through the combination of grain size analysis, XRF data, and tephra shard counting. Complex processes of sediment remobilisation and reworking were identified, raising questions about the integrity of the sediment record between 29 and 28.5 m depth as a primary, undisturbed, deposit. The presence of noncontemporaneous tephra within younger sediments suggests in-wash and redeposition of older material from the edge of the lake further into the lake. Determining the age offset between the age of the redeposited material and contemporaneous sedimentation would be a challenge, further highlighting the importance of sediment taphonomy in Quaternary reconstructions.
The absence of tephra evidence for a number of widespread Quaternary eruptions in the I-08 sequence may be due to preferential deposition of tephra deposits near inflows and internal redistribution through currents (Pyne-O'Donnell et al., 2008). For example, internal processes can result in tephra deposits being reworked to below detection levels in some areas of the lake, however, such processes are complex and dynamic (Dugmore & Newton, 2012;Watson et al., 2016). Spatial variation in intra-basin focussing of tephra over time may explain why cryptotephra are present within the sequence over the Late Glacial and Early Holocene but not during other periods.

Methodological recommendations
Syn-depositional and post-depositional tephra transportation have been demonstrated in a range of studies undertaken in lake environments, often revealing large intra-basin variability in visible and/or crypto-tephra layer thickness (Mangerud et al., 1984). In the case of the Ioannina record, where tephrostratigraphic analysis was carried out only on a single core containing evidence of reworked sediment, the record of tephra deposited at lake site may not have been fully captured (Watson et al., 2016). Thus, replication of this study on future cores from the site may provide a more complete picture of tephra deposition at Ioannina.
The omission of key regional tephra markers (e.g. the Y-3) may be a result of small gaps between the core segments, an artefact of the coring methodology applied. Gaps in the stratigraphy mean it is possible that key isochrons may be missing (Lowe, 2011). Furthermore, gaps in the core complicate the identification of the primary ash fall layer and may results in a reworked layer being identified as a primary air-fall deposit, which can result in age attribution of an eruption age to an incorrect depth. Detailed stratigraphic analysis, facilitating the identification of horizons where reworking is likely, can help in developing a robust tephrochronology. We, therefore, highlight the importance of taking overlapping, parallel cores wherever possible, which minimise gaps in the stratigraphy.

Conclusions
The development of an extended independent chronology for the I-08 core is a step forward in the dating of an important palaeoenvironmental record. The integration of the Ioannina site into the Mediterranean tephra framework provides direct correlations with a range of Quaternary sequences in the Mediterranean region. The potential of tephra studies at Ioannina is by no means limited to the last glacial cycle, and tephrochronologies provide opportunities for dating and correlation of records over multiple glacial cycles. Importantly, tephra studies may act as independent checks of earlier, tuned chronologies for Ioannina sequences.
The chronology developed here also opens up opportunities for future work on the I-08 core, particularly the potential for increased resolution, sub-millennial scale study of the proxy record contained within the sediments. Pollen analysis of Ioannina core I-284 has revealed marked vegetation responses to millennial-scale climatic oscillations (Tzedakis et al., 2004). The three secure isochrons, the last glacial Campanian Ignimbrite (CI/Y-5; ca. 39.8 ka BP) and Pantelleria Green Tuff (PGT/Y-6; ca. 45.7 ka BP) and the Holocene Vallone del Gabellotto cryptotephra marker (E-1; ca. 8.3 ka BP) identified in the Ioannina sequence provide opportunities for direct correlation of this pollen record with palaeoenvironmental, palaeoclimate, and archaeological sequences throughout the eastern Mediterranean. In particular, the CI/Y-5 marker provides an opportunity to undertake detailed inter-site comparisons of the timing of ecosystem responses to Heinrich stadial 4 (40.1-28.3 ka BP sensu. Sanchez Goñi & Harrison, 2010).
We have highlighted a significant challenge in Mediterranean tephra studies and, indeed, tephra studies globally, which is the identification of primary airfall cryptotephra layers against a background of sediment (and therefore tephra) reworking, particularly in areas where multiple eruptions produce huge volumes of glass shards with similar morphologies and geochemical signatures. In this work, whilst it is possible that the record contains distal deposits from multiple eruptions from Campi Flegrei, particularly those following the Campanian Ignimbrite eruption, it has not been possible to produce reliable correlations through geochemical analysis. The main barrier preventing geochemical attribution of these deposits to subsequent eruptions is the broad geochemical envelope of the large Campanian Ignimbrite eruption which has multiple phases , many of which are identified at Ioannina. Refining and developing techniques which allow for the further discrimination of these different eruptions will be a crucial step in developing and expanding the Mediterranean tephra chronostratigraphic framework.

Supporting information
Additional supporting information can be found in the online version of this article. Figure S1. Major element geochemistry for tephra glass shards from visible tephra layer I08T_31.93, showing: (a) total alkali vs. silica plot (Le Bas et al., 1986) and (b) FeOt vs AlO3, plotted alongside geochemical data from proximal Pantelleria Green Tuff deposits and distal occurrences of the Y-6 tephra marker. Figure S2. Particle size distributions for visible tephra layer I08T_30.14, showing upward fining. Figure S3. Major element geochemistries for tephra glass shards from visible tephra layer I08T_30.14 plotted alongside geochemical data from proximal Campanian Ignimbrite deposits and distal occurrences of the Y-5 tephra marker. Figure S4. Biplot showing major element glass compositions of cryptotephra I08T_31.17 and I08T_30.74 m, showing the geochemical field of the visible I08T_30.14 tephra marker, and medial-distal products of Campi Flegrei with a stratigraphic position below the CI. Figure S5. Major element geochemistry of tephra glass shards from I-08 cryptotephra layers between 29.88 and 28.51 m depth, showing glass shard concentrations alongside major and minor element compositions for identified peaks. Plotted for comparison are the major and minor element compositions of selected widespread tephra markers originating from Campi Flegrei including proximal , and distal correlatives from Ionian Sea core M25/4-12 , Lago Grande di Monticchio (Wulf et al., 2004) and Tenaghi Philippon (Wulf et al., 2018). Figure S6. Scatter plot matrix showing the full (wt%) major and minor element geochemical composition of I-08 tephra layers I08T_28.98 and I08T 30.14, in addition to the composition of the Y-3 tephra marker from the type site in the Ionian Sea . Figure S7. Probability density function for the remodelled Caron et al. (2012) age of the Vallone del Gabelloto (VdG/E-1) eruption (with 95.4% highest probability density ranges shown).