Central Mediterranean tephrochronology between 313 and 366 ka: New insights from the Fucino palaeolake sediment succession

Thirty‐two tephra layers were identified in the time‐interval 313–366 ka (Marine Isotope Stages 9–10) of the Quaternary lacustrine succession of the Fucino Basin, central Italy. Twenty‐seven of these tephra layers yielded suitable geochemical material to explore their volcanic origins. Investigations also included the acquisition of geochemical data of some relevant, chronologically compatible proximal units from Italian volcanoes. The record contains tephra from some well‐known eruptions and eruptive sequences of Roman and Roccamonfina volcanoes, such as the Magliano Romano Plinian Fall, the Orvieto–Bagnoregio Ignimbrite, the Lower White Trachytic Tuff and the Brown Leucitic Tuff. In addition, the record documents eruptions currently undescribed in proximal (i.e. near‐vent) sections, suggesting a more complex history of the major eruptions of the Colli Albani, Sabatini, Vulsini and Roccamonfina volcanoes between 313 and 366 ka. Six of the investigated tephra layers were directly dated by single‐crystal‐fusion 40Ar/39Ar dating, providing the basis for a Bayesian age–depth model and a reassessment of the chronologies for both already known and dated eruptive units and for so far undated eruptions. The results provide a significant contribution for improving knowledge on the peri‐Tyrrhenian explosive activity as well as for extending the Mediterranean tephrostratigraphical framework, which was previously based on limited proximal and distal archives for that time interval.

The intense and persistent Middle to Upper Pleistocene explosive volcanism in the central and eastern Mediterranean region entailed the widespread dispersal of volcanic ash and thus deposition in numerous sedimentary archives. Hence, this region provides favourable conditions for tephrostratigraphical and tephrochronological studies. The first detailed tephrostratigraphical framework based on marine sediment successions was established in the Mediterranean region in the 1970s (Keller et al. 1978) and has been continuously refined since then by including terrestrial sequences (e.g. Paterne et al. 1988Paterne et al. , 2008Narcisi & Vezzoli 1999;Wulf et al. 2004Wulf et al. , 2008Wulf et al. , 2012Calanchi & Dinelli 2008;Albert et al. 2012;Tamburrino et al. 2012Tamburrino et al. , 2016Giaccio et al. 2013a;Insinga et al. 2014;Petrosino et al. 2014Petrosino et al. , 2019Bourne et al. 2015;Zanchetta et al. 2019). Tephrostratigraphical and tephrochronological studies in the Mediterranean are often applied for chronological purposes within the scope of palaeoenvironmental, palaeoclimatic and/or archaeological investigations (e.g. Zanchetta et al. 2016;Lane et al. 2017;Wagner et al. , 2022. The applicability and reliability of this dating-correlation tool depends on the completeness and the quality of the reference geochemical and geochronological dataset, which in turn requires integration of investigations of both near-vent and distal tephra, an approach that has also proven crucial for assessing the frequency of eruptions and associated volcanic hazards (e.g. Sulpizio et al. 2014;Albert et al. 2019;Monaco et al. 2021).
Both the completeness of explosive volcanic eruption stratigraphies and the availability of geochemical and precise and accurate chronological data of the Mediterranean tephrostratigraphical framework differ throughout the Quaternary. Single distal archives have pushed and the Ionian Sea record ODP964 to c. 800 ka (Vakhrameeva et al. 2021). However, only specific time intervals older than 200 ka were studied in a few records at high resolution (Giaccio et al. , 2015a(Giaccio et al. , 2021Vakhrameeva et al. 2018Vakhrameeva et al. , 2019Pereira et al. 2020;Monaco et al. 2021Monaco et al. , 2022b. Records in proximity to the peri-Tyrrhenian volcanoes indicate frequent volcanic activity with powerful eruptions during the period 200-370 ka (Rouchon et al. 2008;Sottili et al. 2010;Palladino et al. 2014;Marra et al. 2020aMarra et al. , b, 2021b. Unfortunately, near-vent proximal archives are often discontinuous and incomplete, especially for the older activity, because of subaerial processes (pedogenetic, erosional) and burial by subsequent younger volcanic activity, as demonstrated by numerous eruptions identified only in distal archives Petrosino et al. 2014Petrosino et al. , 2019Leicher et al. 2021;Vakhrameeva et al. 2021). Long and continuous distal records can provide a more complete stratigraphical order of the peri-Tyrrhenian volcanic activity between 200 and 370 ka. Combining proximal volcanological information with distal sedimentary records allows the assessment of the relative (climato-) stratigraphic position of prominent eruptions and the integration of existing eruptive ages and newly identified eruptions into the tephrostratigraphical framework. This is crucial to extend the understanding of the history of peri-Tyrrhenian volcanic provinces as well as the activities of individual volcanic centres. Further, identifying these eruptions within palaeoenvironmental records would allow determination of their climatostratigraphical position, which can then be used to investigate leads and lags of spatial and temporal environmental change (e.g. Blockley et al. 2014;Regattieri et al. 2015;Zanchetta et al. 2016;Giaccio et al. 2021). Equivalents of known eruptions in the period 200-370 ka are currently only partially identified in distal archives (Vakhrameeva et al. , 2021Leicher et al. 2019). A comprehensive and continuous high-resolution tephrostratigraphical record covering this period comes from the Fucino Basin in central Italy. Its tectonic evolution allowed the deposition of a continuous sedimentary succession recording past environmental and climatic changes. Moreover, its geographical position in a good range (i.e.~50-200 km) downwind of most volcanic systems from central Italy makes it a cornerstone to study proximal and distal tephra deposits. Previous drillings in the Fucino Basin have demonstrated the exceptional potential of the record for the last 190 ka (F1-F3 record, FUC S5-6; Giaccio et al. 2017a;Di Roberto et al. 2018;Mannella et al. 2019;Del Carlo et al. 2020). The F4-F5 record was drilled in 2017 and covers the last c. 425 ka . Recent studies of this record already greatly improved the Mediterranean tephrostratigraphy for the intervals 366-425 ka (~MIS 11, Monaco et al. 2021) and 250-170 ka (~MIS 6-8, Monaco et al. 2022b). Here, the investigation of the F4-F5 record focuses on the period 313-366 ka (MIS 9-10), presenting new geochemical and chronological data and their integration within the regional tephrostratigraphical framework.

Fucino Basin
The Fucino Basin (42°00 0 00 00 N; 013°30 0 00 00 E) is located 650 m a.s.l. in the Abruzzo region in central Italy and is the largest extensional tectonic basin of the centralsouthern Apennine chain (Fig. 1A). The basin opened along E-W, NE-SWand NW-SE oriented faults (i.e. the Fucino Fault System, Fig. 1B) since the Late Pliocene to Early Pleistocene (e.g. Galadini & Galli 2000;D'Agostino et al. 2001;Giaccio et al. 2012;Amato et al. 2014). Seismic investigations of the Fucino Basin revealed a semi-graben architecture with an increasing sediment infill of up to~900 m thickness from West towards the depocentre in the East (Patacca et al. 2008). The Plio-Quaternary deposits unconformably overlay both late Messinian terrigenous deposits and the Meso-Cenozoic carbonate basement (Cavinato et al. 2002;Giaccio et al. 2019;Mondati et al. 2021). The Plio-Pleistocene sedimentation is believed to have started before 2.0-3.0 Ma (Giaccio et al. 2015b;Mondati et al. 2021) and appears to be continuous, at least in the central part of the basin (Giaccio et al. 2017aMannella et al. 2019). The basin was constantly endorheic (Lanari et al. 2021) and hosted a lake (Lacus Fucinus) until it was artificially drained in historic times.
volcanic district (CAVD) terminates at c. 350 ka (Karner et al. 2001;Giordano et al. 2010;Giaccio et al. 2013b) followed by a c. 60-ka-long period of dormancy (Marra et al. 2016b). Within the Volsci Volcanic Field (VVF) the main eruptive phase occurred between 350 and 430 ka, potentially lasting until 330 ka (Marra et al. 2021a). At the Roccamonfina volcano, the Brown Leucitic Tuff (BLT) stage extended c.400-350 ka, followed by the White Trachytic Tuff (WTT) stage at c. 330 ka, both including numerous eruptions (Rouchon et al. 2008). The oldest accessible volcanic products within the Campanian Province dates to c. 250 ka (Campanian Volcanic Zone of Rolandi et al. 2003;Belkin et al. 2016), but proximal well-log data (Brocchini et al. 2001) and ash layers preserved in distal archives suggest an older activity (Petrosino et al. , 2015Vakhrameeva et al. 2018Vakhrameeva et al. , 2021Leicher et al. 2019Leicher et al. , 2021. At the Aeolian Islands the oldest proximal deposits date back to 270 ka (Forni et al. 2013), but evidence of older explosive eruptions is found in distal archives Leicher et al. 2019;Vakhrameeva et al. 2019Vakhrameeva et al. , 2021. Little is known about the activity of Mt Etna and Pantelleria of the Sicily Province during the time period here in focus, but the oldest known deposits date back to c. 500 ka (Peccerillo & Frezzotti 2015) and 320 ka (Mahood & Hildreth 1986), respectively. Mount Vulture was dormant between c. 160 and 490 ka according to proximal and distal archives (Villa & Buettner 2009).

