Reconstruction of the 1908 Messina gravity flow (central Mediterranean Sea) from geophysical and sedimentological data

Earthquakes


Introduction
Sediment gravity flows are some of the most important geological processes on Earth (Mosher et al., 2010;Mountjoy et al., 2018).These laminar to turbulent suspended density flows transport large quantities of sediment from continental margins to deep sea basins (Talling et al., 2012).They largely influence canyon and channel development, and play a key role in the development of sedimentary basins and facies distribution (Talling et al., 2013;Mountjoy et al., 2018;Stevenson et al., 2018).They also pose a geohazard to seafloor infrastructure such as telecommunication cables, as documented during the 1929 Grand Banks event (Heezen and Ewing, 1952;Piper et al., 1999), the 2003 Boumerdes Algeria earthquake (Babonneau et al., 2017) and the 2006 Pingtung Earthquake off SW Taiwan (Hsu et al., 2008).It has further been suggested that gravity flows may contribute to tsunami generation (Assier-Rzadkieaicz et al., 2000;Fine et al., 2005).In spite of their significance, sediment gravity flows remain poorly studied and understood, especially in terms of their flow properties and evolution and the spatial variability of erosional and depositional processes.Various processes (e. g., earthquakes, submarine slides, floods, storms, tsunamis, anthropogenic activities) are known to trigger sediment gravity flows (e.g., Piper et al., 1999;Arai et al., 2013;Talling, 2014).It is, however, difficult to measure and monitor gravity flows in situ, as it is nearly impossible to predict when and where these events occur (Paull et al., 2003;Hage et al., 2018Hage et al., , 2019;;Mountjoy et al., 2018;Clare et al., 2021;Talling et al., 2022).Hence, flow properties of large gravity flows are mainly reconstructed based on the analysis of their deposits and modelling (e.g., Meiburg and Kneller, 2010;Talling et al., 2012;Stevenson et al., 2018).
A well-known historic event is the 1908 Messina turbidity current in the western Ionian Basin, offshore eastern Sicily, which is one of the most seismically active areas of the Mediterranean Sea (Ryan and Heezen, 1965) (Fig. 1A).A M w 7.1 earthquake on 28 December 1908 at 5:21 (local time), which caused >60,000 casualties and a tsunami, is known to have triggered the Messina turbidity current (Omori, 1909; Fig. 1.Bathymetric maps highlighting the research area of this study, the western Ionian Basin.Prominent tectonic features in this region as a result of the active Calabrian subduction are highlighted in (A).Location of (A) is shown in the globe, which was created using ArcGlobe.Geomorphological features and the location of telegraph cables in 1908 (white-dotted line) and cable breaks (black circles) are shown in (B) (black box in (A)), which is the defined research area of this study.Acronyms W-lobe stands for the separated western lobe of the Calabrian Accretionary Wedge and A.S. for Alfeo Seamount.The transparent red polygon in (A) indicates the location of Messina Straits.The tectonic map is modified from Gutscher et al. (2016Gutscher et al. ( , 2017)).Note: symbols indicating the location of various cities shown in (B) will be relevant for the results sections.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)Ryan and Heezen, 1965;Tinti and Armigliato, 2003).The source of both the earthquake and tsunami is argued to be somewhere in the Messina Straits, but remains highly debated in the literature (e.g., Tinti and Armigliato, 2003;Billi et al., 2008;Argnani et al., 2009a;Barreca et al., 2021).Tsunami simulations based on proposed earthquake faults do not match observations (e.g., Tinti and Armigliato, 2003;Billi et al., 2008).A submarine landslide has been suggested to have contributed to tsunami generation (e.g., Billi et al., 2008;Favalli et al., 2009;Schambach et al., 2020), but such a contribution of a landslide is doubted by others (Argnani et al., 2009b;Gross et al., 2014).The event severed two telegraph cables, which are thought to have been the result of the gravity flow (Fig. 1B) (Ryan and Heezen, 1965) that was either caused by the earthquake directly or by one or several submarine landslides.The Gallico-Gazzi cable in the northern part of the Messina Straits broke at the time of the earthquake (Fig. 1B) (Ryan and Heezen, 1965).Three additional cable breaks were recorded across the Malta-Zante telegraph cable up to 18 h after the earthquake: one at 15:15 in the external Calabrian accretionary wedge (CAW) followed by a second break at 15:45 in the same location (Fig. 1B) (Ryan and Heezen, 1965).The third break occurred at 23:20 closer to the Malta Escarpment (Fig. 1B) (Ryan and Heezen, 1965).
Since the first detailed description by Ryan and Heezen (1965), sediment cores showing evidence for the 1908 turbidity current have been collected from the western Ionian Basin (Fig. 2A).The turbidite is found in cores from a syn-tectonic basin located between the internal and external CAW and in proximity to the Ionian Abyssal Plain (IAP) (Fig. 2A) (Polonia et al., 2013(Polonia et al., , 2017(Polonia et al., , 2021)).The distribution of the cable breaks, approximate location of the earthquake, and mineral composition of sediments in the cores allowed a first estimate of the source area, extent, flow direction and speed of the 1908 gravity flow (Ryan and Heezen, 1965;Polonia et al., 2013Polonia et al., , 2017Polonia et al., , 2021)).Ryan and Heezen (1965) suggested an average flow velocity of ~6.2 ms − 1 assuming the source to be west of Gallico (Southern Calabria) within the Messina Straits (Fig. 1B).Polonia et al. (2021) concluded that the gravity flow involved multiple source areas and suggested the seafloor offshore Mt.Etna (north-eastern Sicily) and southern Calabria, although no specific location is provided.They identified a canyon system southeast from Mt. Etna as the main passageway (Fig. 1B) (Polonia et al., 2017).Here we hypothesise that the main passageway of the 1908 gravity flow was not directly connected to Mt. Etna.The previous attempts to reconstruct the 1908 gravity flow and older events were either based on low resolution bathymetry or sub-bottom and sediment core data with sparse coverage (Ryan and Heezen, 1965;Polonia et al., 2013Polonia et al., , 2017;;San San Pedro, 2016).There are large variations for proposed passageways and none of the studies identified any erosive or depositional bedforms along these passageways to support their findings (Fig. 1B).Thus, despite being one of the few known historic large run-out events, knowledge gaps regarding the source region, flow evolution and direction, and the type and scale of the erosional and depositional processes remain.
The aim of this study is to improve the current understanding about the 1908 gravity flow by inferring the: 1) potential passageways; 2) erosional and depositional bedforms along channel floors; 3) flow thickness; 4) flow velocity; and 5) potential source areas.Multiple data sets acquired over the past decade, consisting of multibeam echosounder data, sub-bottom profiles and sediment cores, have been used to address these objectives.Here we have one of the few known large run-out gravity flows associated with one of the biggest earthquakes (M w 7.1) known to have affected the Central Mediterranean Sea (e.g., Tinti and Armigliato, 2003;Polonia et al., 2013;Gutscher et al., 2016Gutscher et al., , 2017)).The absence of large-scale earthquake events (M w > 6) in this region since 1908 (Rovida et al., 2022) implies that erosional and depositional features associated with the 1908 gravity flow are likely still preserved and have not been altered significantly by younger gravity flows.The largest earthquake in the study area since 1908 was the 1990 M L 5.4 earthquake, offshore Augusta (Amato et al., 1995), which is not associated with any known tsunami or submarine landslides.The western Ionian Basin is, therefore, an ideal study site to improve our current understanding about large run-out events, especially as it is also one of the few events where information about cable breaks is available.The new findings can further be used for tsunami modelling to understand the potential involvement of gravity flows in tsunami generation.Better understanding of this gravity flow will have implications for similar events worldwide, and for better assessing marine geohazards offshore eastern Sicily.

