The Medial Offshore Record of Explosive Volcanism Along the Central to Eastern Aegean Volcanic Arc: 1. Tephrostratigraphic Correlations

The Milos, Christiana‐Santorini‐Kolumbo (CSK) and Kos‐Yali‐Nisyros (KYN) volcanic complexes of the Aegean Volcanic Arc have repeatedly produced highly explosive eruptions from at least ∼360 ka into historic times and still show recent unrest. We present the marine tephra record from an array of 50, up to 7.4 m long, sediment cores along the arc collected in 2017 during RV Poseidon cruise POS513, which complements earlier work on distal to ultra‐distal eastern Mediterranean sediment cores. A unique set of glass‐shard trace element (LA‐ICPMS) compositions complements our major element (EMP) data on 220 primary ash layers and 40 terrestrial samples to support geochemical fingerprinting for correlations with 19 known tephras from all three volcanic complexes and with the 39 ka Campanian Ignimbrite from the Campi Flegrei, Italy. The correlations include 11 eruptions from CSK (Kameni, Kolumbo 1650, Minoan, Cape Riva, Cape Tripiti, Upper Scoriae 1 and 2, Middle Pumice, Cape Thera, Lower Pumice, Cape Therma 3). We identify a previously unknown widespread tephra from a plinian eruption on Milos (Firiplaka Tephra). Near the KYN we correlate marine tephras with the Kos Plateau Tuff, the Yali 1 and Yali 2 tephras, and the Upper and Lower Pumice on Nisyros. Between these two major tephras, we found two tephras from Nisyros not yet observed on land. The four Nisyros tephras form a systematic trend toward more evolved magma compositions. In the companion paper we use the tephrostratigraphic framework established here to constrain new eruption ages and magnitudes as a contribution to volcanic hazard assessment.


Geological Setting
The Aegean Sea, as a part of the Mediterranean Sea, is characterized by a complex arrangement of subduction zone segments at which the African (Nubian) plate subducts beneath the Eurasian plate. The subduction was initiated in the Miocene (Zhu et al., 2006) and is continuing until today. At the Hellenic trench south of the Aegean Sea, the eastern Mediterranean lithosphere (frontal part of the African plate) subducts with a velocity of ∼4 cm/ yr and a mean dip angle of 25° beneath the continental Aegean microplate, resulting in one of the tectonically most active regions of western Eurasia (e.g., McKenzie, 1972;Papazachos et al., 2005;Shaw & Jackson, 2010). The overriding Aegean continental microplate is part of the Hellenic (Alpine) orogen and is dominated by Early Triassic metamorphic rocks (e.g., Cycladic Blueschists; Bonneau & Kienast, 1982). Slab retreat and rapid southward migration of the Hellenic trench since the Miocene (Zhu et al., 2006) led to extension in the southern and central Aegean region causing deep basins (e.g., the Cretan basin) and seismically active faults as well as horst and graben structures on the Aegean seafloor (Le Pichon & Kreemer, 2010;Royden & Papanikolaou, 2011).
The Aegean Volcanic Arc stretches from the Gulf of Saronikos in the northwest to an area close to the Turkish coast in the East and includes the active volcanic centers at Methana, Milos, Santorini, and Nisyros ( Figure 1). The volcanic structures are hosted within extensional neotectonic basins/grabens cross-cutting the Plio-Quaternary sedimentary sequences. The large marginal faults of the basins are typically normal faults locally associated with dextral NE-SW strike-slip zones. The prevailing tectonic orientation is NE-SW at Santorini and Nisyros and NW-SE at Methana and Milos, following the general geometry of the volcanic arc. East-west faults are observed to disrupt the previous arc-parallel NW-SE structures at Methana and Milos (e.g., Piper & Perissoratis, 2003;, Nomikou, Hubscher, et al., 2016, Nomikou, Hubdcher, Papanikolaou, et al., 2018. Faults facilitate the ascent of magmas along the active Aegean Volcanic Arc, since the distribution of volcanic centers and the linear alignment of eruptive vents follow fault orientations (Pe- Piper & Piper, 2005). Whereas large, composite volcanoes with caldera structures (Santorini and Kos-Yali-Nisyros) can be found in the central and eastern sectors of the arc, small, commonly monogenetic eruptive centers prevail in the western sector (Francalanci et al., 2005). The similar distance of all volcanoes to a Benioff zone at ∼130 km depth provides clear evidence for the role of subduction in the genesis of this volcanism (e.g., Papazachos et al., 2005). Accordingly, typical calc-alkaline compositions can be found at all volcanic centers although variations in abundance of different magma types in space and time along the arc can be observed (Pe-Piper & Piper, 2005). 10.1029/2021GC010010 5 of 32

Milos Volcanic Complex
The islands of Milos, Antimilos, Kimolos and Polyegos, form the westernmost volcanic complex in our study area (Figure 1b). The geology of Milos can be divided into four main units: (a) Mesozoic metamorphic basement, (b) Neogene sedimentary rocks, (c) Pliocene and Pleistocene to recent volcanic sequences and (d) the alluvial cover (e.g., Fytikas et al., 1986).
Explosive rhyolitic volcanism on Milos during the Pliocene and Quaternary is documented by a well-exposed succession of pumiceous deposits (Rinaldi & Campos Venuti, 2003). The volcanic activity started at 2.66 ± 0.07 Ma with submarine rhyolitic explosive eruptions before the volcanic activity emerged around 1.44 ± 0.08 Ma and formed scattered subaerial andesitic and dacitic to rhyolitic centers (Fytikas et al., 1986;Stewart & McPhie, 2006). The Pliocene-Recent volcanic sequence of Milos has been subdivided into five units by Fytikas et al. (1986) and further investigated by Stewart and McPhie (2006). It comprises deposits from submarine rhyolitic cryptodomes and pumice cones at the base (unit 1), followed by submarine dacitic and andesitic domes and lavas (unit 2), subsequent submarine to subaerial lavas, domes, and pyroclastic deposits of variable compositions (unit 3), then the rhyolitic lava-pumice complexes of Trachilas and Firiplaka (unit 4), and finally the locally confined deposits from phreatic activity formed until recent times (unit 5). The eruption products cover several square kilometers on the island and include pyroclastic sequences up to ∼200 m thick and small-volume, biotite and quartz bearing rhyolite lavas up to 100 m in thickness ( Figure 2; Fytikas et al., 1986;Campos Venuti & Rossi, 1996;Stewart & McPhie, 2006).