F4-F5 record
Two parallel cores were obtained from the F4-F5 drill site (42°00 0 06.22 00 N, 13°32 0 17.79 00 E), which is located close to the centre of the Fucino Basin (Fig. 1A). Seismic information (Cavinato et al. 2002) and sedimentation rate estimations based on adjacent previous drilling campaigns (Fig. 1B;GeoLazio, SP cores, F1-F3;Giaccio et al. 2015bGiaccio et al. , 2017aMannella et al. 2019) suggested this drill site as ideal for reaching older sediments in relatively shallow depth. The 1.5-m-long core segments from the two boreholes were correlated based on core images, X-ray fluorescence (XRF) data and palaeomagnetic information to the 98.11-m-long composite section F4-F5 . Grey-whitish lacustrine calcareous marls dominate the sediments and alternate with darkish clay lacustrine muds. Tephra layers are frequently intercalated. The variability of the calcareous content in the F4-F5 sediments is mainly related to variations in the lake's primary productivity and precipitation of endogenic calcite, which depend on temperature and hydrology and thus on glacial-interglacial and sub-orbital climatic variability . The uppermost~35 m of the F4-F5 record were successfully correlated with the tephrostratigraphical framework of the F1-F3 record ) that spans the last 190 ka (Giaccio et al. 2017a). 40 Ar/ 39 Ar dating of a tephra layer (TF-126) from the bottom of the F4-F5 succession revealed that the entire record covers the last 425 ka (MIS 12, Giaccio et al. 2019). Detailed investigations of the intervals 80. 52-98.11 and 31.74-49.02 m correlated depth (m c.d.) presented a comprehensive tephrostratigraphical framework spanning the MIS 11/12 transition until the beginning of MIS 10 (366-425 ka; Monaco et al. 2021) and the late MIS 8 to early MIS 6 (c. 170-250 ka, Monaco et al. 2022b). The focus of the present study is the succeeding interval 60.42-80.52 m c.d., covering the time interval 313-366 ka (MIS 9-10).

XRF core scanning
Individual core halves of the F4-F5 record were scanned using an Itrax XRF scanner (Cox Analytical Systems, Sweden) at the Institute of Geology and Mineralogy of the University of Cologne, Germany. XRF scans were made as described in Giaccio et al. (2019) using a Chromium tube set at 55 kVand 30 mAwith a dwell time of 10 s and a step size of 2.5 mm.

Tephra identification and sample preparation of F4-F5 tephra layers
Tephra layers were identified during visual core description and subsequent inspection of high-resolution linescan images. Tephra layers were sampled over their entire thickness as a bulk sample or across subunits, if recognized (e.g. based on colour or clast morphology and grain-size change, cf. Table 1). Tephra layers of <0.5 cm thickness or mixedwith larger amounts of lake sediments were prepared following the preparation protocol described in Leicher et al. (2021) using HCl for carbonate removal, sieving and heavy liquid separation. If samples were pre-treated, respective information of the analysed fraction is given along with the geochemical data at the EarthChem repository (Leicher et al. 2022).

Sampling of proximal equivalents
The Magliano Romano Plinian Fall was resampled north of the Sacrofano caldera of the Sabatini Volcanic District. Sample SAB-01rs corresponds to sample SAB-01 described in Sottili et al. (2010). At the Roccamonfina volcano, three subsamples of Unit E and two subsamples of Unit D of the WTT were collected in the same type localities as described by Giannetti & De Casa (2000). Details on section locations and sampled subunits are provided in Table 2.

EPMA-WDS
Major and minor elements of individual glass fragments were analysed by electron microprobe wavelength dis-  Leicher & Giaccio 2021). A Jeol JXA-8900RL electron microprobe equipped with five-wavelength dispersive spectrometers was used for analyses at the University of Cologne. The operation conditions were set to 12 kV accelerating voltage, 6 nA beam current and 5 lm beam diameter. Full details of calibration and measuring conditions are given in Leicher (2021). Information on which laboratory the respective sample was measured in is given along with individual geochemical data at the EarthChem repository (Leicher et al. 2022 Ar and 38 Ar isotopes were measured simultaneously using three ATONAâ amplifiers together with an electron multiplier for 36 Ar isotopes, which was also used for 37 Ar in a second run. Each isotope measurement corresponds to 15 cycles of 20 s integration time. Peak intensity datawere reduced using ArArCALC V2.4 (Koppers 2002). The neutron fluence J factor was calculated using co-irradiated Alder Creek sanidine (ACs) standard ACs-2 associated with an age of 1.1891 Ma (Niespolo et al. 2017) according to the K total decay constant of Renne et al. (2011) (k e.c. = (0.5757AE0.016)910 À10 a À1 and k b¯= (4.9548AE 0.013)910 À10 a À1 ). To determine the neutron flux for each sample, at least six flux monitor crystals from pits framing the samples in each irradiation disc were used. The J-values are the following: TF-62, 0.00056060AE0.00000045; TF-68, 0.00056080AE0.00000062; TF-69TOP, 0.00056060AE 0.00000039; TF-69BOTTOM, 0.00056080AE0.00000056; TF-75, 0.00055710AE0.00000072; TF-81, 0.00055750AE 0.00000061; and TF-82, 0.00055800AE0.00000084. To ensure the detector linearity, mass discrimination was daily monitored by analysis of at least 15 air shots of various beam sizes ranging from 5.0 9 10 À3 to 2.0 9 10 À2 V (one to four air shots) automatically generated during the nights before and after the unknown measurements. Discrimination was calculated according to the 40 Ar/ 36 Ar ratio of 298.56 (Lee et al. 2006). Procedural blank measurements were achieved after every three unknowns. For a typical 5 min time blank backgrounds are between 2.1 and 3.2 10 À4 V for 40 Ar and 60-90 cps for 36 Ar (about 1.0-1.3 10 À6 V equivalent). Full analytical data for each sample can be found in Table S1.

Age-depth modelling
Age-depth modelling was performed using the software package Bacon v. 2.5.7 (Blaauw & Christen 2011) within the open-source statistical environment R (Team 2022). It was based on 40 Ar/ 39 Ar ages of tephra layers identified within the sediment succession between 60.52 and 80.52 m c.d. (Fig. 2). The entire succession was divided into 10 cm vertical sections for modelling individual accumulation rates at a 99% confidence interval, which provide the basis for the age-depth model. Major sedimentological changes identified during core description andwithin the XRF-downcore datawere considered via the 'boundary' function within the model. The resulting succession and subsequent profile depth were adjusted for tephra layers with a thickness of more than 5 cm to avoid bias of accumulation rates by these rapid depositional events. An individual age was calculated for each tephra layer.