Regional setting
Our study area comprises a part of the western Ionian Basin, from the Messina Straits down to the former location of the Malta-Zante telegraph cable, and from the Malta Escarpment to the location of the 4th cable break (Fig. 1A-B).The western Ionian Basin is a ~ 120 km-wide and ~ 280 km-long basin located in the central Mediterranean Sea south of Italy.It extends from the Messina Straits between Calabria and Sicily down to the IAP (Fig. 1A-B).The entire region is strongly affected by the Europe-African plate convergence, with the western Ionian Basin being part of the south-eastward trending Calabrian subduction zone (Anzidei et al., 1997;Goes et al., 2004;Barreca et al., 2021).This subduction zone is characterised by a complex tectonic regime in the form of coexisting compressional, extensional and uplifting deformation.As a consequence, the geomorphology of this region is strongly affected by the tectonic deformation, as apparent through differences in the accretionary wedge (Goes et al., 2004;Gutscher et al., 2017;Camerlenghi et al., 2020).The external, evaporitic wedge contains anticlines with confined sedimentary basins as a consequence of compression and Messinian salt deformation, while the more rigid internal, clastic wedge shows a mottled morphology with numerous submarine channels and basins (Fig. 1A) (Polonia et al., 2011;Gutscher et al., 2017).The internal and external CAW is separated by an up to 1 km-high escarpment, otherwise known as "the wall" (Fig. 1B) (Gutscher et al., 2017).The Alfeo Seamount (AS) and the westernmost lobe (W-lobe) of the CAW are noticeable features, as they appear as separated morphological structures within this tectonic setting (Fig. 1A-B) (e.g., Gutscher et al., 2017).
The Messina Straits comprise the northernmost part of the western Ionian Basin and is generally considered to be a northeast-southwest trending graben structure extending between north-eastern Sicily and southern Calabria (Fig. 1A) (Valensise and Pantosti, 1992).This extensional component is accompanied by regional uplift of the adjacent continental margins of northern Sicily (Peloritani Massif) and southern Calabria (Aspromonte Massif), as evident through the occurrence of marine terraces along the mainland up to 1300 m above sea level (Valensise and Pantosti, 1992).The uplift occurs at rates of 0.9 to 1.4 mm yr − 1 and can be dated back to ~1 Ma BP (Valensise and Pantosti, 1992).This ongoing uplift results in large onshore erosion and consequent high sediment discharge into onshore river systems (Goswami et al., 2014;Ridente et al., 2014).The presence of narrow (<1 km-wide) continental shelves allow these onshore river systems to almost directly discharge sediment into the canyon heads (Goswami et al., 2014(Goswami et al., , 2017;;Ridente et al., 2014).Sedimentation within the western Ionian Basin is further influenced by bottom currents (Sparnocchia et al., 2011;Micallef et al., 2016;Rebesco et al., 2021).Sediment waves at the foot of the Malta Escarpment are interpreted as a direct result of these currents (Rebesco et al., 2021).Measurements at localised mooring stations in this area showed a predominant south-south-westward directed bottom water flow with current velocities of >10 cm s − 1 (Rebesco et al., 2021).
The entire region is prone to earthquakes, many of which are associated with tsunamis (e.g., 1169,1693,1908) and sediment gravity flows, that over the past few centuries caused significant fatalities and infrastructural damage (Omori, 1909;Ryan and Heezen, 1965;Anzidei et al., 1997;Tinti and Armigliato, 2003;Polonia et al., 2013).The sources of many of these events are still debated (e.g., Tinti and Armigliato, 2003;Billi et al., 2008;Argnani et al., 2009b).Barreca et al. (2021) just recently proposed the 1908 M w 7.1 causative earthquake fault to be located within the Messina Straits and the northern portion of southern Calabria considering the distribution of the marine terraces.Different scenarios have also been proposed and tested to explain tsunami generation in this region.Simulations on the 1908 tsunami, for example, considered a submarine landslide in addition to the earthquake source.Most successful simulations considered landslides offshore Nizza (Favalli et al., 2009) or Fiumefreddo (Billi et al., 2008) of eastern Sicily and multiple smaller failures along the southern Calabria continental slope (Schambach et al., 2020) (Fig. 1B).A submarine landslide would also be a likely source for the formation of the 1908 sediment gravity flow, but a large sediment mass failure has not been identified in these regions (Argnani et al., 2009b;Gross et al., 2014).Many of the proposed landslide locations were based on morphological observations without confirmation by dating and sub-bottom or high-resolution seismic reflection data.
Multibeam bathymetry and backscatter data were used to characterise the morphology of the western Ionian Basin and to define different conduit systems and source areas for the 1908 gravity flow (Fig. 4).ArcMap was used to map geomorphological features such as the channel floor and thalweg, basins, interchannel heights, escarpments and bedforms.Backscatter data, which are a measure of the reflectivity of a surface, are representative of surface roughness, slope gradient, and sediment grain size (Augustin et al., 1996) (Fig. 3).