Christiana-Santorini-Kolumbo Volcanic Complex
The Christiana-Santorini-Kolumbo (CSK) volcanic complex is located in the central sector of the 450-km-long Aegean Volcanic Arc ( Figure 1) and comprises three distinct volcanic centers along a 40-km-long, 040°-045° striking fault zone Nomikou, Carey et al., 2013), crossing the island arc. At the SW end of this zone and 23 km from Santorini lies the extinct Christiana Volcano with its two main islets Askani and Eschati reaching up to 283 m above the sea level. Christiana Volcano has produced lavas and tuffs of unknown ages (Aarbourg & Frechen, 1999).
Santorini at the center of the CSK complex has produced volcanic deposits since the Mid-Pleistocene that overlie the Mesozoic to Tertiary basement (Jackson, 1994). The Santorini volcanic center consists of the islands Thera, Thirasia and Aspronisi that are arranged as a dissected ring around a flooded caldera. The caldera, a 11 × 7 km composite structure, resulted from at least four collapse events over the last 200 kyrs and contains in its center the post-caldera volcanic islands Palea Kameni and Nea Kameni (Druitt & Francaviglia, 1992;Pyle & Elliott, 2006;Nomikou et al., 2014). The oldest dated volcanic rocks are early Quaternary submarine tuffs near Akrotiri in SW Thera, which are speculated to originate from vents at Akrotiri and Christiana (Francalanci et al., 2005). Subaerial volcanism at Santorini began about 650 ka ago (Druitt, 2014;Druitt et al., 1999), and has continued until the present day. More than a hundred explosive eruptions within the last 250,000 years, triggering at least four caldera collapses, make Santorini one of the world's most active arc volcanoes (Druitt et al., 1999). The succession of major tephra events, including at least 12 large plinian eruptions, is summarized in Figure 2. This succession has been interpreted to represent two cycles (∼170 ky per cycle) of increasing eruption intensities and magnitudes, one from Cape Therma 1 to Lower Pumice 2, and the other from Cape Thera to the Minoan eruption (Druitt et al., 1989). Each cycle started with eruptions of mostly intermediate-composition magmas and developed to large silicic eruptions at the end.
The last caldera collapse was caused by the late Bronze Age (3.6 ka) Minoan eruption (Druitt, 2014), which discharged several tens of km 3 DRE (Johnston et al., 2014;Pyle, 1990) of rhyodacitic magma, distributed ash over a large area of the eastern Mediterranean and Turkey, and generated tsunamis that may have contributed to the downfall of the advanced Minoan civilization (e.g., McCoy & Heiken, 2000;Nomikou, Druitt, et al., 2016;Vougioukalakis & Fytikas, 2005). After the Minoan eruption, volcanic activity continued inside the caldera where effusive and mildly explosive eruptions built up the dacitic Palea Kameni and Nea Kameni islands between 197 BC and 1950 AD .
About 7 km off the north-eastern coast of Thera, the ∼480-m high submarine Kolumbo Volcano and 23 other submarine cones formed within the fault-bordered Anydros basin . Seismic profiles across Kolumbo revealed a complex internal structure of five volcaniclastic units ( Figure 2) probably due to repeated phases of construction and destruction (Hübscher et al., 2015;Hooft et al., 2017;Nomikou et al., 2019). In 1650 AD the Kolumbo Volcano exploded in a paroxysmal event which ejected huge quantities of volcanic ash, forming a 1.7-km-diameter summit crater (Carey, Nomikou, Croff Bell, & Ballard, 2013;, and generating a devastating tsunami Ulvrova et al., 2016). This event was the most hazardous and powerful eruption in the Greek territory in historic times killing more than 70 people on Thera and causing damages within a radius of 150 km (Fouqué, 1879;Vougioukalakis & Fytikas, 2005). The submarine to emergent rhyolitic pumice deposits are ∼250 m thick in the crater walls whereas distal fall deposit is found as far away as on the Turkish mainland and forms the youngest widespread ash layer on the regional seafloor. Cantner et al. (2014) investigated the stratigraphy of the proximal submarine deposits and the pre-eruptive storage conditions of the rhyolitic magma, and concluded that the explosive eruption started submarine but finally emerged above the sea surface. Fuller et al. (2018) sampled and analyzed the fine-grained ash deposits on the seafloor around Kolumbo which form a strong contrast to the fines-poor breccias found at the crater. The fine-grained ash deposits were emplaced by various forms of vertical and lateral ash-rich gravity currents. Today, the shallowest point of Kolumbo Volcano lurks only 18 m below sea level, and there is continuing CO 2 degassing and hydrothermal activity in the north-eastern part of the crater (Carey, Nomikou, Croff Bell, Lilley, et al., 2013;Kilias et al., 2013). The hydrothermal vent field north of Kolumbo, discovered in 2006 by Sigurdsson et al. (2006), developed in the latest Pleistocene mainly during explosive events (Hübscher et al., 2015).

