Submarine landslides and canyons: slope degradation and shelf-edge indentation in a tectonically active margin (Finale Basin, Northern Sicilian Margin, Tyrrhenian Sea)

Gamberi, F.; Scacchia, E.; Ferrante, V.

doi:10.3389/feart.2025.1594311

ORIGINAL RESEARCH article

Front. Earth Sci., 05 January 2026

Sec. Quaternary Science, Geomorphology and Paleoenvironment

Volume 13 - 2025 | https://doi.org/10.3389/feart.2025.1594311

This article is part of the Research TopicSculpting Processes in Shelf-Margin Canyons: Implications for Margin Architecture, Resource Potential and Environmental ManagementView all 5 articles

Submarine landslides and canyons: slope degradation and shelf-edge indentation in a tectonically active margin (Finale Basin, Northern Sicilian Margin, Tyrrhenian Sea)

F. Gamberi¹*

E. Scacchia²

V. Ferrante¹

¹ISMAR–Istituto di Scienze Marine, Consiglio Nazionale delle Ricerche, I, Venezia, Italy
²ISPRA-Italian Institute for Environmental Protection and Research, Rome, Italy

Submarine canyons and landslides are closely related geological features of continental margins. They prevail on the continental slope and often reach the shelf edge, indenting the continental shelf. The character of the linkage between submarine canyons and landslides varies depending on their geological and geodynamic setting. Consequently, a wide range of geomorphologies and processes develops in association with landslides and canyons. They reflect different evolutionary pathways that ultimately lead to continental margin degradation. The latter is studied in our article through the analysis of multibeam bathymetric data and seismic lines in the Finale Basin in the northern Sicilian Margin. We identify landslides with complex histories and characters, with their main scarps located on the slope or indenting the shelf edge, often associated with debris flows. More importantly, we infer that depending on the landslide’s style, two types of canyons develop. “Worm canyons” nucleate in the lower slope and develop through successive small rotational slides migrating upslope along narrow corridors. “Racket canyons” develop in the depressions created by larger rotational slides in the upper slope and evolve through further downslope excavation due to debris-flow processes. Both canyon types can result in shelf-edge indentations. The latter are sometimes enhanced by duck-foot bowls, triangular depressions that form in the upper reaches of some of the canyons. The observations in our study area suggest that the northern Sicilian Margin is responding to out-of-grade conditions through canyon and landslide formation due to continental margin uplift and tilting. We highlight that canyon nucleation can occur in either the lower or the upper slope. We also show that, regardless of canyon type, debris flows are the most important processes for slope excavation, canyon enlargement, and canyon propagation. Finally, we observe a hierarchical pattern in the development of the elements and processes that lead to shelf-edge indentation. Relatively large indentations form coincidentally with a canyon-generating landslide; smaller indentations relate to bowl-shaped source areas of channelized debris flows.

1 Introduction

Submarine canyons and landslides are the most important products of the continental margin degradation. Recent studies show that canyons have a wide range of morphologies and provide important generalizations concerning the controls on canyon genesis and evolution at the margin scale (Harris and Whiteway, 2011; Harris et al., 2014; Bührig et al., 2022b). In addition, numerous articles have documented the importance of local factors, such as tectonics, oceanographic parameters, distance from the coastline, and gradient, in modifying the canyon’s nature at a local scale (Casalbore et al., 2011; Puga-Bernabéu et al., 2013; Gamberi et al., 2015; Kelner et al., 2016; Smith et al., 2018). In addition to canyons, landslides are widespread geomorphic elements of offshore regions. They have a wide range of dimensions and character and are a major process contributing to the erosional degradation of continental margins worldwide.