2D sub-bottom profiles
Sub-bottom data were acquired during research expeditions M86-2 (2011/12) and CIRCEE-HR leg 1 and leg 2 (2013) using Chirp and Atlas Parasound P70 acquisition systems (Fig. 2A).The Chirp data were acquired using a signal frequency bandwidth of 1.8 to 5.3 kHz (Gutscher et al., 2013).This system images the upper 50 to 80 m of the sub-bottom with a minimum horizontal resolution of 11 to 20 m in 1000 mwd and 20 to 40 m in 4000 mwd (Fresnel zone) (Gutscher et al., 2013).The vertical resolution of this system is ~0.75 m (San Pedro et al., 2017).The Parasound system is a parametric echo sounder operating at 4 kHz; the system allows imaging of the upper 100 m of the sub-bottom at a vertical resolution of 0.15 m (Teledyne Marine, 2017).These parametric echo sounders have a narrow opening angle of 4.5 • , which allows data acquisition with fewer side-echoes when compared to conventional systems (e.g., boomer, chirp) (Teledyne Marine, 2017).The minimum horizontal resolution is 70 to 280 m between 1000 and 4000 mwd (Spieß, 1993).Detailed information about specifications of the systems used are available in the cruise reports (e.g., Krastel et al., 2014;Gutscher et al., 2013).Sub-bottom profiler data were used to map the distribution and relationships of acoustic facies (e.g., high/low amplitude sub-bottom echoes, onlapping) according to Damuth (1975Damuth ( , 1980) ) and Damuth and Olson (2015).This information was used to infer sediment grain size, erosion, and long-term deposition in order to reconstruct erosional and depositional processes along the upper sediment strata as a consequence of recent gravity flow activity.A map showing the acoustic facies distribution was generated.We attempted to interpolate facies between adjacent profiles if the spacing between the sub-bottom profiles was not too large (<10 km) and where acoustic facies were associated with specific morphological features (e.g., sediment waves, basins, steep slopes).IHS Kingdom Suite™ was used for the data analysis of the sub-bottom data.A sound velocity of 1500 ms − 1 was used to convert from two-way travel time into metres.

Sediment cores
Sediment gravity cores were collected from the mid-to south western Ionian Basin during research expedition CIRCEE-HR (Figs. 2A, 3, Table 1) (Gutscher et al., 2013).The sediment cores were visually described in the laboratories of the European Institute for Marine Studies at the University of Brest (IUEM, France).The visual core description included a re-assessment of sediment lithology, sediment structures, stiffness and colour.Oher core analysis (e.g., density, grain size, age) carried out by and published in San San Pedro, 2016 and San Pedro et al. ( 2017) were considered in this study.The information was Fig. 4. Shaded multibeam bathymetry relief with geomorphic interpretation showing the distribution of channel floors, basins and various erosional and depositional bedforms.Coloured lines (yellow, green, orange, red and blue) show mapped canyons and channels.Note that blue lines represent a continuation of red and orange.This geomorphological map of the western Ionian Basin was generated by integrating information from the various data sets and results from Gutscher et al. (2016Gutscher et al. ( , 2017)), Polonia et al. (2013Polonia et al. ( , 2017) ) and San San Pedro, 2016.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)used to ground truth findings in the sub-bottom data and to reconstruct the depositional behaviour of the 1908 gravity flow.

Results
The western Ionian Basin is characterised by numerous, deeply incised (~100 m-deep) channel systems.Three main passageways for gravity flows, the western, central and eastern canyon-channel systems (C1, C2, C3), were distinguished based on their morphological characteristics (e.g., extension, continuity) (Fig. 4).In this study, the term "canyon" refers to conduits along the continental slopes as defined by Amblas et al. (2018).Goswami et al. (2014) previously distinguished different canyon systems for the continental slope adjacent to the Messina Straits.Here we modified this previous classification and expanded it southward of the Messina Straits (Fig. 3).The term "channel" refers to conduits extending beyond the base of the continental slope.The term "canyon-channel systems" refers to continuous systems of canyons, channels and adjacent distributaries.The term acoustic facies or facies is used to describe observations from sub-bottom profiles, while sedimentary features are described as sediment sequences with units.4.1.1. Canyons 4.1.1.1. Sicily and Malta Escarpment.The eastern Sicilian continental slope (7 • ) and Malta Escarpment (~11 • ) show various canyons that have second and third order tributaries (Fig. 5).The canyons generally widen from 250 to 2100 m towards Mt.Etna, but are narrower (500-1000 mwide) with less tributaries along the Malta Escarpment (Figs. 1; 3A, 3C).The largest canyon system within the research area is Fiumefreddo Valley, offshore Fiumefreddo (2.5 km-wide, ~150 m-deep) (Fig. 3D).Differences in the geomorphology and backscatter intensities allow us to distinguish four zones of canyon systems along the eastern Sicilian margin; termed SiS1 to SiS4 (Figs. 3,5).Individual canyons are easily recognizable within all zones except SiS1, which shows a high density of shallow, ~30 m-deep canyons.High to moderate backscatter patches are present, but generally confined to the canyon floor, while the background morphology shows low to moderate backscatter patches (Figs.3A, C, D).

Southern Calabria.
Canyon systems of the southern Calabrian slope (6-8 • ) can be distinguished into four zones, termed CS1 to CS4 (Figs. 3, 5).There are fewer, but generally broader (~800 m-wide) canyons along the Calabrian slope in comparison to the opposite eastern Sicilian slope (Figs.3A-B).Canyons in all four zones show similar incision depths (50-100 m-deep) and often have second and third order tributaries (Fig. 5).CS1 and CS4 show high backscatter patches along canyon floors (Fig. 5).CS2 and CS3 show moderate to high backscatter patches along the entire slope, which makes it difficult to distinguish single canyons from the background morphology on the basis of backscatter data alone (Figs. 3A-B, 5).

Eastern canyon-channel system
The eastern canyon-channel system (C3) extends from the northern part of the Messina Straits down to the external CAW and leads directly to the first two cable breaks of the Malta-Zante telegraph cable (Figs. 4, 6).Tributaries into C3 are almost exclusively from northeastern Sicily (SiS1 to SiS3) and Southern Calabria (CS1 to CS4).C3 is divided into three sub-systems, each showing an abundance of characteristic bedforms (Figs. 4, 8A-C).Numerous localised scarps are evident along channel walls especially along the upper part of C3 (Fig. 8).
C3-1 and C3-2 merge into C3-3, which is a southward extending channel system with numerous distributaries (Fig. 4).The majority of channels follow linear features leading in and out of two large and flat (<0.1 • ) basins (basin 2, 624 km 2 & basin 3, 324 km 2 ) (Fig. 4).A welldeveloped train of large-scale crescent-shaped depressions (up to 50 mdeep, 500-1000 m-long) is evident for 18 km along the channel floor leading into basin 2 (Figs.4B, 9).Almost all distributaries of C3-3 are hanging valleys with elevation differences of 30 to 70 m to the main channel (Fig. 6).Slope gradients of <0.2 • are observed along up to 40 km-long sections of these channels (Fig. 6).One of the hanging valleys (C3-cable) extends from basin 2 towards the cable breaks (Fig. 4).It crosses a 200 m-high knickpoint that is part of the escarpment separating the external and internal CAW (Figs. 1, 6).Downslope of the knickpoint this distributary bifurcates into a western and southern branch (Fig. 4).The southern branch leads directly to the cable breaks, while the western branch leads into C2-4 (Fig. 4).The distance from CS4 and SiS2 to the cable break location via this southern branch is