Kos-Yali-Nisyros Volcanic Complex
The Kos-Yali-Nisyros volcanic complex (KYN) is located at the eastern end of the Aegean Volcanic Arc and its volcanic activity reaches back to about 3 Ma (e.g., Allen & Cas, 2001;Bachmann et al., 2012). The island of Kos is covered by the Kos Plateau Tuff, product of the largest eruption in the upper Pleistocene (161.3 ± 1.1 ka; Smith et al., 1996) in the Aegean region.
The vent for the paroxysmal caldera-forming KPT phreatoplinian eruption lies somewhere between the islands of Kos, Yali and Nisyros, and ejected an initial fall deposit and subsequent huge pyroclastic density currents that traveled over the sea and emplaced thick ignimbrites on Kos and other islands Cas, 1998, 2001). Six major stratigraphic units have been identified, including a widespread vitric ash fall deposit at the base, followed by several phreatomagmatic surge-like pyroclastic density current deposits and ignimbrites extending up to ∼160 km radially from the volcano (Allen et al., 1999). About 60 km 3 DRE of rhyolitic magma were erupted and are thought to have triggered subsidence of a submarine caldera south of Kos (Allen et al., 1999;. Subsequently, this structure was re-filled by the Nisyros composite volcano (e.g., Limburg & Varecamp, 1991), the Yali pumice cone (Allen & McPhie, 2000) and the Pleistocene-Recent Strongyli basaltic andesite cone as well as several submarine volcanic cones, which all have formed under strong tectonic control in the complex horst and graben structure Nomikou & Papanikolaou, 2011;Papadopoulos et al., 1998;Papanikolaou & Nomikou, 2001;Tibaldi et al., 2008).
Nisyros Island (8 km diameter) lies south-southeast of Kos Island. Nisyros Volcano grew on a basement of Mesozoic limestones and Neogene sediments and has a 4-km-wide and 450-m-deep caldera as well as 700-m-high post-caldera domes of Late Pleistocene age (Francalanci et al., 1995). The volcanic edifice formed in two cycles of explosive and effusive eruptions fed initially by calc-alkaline basaltic-andesitic and subsequently by rhyolitic magmas (Figure 2b; Volentik et al., 2002;Longchamp et al., 2011;Tomlinson et al., 2012;Dietrich & Lagios, 2018). The first cycle started as a basaltic-andesitic submarine volcano which later emerged and effusively and explosively discharged basaltic-andesites to rhyolites (Kyra pyroclastics and dome collapse breccia, Figure 2). The second cycle started with two major caldera-forming plinian eruptions that destroyed the pre-existing volcano and emplaced the Lower and Upper Pumice layers, each followed by post-caldera dome and lava extrusion as well as erosion and paleosol development after the Lower Pumice ( Figure 2; Francalanci et al., 1995;Hardiman, 1999). Finally, Nisyros was covered by the Yali 2 pumice fall deposit. A major issue with the Nisyros stratigraphy is that published radiometric ages of some deposits vary strongly and do not fit the observed stratigraphic succession (see also part 2 as well as Volentik et al., 2002;Tomlinson et al., 2012;Popa et al., 2020).
Yali Island is located 10 km south of the island of Kos and 5 km north of Nisyros ( Figure 1) and lies within an extensional setting (the Kerme Graben; Dewey & Sengör, 1979) and close to or inside the postulated Kos Plateau Tuff caldera (Allen & Cas, 2001;Keller et al., 1990). Yali is the source of the youngest highly explosive eruption of the Kos-Nisyros-Yali volcanic complex, which emplaced the Yali-2 (or Yali-C) subaerial pumice fall deposit ∼31 kyrs ago (Figure 2b; Allen & McPhie, 2000). This deposit is locally covered by the 21 ka Kamara obsidian lava and pumice deposit. Yali-2 overlies bioclastic sandstone and limestone units that cover the >150 m thick Yali-1 submarine pumice breccia succession of yet unknown age and hidden basal contact (Allen & Mc-Phie, 2000). The various facies of the Yali-1 tephra formed during submarine eruptions and resulted from deposition of pumice clasts from aqueous suspension, emplacement of submarine gravity currents and re-sedimentation of primary tephra (Allen & McPhie, 2000).

Previous Offshore Tephrostratigraphic Work
Given the relatively small size of the Aegean volcanic islands, even moderate size explosive eruptions on land can be expected to produce deposits on the surrounding seafloor. Since plinian eruptions inject volatiles and ash into the stratosphere their dispersal is mainly controlled by the dominantly strong (up to >60 km/hr) westerly stratospheric winds in the Aegean region. These winds transport the eruption products preferentially towards the east-southeast during most of the year (e.g., monthly record of stratospheric winds (18-27 km height) for the last five years; https://earth.nullschool.net/about.html). Exceptions occur in the summer months May to August, when northerlies to southerlies are present at moderate strength (5-30 km/hr). Such winds may be an explanation for observed tephra dispersal to the north. Additionally, the tropospheric wind regimes are very variable throughout the year (e.g., Tyrlis & Lelieveld, 2013) and may have influenced the ash dispersal of smaller eruptions as well as specific tropospheric phases (e.g., co-ignimbrite plumes) of large eruptions. The sediments of the Aegean Sea and the central to eastern Mediterranean Sea, (hemi-)pelagic sediments periodically interrupted by sapropels representing periods of deep-sea anoxia (Rohling et al., 2015), therefore contain abundant visible tephras and invisible cryptotephras from the Aegean Volcanic Arc but also from Italian and Anatolian volcanoes. The marine tephrostratigraphy of the Mediterranean Sea is a well-established grid of widely dispersed marker tephra layers that have been subdivided into foraminifera zones from V to Z (e.g., Federman & Carey, 1980;Keller et al., 1978;McCoy, 1981). Thus, tephra layers have been named for example, Z2 (Minoan), Y2 (Cape Riva), X1 (Upper Scoriae 1), or W3 (Kos Plateau Tuff). The advantage of the Mediterranean marine tephrostratigraphy over many other arc regions is that the different deposits and source volcanoes can be reliably distinguished by geochemical compositions (e.g., Aksu et al., 2008;Federman & Carey, 1980;Hardiman, 1999;Karkanas et al., 2015;Keller et al., 1978;Margari et al., 2007;Narcisi & Vezzoli, 1999;Tomlinson et al., 2015Tomlinson et al., , 2012Vinci, 1985;Wulf et al., 2020Wulf et al., , 2002. The major element compositions of Italian alkaline and peralkaline volcanic rocks differ significantly from calc-alkaline products of the Aegean Arc and Anatolian arc volcanoes (e.g., Clift & Blusztajn, 1999). The available geochemical data base for late Quaternary tephras is particularly detailed, especially for Italian volcanic centers (Wulf et al., 2004(Wulf et al., , 2008(Wulf et al., , 2012Tomlinson et al., 2012Tomlinson et al., , 2015 but also for some volcanoes in the Aegean Sea (Satow et al., 2015;Tomlinson et al., 2015;Wulf et al., 2020). However, so far only the largest and most widespread tephras of the Aegean Volcanic Arc have been included in this database. This deficiency was also noted by Wulf et al. (2020) and is particularly relevant for this study because we expect to find tephras from inter-plinian activity, at least from Santorini, in addition to the medial deposits from the large eruptions, because our cores are located at proximal to medial distances from the volcanic islands.

Marine Core Sampling and Logging
During RV POSEIDON cruise POS 513 (2017), 47 gravity cores and three box cores were recovered between Milos in the west and Kos in the east (Figure 1b). Recovered core length varied between ∼30 cm and 7.5 meters (box cores 10-40 cm) at water depths of ∼100 to ∼1,200 meters (Freundt, 2017).
Visual core descriptions were completed on board using the IODP description scheme from Mazzullo and Graham (1988). The archive-half sections of the sediment cores were visually described for lithologic and sedimentary features aided by 20× wide-field hand lens and binocular microscope. Visual inspection and smear slide analysis yielded information about variations in lithologic components, color, sedimentary structures, and occasionally drilling disturbances. Lithologic components comprise tephra particles (glass, minerals) as well as various rock fragments and microfossils. For tephra analysis we sampled suspected ash layers on board and enriched the volcanic components from mixed sediment in the laboratory by wet sieving into 63-125 μm and >125 μm fractions. The grain-size fractions were then used for geochemical microanalyses.