Sediment lithology and texture and geochemical composition of tephra layers
The sediment between 60.42 and 80.52 m c.d. can be split into two main lithological units. The lower unit (67.50-80.52 m c.d.) consists of carbonate-poor, greenishbluish silty clays, whereas carbonate-rich, brownishochre clayey silts characterize the upper unit (60.42-67.50 m c.d.). Within the upper unit, no sediments were recovered between 63.32 and 63.66 m c.d. The interval 66.6-67.5 m c.d. marks a transition between the two units. Overall, 32 horizons containing volcanic fragments (e.g. juvenile glass fragments and/or pyrogenic minerals) were identified. The position of these tephra horizons and their characteristic lithological features, i.e. thickness, colour, morphological appearance, type of juvenile fragments and mineral assemblage, are given in Table 1. Tephra horizons are intercalated in fine-grained lake sediments as massive discrete layers and single or arrays of lenses, and vary in thickness between 0.4 and 24.3 cm. Bioturbation, sediment load and/or drilling disturbance structures are present, but had only a minor effect on the preservation of most of the tephra layers (e.g. for determining the position of the isochrone or internal structures). Tephra layers TF-65a/-70/-72, were affected by drilling-induced disturbances (centimetrescale displacement).
Of the 32 identified layers, 25 tephra layers contained sufficient amounts of well preserved and unaltered glass fragments suitable for EPMA-WDS analyses. Two of these tephra (TF-62/-85) were analysed in previous studies Monaco et al. 2021), of which TF-62 is re-examined here. Among the seven other tephra, tephra layers TF-72/-74d contained only one and three fresh glass shards, respectively. The other five tephra layers (TF-73/-74/-74b/-74c/-80) contained exclusively minerals and/or heavily weathered glass fragments, and thus were not suitable for the determination of their glass geochemical fingerprints.
The full dataset of EPMA-WDS analyses is available at the EarthChem repository (Leicher et al. 2022). The individual classification of tephra layers according to the total alkali vs. silica (TAS) diagram after Le Bas et al. (1986) is summarized in Table 1 and visualized in Fig. 3. Tephra layers show most of the compositional features typical for volcanic rocks known from peri-Tyrrhenian volcanoes including tephrites, phonotephrites, tephri-phonolites, phonolites, foidites, trachytes and rhyolites. Based on their compositional differences, the newly investigated tephra layers from Fucino can be split into three compositional groups (Table 1, Fig. 3): CG1 (tephrite-phonolites, n = 16), CG2 (phonolitetrachyte-rhyolites, n = 5) and CG3 (tephrite-phonotephrites and phonolite-trachytes, n = 5). The definition of the compositional groups is given along with the discussion about the volcanic origin of tephra layers below.

Ar/ 39 Ar chronology
The results of 40 Ar/ 39 Ar dating of individual tephra layers are presented as probability diagrams (Fig. 2). Weighted mean age uncertainties are reported at 2r, including J uncertainty, and were calculated using Isoplot 4.1 (Ludwig 2009). Inverse isochrones of individual samples are characterized within uncertainties by an atmospheric 40 Ar/ 36 Ar initial intercept, suggesting that dated crystals are without detectable excess argon.
TF-62. -15 sanidine crystals were individually dated for this layer. The probability diagram ( Fig. 2A) is multimodal with four clearlyolder crystals ranging from about 327.5 to 437.9 ka. The youngest population, including the majority of the sanidines (11 crystals), is interpreted as juvenile and provides a weighted mean age of 313.5AE1.4 ka (Mean Squared Weighted Deviation (MSWD) = 0.47, p = 0.92).
TF-69. -TF-69 was split based on visual differences into two distinct subunits (Table 1), which were treated and dated separately. A total of 15 crystals were dated for   subunit TF-69BOTTOM. The Ca/K ratios of the analysed crystals demonstrate that this subunit mainly comprises sanidines (12 crystals) and only a small proportion of leucites (three crystals). The related probability diagram (Fig. 2C) displays two obviously older crystals (352AE8 and 387AE9 ka). The main mode that is interpreted as juvenile crystals (13 crystals TF-75. -Fifteen leucite crystals were individually dated. Owing to the small size of the crystals (<350 lm), individually calculated ages are less precise than for other tephra layers. However, all crystals analysed gave within uncertainty the same age, resulting in a Gaussian probability diagram (Fig. 2F). Based on this very homogeneous crystal population, a weighted mean age of 350.9AE3.0 ka (MSWD = 0.16, p = 1.00) was calculated.

Volcanic origin of tephra layers
General compositional features and temporal span of investigated tephra layers Homogeneous compositions or continuous compositional trends are observed for most (20/27) of the analysed tephra layers and indicate individual volcanic sources and events (Fig. 3). However, seven tephra layers have a heterogeneous composition with distinct geochemical populations. Tephra layers of CG3 (TF-63/-65a/-76/-81/-82) show a bimodal composition, with a primitive tephritic-phonotephritic and a more evolved phonolitic-trachytic population, which are separated by a silica gap (populations with mean-SiO 2 of 46 and 60 wt%, respectively). The bimodal composition can be interpreted as compositional gap of a single event or by mixing of different eruptions in one layer. However, for TF-65a, TF-76, TF-81 and TF-82 a reworked origin and thus a mix of different eruptions appears unlikely. Their specific compositions do not reflect those of the adjacent tephra layers, nor do they show a more heterogeneous mix of several eruptions (n >2), which would be expected for reworked tephra (e.g. wash-in of several different tephra deposited from the catchment). Furthermore, their individual compositions follow the same magmatic evolutionary trend. This trend is not seen in the composition of TF-63, which may represent two different volcanic origins. The other multicompositional tephra  owing to either natural (e.g. bioturbation, wash in of reworked material) or artificial processes (e.g. drilling, core processing). The only volcanic sources identified for F4-F5 tephra layers in previous studies were the proximal peri-Tyrrhenian volcanoes (Giaccio et al. 2017aDel Carlo et al. 2020;Monaco et al. 2021Monaco et al. , 2022b. Also the spectrum of geochemical compositions analysed within the interval 313-366 ka from the F4-F5 record overlaps with those of the Roman, the Roccamonfina and the Campanian Provinces (Provinces after Peccerillo & Frezzotti (2015), see also