Sub-seafloor acoustic facies distribution and relationships
Six main sub-seafloor acoustic facies are distinguished and mapped across the study area following the classification scheme provided by Damuth (1975Damuth ( , 1980) ) and Damuth and Olson (2015) 13).The facies are distributed as follows: 1.  southern portion of C2-2 towards C1-1 shows six sequences of stacked sediment units within the upper 50 cm (Fig. 10).Each sequence is <10 cm-thick with a sandy base unit being <4 cm-thick.KCIR-01 collected along the channel floor of C1-1 proximal to the third cable break shows a similar characteristic as KCIR-04 (Fig. 10).There are up to five sequences of stacked sediment units in the upper 50 cm (Fig. 10).In comparison, KCIR-08 that was collected further downslope of C1-1 towards the IAP shows four >10 cm-thick sequences (Fig. 10).KCIR-03 was collected within the southern part of C2-4 in proximity to CQ14_02 described in Polonia et al. (2017) (Figs. 2A, 11).There are three different coloured units in the upper 1.09 m (T1; Fig. 11).The base of T1(Unit 1, 41-109 cm) is a massive, dark grey medium to coarse sand with a high abundance of mica, which is overlain by a 41 cm-thick mud cap (0-41 cm) (Fig. 11).KCIR-02 was collected in proximity to KCIR-03 within a basin of the external CAW (Fig. 2A).The upper 40 cm contains three sequences of stacked sediment units with a 4 cm-thick unit of dark grey to black, stiff, fine sand at the top that resembles Unit 1 of KCIR-03 (Fig. 11).

Geophysical and sedimentological indicators to reconstruct recent erosional and depositional processes
Facies A and B-1 along canyon and channel floors are associated with high backscatter patches and often but not necessarily with crescentic-shaped depressions (e.g., along SiS2 canyon floors) (Figs.3C, 9, 13A).SiS2, SiS4, CS1 and CS4 show the highest backscatter intensities along the continental slope (Figs. 3, 5).River systems discharge high amounts of sediment from onshore erosion and flash floods almost directly into the canyon heads of these systems (Goswami et al., 2014;Ridente et al., 2014).We interpret crescentic-shaped depressions such as those observed along SiS2 and C3 channel floors as scours (Fig. 9) (cf., Paull et al., 2014;Slootman and Cartigny, 2020).Scours are described to often contain coarser grained material (e.g., Hage et al., 2018).These net-erosional cyclic steps are observed along canyon and channel floors worldwide (e.g., Symons et al., 2016;Covault et al., 2014Covault et al., , 2017)).They are widely interpreted as a result of erosion through bypassing turbidity currents as cyclic hydraulic jumps cause a shift from Froude supercritical to subcritical flow conditions (e.g., Fildani et al., 2006;Covault et al., 2014;Slootman and Cartigny, 2020).Facies A and B-1 with high backscatter patches are, therefore, interpreted to represent either coarse grained material or small to medium sized erosional or depositional bedforms as a result of high energy events such as river discharges, storm surges or gravity flows (cf., Damuth, 1975Damuth, , 1980;;Goswami et al., 2014Goswami et al., , 2017)).Bedforms of <30 m in size, however, may not be resolved in the bathymetry data due to restrictions in the horizontal resolution (30 m).Facies B-2 reflections are caused by a diffraction of the acoustic signal often associated with steep slopes and rough morphology such as channel walls and scarps found along C3 (c.f., Damuth and Olson, 2015).
Facies C and Facies D are often associated with low to moderate backscatter patches and are abundant across basins and along channel floors of the southern part of the western Ionian Basin (Figs. 3,11,13B).Facies C reflections and this type of backscatter intensities can be interpreted as coarse-grained sediments with high silt and sand contents deposited from high velocity currents (Damuth, 1980).We interpret Facies C, therefore, as a result of deposition from a diluted debris flow or high-density turbidity current given the distance from potential source areas and distribution of this facies (Fig. 13).Sediment cores collected from regions of Facies D contain stacked units of different sediment colour and grain size that are identified as turbidites (Fig. 11) (cf., Köng et al., 2016;Polonia et al., 2013Polonia et al., , 2017;;San Pedro et al., 2017).We, therefore, interpret that Facies D associated with low to moderate backscatter patches represents stacked turbidites in the research area (e. g., Damuth, 1975Damuth, , 1980)).
Facies E occurs locally throughout the research area in vicinity to steep slopes.This facies is typical for mass transport deposits (MTD) (e. g., Damuth, 1975Damuth, , 1980)).Noticeable is a ~ 10 m-thick sheet-like Facies E deposit along the lower slope of CS2, which suggests sediment failures along the upper slope (Fig. 13A).The absence of an overlying acoustic facies (Facies C or D) indicates fairly recent deposition of this MTD.It is, therefore, likely a result of the last known large event (M w > 6) affecting the region, which was the 1908 earthquake (Rovida et al., 2022).Facies F, which occurs as straight to sinuous ridges and troughs along the present seafloor, is interpreted as sediment waves deposited either from bottom currents or turbidity currents or a combination of both (Figs. 7, 13; 14) (e.g., Damuth and Olson, 2015;Rebesco et al., 2021).
A decrease in the abundance of turbidites within a sequence of downslope sediment cores (KCIR-04, KCIR-01, KCIR-08) supports our finding (Fig. 10).High backscatter patches of Facies B-1 and E along the foot of the Malta Escarpment indicate the deposition of coarse-grained sediment (Fig. 3E, 13) (cf., Damuth, 1975Damuth, , 1980)).We therefore interpret turbidites found in sediment cores along C1-1 as a result of sediment failures and consequent gravity flows from Malta Escarpment and Alfeo Seamount rather than from massive gravity flows passing from Mt. Etna through C1-1 (Fig. 4).These findings are corroborated by findings from Spatola et al. (2020) who showed several large sediment failures across the Malta Escarpment.Sediment waves (Facies F) at the foot of the Malta Escarpment along the southern portion of C1-1 have been described to be either a result of turbidity or bottom currents (Fig. 4, 13B, 14) (Gutscher et al., 2016;Micallef et al., 2016;Rebesco et al., 2021).Rebesco et al. (2021) interpreted them as a result of bottom currents rather than large-scale turbidity currents given the high bottom current velocities (>10 cm s − 1 ) and absence of progressive downslope thinning and decrease in dimension (Fig. 14).We agree with the observations by Rebesco et al. (2021) and that bottom currents may be the main factor forming these sediment waves.The presence of sediment failures along the Malta Escarpment and abundance of turbidites in sediment cores down to the IAP, however, proofs the occurrence of turbidity currents within C1-1 (KCIR-04, KCIR-01, KCIR-08) (Polonia et al., 2013(Polonia et al., , 2017;;Köng et al., 2016;San Pedro et al., 2017).It suggests that turbidity currents evolving from these failures may have influenced the development of the sediment waves (Facies F) in addition to bottom currents (Figs.3E, 13B) (cf., Migeon et al., 2001;e.g., Gutscher et al., 2016).Large-scale gravity flows (>15 km-wide) emerging from an amalgamation of several failures along eastern Sicily and the Malta Escarpment possibly have enough energy to reach the IAP.