Primary Versus Reworked Ash and Ash Pods
Pyroclasts can be incorporated into marine sediments by a variety of processes. These processes include (a) water-settled primary fall emplaced on the seafloor either by settling of single particles or by vertical density currents released from ash concentration zones at the ocean surface, (b) eruption-fed submarine density currents formed either by submarine eruptions or by subaerial pyroclastic density currents entering the ocean, and (c) submarine density currents that entrain and re-deposit pyroclasts (e.g., Cassidy et al., 2014;Freundt et al., 2021;Hopkins et al., 2020;Schindlbeck et al., 2013;Schneider et al., 2001).
We use a combination of criteria to distinguish ash produced by explosive eruptions, herein referred to as "primary," from "secondary" volcaniclastic deposits emplaced by reworking of primary tephra (see Freundt et al. (2021) for more extensive discussion). Heterogeneous glass-shard compositions that do not show a clear magmatic differentiation trend have been excluded from the final data set. On the other hand, massive or size-graded deposits composed of glass shards of homogeneous compositions are considered primary deposits from explosive eruptions, regardless of the details of final transport and deposition.
Ash pods, that is, isolated lensoid or irregularly shaped bodies of ash in marine sediments, can be formed by a variety of processes but bioturbation and resedimentation of marine ash deposits, sometimes nearly instantaneous, are the most common (Cambray et al., 1993;Fujioka, 1986;Hunt & Najman, 2003). Where ash pods of homogeneous glass composition are arranged in an ash-pod layer, it is reasonable to assume that this layer marks the position of the original primary ash layer. Single ash pods, however, can occur in a position significantly above or below the position of the original ash layer, depending on the mechanism and timing (syn-or post-depositional) of their formation. Single ash pods were not included in the database.

Chemical Analyses
For comparison with marine tephra glass-shard compositions we established a glass-compositional reference data base for Aegean arc tephras using samples from all relevant tephra units that have been stratigraphically well defined by previous studies (Allen & Cas, 1998;Allen & McPhie, 2000;Cantner et al., 2014;Druitt et al., 1999) or by own field work (e.g., Milos, Kolumbo) (see Table S1 for available sample coordinates and Table S2a in Data Set S1 for sampled respective stratigraphic units and geochemical data). The tephra samples have been collected by authors of those studies and come from Milos and Kolumbo (sampled by Nomikou and colleagues), from Santorini (sampled by Druitt and colleagues), and from Kos, Nisyros and Yali (sampled by Allen and Mc-Phie). Samples from Santorini are from near-vertical cliffs and we therefore recommend to refer to the geological map and/or contact T. Druitt for details of sampling sites in order to avoid sampling errors. We crushed samples from all major tephras on land to obtain glass fragments that we analyzed by the same methods as the marine glass shards. The 63-125 μm fraction of both types of samples was embedded with epoxy resin into 12 pre-drilled holes in acrylic mounts and polished to facilitate measurements with the electron microprobe (EMP) and the Laser Ablation Inductively Coupled Plasma Mass Spectrometer (LA-ICP-MS).

Electron Microprobe (EMP) Analyses
Glass shards (∼7,000 in total) were analyzed for major and minor elements on 474 epoxy embedded samples using a JEOL JXA 8200 wavelength dispersive EMP at GEOMAR, Kiel, utilizing the methods of Kutterolf et al. (2011). A calibrated measuring program was used based on international standards with a 10 μm electron beam to minimize sodium loss. Oxide concentrations were determined using the ZAF correction method. Accuracy was monitored by two measurements each on Lipari obsidian (Lipari rhyolite; Hunt & Hill, 2001) and Smithsonian basaltic standard VGA99 (Makaopuhi Lava Lake, Hawaii; Jarosewich et al., 1980) after every 60 single glass shard measurements (∼15 per sample). Standard deviations are <0.5% for major and <10% for minor elements (with the exception of P 2 O 5 and MnO 2 in samples >65 wt% SiO 2 ). All analyses with totals >90 wt% were normalized to 100% in order to eliminate the effects of variable post-depositional hydration and minor deviations in focusing of the electron beam. The acceptable analyses of each sample were then averaged in order to characterize the elemental compositions of each tephra. Layers with heterogenous compositions but inside differentiation trends were included with their intra-sample deviation in the data base whereas widely heterogenous compositions outside any possible differentiation trend were excluded. All of the resulting major and trace element data and their respective errors are listed in Tables S2 and S3a in Data Set S1.

Laser Ablation-Inductively Coupled-Mass Spectrometry (LA-ICP-MS)
The trace element concentrations of ∼1,500 glass shards from marine and onland tephras (150 samples, ∼5 shards per sample) were determined by LA-ICP-MS between September 2018 and March 2021 at the Academia Sinica in Taipei, Taiwan, providing the first complete trace element reference data base for Aegean arc tephras. The LA-ICP-MS instrumentation comprises a 193 nm excimer laser (Teledyne CETAC Analyte G2) set to a spot size of 24-30 μm (using 5-10 J/cm 2 energy density at 4-10 Hz repetition rate), coupled to a high-resolution ICP-MS (Agilent 7900). Detailed machine setups, procedures, and methods of the laboratories are described in Schindlbeck et al. (2016). Following 45 s of blank acquisition, typical ablation times were around 75 s. Data reduction was performed using Version 4.0 of "real-time on-line" GLITTER© software (van Achterberg et al., 2001), immediately following each ablation analysis. Individual silica and calcium concentrations, measured by EMP on the same glass shards, were used as an internal standard to calibrate each trace element analysis. An international glass standard (BCR-2g) was measured every 10 sample measurements in order to monitor accuracy and to correct for matrix effects and signal drift in the ICP-MS, and also for differences in the ablation efficiency between sample and reference material (Günther et al., 1999). The concentrations of NIST SRM 612, needed for external calibration, were taken from Norman et al. (1996). The limit of detection (LOD) for most trace elements was generally <100 ppb. For Rare Earth Elements (REE), the LOD is generally around 10 ppb. The analytical precision is generally better than 10% for most trace elements.