Compositional group 1 (CG1)
Tephra layers of CG1 are characterized by a composition evolving from tephrites to phonolites or trachytes and represent the majority (16/27) of tephra layers analysed within this study. CG1 tephra layers plot within the overlapping TAS fields of known compositions from the Campanian, the Roman, and Roccamonfina and Volsci volcanoes (Fig. 3A). According to the CaO/FeO vs. Cl diagram (Fig. 3F-H) CG1 compositions plot within the fields of Roccamonfina and, more pronounced, volcanoes of the Roman Province, while an origin from the Campania-Neapolitan volcanoes can be ruled out. Owing to the wide compositional spectrum of CG1 tephra layers, the group can be subdivided into three subunits (CG1A, -1B, -1C) based on their different degree in magmatic evolution (cf. Fig. 3A, B dominant rock classification in the TAS and alkali-FeO-MgO diagram).
Compositional group 1A (CG1A). -TF-62/-64b/-68/-69/-71 form the sub-compositional group CG1A, which encompasses the most evolved compositions of CG1. The majority of analysed shards have phonolitic and/or trachytic compositions with only minor portions of less evolved products being classified as tephriphonolitic, phonotephritic or tephritic (TF-62 only). All tephra layers have alkali ratios (K 2 O/Na 2 O) > 1, reaching high values of 1.5-2.6 for the dominant phonolitic and trachytic populations. Regarding their CaO/FeO ratios (mean = 1.1-1.3) and characteristic low Cl contents (mean = 0.1-0.2 wt%), tephra layers are similar to known compositions of eruptions from the Roman Province, of which only the SVD and the VVD were active during the time window explored here.  is a slightly older eruptive event of the Sacrofano area of the SVD and matches with its age of 312.8AE2.0 ka ) that of TF-62. Also the lithological zonation of the MRPF is similar to that of TF-62, with a basal black scoria subunit, topped by the main whitish, well vesicular, leucite-bearing tube pumice lapilli, and an upper subunit of dark grey, moderately vesicular, leucitebearing pumice lapilli . The new obtained geochemical data of the MRPF further support the correlation with TF-62, as they are representative of the main unit displaying the most evolved phonolitic compositions (Figs 4A-C, S1A, B). It is likely that the less evolved compositions observed for TF-62 refers to the black scoria MRPF subunit as described in proximal settings by Sottili et al. (2010). However, this subunit was not analysed within this study.
TF-64b (modelled 318.8AE3.2 ka). -Tephra TF-64b has a phonotephritic-tephriphonolitic-phonolitic composition (mean K 2 O/Na 2 O = 2.0), which indicates, according to the CaO/FeO vs. Cl diagram, an origin in the SVD field (Fig. 3F). Volcanic activity of similar age within the SVD is known from the Bracciano caldera with the Tufo di Bracciano eruptive phase, which represents the main caldera forming event . The age of the eruption is imprecisely constrained in proximal settings between 310.0AE5.0 and 319.1AE6.0 ka. , whereas associated equivalent volcanic deposits found in the archaeological site of La Polledrara di Cecanibbio suggest an age of 323.9AE2.0 ka . Minor explosive activity of the SVD is also known from Monte Musino (319.1AE6.0 ka) scoria cone of the Sacrofano caldera ), but glass geochemical datasets for these activities are not available.
Thus, a precise correlation of TF-64b cannot be provided at this time.   lithological zonation with a greyish-whitish bottom and a black top subunit (Table 1). TF-68 instead includes more evolved tephriphonolitic and trachytic components and a twofold lithological zonation (Table 1) with a fine-grained whitish bottom and coarse-grained, normal-graded greyish top unit. Within the CaO/FeO vs. Cl diagram both layers plot within the fields of the Vicovolcano, the SVD and the VVD, with the majorityof analyses indicating an origin from the latter (Fig. 3F). . Within the TAS diagram, the WOB composition shows a narrow phono-trachytic trend , which is wider and more evident in TF-68 than in TF-69 (Fig. 4D). However, considering all other element compositions, the geochemistry data do not allow an unambiguous preferable correlation (Figs 4D-F, S1C, D). The ages for the WOB eruption products range between 307 and 335 ka (Turbeville 1992;Nappi et al. 1995;Marra et al. 2020b), whereof Marra et al. (2020b) suggest an age of 335.5AE1.5 ka for the main eruptive event, which is close to those of TF-68/-69. The bottom units of the WOB deposits described in proximal settings consist of whitish-light grey pumice Plinian fallout and flow deposits rich in sanidines, which are overlain by a unit of moderately vesicular, dark grey-black scoria blocks very rich in leucites . The same pattern is observed for TF-69 (bottom sanidines, top leucites, Fig. 2). Along with its slightly older age, a correlation of TF-69 with the main eruptive phase of WOB is proposed. Thus, the slightly younger layer TF-68 most likely corresponds to the subsequent WOB activity, which is also evident in the scattering ages obtained on proximal deposits (Marra et al. 2020b) and Fucino tephra layers TF-66/-67.
TF-71 (modelled 338.5AE1.7 ka). -The glass composition of tephra TF-71 is mainly phonolitic, but also encompasses some tephriphonolitic glasses with relatively high K 2 O/Na 2 O ratios between 1.5 and 3.1 (Fig. 4E). Within the CaO/FeO vs. Cl diagram, TF-71 plots in the overlapping fields of the SVD and VVD similar to TF-68 and -69 (Figs 3F, 4D-F). No explosive volcanic activity is known from the SVD at that time, but the age of TF-71 falls between two active pulses of the VVD. TF-71 is slightly older than the oldest weighted mean ages given for the WOB activity (331-335 ka) and slightly younger than those of the Ponticello Pumices TF-65 (modelled age 320.8AE3.3 ka). -TF-65 has a main tephritic-foiditic composition (mean K 2 O/Na 2 O = 1.15), which is accompanied by a second minor foiditic shard (cf. Fig. 3D). The major element compositions of the second population are clearly Na-dominated (K 2 O/ Na 2 O = 0.7) and have significantly higher Al 2 O 3 and lower TiO 2 , MgO and CaO concentrations compared with the main population (Fig. S2). Based on the CaO/ FeO vs. Cl discrimination, most shards of the main population indicate an origin from the SVD, whereas some plot in the VVD field (Figs 3G, S2). The Bracciano-Sacrofano volcanic activity of the SVD is of similar age ), but has not been geochemically investigated because of strong thermal alteration and/or zeolitization (cf. similar aged TF-64b). The main eruptive event, the Tufo di Bracciano, is described as a major caldera-forming eruption ( TF-83/-84 (modelled 365.3AE3.0 ka; 365.5AE2.9 ka). -TF-83 has a homogeneous tephritic main composition and some foiditic-tephritic shards (Fig. 5A). The tephrites are characterized by higher mean concentrations in MgO (7.98 wt%), CaO (15.31 wt%) and K 2 O/ Na 2 O ratios (1.92), whereas the foiditic-tephritic shards have mean values of 6.46 wt%, 14.32 wt% and, 1.39 respectively. TF-84 has a tephritic-phonotephriticshoshonitic composition, which is accompanied by foidite-tephrite-phonotephrites. Although both TF-84 compositional groups differ in their alkali ratio, with the latter having a mean alkali ratio of 1.48 and a major group of 2.65, the other elemental compositions with regard to their SiO 2 content lie on the same evolutional trend and suggest a common volcanic source. Volcanic activity with low evolved compositions and/ or SiO 2 -poor is reported for the VVF, Roccamonfina and CAVD at that time. At the VVF, the Selva Piana products are of similar age (362.4AE11.0 ka, Boari et al. 2009b), but do not show high K series compositions. The BLT stage marks the end of the HKS phase of the Roccamonfina volcano, but whole-rock compositions of similar aged lavas (RMF4, RMF3, 89X, RMF7; 369-355 ka; Rouchon et al. 2008) slightly differ in their geochemical compositions (more evolved HKStype lavas, slightly younger KS-lavas, Fig. 5A-C). Minor (strombolian and hydromagmatic) activity of post-calderic centres within the CAVD marks the end of the Tuscolano-Artemisio phase with the deposition of the Madonna degli Angeli succession, whose products have been dated between 355 and 366 ka (Table 3 remains unknown, as ages are indistinguishable within uncertainties among themselves and from the age of the Villa Senni eruption (Marra et al. 2009). Whole rock geochemical compositions of associated lava flows Gaeta et al. 2016) and from Prata Porci (Sottili et al. 2009) suggest a similar composition to TF-83/-84 ( Fig. 5A-C (Fig. 3H).
TF-66/-67 (modelled ages 327.5AE3.0/328.5AE2.8 ka). -TF-66/-67 are two distinct tephra layers with no indication of reworking of the younger TF-66. They have a similar age and occupy an almost identical position within the tephriphonolitic field of the TAS diagram, with some data scattering in the phonotephritic and shoshonitic fields (Fig. 4D-I). The similarity is also seen in the other elemental compositions and lithological characteristics (Figs 4D-F, S1C, D; Table 1). The CaO/ FeO vs. Cl diagram suggests an origin from the SVD or VVD (Fig. 3H) cluster covers a wide spectrum in the TAS diagram, with silica ranging from 45 wt% (especially TF-74a) to 55 wt% (Fig. 6A). TF-74a has a tephritic, phonotephriticdominated composition (mean K 2 O/Na 2 O = 1.0) overlapping with that of TF-74d, for which only two phonotephritic (K 2 O/Na 2 O = 0.5) and one trachyandesitic shard (K 2 O/Na 2 O = 0.9) have been analysed. The compositions of TF-75/-77 are mainly phonotephritictephriphonolitic, of which the one of TF-75 also reaches the tephritic field, with both layers having a mean K 2 O/ Na 2 O ratio of 0.9. TF-74a/-75/-77 plot in a similar overlapping position between the SVD and Roccamonfina fields within the CaO/FeO vs. Cl diagram, whereas TF-74d plots in both the Vico and Roccamonfina fields (Fig. 3H). Owing to the volcanic inactivity of the SVD and the Vico volcano between 330 and 440 ka (Marra et al. 2020a) and between 305 and 390 ka (Perini et al. 2004;Monaco et al. 2021 (Boari et al. 2009b). From the Roccamonfina volcano, the BLT sequence is imprecisely framed between 344 and 353 ka (Rouchon et al. 2008;Scaillet et al. 2008;Santello 2010). Luhr & Giannetti (1987) and Santello (2010) described these deposits ranging from tephritic to phonolitic in composition, similar to those observed for TF-74a/-74d/-75/-77 ( Fig. 6A-E). Their low alkali ratios, and especially those of TF-74d, match those observed for the BLT pumices with CaO >5.6 wt%. The BLT pumices show increased Na/K ratios owing to the analcimization of leucite (Luhr & Giannetti 1987). However, the analcimization of proximal deposits and weathering, as expressed by high loss on ignition values (7-10%), limit any tentative geochemical correlation between the whole-rock XRF data and the single-grain EPMA-WDS data. Nevertheless, these data demonstrate that also less evolved composition could