Activity of C2
C2 is interpreted as a net-sink for gravity flows rather than a system that facilitates prolonged sediment transport.It is the only system, however, which is not directly connected to canyons from eastern Sicily or Calabria (Fig. 4).Gravity flows from C1 and C3 need to be >30 mthick in order to bypass the 30 to 60 m-high hanging valleys leading into C2 (Fig. 6).Sub-bottom profiler data indicate that the upper part of C2 is dominated by coarser grained sediment material (Facies C, C2-1) and up to 20 m-thick MTDs (Facies E, C2-2 & C2-3) (Fig. 13B).The sediment appears to be finer grained downslope towards C1-1 and C2-4, as there is a change in acoustic facies from Facies E and C into Facies D (Figs. 11,13B).Sediment cores from the southern part of C2-4 contain stacked units of turbidites (Fig. 11) (cf., Köng et al., 2016;Polonia et al., 2017;San San Pedro, 2016).Seafloor gradients are <0.1 • for >20 km along the up to 6 km-wide channel floor (Fig. 6).These findings suggest that gravity flows undergo flow transformation within C2.We interpret that the low seafloor gradient, presence of seafloor relief (up to 80 mhigh) along the channel floor and absence of confinement decelerate gravity flows entering C2, especially those that continue down C2-4 (cf., Talling et al., 2007;Wynn et al., 2012;Gavey et al., 2017).This flow deceleration results in deposition from the flow and consequent flow transformation either from a debris flow or higher density turbidity currents into more dilute turbidity currents (cf., Damuth, 1975Damuth, , 1980;;Talling et al., 2007).Sediment waves (Facies F) along C2-4 with a downslope decrease in dimension are thus interpreted as turbidite sediment waves (Fig. 14) (cf., Rebesco et al., 2021).The ~20 m-high vertical offset along these sediment waves is part of a fault system (Figs.1A, 14) (e.g., Gutscher et al., 2016).Turbidity currents crossing this offset likely accelerate as a result of hydraulic jumps (cf., Chen et al., 2021).Sediment waves are redeveloped at the foot of the fault indicating rapid flow deceleration and consequent sediment deposition (Fig. 14).The MTDs (Facies E) apparent along C2-2 and C2-3 are interpreted as a result of sediment failures along the Western Lobe or Alfeo Seamount give their location along linear faults of the external CAW (Figs. 1A, 4).
3. Coarse-grained sediment deposits (Facies B-1 with high backscatter patches) along the channel floor and in proximity to scours.Validation from sub-bottom profiles in areas with high backscatter patches is not always possible.In these areas high backscatter patches may represent coarse grained material from the deeper subbottom rather than the seafloor, given the probability that 12 kHz backscatter data image sediment down to 4 m below the seafloor (Hillman et al., 2018).4. Localised scarps along channel walls that indicate sediment failures caused either by ground shaking or slope undercutting from bypassing gravity flows (Fig. 8) (e.g., Goswami et al., 2014).
C3 is deeply incised and narrow compared to the other two canyonchannel systems, which allows a relatively long confinement of sediment gravity flows until basin 2 (Fig. 4).Despite the occurrence of scours (25 m-deep, up to 1.9 km-long) below the confluence with C1-2, the overall low backscatter patches and presence of scattered, small to medium sized scours (~10 m-deep, 200-300 m-long) in the lower portion of C3-1 suggest that this channel was not affected by recent gravity flows (cf., Slootman & Cartney, 2020) (Figs. 3, 9).A higher abundance of erosional and depositional features along C3-2 such as large-scale trains of scours (~25 m-deep, 1-3.9 km-long) are interpreted either as a result of repeated gravity flow events or a massive and recent gravity flow (Figs. 8,9).Sediment flows probably quickly decelerate due to the low and unconfined relief of basin 2, resulting in deposition of coarse to fine-grained sediments, as indicated by the presence of low and localised high backscatter patches (Fig. 3) (cf., Augustin et al., 1996;Wynn et al., 2012).Only high energy and large gravity flows (>60 m-thick) would be able to enter the ~60 m-high hanging valleys that extend downslope from basin 2 (Fig. 6).It is possible that the increase in slope gradient from basin 2 (<0.1 • ) to almost double (<0.2 • ) within hanging valleys may have been sufficient to cause flow acceleration, erosion and subsequent flow confinement as shown elsewhere (e.g., Wynn et al., 2012).This change in flow behaviour would indicate that the flow was able to keep most of its fine-grained sediment load carrying and depositing it further downslope within C2-4 and explains the thick mud cap observed on top of KCIR-03 Unit 1 and CQ14_2 Sta1 (Figs. 6, 11) (cf., Wynn et al., 2012).It also would explain the abundance of numerous distributary channels extending downslope from basin 2 (Fig. 4).The scours within C3-2 and C3-3 appear relatively fresh, showing sharp, defined edges that do not appear to be masked by hemipelagic deposition, which indicates that those bedforms were formed or altered by a recent high energy event (Figs.8; 9) (cf., Paull et al., 2014;Slootman & Cartney, 2020).

Evidence for the 1908 turbidite in the middle to southern western Ionian Basin
Polonia et al. (2013, 2017, 2021) reported the 1908 turbidite in various sediment cores collected along C1-1 near the IAP and C2-4 based on radiocarbon dates (Fig. 2A, e.g., CALA-04, CQ14_02).Turbidites in the CIRCEE-HR cores resemble the turbidites described by Polonia et al. (2013Polonia et al. ( , 2017) ) (Fig. 11).San San Pedro, 2016 andSan Pedro et al. (2017), however, also using radiocarbon dates classified the uppermost turbidite in the CIRCEE-HR cores resulting from either the 1693 southern Sicily earthquake (e.g., KCIR-08) or 1444 Etna eruption (KCIR-02) (Fig. 11).Given the strong similarities (colour, thickness, grain size) and proximity of the KCIR-03 core to the CQ14_02 core of Polonia et al. (2017), we surmise that T1 of KCIR-03 and Sta1 of CQ14_02 may have resulted from the same event (Figs.2A, 3, 11).We found that the upper deposit in KCIR-02 resembles Unit 1 of KCIR-03.These turbidites are likely related to 1908 rather than 1693 or 1444, given the proximity to the 1908 cable breaks and the absence of overlying turbidites (Fig. 11) (cf., Ryan and Heezen, 1965;Polonia et al., 2017).If it would indeed be the 1693 deposit, then this would suggest that the 1908 turbidity current was either too powerful, meaning it must have been highly erosional without leaving any deposits within a basin otherwise characterised by long term turbidite deposition, or it took a different passage.Both scenarios are unlikely, as in this case: A) the 1693 deposit would have likely been eroded; and B) cable breaks occurred downslope of these cores, which shows that the gravity flow must have bypassed this location.In addition, the 1444 event is not recorded in any of the cores from Polonia et al. (2013Polonia et al. ( , 2017)).
The discrepancy in the age estimates could be associated with uncertainties (e.g., marine reservoir effect, analytical precision, sample contamination) in the radiocarbon dating used by the different groups of researchers (Lowe et al., 2007;Urlaub et al., 2013).In addition, dating of turbidites is often difficult as an undisturbed hemipelagic sample is needed, but may not be available or too thin (<1 cm) between stacked turbidites to allow proper sampling (e.g., Urlaub et al., 2013).Another problem may be that gravity cores not always sample the uppermost sediment layers (Morton and White, 1997;Skinner and McCave, 2003).It could therefore be argued that the 1908 turbidite was not or only partly sampled in the CIRCEE cores (e.g., KCIR-08) and that in both cases the dates are correct.We disregard this scenario for KCIR03, given that the upper 2 m show the same pattern (e.g., number and colour of turbidites) as CQ14_02 (Polonia et al., 2017) (Fig. 11).