Correlation Techniques
Ash-layer correlations are based on glass-shard compositions, supplemented by modal compositions (e.g., crystals, lithic fragments, micro-and macrofossils), sedimentary structures, stratigraphic relationships and geographic location. Supporting data include the shape, vesicularity and vesicle texture of glass shards and pumiceous fragments, and also the mineral content of the ash layers as determined in smear slides (Figures 3 and 4). In order to correlate marine ash beds with tephras on land, we analyzed the glass compositions of >3 juvenile clasts from one to four different units and/or samples of 24 onshore tephras at all relevant eruptive centers and complemented these analyses with published data.
The first step in onshore-offshore correlations is the graphical comparison of the average compositions of each marine ash bed with the compositional fields defined for the onshore tephras. A correlation is accepted when these compositions overlap within the respective analytical errors in all variation diagrams that usefully discriminate tephra glass compositions. Next to major and trace element concentrations, this approach uses trace element ratios that minimize the influences of both analytical errors and alteration.
The second step is the application of a hierarchical cluster analysis (HCA; e.g., Kutterolf et al., 2008) using Origin 2021 to statistically constrain correlations within a combined marine and onshore tephra data base. This step is particularly useful where visual inspections of core-to-core and offshore-onshore in the first step remained ambiguous. The HCA creates different cluster solutions between 1 (all samples in one cluster) and n (each sample in a single cluster) based on the parameters applied to the modeling and the data set. The result of each modeling is a dendrogram, a graphical summary of the cluster solution. Before applying the cluster analysis, we reduced the compositional variables (major and trace elements as well as trace element ratios) by performing a principal component analysis that reduces the variables to 10 components which account for 96.57% of the variance in the samples. We used the nearest and farthest neighbor linkage as well as the Ward clustering method and the single and squared Euclidian distance measuring intervals to test and achieve the best possible correlations for our data set that features both a large compositional variety among the investigated tephras and small differences for some tephras from the same eruptive center. Several HCA's were carried out, compared and evaluated with regard to the observed stratigraphic relationships. The Ward clustering method in combination with a squared Euclidian distance measure were the most reliable cluster solutions for the grouping and correlation of the studied samples.  The combination of different discrimination diagrams in combination with HCA, observed stratigraphic order, and ages has the potential to result in unambiguous correlations. In the end, ash layers in single cores that can be correlated across different sites and/or with tephras on land are grouped as a "tephra layer" and are named from young to old (CA 1-20). Thus, a "tephra layer" includes all ash layers in different cores that have been emplaced by the same single volcanic eruption.

General Description of Cores
The hemipelagic background sediment is composed of layers in the dm-thickness range that consist of clay to silty sand and volcaniclastic clay to sand. Coarser sediments (sand to granule) are dominated by bioclasts like foraminifera, nannofossils, sponge spicules, and echinoderms mixed with some phyllosilicate minerals; clays and silty clays contain foraminifera and nannofossils. Overall volcaniclastic matter and reworked sedimentary clasts (consolidated clay and silt agglomerates) are strongly admixed in the background sediment.
The cores collected in the central and eastern Aegean Sea contain a total of 348 ash/lapilli bearing horizons, which occur as primary ash layers, ash pod horizons, ash pods and dispersed ash. They are intercalated within the hemipelagic sediment and range from 1 to 65 cm in thickness (Figures 3 and 4). Sorting out reworked tephra-bearing layers and combining multiple beds within the same layer, we classify the remaining 220 ash layers as primary pyroclastic deposits. These include 149 pale gray to white felsic tephras, 54 gray beds of probably intermediate composition and 17 black layers of mafic compositions (e.g., Figures 3 and 4).
The upper boundaries of the tephra layers are commonly irregular, horizontal to inclined, and either sharp or gradational with the overlying sediment. The lower boundaries are predominantly very sharp, horizontal to inclined, and locally show features of erosive cutting into the underlying hemipelagic sediments, creeping or slumping (Figures 3a-3f). Load casts are found especially where coarser ash to lapilli overlie fine sediment. The tephra layers are predominately unstratified with normal grain-size gradation, but also planar and cross stratified layers with overall reverse grading can occur, especially above erosive boundaries. Some ash layers are significantly more competent compared to their host sediment due to diagenetic processes and were tilted, bent and dissected during coring. Generally moderate, but partly extensive bioturbation has locally destroyed the primary sediment structures.
Juvenile particles of white to light gray ash beds are mainly (>90 vol%) composed of fresh, clear, colorless, glass shards varying from highly vesicular, pumiceous textures with commonly tubular and elongated bubbles to angular blocky, cuspate, flat and y-shaped shards with very few vesicles (Figures 3 and 4). Grain sizes range from fine ash to coarse lapilli. The mineral assemblages comprise, in order of abundance, plagioclase, clino-and orthopyroxene, quartz, amphibole (mostly of hornblende composition), ilmenite and magnetite, as well as biotite and sanidine, but modal proportions vary among tephras. Generally, the crystals do not exceed 1 to 10 vol% in total.
Dark-gray mafic ash layers consist predominantly of dark brown glass shards. Most of these have blocky shapes and are moderately to poorly vesicular (Figure 3h). More mafic, dark gray to black tephra layers commonly have brownish glass shards with round to elliptical vesicles but tachylitic fragments are also observed. Some black ash layers also contain highly vesicular, tubular, brownish glass shards. The mineral assemblages of the dark tephras include plagioclase, pyroxene, Fe-Ti oxides and some olivine.
Glass-shard compositions of the Aegean marine tephras mostly range from andesitic to rhyolitic compositions with a few basaltic-andesitic samples (Figure 5a). Two glass-shard groups deviate significantly in having trachybasaltic and trachyphonolitic compositions (Figure 5a). Apart from these two groups, all other samples straddle widely around the dividing line between between normal and high-K calc-alkaline compositions (Figure 5b).

Juvenile Glass Compositions of Proximal (Onshore) Tephras
We present the first published major and trace element glass compositions for Milos tephras. Moreover, for many other tephras we provide the first published trace element glass compositions. Figure 6 illustrates the compositional variabilities within and among the onshore tephras and eruptive centers. In many cases, tephra glass 10.1029/2021GC010010 13 of 32 compositions can be well discriminated by major element concentrations but in other cases there is significant overlap. Further distinctions can then be made on the basis of trace element ratios. Figures 7 and 8 show a compilation of glass geochemical variation diagrams that are most useful for the purpose of correlation. All correlations are also compiled in Table 1. These correlations are supported by results from hierarchical cluster analyses (Figure 9) using 88 chemical elements and element ratios. Cluster analysis   successfully reproduces known stratigraphic relations of the tephras. The compositions of glass shards suggest that pods and sediment layers that contain dispersed ash commonly occur in the correct stratigraphic context, implying that redeposition of ash having occurred soon after or contemporaneous with the primary emplacement or that disturbance was by in-situ bioturbation.

Correlation of Marine Ash-Layers With Their Volcanic Source
In summary, we have been able to correlate 209 marine tephra layers to 19 Aegean tephras onshore as well as one tephra from the Campi Flegrei, Italy. Correlated marine tephra layers are labeled CA1 through CA20 in stratigraphic order from young too old. The remaining (11) uncorrelated ash horizons have compositions not represented in our database.