be associated with the BLT. Based on the chronological and general geochemical similarities with the eruptions associated with the BLTsequence, the origin of TF-74a/-74d/-75/-77 is presumably the Roccamonfinavolcano. In distal settings, a similar wide compositional tephrite to evolved tephri-phonolite spectrum is observed for the BAG tephra (Fig. 6A). This tephra is a widespread marker horizon found in several Loess Palaeosol sequences of the Danube basin in eastern Europe  (Fig. 1, Pouclet et al. 1999;Laag et al. 2021;Jordanova et al. 2022). Previous correlations of the BAG tephrawith the Villa Senni eruption have been disproven by Monaco et al. (2021), who proposed an alternative correlation with the older F4-F5 tephra cluster TF-102/106 (397-399 ka). However, new chronostratigraphical information of the BAG tephra in both the Zenum and Suhia Kladenetz sequences (Laag et al. 2021;Jordanova et al. 2022) suggests a deposition in MIS 10 glacial deposits, which is in conflict with the position of the TF-102/106 tephra during MIS 11. Based on the glass geochemistry obtained at the P aszt o site (Pouclet et al. 1999), TF-75/-77/-79 show strong geochemical similarities with the BAG tephra ( Fig. 6A-E), of which the best match is given by TF-77. Tephra T32 from the Pi anico-S ellere palaeolake sequence in northern Italy was suggested to correlate also with the BAG tephra and to be an equivalent of the BLT eruptive series (Brauer et al. 2007b). However, the age of the sequence and the supposed tephra correlation are controversially discussed (Pinti et al. 2007;Brauer et al. 2007a), as palaeomagnetic information (Scardia & Muttoni 2009) and a K-Ar age of an underlying tephra (779AE13 ka, Pinti et al. 2001) suggest deposition of the sequence within MIS 19. Further, isotopic analyses of T32 rather suggest the palaeo-SVD as the volcanic source (Roulleau et al. 2009;Sottili et al. 2019). This is further corroborated by a proposed correlation with tephra SUL2-11 from the Sulmona MIS 19 deposits (Giaccio et al. 2015a), which was also assigned to the palaeo-SVD activity .
TF-78 (modelled age 353.2AE3.6 ka). -Tephra layer TF-78 has a mainly phonotephritic-tephriphonolitic composition with some basaltic-trachyandesitic and andesitic components (Fig. 4G). The alkali ratio increases from 0.8 to 3.1 with increasing SiO 2 content, and MgO concentrations are higher and Al 2 O 3 lower compared with the generally similar TF-74a/-75/-77 tephra layers. Therefore, a correlation with the BAG tephra found in Hungarian loess deposits appears unlikely (cf. TF-75/-77/-79; Fig. S3). Based on the CaO/FeO vs. Cl diagram an origin from the Roman Province is likely, with most data plotting within the field of the VVD (Fig. 3H). The Ponticello Pumices is a major Plinian fall eruption from the Bolsena area of the VVD and dated at 350.9AE4.0 ka (Nappi et al. 1995), whereas a more recent dating is showing a significantly younger age (346.6AE1.4 ka, Marra et al. 2020b). Although white pumice lapilli are described as trachytes (Nappi et al. 1994), these deposits also contain a mafic component that could match the less evolved composition of TF-78. However, with respect to the age discrepancy, TF-78 is classified as an undefined eruption associated with the VVD, which is also possibly represented by 40 (Luhr & Giannetti 1987), the phonolitic part of TF-79 overlaps with the evolved composition described for the BLT (Fig. 6A-E). Moreover, the similarity to TF-74a/-75/-77 and their common position within the CaO/FeO vs. Cl diagram (Fig. 3H) strengthen the tentative correlation of TF-79 with an early BLT phase of the Roccamonfina volcano. Luhr & Giannetti (1987) described deposits from a Plinian eruption, directly underlying those of the BLT eruption, which may represent a potential counterpart for TF-79. As discussed for TF-75/-77, TF-79 is another potential equivalent of the BAG tephra, although its compositional spectrum is shifted towards more evolved compositions (SiO 2 50-58 wt%), compared with the less evolved BAG composition (SiO 2 mainly 48-54 wt%) (Fig. 6A-E).
Province preceding the Campanian Ignimbrite eruption (>40 ka; Giaccio et al. 2017b) is only fragmentarily documented and the precise volcanic origin of older products within the Campanian Province remains unclear at present (Rolandi et al. 2003;Wulf et al. 2004;Munno & Petrosino 2007;Paterne et al. 2008;Wulf et al. 2012;Petrosino et al. 2014;Leicher et al. 2019Leicher et al. , 2021Vakhrameeva et al. 2021;Monaco et al. 2022a). However, some widespread Mediterranean tephra were recently correlated with Plinian deposits found in the Campanian Plain and attributed to a Campi Flegrei activity of 92-109 ka (Monaco et al. 2022a), suggesting that the Campi Flegrei was probably the source of the Campanian Province-like tephra of the Middle Pleistocene. Known volcanic activity in the age range of CG2 tephra layers (316-348 ka) is also reported for the Roccamonfina volcano with the emplacement of two main eruptive sequences of the BLTand WTT (Giannetti & De Casa 2000;Rouchon et al. 2008).
TF-64/-64a (modelled ages 316.0AE2.9 ka / 316.2AE 2.9 ka). -TF-64 has a mainly LAR-type trachytic composition (mean K 2 O/Na 2 O = 0.98) with a minor phonolitic component and is further characterized by low CaO concentrations (mean = 1.3 wt%). TF-64 plots within the CaO/FeO vs. Cl diagram in the compositional fields of Ischia and Campi Flegrei (Campanian Province), with some data scattering also in the Roccamonfina field (Fig. 3J). TF-64a has a phonolite-dominated composition with only few shards classified as trachytes and a mean alkali ratio of 0.9. Trachytic shards plot within the CaO/FeO vs. Cl diagram in the field of the Campanian Province, whereas phonolites both plot in Roccamonfina and Vico fields (Fig. 5D-F). The trachytic geochemical composition is very similar to the composition of the closely overlying (<3 cm distance) TF-64, which may suggest a dislocation of some shards either by bioturbation or by drilling disturbances. Volcanic activity with the age of TF-64/-64a from the aforementioned sources is only known from the Lower The geochemical composition of the WTT products is similar those of TF-64/-64a (Figs 5, S4D-F), which supports their general assignment with the eruptive sequence of the Lower WTT, including Units A, B, C and D (Giannetti & De Casa 2000). Specifically, the composition of TF-64/-64a suggests a tentative correla-tion with Unit D, despite some differences in geochemistry (e.g. wider compositional spectrum of TF-64/-64a, Fig. 5D-F). On the other hand, the Bojano WTT equivalents (MOL12/13, BOJ-01; Amato et al. 2014;Galli et al. 2017) have a more evolved composition, similar to that of the WTT Unit E (Fig. 5D-F). A correlation of MOL12/13, BOJ-01 with Unit E is further supported by their younger ages of 312.1AE5.0 ka (Amato et al. 2014) and 307-310 ka (Giannetti & De Casa 2000), respectively.
TF-70 (modelled age 335.7AE3.0 ka). -TF-70 has a mainly high-silica trachytic composition (mean SiO 2 = 67.8 wt%, CaO = 1.6 wt%) but includes also few less evolved trachytes (mean SiO 2 = 62.8 wt%, CaO = 2.7 wt%). TF-70 has the highest alkali ratios among CG2, being HAR-type trachytes with means of 2.5 and 2.9, respectively. The CaO/FeO ratios 0.4-0.7 are in then range of known Campanian Province and Roccamonfina products (Fig. 3J). However, TF-70 has Cl concentrations between 0.02 and 0.07 wt% which are untypically low for these volcanic sources and are rather known for Roman products. Among the Roman volcanic sources, the Vico volcano is the only known source for highly evolved trachytic-rhyolitic compositions (Perini et al. 2004;Pereira et al. 2020;Monaco et al. 2021), but the other elemental compositions are different. Moreover, no eruptive events are known for the Vicovolcano in the time period of interest (Perini et al. 2004), which questions this volcano as the potential source. Considering the activityof the Roccamonfinavolcano, the age of the oldest deposits of the WTT is discussed as being between 322.0AE4.0 ka for Unit A (Giannetti & De Casa 2000) and 331.5AE2.0 ka for an undefined unit (Quidelleur et al. 1997;Rouchon et al. 2008). For those deposits no glass compositions are available, but some high-silica trachytes have been associated with the younger Unit E of the WTT (this study and Amato et al. 2014;Galli et al. 2017). If those are compared with TF-70, the latter shows different compositional trends with lower Al 2 O 3 and higher CaO, FeO, MgO and K 2 O/ Na 2 O values at a given SiO 2 concentration (Figs 5D-F, S4D-F). Therefore, only a tentative correlation of TF-70 with the undefined above-mentioned WTT unit is proposed based on general WTT geochemical similarities and chronological constraints.
TF-74e (modelled age 348.3AE3.6 ka). -Tephra TF-74e has a mainly phonolitic-trachytic composition, but also includes three rhyolitic shards (Fig. 3I). Phonolites and trachytes have relatively high alkali ratios of between 1.2 and 2.9, together with high CaO concentrations (2.1-4.1 wt%). Rhyolites are also characterized by high K 2 O/Na 2 O ratios (mean = 2.2) and a mean CaO concentration of 1.6 wt%. Bivariate element plots suggest that both groups form a compositional trend and thus belong to the same volcanic source, even if the low number of rhyolitic shards and missing intermediate compositions hamper an unambiguous correlation. Within the CaO/FeO vs. Cl diagram, TF-74e plots both in the Vico and Roccamonfina fields (Fig. 3J). The ultrapotassic character with a rhyolitic endmember of TF-74e resembles that of older Vico products (Monaco et al. 2021), but an origin from the Vico volcano appears unlikely, as there is no evidence for activity in proximal and distal archives between 305 and c. 400 ka (Perini et al. 2004;Monaco et al. 2021). Further, Na 2 O and K 2 O concentrations of Vico Period I products are respectively lower and higher compared with TF-74e (Fig. 6G). The age of TF-74e matches that of the BLT eruption from the Roccamonfina volcano, and together with their general compositional similarity (Luhr & Giannetti 1987), a correlation with this eruption is tentatively proposed, as discussed for adjacent tephra layers TF-74a/-74d/-75/-77. Within the Mediterranean tephrostratigraphical framework tephra layers ODP964B 2H-5-52.5 from the Ionian Sea (340 ka, Vakhrameeva et al. 2021), TP09-65.835a from Tenaghi Philippon (358 ka, Vakhrameeva et al. 2018) and OH-DP-1527 from Lake Ohrid (358 ka, Leicher et al. 2019) have similar geochemical characteristics to TF-74e (Fig. 6F-K). However, a correlation of TF-74e with OH-DP-1527 and TP09-65.835a is in conflict with the supposed 10 ka older ages and compositional variations of OH-DP-1527 (e.g. >Al 2 O 3 , MgO, Cl) and TP09-65.835a (>MnO, Cl). Although ODP964B 2H-5-52.5 also has higher MgO and Cl values and is slightly younger compared with TF-74e (Fig. 6F), their overall similarities suggest a tentative correlation.