The main route of the 1908 gravity flow
The interpreted 1908 turbidite is thickest (up to 1.34 m thick) within C2-4 and in proximity to the first two cable breaks of the Malta-Zante telegraph cable (Figs. 4,11) (Ryan and Heezen, 1965;Polonia et al., 2013Polonia et al., , 2017;;San San Pedro, 2016).This basin is supplied with sediment from both C1 and C3 via numerous possible passageways (Fig. 4).C3 with C3-2 and C3-3 was a main passageway for past events as indicated by the abundance of erosional and depositional bedforms (e.g., scours, turbidite sediment waves) (Figs. 4,6,8,9).Although the age of the scours within C3-2 and upslope of basin 2 is unknown, they appear fresh, which suggests that they were affected recently by the passage of a large and erosive gravity flow (Figs. 3, 9) (cf., Paull et al., 2014;Slootman & Cartney, 2020).In addition, there are moderate to high backscatter patches within basin 2 along paths leading to various hanging valleys.These backscatter patches may be a result of various past gravity flows, but their abundance highlight the activity of C3 as a passageway for large sediment flow events.The 1908 event is the most recent known large event in the study area and C3-3 leads directly to the location of the first two cable breaks (Fig. 4, C3-cable) (cf., Tinti and Armigliato, 2003;Polonia et al., 2013).C3 (C3-2 & C3-3) and possible C2 or parts of it are, therefore, interpreted as the main passageway for the 1908 gravity flow.San San Pedro, 2016 previously suggested similar passageways to the ones presented in this study on the basis of bathymetry and sediment core data, but for events predating 1908.Using XRF and grain size analysis San San Pedro, 2016 and San Pedro et al. ( 2017) Fig. 13.Acoustic facies distribution of the uppermost sub-bottom reflections along the western Ionian Basin based on the classification scheme of Damuth (1975Damuth ( , 1980) ) and Damuth and Olson (2015).It has to be noted that Facies B-2 covers almost all of the Malta Escarpment and sides along canyons and channels.This facies was not interpolated across these regions, especially the canyon and channel walls, as it would have distracted from the other facies.Black dotted lines show the location of sub-bottom data used within this study.Examples of sub-bottom profiles are shown in Figs.10-12.concluded that past-turbidites in cores from C2-4 originated from Messina Straits, which supports our interpretation of C3 as the main passageway.
The cable breaks occurred with a delay of ~30 min suggesting that the gravity flow split across different passageways ().Using the distance from potential source areas (CS4 and SiS2) to the first two cable breaks via C3 (197 to 222 km) and the timing of the cable breaks (~9 h) provides an average flow velocity of 5.6 to 6.3 ms − 1 (Figs.4, 6).It is possible that knickpoints along C2 and C3 caused a hydraulic jump with flow acceleration given the sudden change in slope angle and height differences of up to 180 to 300 m (Fig. 6) (cf. Chen et al., 2021).This reacceleration also may explain that the gravity flow had enough power to cause the cable breaks, despite the potential split along different passageways (cf., Wynn et al., 2012).
The third cable break occurred ~18 h after the earthquake (23:20 pm) and nine hours after the first two cable breaks (Ryan and Heezen, 1965).Different scenarios to explain its occurrence have been put forward: A. Direct passage from Mt. Etna (SiS4) through C1-1 (Fig. 15B).This scenario would match suggestions by San San Pedro, 2016 and Polonia et al. (2013Polonia et al. ( , 2017)), who identified C1-1 as passageway for 1908 and previous events.It also matches findings from Billi et al. (2008) and Tinti and Armigliato (2003), who suggested a submarine landslide off Fiumefreddo to explain the observed tsunami run-ups.If such a landslide did occur, it would be a likely source for the 1908 gravity flow (e.g., illustrated elsewhere by Piper et al., 1999;Talling et al., 2012Talling et al., : 2013)).Seismic reflection data, however, show no indication for a recent landslide within the proposed failure area offshore Fiumefreddo (Argnani et al., 2009b;Gross et al., 2014).Moreover, the distance (~205 km) and the timing of the cable break suggests an average flow velocity of ~3 ms − 1 (Fig. 6).This velocity would likely be too slow to cause cable breaks (Gavey et al., 2017;Talling et al., 2022).B. Multiple localised and delayed sediment failures along the Malta Escarpment, the Western Lobe and Alfeo Seamount as a result of aftershocks or the impact of the tsunami wave causing secondary turbidity currents (Fig. 15C).Polonia et al. (2021) previously suggested localised failures along the Malta Escarpment to explain the tsunamite cap found in sediment cores.It is possible that both the tsunami deposit and up to 13 cm-thick sandy unit of the 1908 turbidite that underlies the tsunami unit were caused by these delayed, localised turbidity currents.C. Overflow of the turbidity current that also caused the first two cable breaks via the external CAW (Fig. 15A).Sub-bottom data (Facies D) and turbidites found in KCIR-02 provide evidence that past turbidity currents deposited sediment along large portions of the accretionary wedge and likely also the 1908 turbidite (see 5.3.1,Figs.11, 13B).
The question remains whether a turbidity current crossing the external CAW with its basin-ridge morphology would have still been powerful enough to cause the third cable break (Fig. 15A).D. The timing of breaks along the Malta-Zante telegraph cable are wrong.This possibility cannot be ignored as the only information about the timing of cable breaks is provided by Ryan and Heezen (1965) with the source about the cable repair stated as "unpublished cable repair records".A change in the sequence of cable breaks could imply that the source for gravity flows along C1 may have been indeed further upslope and closer to Mt. Etna.A main passage through C1, however, contradicts the morphological findings presented in this study, which strongly suggest C3 as main passageway for the 1908 gravity flow.