Milos Volcanic Complex
The pyroclastic sequences in unit 4 on Milos formed during at least three major eruptive events (e.g., Figure 2; Fytikas et al., 1986). From bottom to top, the succession of the past ∼370 ka comprises the Trachilas tephra(s?) in the north, and the lower and upper Firiplaka tephra(s?) in the south (Campos Venuti & Rossi, 1996;Fytikas et al., 1986;Stewart & McPhie, 2006). Ages and distribution of the explosive and effusive eruption products beyond the island are not well constrained but new Ar/Ar ages for the younger pyroclastic sequence are 317 ± 4 ka for the Trachilas complex and 62 ± 3 to 70 ± 10 ka and 110 ± 20 ka, for the upper and lower Firiplaka tephras, respectively (Zhou et al., 2021). Furthermore, Campos Venuti and Rossi (1996) distinguished three members within the Firiplaka sequence comprising ∼50 pyroclastic flow units in total.
Phenocrysts in the crystal-poor (<10%) pumices of the onshore Firiplaka deposits comprise abundant plagioclase, sanidine, quartz, and lesser biotite and amphibole (Fytikas et al., 1986), which is consistent with the mineral assemblage and the order of abundance we found in the marine ashes (<5% crystals).
The major element glass composition of a 0.3 cm thick marine ash layer 133.1-133.4 cm bsf (below sea floor) in core TR172-26 (Figure 1b) described by Hardiman (1999) is a possible match with Firiplaka tephra. With this and our core POS513-54, the Firiplaka tephra extends at least 200 km to the southeast from Milos.

Santorini
The pyroclastic sequence emplaced on Santorini during the past ∼360 kyrs contains 13 tephras from major, dominantly silicic and intermediate eruptions (Figure 2), including at least four caldera-forming events (Lower Pumice 2, Upper Scoriae 1, Cape Riva, Minoan; Druitt et al., 1989Druitt et al., , 1999Druitt & Francaviglia, 1992). All Santorini tephras have more or less similar mineral assemblages and minor to moderate crystal contents (5%-20%) consisting of dominant plagioclase and subordinate clino-and orthopyroxene, Fe-Ti oxides and ± pyrrhotite and apatite; some dacitic and rhyodacitic tephras also have rare amphibole whereas andesites also may include olivine (Druitt et al., 1999;Vespa et al., 2006). Numerous deposits from smaller, inter-plinian eruptions are intercalated between the major tephras , some of which are postulated to be present in marine cores by Wulf et al. (2020). Since trace element glass compositions of inter-plinian tephras are not available, we tested possible correlations of marine tephras with the inter-plinian units by comparing major-element glass compositions Figure 7. Glass-shard compositions of marine ash layers (normalized to 100 wt% anhydrous compositions) recovered from the western part of our research area around the CSK volcanic complex, compared with glass-composition fields for Santorini onshore tephras shown in Figure 6. Marine data symbols are averages of all analyses for each ash layer and the attached gray bars indicate the standard deviation around the mean. The analytical data are given in the supplementary tables in Data Set S1. Marine ash samples identified by symbols are listed at right by core number, depth below seafloor, and correlated tephra layer CAx. The color coding groups ash layers that belong to the same tephra layer CAx. The legend at top identifies colored composition fields of the onshore tephras. with their data (see Figure S1) but mostly found no unambiguous correlation. One exception is marine tephra layer CA5 which would fit the major element composition of inter-plinian tephra associated with the Cape Tripiti tephra ( Figure S1; Wulf et al., 2020). However, as we will show below, we prefer to correlate CA5 with the Cape Tripiti tephra itself because of the good trace element fits. It is possible, however, that Cape Tripiti tephra and associated inter-plinian tephras are indistinguishable, also in trace element compositions.
Starting with the oldest Santorini tephra in the marine sequence, the 30-cm-thick ash layer CA20 in core POS 513-28 (266-297 cm bsf) has a glass-chemical composition matching that of Cape Thera, Cape Therma 3 and Middle Pumice, which are all very similar (Figures 7 and 9, Tables 1 and S3 in Data Set S1). However, there is a general decrease in silica from bottom to top of the inversely graded marine ash layer that may reflect the compositional zonation observed in the Cape Therma 3 tephra on land, where a basal greyish-orange pumice fall is overlain by gray to pink to black andesitic scoria flow deposits (Druitt et al., 1999;Wulf et al., 2020). Additionally, CA20 occurs stratigraphically below Lower Pumice 1 (186 ka) in the same core, excluding a possible correlation with Cape Thera and Middle Pumice which are both younger than Lower Pumice 1. We therefore favor a correlation of CA20 with the 200.2 ± 0.9 ka old  Cape Therma 3 tephra. Wulf et al. (2020) also geochemically correlated a 2.5 cm thick ash layer at 639.5 cm bsf in RV Meteor M40/4 core KL51, which also has dacitic to andesitic glass compositions, to the Cape Therma 3 tephra.
Lower Pumice 2 tephra (LP2) consists of a ∼20-m-thick, coarse grained, normally graded fall deposit of white pumice clasts in the western cliffs of Thira, overlain by several meter-thick pyroclastic flow deposits (Druitt et al., 1999;Keller et al., 2014;Simmons et al., 2017). We did not find equivalents of the 176.7 ± 0.6 ka old  LP2 tephra in our cores, but Wulf et al. (2020) correlated a layer in core M40/4-KL51 (529 cm bsf) with LP2 and they reported that this layer has previously been correlated by Hieke et al. (1999) to Keller's V-1 tephra (Keller et al., 1978). Two-to 4-cm-thick tephra layer CA17 is found in cores POS 513-28 (183-185 cm bsf), and −40 (491-495 cm). Its glass composition falls in the overall cluster of Middle Pumice, Cape Therma 3 and Cape Thera (CTh,Figures 7 and 8; Tables 1 and S1 in Data Set S1). However, there are additional observations that lead us to favor a correlation with the 156.9 ± 2.3 ka Cape Thera tephra. First, dark marine tephra layer CA17 contains transparent to brownish glass shards that have a broad range in compositions (e.g., 55 to 70 wt% SiO 2 , mean 63 wt% SiO 2 ) which reflects the compositional heterogeneity of the trachy-andesitic to trachy-dacitic compositions found in the pumice fall and scoria flow deposits on land (Druitt et al., 1999;Wulf et al., 2020). Second, CA17 assumes a stratigraphic position in the cores between ash layers that correlate with Lower Pumice 1 and Middle Pumice or Upper Scoriae 1, respectively. CTh has also been assigned to ash layers in cores KL51 (498.5 cm bsf) and LC21 (1,077-1,119 cm bsf) by Wulf et al. (2020). Applying our major element correlation fields to the glass data of Hardiman (1999) suggests also correlation with an ash layer in core TR172/27 (305-307 cm bsf).
Tephra layer CA16 is found as dispersed ash in core POS 513-28 (146-155 cm bsf) and as a distinct layer of 1-14 cm thickness (Figure 3f) in cores POS 513-40 (445-446 cm bsf) and POS 513-37 (296-310 cm bsf). It correlates compositionally with the Middle Pumice (MP, Figure 7; Tables 1 and S3 in Data Set S1) and all three marine-bed compositions are confined to the same cluster branch as the onshore sample from the upper part of the  Figure 6. Marine data symbols are averages of all analyses for each ash layer and the attached gray bars indicate the standard deviation around the mean. The analytical data are given in the supplementary tables in Data Set S1. Marine ash samples identified by symbols are listed at right by core number, depth below seafloor, and correlated tephra layer CAx. The color coding groups ash layers that belong to the same tephra layer CAx. The legend at top identifies colored composition fields of the onshore/proximal tephras. Yali 2a