Compositional group 3 (CG3)
Compositional group 3 comprises tephra layers with bimodal compositions, whose individual geochemical populations share features of different compositional groups described before. Tephra TF-63 has two populations that have CG1B and CG2 characteristics, whereas populations of tephra layers TF-65a/-76/-81/-82 show a mix of CG1A-C characteristics.
TF-63 (modelled age 315.3AE2.8 ka). -Tephra layer TF-63 has a twofold composition (Fig. 3K) with a CG1Btype and a CG2-type geochemical population (Figs 5, 6). The CG1B-type population (POP1) follows a tephriticphonolitic trend dominated by phonotephritic shards, has K 2 O/Na 2 O ratios between 0.8 and 1.4 and plots within the Roman Province field (SVD, VVD) of the CaO/FeO vs. Cl diagram (Fig. 3L). The second population (POP2) is made up of phonolitic-trachytic shards (CG2-type) with K 2 O/Na 2 O ratios between 0.7 and 1.14, which indicate an origin from the Campanian or Roccamonfina volcanoes within the CaO/FeO vs. Cl diagram (Fig. 3L). Based on the different compositional trends of CG1B and CG2 type populations and thus their supposed different volcanic origins, TF-63 most likely represents two volcanic events of similar age that were mixed. The CG1B-type tephra is compatible with the low-evolved compositions of the overlying TF-62, which is correlated with the MRPF eruption of the SVD (Figs 4A-C, S1A, B). Owing to the geochemical and chronological similarity, population 1 of TF-63 may represent a pre-eruption phase of the main MRPF eruption. Population 2 of TF-63 is similar to the compositions of underlying TF-64/-64a (Figs 5D-F, S4D-F), which have both been associated with Unit D of the WTT eruption from Roccamonfina (Giannetti & De Casa 2000). As TF-63 is stratigraphically well separated from TF-64/-64a, this may indicate that the WTT Unit D could include at least two eruptive phases, which were separated by a short time span of less than one millennium.
TF-65a (modelled age 325.3AE3.1 ka). -Tephra layer TF-65a has a trachytic-phonolitic, CG1A-type main population (mean alkali ratio =1.5) and a second minor population of foidites and phonotephrites similar to those of CG1B/-C (mean alkali ratio = 0.9). The structure of the tephra layer is affected by drilling disturbances, which do not allow assessment of whether the different populations may refer to individual layers or may represent a zonation of a single layer. However, bivariate element plots suggest an evolutionary trend between both populations, indicating a common volcanic source (Fig. S2). Based on their CaO/FeO ratios and Cl concentrations, a common origin from the SVD is most likely (Fig. S2). The age of TF-65a (325 ka) iswithin the uncertainties of the oldest age obtained for the Tufo di Bracciano eruption (cf. TF-64b; 323.9AE2.0 ka; Pereira et al. 2017). However, the scarcity of glass geochemical data from proximal SVD equivalents hampers a more reliable correlation.
TF-76 (modelled age 350.9AE3.6 ka). -Tephra layer TF-76 has a twofold composition (Figs 3K, 4G-I) comprising one tephritic-phonotephritic population (CG1Btype, mean K 2 O/Na 2 O = 1.2) and a second phonolitictrachytic one (CG1A-type, mean K 2 O/Na 2 O = 2.3, CaO = 2.39 wt%). Both populations plot within the Vico and VVD fields of the CaO/FeO vs. Cl diagram, similar to the slightlyolder TF-78 suggesting a common volcanic source (Fig. 3L). Further, both tephra layers fall in the same compositional trend, which differs from the one of the similar aged tephra group associated with the Roccamonfina volcano (TF-74a/-74d/-75/-77: <MgO, FeO; >Na 2 O, Cl). The Ponticello Pumices are the only described eruption in proximal settings of the VVD with a similar age to TF-76, being K/Ar dated at 350.9AE4.0 ka (Nappi et al. 1995), whereas 40 Ar /39 Ar methods suggest a more precise age of 346.6AE1.4 ka (Marra et al. 2020b). Nappi et al. (1994) presented geochemical data (wholerock XRF) of white pumices of the Ponticello Pumices eruption, which match the trachytic glass composition of S3). However, a comparison of wholerock XRF data with single-grain EPMA-WDS data may be biased by the influence of phenocrysts  and the ages only partly overlap. A precise correlation of TF-76 therefore remains to be clarified.
Single-shard compositions of tephra layers TF-65/-72/-79 TF-65 (modelled age 320.8AE3.3 ka). -Owing to the distinctive different geochemical composition and the single-shard character, the foiditic component of the predominantly tephritic TF-65 most likely represents displaced material of a second eruption. The composition is not similar to other adjacent tephra layers of the sequence, which excludes reworking ofone of these layers and rather suggests an origin from an undetected cryptotephra horizon. The Na-dominated foiditic composition is unique, but also precludes a correlation with known foiditic volcanic sources, such as the Colli Albani district or the Volsci Volcanic Range. Both source areas are characterized by high K-rocks (Boari et al. 2009a;Giaccio et al. 2013b) and were dormant at that time (Giaccio et al. 2013b;Marra et al. 2021a). Therefore, the precise chronostratigraphic position and origin of the second population remain unclear at present. TF-72 (modelled age 340.8AE3.6 ka). -For tephra TF-72 only a single shard could be analysed, being an LAR-type trachyte (K 2 O/Na 2 O = 0.8), with low CaO concentration (0.9 wt%) plotting in the compositional field of Ischia (Fig. 3J). Evidence for Middle Pleistocene volcanic activity within the Campanian Province is only provided from distal tephra layers, some of which are characterized by low-CaO, LAR trachytes. Among these LAR-type trachytes, a marine cryptotephra layer ODP964B 2H-5-67.5 from the Ionian Sea is of similar age (345 ka, Vakhrameeva et al. 2021), but differs in its total elemental composition from TF-72. The Roccamonfina volcano could be another potential volcanic source, since the limited available geochemical data for the slightly younger WTT deposits (307-332 ka; Quidelleur et al. 1997;Giannetti & De Casa 2000) are also characterized by low-CaO, LAR-type trachytes . Therefore, the origin of TF-72 remains unknown at present, but probably lies within the Campanian Province or the Roccamonfina volcano, predating the oldest known WTT deposits.
TF-79 (modelled age 355.1AE3.6 ka). -The peculiar composition of the single rhyolitic shard from this predominantly tephriphonolitic-phonolitic tephra does not match those of similarly aged Aeolian Arc rhyolitic eruptions identified in Lake Ohrid (OH-DP-1530/-1520; Leicher et al. 2019), as its high K-content rather suggests an origin from the Vicovolcano (Perini & Conticelli 2002;Perini et al. 2004;Pereira et al. 2020). TF-79 overlapswith the compositions of older rhyolitic Vico eruptions found in the sequence (Pereira et al. 2020;Monaco et al. 2021), but also those identified in the younger TF-74e. As only a single shard was found and no volcanic activity from the Vico volcano is known from proximal and distal archives (Perini et al. 2004) for the time of deposition of TF-79, it most likely represents displaced material from an unknown source.
Insights into the peri-Tyrrhenian explosive history during the 313-366 ka interval The eruptive order of tephra layers identified in the F4-F5 site confirms the general pattern of volcanic activity observed in proximal settings (Fig. 7). The high number of identified tephra layers provides more details on the individual eruptive sequences (Fig. 8), with two main eruptive series from the Roccamonfina volcano (c. 355-366 ka, 315-330 ka) and intercalated tephra layers associated with the main eruptive pulses of the Colli Albani (c. 365 ka), Vulsini (c. 330 ka and 350 ka) and .
Based on chronological constraints and a general SVD-like geochemical character, tephra layers TF-64b, TF-65 and TF-65a were associated with the volcanic activity of the Bracciano and Sacrofano calderas of the  SVD between 319 and 323 ka. The Tufo di Bracciano, which is the main SVD eruptive event of this period, is chronologically scattered in proximal settings (310-324 ka; Sottili et al. 2010;Pereira et al. 2017). The oldest age matches that of TF-65a (i.e. 325.2AE3.1 ka), while the slightly younger deposits TF-64b/-65 (318,8AE2.9 ka, 320.8AE2.9 ka) may represent subsequent smaller events of post-caldera eruptive centres located within the Bracciano and Sacrofano area . The correlation of the Magliano Romano Plinian Fall with TF-62 represents the first recognition of this event in a distal archive. TF-62 provides a 40 Ar/ 39 Ar age of 313.5AE1.4 ka that is in agreement with the age of 312.8AE2.0 ka obtained on proximal deposits . Tephra TF-63 (modelled age 315.3AE2.8 ka) may represent an unknown pre-MPRF event from the SVD, yet to be recognized in proximal settings.
Tephra layer TF-76 was tentatively associated with the Ponticello Pumices eruption of the Vulsini volcanic district (347-351 ka; Nappi et al. 1995;Marra et al. 2020b), but the correlation needs to be validated by a comparison with proximal geochemical data, which are currently lacking. TF-78 is a less evolved and slightly older eruption (353.2AE3.6 ka), potentially correlated to VVD predating the Ponticello Pumices. Furthermore, TF-69 represents the first distal equivalent recognized of the prominent WOB eruption of the VVD, which caused a caldera collapse in the sector of the Bolsena caldera ). The recognition of several similar tephra layers post-(TF-66/-67/-68) and pre-dating (TF-71) the actual WOB eruption supports the findings of Marra et al. (2020b), who identified several previously undefined eruptions by assessing clusters of singlecrystal 40 Ar/ 39 Ar ages of pyroclastic deposits of the VVD. This supports the idea of a complex eruptive history and helps to define the age of the main WOB eruption. Marra et al. (2020b) identified in different deposits two clusters at 330.8AE1.0 ka (cluster CB-2-3-4-260) and 337.9AE1.2 ka (cluster CB-3), resulting in a combined age of 335.5AE1.5 ka (CB-3 combined) for the WOB eruption. The new findings of the Fucino sediment succession allow a better discrimination of individual eruptions. TF-69 (331.5AE2.2 ka) and TF-71 (338.5AE3.3 ka) match the two clusters identified by Marra et al. (2020b). TF-68 with a 40 Ar/ 39 Ar age of 329.2AE3.4 ka and TF-66/-67 with modelled ages of 327.5AE3.3 and 328.5AE2.8 ka postdate the WOB eruption, as also indicated by a much younger age cluster of 322.3AE0.9 ka (cluster CB-4-5, Marra et al. 2020b).