Pioneering marine geology studies have recognized the important role played by erosion and landsliding in submarine canyon development (Shepard, 1933; Daly, 1936). Since then, two distinct models have been proposed, linking the evolution and progressive growth of canyons, respectively, to the upslope or downslope propagation of erosional processes. The upslope erosion model implies that mass failure on the lower slope promotes submarine canyon initiation and is followed by retrogressive erosion, leading to canyon shelf indentation (Twichell and Roberts, 1982; Farre et al., 1983). The downslope erosion model proposes that an upslope mass movement initiates canyon excavation and then fosters the downslope propagation of erosion through sediment-gravity flow focusing (Pratson et al., 1994; Krastel et al., 2001; Green et al., 2007; Krastel et al., 2011; Rise et al., 2013). Accordingly, the upslope erosion model implies that canyon landward propagation occurs mainly through mass movements, whereas the downslope erosion model assigns a predominant role to the erosional behavior of downslope gravity flows (Wan et al., 2022). The two models can also combine during the evolution of single canyons, with initial phases of canyon formation dominated by headward erosion, followed by prevailing downslope erosion in the mature canyon stage (Puga-Bernabéu et al., 2011; Micaleff et al., 2014; Bührig et al., 2022b; Harishidayat et al., 2024).

Various authors have recently classified canyons based on the position of their heads (Harris and Whiteway, 2011; Bührig et al., 2022b; Harishidayat et al., 2024). Shelf-incising canyons have heads that cut across the shelf break and are associated with landward-deflected isobaths on the continental shelf. In contrast, blind canyons have heads wholly confined to the slope, below the shelf break. Fluvial processes play an important role in the evolution of many canyons that incise the continental slope and face rivers (Bührig et al., 2022a; Bührig et al., 2022b). In these cases, downslope-eroding sediment gravity flows, sometimes linked to hyperpycnal flows, can materially contribute to canyon excavation (Casalbore et al., 2011). Downslope-eroding sediment gravity flows, connected with rivers, could have been more effective in canyon development in the past. During past low-stand periods, incised valleys on the shelf and river mouths, at or close to the shelf edge, could have contributed to the formation of canyons with heads that are now stranded at the shelf edge or on the upper slope (Baztan et al., 2005). On the other hand, the link between retrogressive landslides and canyons is evident in many margins worldwide (Micallef et al., 2012; Harishidayat et al., 2024). The association between landslides and canyons is particularly apparent in margins with active tectonics, where earthquakes, vertical movements, and tilting favor slope instability, eventually leading to canyon formation (Bührig et al., 2022a).

An example of such an active tectonic setting is the northern Sicilian Margin, where numerous canyons dissect the slope and sometimes incise the narrow continental shelf, occasionally reaching very close to the coastline (Sulli et al., 2013; Gamberi et al., 2014; Lo Iacono et al., 2014; Gamberi et al., 2017; Gamberi, 2020; Lo Presti et al., 2022; Scacchia et al., 2025). The canyons in the western part of the margin have been interpreted as resulting from both retrogressive landsliding processes and downslope turbidity currents during the last glacial maximum (Lo Iacono et al., 2014). In the central and eastern parts of the margin, many of the canyons are shaped by processes connected with river deltas, longshore currents, and other coastal oceanographic processes (Gamberi et al., 2015; Gamberi et al., 2017; Lo Presti et al., 2022). Similar processes at canyon heads were active during past periods of lower sea-level stands, as shown by the distribution of past, now submerged, coastal systems on the continental shelf (Gamberi, 2020). However, some of the canyons are now stranded at the shelf edge, and proof of their past connection with rivers is missing. This type of canyon is dominant in the Finale Basin, located in the central part of the northern Sicilian Margin, where, in association with landslides, canyons form pervasive erosional systems with different morphologies. The morphological details of the elements that contribute to the degradation of this portion of the northern Sicilian erosional margin are studied here with the aim of detailing the processes that link submarine landslides and canyons. In particular, we focus our research on the analysis of the relationships between the location and the entity of initial sediment failures and the subsequent evolution of landslides and canyons.

2 Tectonic and physiographic setting

The Finale Basin, the subject of our study, is located along the southeastern margin of the Tyrrhenian Sea and is part of the Capo d’Orlando Basin, an intraslope basin bounded seaward by the Aeolian Island Arc (Figure 1) (Bacini Sedimentari, 1980). The continental shelf is always very narrow; it is never wider than 15 km. Offshore from the towns of Capo d’Orlando and Torremuzza, the continental shelf displays its widest extension with a relatively linear and continuous E–W slope (Figure 1C). The continuity of these physiographic elements terminates westward, where the depression of the Finale Intraslope Basin causes a narrowing of the continental shelf and a southward shift on the continental slope (Figure 1C).