Potential source region
North-eastern Sicily and southern Calabria (or portions of it) were previously identified as potential source regions (Ryan and Heezen,  , 9).These bedforms could be a result of erosion from hyperpycnal river discharge and consequent dilute turbidity currents or sediment destabilisation in 1908 (e.g., Casalbore et al., 2011;Goswami et al., 2014).An absence of sub-bottom and sediment core data, however, does not allow to clearly identify these bedforms (Figs.2A, 12).2) SiS4 canyons merge into C1-1 and partially C3-1 via C1-2 from Fiumefreddo Valley (Fig. 4).C3-1 shows no indication for the passage of a recent and massive gravity flow, while sediment transport within C1 appears limited (<25 km) in its downslope extent (Figs.3D-E, 4).Multiple failures, however, likely occurred along various canyons of the Malta Escarpment (SiS4) as a consequence of the impact from the 1908 tsunami or aftershocks (Fig. 15, see 5.3.2) (cf., Arai et al., 2013;Polonia et al., 2017).3) CS1 shows numerous scarps resulting from slope failures, which may be potential sources for the 1908 gravity flow (e.g., Goswami et al., 2014).Monaco and Tortorici (2000) reported that the 1908 earthquake caused several landslides between Reggio Calabria and Scilla that are part of the catchment area of CS1.The occurrence of numerous landslides onshore may indicate that there were also sediment failures offshore.A cable break offshore Gallico was reported at the time of the earthquake (Fig. 1B) (Ryan and Heezen, 1965).It is further suggested that the 1908 causative fault extends from Messina Straits to the northern part of southern Calabria (CS1) (Barreca et al., 2021).4) The presence of MTDs (Facies E) along the lower slope of CS2 suggests recent sediment failure along the upper slope of CS2 and CS3 (Fig. 13B) (see 5.1).This observation is consistent with Goswami et al. (2014), who identified slides within CS2 and CS3, but interpreted high backscatter intensities within CS2 and CS3 as bedrock outcrops.Omori (1909) further reported onshore landslides between Pellaro and Lazzaro located within CS2 and CS3.Sparse data coverage along the upper slope of CS2 and CS3 does, however, not allow to identify a head scarp.5) CS4 shows high backscatter patches but restricted to the canyon floors (Fig. 3B).No major scarp can be identified and the absence of sub-bottom profiles does not allow to indicate any MTDs along the lower slope.
Given above arguments, we interpret that CS2 and potentially CS3 are the most likely areas for mass failure in 1908 and thus the most likely main source region for the gravity flow.CS1 and CS4 can not be discarded as 1908 source regions given the occurrence of cable breaks within CS1 and sparse data coverage along CS4.The fault system suggested by Barreca et al. (2021) would explain sediment failure along CS1, CS2 and CS3 as a consequence of the 1908 earthquake.Additional sediment cores from C3 and further geochemical analysis on existing sediment cores from Messina Straits and the depositional area are needed for a better understanding of the exact source area or areas.

Potential flow thickness and transformation
The height of the backscatter patches within C3-2 and the presence of overspill turbidite sediment waves on interchannel heights suggest that past sediment flows exceeded a thickness of >170 m within this system (Fig. 6) (cf.,Stevenson et al., 2018).We interpret that the 1908 gravity flow was also >170 m-thick given its large run-out and destructive nature >200 km southwards and the distribution of erosional and depositional bedforms (Figs. 7,9).The presence of scours along C3-2 further indicates that the 1908 flow must have had already a suspended upper layer as needed to form these features (Fig. 8) (cf., Slootman & Cartigney, 2020).Scours towards basin 2 have a similar length to those observed further upslope within C3-2, but are more deeply incised, which may imply formation by a more erosive gravity flow (Fig. 9).This finding suggests flow transformation between these two locations.Two scenarios are possible: 1) The gravity flow incorporated coarse-grained sediment from the channel floor turning it into a denser and more erosive flow, which explains the deeper eroded scours (cf., Covault et al., 2017); or 2) it transformed from a higher density current with a thin turbulent layer into a lower density current with a thicker turbulent layer that somehow had a highly erosive, dense base layer.The gravity flow would have needed to be >60 m-thick to overspill into the hanging valleys surrounding basin 2 to continue its downslope movement towards the first two cable breaks (Figs. 4,6).It likely accelerated passing from basin 2 (<0.1 • ) into the hanging valleys (<0.2 • ) as a result of increased gradients, which allowed it to keep most of its fine-grained sediment load in suspension (Fig. 6) (cf., Wynn et al., 2012).In the vicinity of the first two cable breaks, the turbidity current must have been >140 m-thick in order to reach their location along the external CAW (Fig. 6).It must have been a fully developed turbidity current as evident through turbidites found in sediment cores along C2-4 (Fig. 11).
A > 60 m-thick gravity flow within basin 2 indicates that part of the gravity flow entered into C2 (C2-1).The distribution of echocharacteristics (Facies C and E into Facies D) suggests flow transformation from a debris flow or high-density turbidity current into a more dilute turbidity current (Fig. 13) (cf., Piper et al., 1999).Gradual downstream sediment deposition from the different flow types are evident through the acoustic facies changes and sediment cores collected within the elongated, 6 km-wide and almost flat (<1 • ) basin of C2-4 (Fig. 11:) (cf., Chen et al., 2021).

Conclusions
We identify three main canyon-channel systems within the western Ionian Basin offshore Eastern Sicily; the western (C1), central (C2) and eastern (C3) system.These systems show different activity in terms of sediment transport.C1 is a highly active system in terms of sediment influx, but transport appears limited <25 km downslope from the canyons due to a lack in confinement along the main channel floor.C2 is interpreted as the main depositional centre for sediment flows.It shows a long turbidite record.C3 shows numerous erosional and depositional bedforms (e.g., scours, turbidite sediment waves, channel wall collapse).These bedforms appear fresh, indicating that they were formed or at least altered by a more recent gravity flow event.The last known big event in this region was the 1908 Messina turbidity current.C3 is therefore interpreted as the main passageway for this gravity flow.It also directly leads to two out of three cable breaks (Ryan and Heezen, 1965).A direct flow from the Mt.Etna area is unlikely given the missing confinement and sequence of cable breaks.Multiple sediment failures along north-eastern Sicily and southern Calabria due to the initial earthquake, and/or secondary failures along Malta Escarpment due to aftershocks or impact from the tsunami wave, are needed to sufficiently explain the sequence of observed cable breaks and distribution of the 1908 turbidite.
The following characteristics about the 1908 gravity flow are derived: 1) Source region: Southern Calabria was the most likely source region based on gravity core comparison, morphological characteristics and sub-bottom profile analysis.The Calabrian slope (CS2, CS3) off San Leo and Bocale shows up to 10-m thick MTDs along the seafloor, making this area the most likely main source area for 1908.Northeastern Sicily (SiS2), however, cannot be fully excluded as additional source region.
2) Flow thickness: At least 170 m near the source region, and > 140 mthick upon reaching the location of the cable breaks.3) Flow evolution: The gravity flow likely already had a suspended second layer near the source area in the upper portion of C3, as indicated by the observation of turbidite sediment waves on adjacent interchannel heights.Deeply eroded, large-scale scours (up to 50 mhigh, 500-1500 m-long) towards basin 2 indicate that the gravity flow had potentially evolved into a denser flow such as a debris flow or high-density turbidity current.It is possible that the initial gravity flow incorporated sediment from the channel floor towards basin 2 (<0.1 • ) turning it into a more erosive flow.The flow potentially also accelerated within subsequent hanging valleys (<0.2 • ) allowing it to carry most of its fine-grained load into C2-4.Towards the third cable break the flow must have been a fully developed turbidity current as evident through turbidites found in sediment cores.4) Flow velocity: Estimated to be 5.6 to 6.3 ms − 1 based on distance to the cable and timing of the breaks, which agree with previous estimates of Ryan and Heezen (1965).