YA2
Yali   Note. D: radiometric and/or sedimentation rate ages associated with sapropels from Druitt et al., 1999, Bronk Ramsey et al., 2015, data from Keller cited in Vespa et al., 2006and Gertisser et al., 2009, Giaccio et al., 2017, Satow et al., 2015 ages. This only important in the age column where bold/italic are radiometric ages and only italic ones are estimates. Table 1 Continued MP fall deposit (Figure 9). MP was additionally found by Wulf et al. (2020) in core KL51, 452.5-455 cm bsf, by Federman and Carey (1980) in TR172/19 and 27, and by Vinci (1985) in core KB28 327-330 cm bsf.   Figures 7 and 8 are verified by the respective cluster grouping. Only the glass compositions of Middle Pumice, Cape Therma 3 and Cape Thera tephras are too similar to be discriminated. Note that distance scales are interrupted for clarity.

31-33 ka
The 3-to 32-cm-thick tephra layer (Figures 3e and 3h) correlates with the Upper Scoriae 1 deposit onshore (US1, Figure 7; Tables 1 and S3 in Data Set S1) that is described as a major dark scoria fall overlain by a series of massive pyroclastic flow deposits containing coarse-grained spatter agglomerates (Druitt et al., 1999). Tephra layer CA15 is mainly composed of brown glass shards with characteristic andesitic composition that falls into one cluster subgroup together with the onshore sample Upper Scoriae 1a ( Figure 9). Moreover, the marine tephra also contains a few glass shards of trachy-andesitic composition as well as some rare olivine crystals both of which are also observed in its onshore US1 counterpart (Andújar et al., 2016;Wulf et al., 2020). US1 was also recognized in other cores by Wulf et al. (2020) who also assigned layer X-1 from Keller et al. (1978) and ash beds in numerous other cores of Hardiman (1999) and Aksu et al. (2008) to US1, following the suggestion of Hieke et al. (1999) and Schwarz (2000). Our revisiting of the given literature data confirms these correlations.
Another prominent black marine tephra layer, CA10, can be correlated with Upper Scoriae 2 (US2, Figure 7; Tables 1 and S3 in Data Set S1). It is 7-12 cm thick in one core west (POS 513-9, 314-326 cm bsf) and six cores east  of Santorini. Marine and onshore Upper Scoriae 2b samples together fall into one confined cluster subgroup ( Figure 9). Similar to Upper Scoriae 1, supporting arguments for this correlation are the rare occurrence of olivine and the range of glass compositions from andesite to dacite in the marine ash, which reflect the andesitic bulk-rock composition of onshore US2, being dominated by spatter agglomerates (Druitt et al., 1999).
Tephra layer CA5 occurs in cores POS 513-15, −20, −28, −30, −31, −40, and −41 as a 1-to 7-cm-thick distinct ash bed or ash pod layer. It correlates with the up to 1-m-thick rhyodacitic onshore deposits from the 27.5 ± 1.2 ka old Cape Tripiti eruption (CT, Fabbro et al., 2013;Satow et al., 2015) on the basis of major and trace element glass compositions (Figure 7; Tables 1 and S3 in Data Set S1) and is distinct from other large Santorini tephras. Apart from one sample, the glass compositions of Cape Tripiti and Cape Riva tephras are very well discriminated in two related branches of the cluster analysis, supporting the correlation ( Figure 9). As discussed above, minor inter-plinian pyroclastic fall layers at this stratigraphic level have the same major-element glass compositions as Cape Tripiti tephra and their contribution to CA5 cannot presently be excluded (see Figure S1).
One of the tephras most frequently found in our cores is layer CA4, which occurs as ash layers occasionally showing both fall and flow facies (Figure 3b). CA4 is present as 3-to 43-cm-thick ash layers in cores POS 513-8  from Santorini. The rhyolitic glass shards facilitate the correlation with the 22.0 ± 0.3 ka old (Bronk Ramsey et al., 2015), rhyodacitic, Cape Riva Tephra (CR, Figure 7; Tables 1 and S3 in Data Set S1) that is associated with the third caldera-forming eruption of Santorini. This eruption emplaced a thick pumice fall deposit followed by a thick ignimbrite (Druitt et al., 1999). Fragments of andesitic scoria enclaves that are found in the onshore samples are very rarely found in the marine ash layers. The marine samples share a sub-cluster with the onshore Cape Riva samples ( Figure 9); one sample, POS 513-28, 37-40 cm shares the Cape Tripiti sub-cluster but can be distinguished within this core from underlying Cape Tripiti due to its major element composition (Figure 7). On a regional scale, CA4 correlates with the widespread Y2 layer of Keller et al. (1978) and is also recognized in numerous other cores around the Aegean Sea, indicated by Federman and Carey (1980), Vinci (1985), Aksu et al. (2008), Margari et al. (2007), and Wulf et al. (2002Wulf et al. ( , 2020. The most frequently occurring marine tephra in the entire region is tephra layer CA3 (in all cores except POS 513-16, 39 and 54), which correlates with the 3.6 ka old Minoan eruption, the latest caldera-forming event on Santorini (Druitt et al., 1999). The correlation is supported by the fairly unique glass compositions (MT, Figure 7; Tables 1 and S3 in Data Set S1), by the well discriminating cluster analysis ( Figure 9) and by the shallow position mostly <50 cm below the sea floor. CA3 consists of highly vesicular pumiceous particles (Figures 3g and 3l) and has thicknesses from 1 to 61 cm; both fall and flow facies can be observed (Figure 3a). Marine ash of the Minoan eruption (Z2 tephra after Keller et al., 1978) occurs in multiple cores all over the Mediterranean Sea and adjoining onshore regions (e.g., Aksu et al., 2008;Federman & Carey, 1980;Hardiman, 1999;Margari et al., 2007;Satow et al., 2015;Vinci, 1985;Wulf et al., 2020).
The tephra layer CA2 occurs in two cores (9)(10)(11)17-20 cm bsf) within 60 km of Santorini. The glass composition is similar to that of deposits from the larger explosive eruptions of Kameni (KM, Figures 7 and 9; Tables 1 and S3 in Data Set S1) suggesting an origin from that volcano although not necessarily from the specific eruption of the sampled Kameni tephra.