Figure 1

Map showing the Capo D’Orlando Basin area. Panel A depicts the location of the Capo D’Orlando Basin relative to the Tyrrhenian and Ionian Seas near Sicily. Panel B illustrates seismic activity, with colored dots indicating earthquake magnitudes. Panel C provides a detailed Shaded relief map highlighting features such as the Aeolian Islands, Capo D’Orlando Basin Plain, Finale Intraslope Basin, and continental shelf, with elevation indicated by a color gradient.

Figure 1. (A) Location of the Capo d’Orlando Basin in the northern Sicilian continental margin above the Ionian subducting slab, and (B) seismicity of the study area. The map shows earthquakes with moment magnitude (Mw) ≥ 4.0 (including some events with Mw < 4) that occurred between 1000 and 2020. Earthquake data are extracted from the CPTI15 v4.0—Catalogo Parametrico dei Terremoti Italiani (Rovida et al., 2022). (C) Shaded relief map of the northeastern Sicilian continental margin. The continental slope separates the continental shelf from the Capo d’Orlando Basin plain and has an E–W trend. The Finale Basin (dashed blue line) forms the southwestern part of the Capo d’Orlando Basin, coinciding with a southward step of the edge of the slope (yellow to green area) and the continental shelf, corresponding with the red, shallow-water area.

The Finale Basin stretches along the northeastern Sicilian Margin with a length of 20 km (Figure 1C). In the context of large-scale tectonic reconstruction, aimed at understanding the regional tectonic setting of the northern Sicilian Margin, it has been interpreted as a rift structure with main extensional faults trending transverse to the margin (Nigro and Sulli, 1995; Pepe et al., 2000; Gamberi, 2019). Very recent extensional faults (0.5 Ma-recent) separate the Capo d’Orlando Basin, characterized by a strong and active subsidence, from the surrounding coastal areas affected by high uplift rates (Pepe et al., 2003; Antonioli et al., 2009; Billi et al., 2010; Sulli et al., 2013). Recent tectonic activity of the northern Sicilian Margin is expressed by several earthquakes (Billi et al., 2010; Rovida et al., 2022), many of which are localized within the Capo d’Orlando Basin (Figure 1B).

The Finale Basin displays a proximal, narrow continental slope with gradients between 10° and 15° and a less inclined distal part, with gradients between 3° and 5°, which we call a basin ramp (Figures 2A,C). The limit between the continental slope and the basin ramp corresponds to the transition from an essentially erosional sector of the continental slope, with canyons and landslide scarps, to a predominantly depositional area (Figures 2, 3B–D). In the Santo Stefano di Camastra Graben, the relatively continuous E-W-trending limit between the slope and the basin ramp is controlled by the Santo Stefano Tectonic Lineament (Figure 2A). In the other basin sectors, it presents an uneven trend due to alternations of troughs and highs perpendicular to the margin (Figure 2A).

Figure 2

Shaded relief and bathymetric map of the Capo d’Orlando Basin Plain showing canyons, landslides, and tectonic lineaments. Key features include the Pollina Graben, Tusa Horst, and multiple canyon systems. Contour lines indicate depth, while colored lines denote tectonic lineaments, such as the Finale and Miliani. Insets B and C provide cross-sectional views, with details on slope and basin ramp. Various labeled figures highlight specific regions and geological phenomena.

Figure 2. (A) Bathymetric and shaded relief map of the Finale Basin with contours every 100 m. The basin, outlined by the white dashed line, reaches a depth of approximately 1200 m. The main tectonic lineaments trend NNW–SSE and encompass various faults forming distinct tectonic lineaments, marked with different colors. A network of canyons and landslides heavily dissect the continental slope. The latter is replaced basinward by the basin ramp, a less inclined area, which, starting proximally from the dashed yellow line, reaches the distal part of the Finale Basin. The basin ramp is characterized by a rugose seafloor indicative of blocky debris-flow deposits (purple areas). The related debris-flow channels (blue lines), with a relatively smooth seafloor, connect to the canyons upslope and in some cases reach the distal part of the basin. (B) Depth profile parallel to the slope direction, highlighting the horst and graben morphologies of the basin. (C) Depth profile perpendicular to the slope. Note the difference in gradients between the slope (approximately 10°) and the basin ramp (approximately 4°), separated by the basin ramp upper limit.