Based on the results of the Fucino sediment succession, TF-69 best represents the main WOB eruption and provides an age of 331.5AE2.2 ka, which is the most accurate age for WOB at present.
Tephra layers TF-83/-84 have been associated with the waning, post-caldera activity of the Tuscolano-Artemisio/Vulcano Laziale phase of the Colli Albani volcanic district (Marra et al. 2009(Marra et al. , 2016bGiordano et al. 2010;Gaeta et al. 2016). The chronology of this activity in the proximal setting was previously imprecisely constrained as the reported ages were supposedly affected by the presence of xenocrysts from the underlying Villa Senni caldera-forming eruption products, and were within uncertainties chronologically indistinguishable (Gaeta et al. 2016;Marra et al. 2016b). The finding of potential equivalents within the sedimentary succession of the F4-F5 record as distinct tephra layers, supports their proposed slightly younger age of the Villa Senni eruption, but also allows the timing to be constrained to within one millennium (Figs 7, 8).
Two tephra clusters of the F4-F5 record were associated with one of the main eruptive cycles of the Roccamonfinavolcano, i.e. the BLTand the WTT stages, which have been dated in proximal settings at c. 355-440 ka and 307-332 ka, respectively (Giannetti & De Casa 2000;Rouchon et al. 2008). TF-70 from between these two main clusters was also tentatively ascribed to the WTT stage (Fig. 7). The BLT and the WTT stages document a general transitional phase of the magmatic system of the Roccamonfina volcano, which was studied intensively by whole rock data (Luhr & Giannetti 1987;Rouchon et al. 2008). However, glass data availability, especially for pyroclastic deposits of the BLT phase, is scarce and thus limits robust correlations at present.
The youngest cluster of tephra layers (TF-63/-64/-64a) was associated with Unit D of the WTT stage of the Roccamonfina volcano, which provides a more precise chronological constraint of the eruption between 315.3AE2.8 and 316.2AE2.9 ka for this WTT unit. However, it also shows that re-examination of the respective units is needed, as several layers related to one main eruptive unit rather indicate multiple eruptions closely spaced in time (Fig. 7). Based on the new data acquired for proximal deposits of WTT Unit E, the origin of distal tephra layers found in the Bojano basin (BOJ-01, MOL10/14) can now be clarified. This distal layer associated with the WTT Unit E has been dated at 312.1AE5.0 ka (Amato et al. 2014) and thus provides important information for reassessing the chronology of Fig. 8. Synthesis of proximal and distal information of the volcanic history of the Italian volcanoes and the resulting tephrostratigraphical framework of the Central Mediterranean region. All 40 Ar/ 39 Ar ages from literature were recalculated according to ACs at 1.1891 Ma (Niespolo et al. 2017) and the total K decay constant of Renne et al. (2011). See also Table 3 for references and ages. The ages of tephra layers from the distal records are based on archive specific age-depth models (Vakhrameeva et al. , 2021Leicher et al. 2021;Jordanova et al. 2022). Data for the LR04 are based on Lisiecki & Raymo (2005) with the MIS substages of Railsback et al. (2015). XRF data of the Lake Ohrid DEEP site from . The column 'Synthesis of Italian volcanic activity' combines all information. If robust correlations of equivalent eruptive events could be established and several ages were available, only the most reliable age is shown as discussed in the respective subchapter. the WTT eruptive series. TF-70 (335.7AE3.0 ka) in turn is associated with an older unrecognized unit of the WTT series, which supports results by Quidelleur et al. (1997) and Rouchon et al. (2008), who suggested an age of 331.5AE2.0 ka for an undefined WTT unit. The volcanic origin of tephra layer TF-72 (340.8AE3.5 ka) is unclear, as it could correspond to an unknown pre-WTT eruption of the Roccamonfina volcano or to an unknown eruption from the Campanian volcanoes. Products of the Campanian Province are less frequently found within the Fucino succession, but volcanic activity of the Neapolitan volcanoes within the studied time interval is indicated by a proximal drill site (Trecase well, 300-370 ka, Brocchini et al. 2001) and in cryptotephra studies of sediment cores from the Ionian Sea and the Tenaghi Philippon records (Vakhrameeva et al. 2019(Vakhrameeva et al. , 2021. The second cluster of tephra layers associated with Roccamonfina is interpreted as a closely spaced series of eruptions of the actual BLT eruptive series, including the major BLTeruption. Among the series of tephra found in Fucino, TF-77/-79 (351.7AE3.6 ka/355.1AE3.6 ka) may correspond to Plinian deposits predating the main eruptive BLT sequence (Luhr & Giannetti 1987). Tephra layer TF-75 with a 40 Ar/ 39 Ar age of 350.9AE3.0 ka represents the coarsest and thickest deposit of the series and thus probably represents the main caldera-forming eruptive phase. The overlying tephra layers TF-74e/-74d/-74a are interpreted as post-caldera eruptive pulses during the subsequent millennium. Although drilling disturbances biased the core quality of this interval and TF-74b/-c did not contain fresh glass, the different geochemical compositions of the individual tephra layers exclude reworking of older BLT deposits. This series of tephra layers suggests a more complex eruptive behaviour than that suggested by Rouchon et al. (2008), who interpreted the BLT as a single crisis event, and rather supports findings of Santello (2010), who described several subunits of the BLT, being deposited within a few thousand years.
The age of the oldest cluster of tephra layers TF-81/-82 is constrained by two new 40 Ar/ 39 Ar ages (362.7AE5.0 and 362.8AE3.2 ka) and has been tentatively associated with the BLT phase owing to their HKS character and temporal overlap with similar effusive products dated in the proximal area (369.0AE8.0 ka, La Frascara; 367.1AE9.0 ka, Galluccio; 357.1AE8.0 ka, Fontana-Radina; 355.1AE5.0 ka, Scipicciano; Rouchon et al. 2008). However, such pyroclastic deposits have not been described from the proximal area yet and the available data are too scarce to track the origin of TF-81/-82 at the present stage.
The eruptive activity of the Volsci Volcanic Field has not been recognized within the Fucino succession so far. Even if the ages of some of the recently re-dated eruptions (Boari et al. 2009b;Marra et al. 2021a) overlap with tephra layers of the F4-F5 record (TF-81-84, c. 365 ka), the geochemical dataset available for the proximal VVD products does not support such a correlation.
The 313-366 ka Fucino tephra record in the framework of the Mediterranean tephrostratigraphy The overall Mediterranean tephrostratigraphical framework of the time interval studied was previously only known from distal archives from the Ionian Sea (eight Italian tephra layers, Vakhrameeva et al. 2021), the terrestrial archives of Lake Ohrid (three Italian tephra layers, Leicher et al. 2019Leicher et al. , 2021 and Tenaghi Philippon (nine Italian tephra layers, Vakhrameeva et al. 2018Vakhrameeva et al. , 2019. The Fucino F4-F5 interval between 313 and 366 ka now stands as the most detailed and well-dated succession of 27 tephra layers from the Roman and Roccamonfina volcanoes (Fig. 8). An updated tephrostratigraphical framework combining distal archives and near-vent volcanic records is provided in Fig. 8. Both proximal and distal deposits indicate the frequent volcanic activity during the investigated time frame and thus highlight the great potential to find respective equivalents in other archives. However, correlations between different archives are still rare and only one correlation was tentatively proposed between the Fucino and ODP964 records (TF-74e/ODP964B 2H-5-67.5). Interestingly, the other numerous tephra layers associated with the Campanian Province found in the ODP964B and Tenaghi Philippon records could identified neither be in the Fucino nor in the Lake Ohrid successions and thus indicate the potential for cryptotephra analysis of these records. However, the absence of macroscopic evidence in these records may indicate a general shift in predominating wind directions during the course of eruptions with more south-easterly distribution of volcanic products. The proposed correlation of Fucino tephra layers TF-75/-77/-79 with the BAG tephra of the Loess-Palaeosol archives of the Danube basin represents a promising tie point for the alignment of different palaeoenvironmental archives on a larger regional scale. However, further geochemical analysis of the BAG and Fucino tephra layers is required to validate and specify the proposed correlation with a single layer, which based on the present data is most likely TF-77. Nevertheless, owing to the narrow age range (350-355 ka) of potential Fucino equivalents, the chronology of the loess sequences can already by improved (cf. Fig. 8), as previous correlations suggested a much older age for the BAG tephra (368 ka, Laag et al. 2021;c. 398 ka, Monaco et al. 2021;368 ka, Jordanova et al. 2022).
Considering the climatostratigraphic position of tephra layers of the F4-F5 record based on Ti XRF data (Fig. 8), tephra layers TF-68/-69to TF-72deposited during the glacial termination III or the MIS 9/10 transition (337AE5 ka, Lisiecki & Raymo 2005). TF-68/-69 have been directly dated and thus are very suitable as marker horizons for assessing spatial and temporal differences in changes of glacial to interglacial environmental conditions.