Kos
We have not found tephra CA18 as a distinct primary ash layer in our cores but it occurs as dispersed ash (about 50 vol%) in a ∼10-cm-and 6-cm-thick sediment layer (POS 513-37, 340-350 cm bsf, POS 513-54, 142-148 cm bsf, respectively) and in the deformed sample recovered in two core catchers (POS 513-46, 300-310 cm bsf and POS 513-36, 310-312 cm bsf). These core positions lie about 80 km southwest of the KYN volcanic complex, and the glass compositions correlate well with the Kos Plateau Tuff (KPT, Figures 8 and 9; Tables 1 and S3 in Data Set S1) in agreement with the stratigraphic position beneath all other KYN tephras in the cores (Figure 2). Marine ash beds of the Kos Plateau Tuff have also been reported by Federman and Carey (1980), Vinci (1985), Hardiman (1999), Margari et al. (2007), Satow et al. (2015), and Wulf et al. (2020) in the Aegean Sea but also in the Black Sea by Wegwerth et al. (2019). Figure 11. Comparison of average glass compositions of 13 uncorrelated ash beds with Zr/Nb, Nb/U, Ba/Y, and Ce/Hf variations along the Aegean Volcanic Arc (W to E) as discussed in the text. Gray dashed bars represent the geographical position of the volcanic centers along the arc. Note that we do not have glass compositions for Christiana. Along-arc variations are based on our own glass compositions measured in this work. The distance along the arc is measured from Antimilos eastward. Positions of tephras along the x-axis are approximate locations of the cores in which they were found. Rectangles and associated arrows (colors are adapted from the respective marine ashes) at the x-axis represent the resulting best-fit location for the source areas of the unknown tephras based on their compositions.
The two fine-lapilli pumice layers N2 and N3 in core POS 513-52 ( Figure 4a) each have unique glass compositions constraining their origin from separate eruptions ( Figure 10). Their trace element characteristics favor an origin from Nisyros ( Figure 11). Moreover, the glass compositions of N2 and N3 form a systematic trend with those of N1 and N4 ( Figure 10) which suggests eruption of all four tephras from the same evolving magma reservoir. Deposits equivalent to the marine N2 and N3 tephras have not yet been found on land between the Lower and Upper Pumice deposits.

Tephras From Unknown Sources
A number of ash beds have glass compositions that we cannot correlate with specific eruption products on the islands. However, following the example from the Central American volcanic arc (e.g., Carr et al., 2007Carr et al., , 2003Kutterolf et al., 2008), systematic variations along the Aegean Volcanic Arc in geochemical parameters that are relatively insensitive to magmatic differentiation ( Figure 11) can be used to attribute ash-bed compositions to specific regions of the arc. Where such compositional correlation is ambiguous, we favor the source area closest to core position considering the large distances between the Milos-CSK-KYN volcanic complexes.
There are three rhyolitic marine tephras in two cores west of Santorini and close to Milos (POS 513-7, and 8) of which the ash layer in POS 513-8 (507-511 cm bsf) correlates to an ash layer in core POS 513-7 (200-204 cm bsf) in terms of major and trace element glass compositions. The trace element compositions of these layers (red box in Figure 11) are not known from Santorini; some trace element ratios would fit Milos tephra compositions but others lie in-between the Milos and Santorini values.
Ash layers POS 513-46, 215-217 and POS 513-46, 227-232 cm bsf (green square and circle in Figure 11) have trace element ratios similar to the Kyra pyroclastics on Nisyros and an estimated age of ∼120 ka (see part 2) that fits to the given eruptive age, but Ba/Y for example, is too low and also other trace element ratios are close but do not directly match the available onshore composition of Kyra pyroclastics (Table S3 in Data Set S1). Therefore this correlation remains uncertain.
Ash layer POS513-15, 196-206 cm bsf (black triangle in Figure 11) has trace element ratios similar to Kolumbo 1650 tephra but an estimated age of ∼20 ka (see part 2), which does not fit. Proximity to Kolumbo Volcano suggests POS 513-15, 196-206 cm bsf came from an older eruption of Kolumbo but without any other constraints this correlation remains uncertain.

Conclusions
We have established a new reference data base of glass compositions, particularly including trace element glass compositions, of stratigraphically controlled widespread, mostly plinian tephras from land that have been erupted during the last ∼200 kyrs along the Aegean Volcanic Arc. This data base enabled us to correlate 209 ash layers in sediment gravity cores collected during marine cruise POS 513 in the Aegean Sea to 19 eruptions from the Aegean Volcanic Arc, resulting in a tephrostratigraphy for medial to distal sediment cores (Figures 12 and 13). We achieved correlations with 11 major tephras from the CSK volcanic complex (Kameni, Kolumbo 1650, Minoan, Cape Riva, Cape Tripiti, Upper Scoriae 1 and 2, Middle Pumice, Cape Thera, Lower Pumice, Cape Therma 3). Contrary to expectation, considering the proximity of some of our cores to Santorini, we found no clear evidence of the inter-plinian events in the seafloor cores, which may either be due to the present unavailability of trace element glass compositions of the inter-plinian deposits or to easy erosion or bioturbation of such presumably very thin ash beds. However, fine ashes from such smaller eruptions are potentially preserved as cryptotephras in the sediments, which we have not investigated in this study but which may be addressed in dedicated future studies.
We have also discovered new tephras not previously known from land. One is a widespread tephra from Milos which was not previously known to have produced such deposits. The others are two tephras from Nisyros that lie bracketed between the marine equivalents of Upper and Lower Pumice on the island. Including also correlations to the Kos Plateau Tuff and Yali 1 and 2 tephras, all our correlations provide a tephrostratigraphic grid between the volcanic centers from Milos to Nisyros (Figures 12 and 13).

Data Availability Statement
All data produced during this study and defining the core and onland sample locations is under way at Kutterolf, Freundt, Hansteen, et al., 2021 in PANGAEA Data Archiving and Publication (https://doi.org/10.1594/PAN-GAEA.938615). An electronic copy of this data file is provided as Tables S1-S5 in Data Set S1.