Figure 3

Seismic profiles with labeled geological features. Panel A shows Pollina Graben and Tusa Horst with faults marked in red. Panels B to D depict continental shelf to basin ramp transitions, highlighting landslide deposits and fault lines. Specific features like Torremuzza landslides and debris-flow deposits are labeled. Color-coded elements include yellow shelf-edge clinoforms, blue horizons marking landslide bases, and red faults. Each panel, annotated A to D, provides a different cross-sectional view, showcasing geological variations and seismic data interpretation.

Figure 3. Sparker seismic lines in the Finale Basin (see location in Figure 2). (A) Line cutting longitudinally through the Finale Basin. The Sant’Ambrogio Tectonic Lineament is the western limit of the basin. The Milianni Tectonic Lineament separates the Pollina Graben to the west from the Tusa Horst to the east. Further faults occur within the main tectonic elements. Above the blue horizon, the discontinuous character and the high amplitude of the surficial reflectors indicate that they are mostly the deposits of landslides and debris flows. (B) Line showing the eastern part of the basin. The Torremuzza and Tusa landslide bodies display discontinuous and chaotic or highly reflective, seismic facies. They lie above the blue horizon and occupy the depletion area of the main landslides, which is, therefore, not completely evacuated. (C) Line showing the area of the Tusa Canyon System and the Tusa landslide. The yellow area marks the shelf-edge clinoforms, which make up the last low-stand wedge. The discontinuous and transparent seismic facies of the Tusa landslide deposit occupy the landslide depletion area above the blue horizon and in the Tusa 5 canyon head. (D) Line transversally crossing the western side of the basin in the area of the Sant’Ambrogio Canyon System. At the shelf edge, a clinoform (yellow area) formed during the last low sea-level stand. As in previous lines, the blue horizon marks the base of debris-flow deposits.

The shelf edge is, in general, at a depth of 155 m (Figure 4), deeper than in the majority of the northern Sicilian Margin, where it is at depths of 95–140 m to the west (Lo Iacono et al., 2014),140–145 m immediately eastward (Sulli et al., 2013), and 125 m further east (Gamberi et al., 2015).

Figure 4

Slope map showing geological features such as the Tusa Canyon System, Santo Stefano Graben, Tusa landslide, and Torremuzza canyon. Contour lines indicate depths, and various regions are labeled with orange lines and annotations. Key regions are marked with figures numbers, and a scale bar indicates distances.

Figure 4. Shaded relief map of the eastern (A) and western part (B) of the study area. In both areas, the majority of the canyons and landslides indent the shelf edge (orange line). Large landslides that do not reach the shelf edge are also present (Rais Bergi and Tusa landslides). The basin ramp has a prevailing blocky seafloor (purple areas) and linear smoother features, indicative of debris-flow deposits with channels (blue lines).

The shelf edge presents an ENE trend in the area facing the Tusa Horst and the Santo Stefano Graben (Figure 4A) and an E-W trend in the area facing the Pollina Graben (Figure 4B). The trend of the shelf edge, however, presents important lateral changes at a smaller scale. In particular, shelf-edge indentations are common and have a maximum across-shelf width of approximately 5 km (Figure 4). The shelf-edge indentations correspond to the headwalls of the canyons and landslides, whose genesis is the main subject of our article.

3 Data and methods

3.1 Available data

Our study of the Finale Basin was carried out mainly through the interpretation of bathymetric data acquired with different multibeam systems. Data were collected during various cruises conducted by the Istituto di Scienze Marine–Consiglio Nazionale delle Ricerche (ISMAR) in the past 15 years. The oldest multibeam bathymetric data, which still contribute to the final digital terrain model (DTM) in water depths >1000 m, were acquired with a hull-mounted SIMRAD EM12 (13 kHz) model in 1996 and 1999, respectively, aboard R/V Gelendzhik and R/V Strakhov in the frame of a national regional mapping project. High-resolution bathymetric data were acquired during two cruises in the upper slope and shelf areas, conducted aboard R/V Mariagrazia in 2009 and 2010 using a hull-mounted Kongsberg EM3002D (300 kHz) and a pole-mounted Reson 7111 (100 kHz) multibeam system, respectively. In 2011, some sectors of the outer shelf were re-mapped with a hull-mounted Kongsberg EM710 (70–100 kHz) multibeam system on board R/V Urania. All the multibeam data were merged and post-processed using CARIS HIPS and SIPS software, and a 10-m-resolution DTM was produced. In addition, a regional reconnaissance seismic grid consisting of both strike and dip lines acquired in 1973 by the former IGM-CNR institute aboard the R/V Bannock with a single-channel 30 kJ sparker system was also used. The vertical resolution of these datasets is approximately 30 m. All available lines are shown in this study.