Declaration of Competing Interest
The authors declare that they are not aware of any financial or personal relationships that pose a potential conflict of interest for this reserach.

Fig. 2 .
Fig. 2. A) Various data sets acquired across the western Ionian Basin used in this study consisting of multibeam echosounder data, sub-bottom profiles and sediment cores with the background bathymetry (grey shaded relief) from Gutscher et al. (2017).Important landmarks and the location of telegraph cables in 1908 are shown.B) Overview map showing the location of examples shown in other figures as highlighted by the different coloured and outlined boxes.

Fig. 3 .
Fig. 3. Backscatter data covering the western Ionian Basin with locations of telegraph cables (white dotted lines), cable breaks (black dots) and sediment cores.Green dots represent cores with confirmed 1908 turbidite, yellow dots are sediment cores without sufficient information, red dots are described to not contain the 1908 turbidite.Labelled sediment cores are discussed in this study.Labelled blue and red lines along Sicily (SiS1 to SiS3) and Calabria (CS1 to CS4) indicate different zones of canyon systems.White dotted boxes (A-E) are locations of zoomed sections shown to the right.Blue arrows indicate the distribution and interpreted direction of flows with different sizes according to backscatter intensity.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 5 .
Fig. 5. Canyon systems along the Sicilian and Calabrian continental slope divided into different zones according to geomorphological characteristics and backscatter intensities.Examples representative of the characteristics of each system are shown to the right.See Fig. 2B for location of examples (1-8).Note that backscatter examples, to the right of the bathymetry examples, image the same location.

Fig. 6 .
Fig. 6.Geomorphological profiles starting from different source regions (e.g., SiS2) that follow the identified canyon-channel systems (C1, C2, C3) downslope to breaks recorded along the Malta-Zante telegraph cable in 1908.The distance from different potential source areas to the cable breaks are provided along the coloured horizontal arrows.Geomorphological features and changes in the slope gradients are highlighted.Coloured vertical dotted arrows show confluences to other profiles.Inlet (a) shows the trains of crescentic-shaped depression (csd) within canyons of SiS2 and (b) shows acoustic Facies E below C3-knickpoint (see section 4.2).The dotted red line in (b) highlights the lower boundary of Facies E. For profile locations see dotted lines with system names (e.g., C1) shown in the map to the right.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 7 .
Fig. 7. Location and characteristics of ridges and troughs found within different conduit systems of the western Ionian Basin.Images to the right show morphological examples, sub-bottom profiles (where available) and 2D geomorphological profiles for each bedform.For location of each example (I-VI) see Fig. 2B, and for the location of profiles see dashed lines on the morphological image to the left.

Fig. 8 .
Fig. 8.Comparison of bathymetry (A), backscatter intensity (B) and interpreted morphologic regions (C) in the eastern Messina Straits (C3).(A & C) highlight the distribution of different geomorphological features, (B) shows the distribution of high backscatter patches and (C) the interpretation such as location of deeper thalweg incisions.For location within the research area see Figs. 2B and 3.

Fig. 9 .
Fig. 9. Location and characteristics of crescentic shaped bedforms that are deeper than the average seafloor found along canyon and channel systems of the western Ionian Basin.Images to the right show morphological examples, sub-bottom profiles (where available) and 2D geomorphological profiles for each bedform.For location of each example (a-e) see Fig. 2B, for location of profiles see dashed black line on morphological image to the left.

Fig. 10 .
Fig. 10.Sub-bottom profiles showing acoustic facies distribution along C1.Profile locations are highlighted by numbers 1-6 and blue bars on the central overview map, all are W-E oriented.The acoustic facies in the uppermost parts of each profile is highlighted.Note: Location of Facies F is shown as dotted line over each profile where present.Circee sediment cores (KCIR04, 01, 08) are displayed on the right and their locations are shown as dots with letters A-C on the overview map.Orientation is downslope, showing a decrease in the abundance of turbidites in the upper 50-70 cm of the cores.Orange arrows show the base of each turbidite, which is in accordance to findings by San San Pedro, 2016.Pink stars show successful attempts of San San Pedro, 2016 to extract samples for dating with ages shown to the right.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 11 .
Fig. 11.Sub-bottom profiles showing acoustic facies distribution along C2 and especially C2-4.Location of profiles are highlighted by numbers 1-3 and red bars on the central overview map.The acoustic facies in the uppermost parts of each profile is highlighted.Red arrows in (3) highlight onlapping of reflections towards the basin sides evident in C2-4.Circee sediment cores (KCIR02, 03) and CQ14_2 modified from Polonia et al. (2017) are compared to the right.Core locations are shown as dots with letters A-C on the overview map.The base of grey to dark grey sandy units are highlighted with orange arrows.KCIR-03 and CQ14_02 are directly compared as highlighted by orange lines.Zoomed section shows Unit 1 of T1 of KCIR-03.This unit appears as dark grey, coarse sand that shows a high amount of mica.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 12 .
Fig. 12. Sub-bottom profiles showing acoustic facies distribution along C3.Location of profiles are highlighted by numbers 1-6 and black bars on the left overview map.The acoustic facies in the uppermost parts of each profile is highlighted.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 14 .
Fig. 14.Comparison of sediment waves apparent in the research area.A) 2D seismic reflection data imaging sediment waves along the southern part of C1-1, image modified from Rebesco et al. (2021) and B) Sub-bottom profile showing sediment waves apparent along C2-4.Note that in both cases sediment waves tilt upslope as highlighted by the blue dotted lines.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 15 .
Fig. 15.Different scenarios showing different passageways and source regions for the 1908 gravity flow.Note: Scenario 1 predicts similar source regions and passageways as suggested by San San Pedro, for the 1693 event leading to deposition in vicinity to the first two cable breaks.

Table 1
List of CIRCEE-HR sediment cores used in this study.