Conclusions
The volcanic origin of 27 tephra layers in the MIS 9/10 interval of the Fucino F4-F5 sediment succession was investigated based on their lithological, geochemical and chronological characteristics. Direct 40 Ar/ 39 Ar dating of six tephra layers set the basis for a Bayesian age-depth model for the interval 313-366 ka, which in turn provided modelled ages for the other tephra layers. The record provides the first tephrostratigraphical framework based on a continuous sediment succession of that time interval and at a suitable distance from the peri-Tyrrhenian volcanic centres for recording explosive eruptions of different intensity. Based on these results, the volcanic sources and, in some cases, the specific equivalent eruptions or eruptive phases of tephra layers have been identified. This information provides new insights into the eruptive history, mainly for major known eruptions of the Sabatini and Vulsini volcanic districts of the Roman Province and the Roccamonfina volcano. In addition, the resulting tephrostratigraphy of the Fucino succession represents a reference dataset for future correlations with specific (minor) eruptive events not yet conclusively identified in proximal settings. This significantly extends the available chronostratigraphical and compositional frame of Italian volcanism reconstructed from proximal areas. Moreover, the stratigraphically ordered succession also provided the possibility to reassess existing ages of eruptions with high temporal resolution.
Overall, the F4-F5 successions represents the richest tephra archive for this time interval of peri-Tyrrhenian volcanism and can thus be used as a reference record for a subsequent extension of the central Mediterranean tephrostratigraphical framework for this period. The large number of so far unidentified eruptive events and their similar geochemical patterns also indicate the need for (i) improving the proximal geochemical and geochronological data basis to reliably distinguish between closely spaced eruptions and validate proposed volcanic origins and (ii) a detailed geochemical investigation of distal ash layers detected in other archives to establish robust correlations.   (Nappi et al. 1995) and EPMA-WDS data of the BAG tephra (Pouclet et al. 1999).  Table S1. Analytical details of 40 Ar/ 39 Ar measurements.