3.2 Terminology

In the Finale Basin, incised features, that is, depressions elongated perpendicular to the slope, have largely variable dimensions and form a hierarchical assemblage in which smaller features often contribute to the overall morphology of larger ones. We term “canyons” the largest incised features, which, as in their original definition (Shepard, 1933; Daly, 1936), represent troughs formed by erosional processes into the continental slope. The smaller incised features are here called “channels” and are identified as smaller elements almost ubiquitous within the canyons. In addition, we call “blind” the canyons and landslides that have their headwalls in the slope and do not reach the shelf edge. Landslides are distinguished from canyons because they form depressions that, unlike canyons, lack a clear and continuous axial channel. We applied the scheme and the concept for land slope movements proposed by Varnes (1978) to classify and reconstruct the character of the landslides and of their components.

In many articles, different canyon morphologies are used to classify canyons into distinct types. To provide a clearer and more descriptive framework, and to distinguish our reconstruction from those already available, we name the two types of canyons identified in the study area as “worm” and “racket” canyons, which, although unconventional, offer an intuitive representation of their morphology and help to avoid confusion with previous classifications.

In our article, we describe a margin characterized by a very narrow shelf, on the order of 5 km of width. The other geomorphologic features, such as canyons and landslides, are small when compared with similar features on other continental margins. The canyons are short features that develop exclusively on the continental slope and are on average 2 km long. As a consequence, although referring to very small-scale geomorphologic elements in absolute terms, we use “large” and “small” to characterize the dimensions of the observed features in the relative context of the study area. Thus, we refer to “large landslides” as those with an along-slope width of their main scarp larger than 1 km.

The hierarchy of shelf-edge indentation proceeds from the largest one, which is assigned the first order, to the smallest third-order ones. However, our hierarchical subdivision is not connected to their absolute size, but rather to their relative dimensions in relation to the other indenting features within a single canyon. Thus, the second-order indentations of one canyon can have dimensions similar to the first-order indentations of another canyon. Because the indentations are often small features, they are not indicated in their real position on the maps, but rather by their landward projection so as not to overprint them with lines. Thus, we remember that the lines that mark the indentations clearly depict their along-shelf extent, but not their real position and shape.

4 Results: observations and interpretation

4.1 Structural and stratigraphic interpretation

Detailed interpretations of the geological setting of the Finale Basin were not available. Thus, as a first step in our research, we undertook an analysis of the structural elements responsible for the basin physiography. The Torremuzza Tectonic Lineament, with its NW–SE-trending fault segments, is the eastern boundary of the basin (Figure 2A). The western boundary of the basin, outside our data coverage, corresponds to an NNW–SSE-trending slope with an eastward bathymetric drop, called the Sant’Ambrogio Tectonic Lineament (Figure 2A). Further major tectonic lineaments separate the Finale Basin into distinct structural elements. The Milianni Tectonic Lineament, with a NNW–SSE trend, separates the Pollina Graben to the west from the Tusa Horst in the central part of the basin (Figures 2A,B, 3A). To the northeast, the Tusa Tectonic Lineament, composed mainly of NW-oriented fault segments, separates the Tusa Horst from the Santo Stefano di Camastra Graben and disappears southward in the bathymetry (Figures 2A,B). The Tusa Horst and the Santo Stefano di Camastra Graben terminate northward along the E–W trending Finale Offshore Tectonic Lineament, which forms a prominent escarpment, limiting southward the Capo d’Orlando Basin; this is one of the main tectonic structures of the southern Tyrrhenian Sea (Figures 1, 2A). The Finale Lineament dies out westward so that the Pollina Graben connects to the flat plain of the Capo d’Orlando Basin through a trough bounded eastward by the NW–NW-SE-trending faults of the Milianni Tectonic Lineament (Figure 2A).

Although very sparse, the available seismic lines have contributed to distinguishing the processes and depositional environments of the study area through seismic facies analysis. Debris-flow and landslide deposits exhibit transparent or chaotic seismic facies (Figure 3A) or are characterized by strong and discontinuous reflections (Figures 3B,C). Occasionally, panels with coherent reflections are present within the main chaotic facies, corresponding to landslide blocks with relatively well-preserved stratigraphy. Such a setting is particularly evident in the Tusa landslide, where a proximal transparent facies is replaced downslope by packages of relatively coherent reflectors (Figure 3C). It shows that the landslides in the study area exhibit different rheologies and types of movement, resulting in deposits with a range of internal facies, from completely chaotic bodies to less deformed units. The seismic line that runs parallel to the margin in the proximal part of the basin ramp (Figure 3A) shows a transparent body representing the upper part of the seismic succession. The area crossed by the line corresponds to a blocky seafloor in the bathymetry (Figure 2). A similar seafloor bathymetric pattern is found further downslope across the majority of the basin ramp (Figures 2, 4). Thus, the integration of seismic and bathymetric data information suggests that the landslides in the study area form aerially widespread bodies that occupy the majority of the basin. Areas with a blocky seafloor form single or multiple lobes in planform and present linear depressions that are bounded by subtle scarps, which are best interpreted as channels (Figures 2, 4). As a whole, therefore, they are best interpreted as channelized, blocky debris-flow deposits connected to the failures that produced the slope landslides.

Only one of the canyons, the Tusa 5 canyon, is imaged by the available seismic line (Figure 3C). Its head sits above a landslide body, marked by a very strong reflection, overlying a highly reflective unit. This setting is here interpreted as evidence that the canyon occupies a depression formed within the depletion zone of a rotational slide that has not completely evacuated, with part of its deposit preserved above the basal rupture surface.

The lower, larger part of the basin infill in both the slope (Figures 3B,C) and in the basin ramp (Figures 3A,D) consists of a highly reflective unit with parallel, continuous reflections, which can be interpreted as resulting from widespread, basin-wide turbidite deposition. Landslide and debris-flow bodies are present only as a relatively thin surficial package, which is, however, present in both the slope and in the ramp (Figure 3).

A package of prograding clinoforms is present at the shelf edge (Figures 3B–D). These are similar to the shelf-edge wedges shown in other parts of the Sicilian Margin, which are interpreted as depositional bodies, with presumably relatively coarse grain sizes, formed in coastal and shallow-water environments during the last low sea-level stand (Gamberi et al., 2020). A similar character is therefore inferred here for the prograding clinoforms in the study area.

4.2 Landslides

Two large depressions dissect the basin margin and indent the shelf edge of the Santo Stefano Graben (Figures 4A, 5A). The western depression, the Torremuzza western landslide, indents the shelf edge with an along-shelf width of approximately 3 km and penetrates landward for approximately 1 km (Figure 5A). In its western proximal part, it presents alternations of terraces and scarps, often coalescing into larger structures with a mainly arcuate, downslope-concave planform (Figures 5A,B). This setting is interpreted as indicative of a rotational slide body consisting primarily of rotated blocks. In contrast, the rugose seafloor of the proximal eastern part of the depression reveals a more chaotic slide body (Figure 5C). As a whole, the depression corresponds to the depletion zone of a relatively large rotational slide, with a deposit consisting of both rotated blocks and more disrupted units.

Figure 5

Slope map of Torremuzza western and eastern landslides, featuring various geological features indicated by colored lines and labels. There are main landslide scarps (red), minor landslide scarps (dotted red), and channels of different features using several colors. Insets B, C, and D show detailed depth profiles and specific areas like Rugose seafloor and shelf-edge indentations. Elevations and slopes are marked with contour lines. A scale bar and slope legend are provided.

Figure 5. (A) Slope shader with bathymetric contours every 5 m of the Torremuzza canyon and landslides. The Torremuzza western landslide presents a terraced seafloor, evidence of a main landslide body composed of rotated blocks separated by minor scarps. The Torremuzza eastern landslide presents a smoother seafloor with longitudinal minor scarps indicative of complex successive movements and debris-flow processes. The Torremuzza central canyon is a racket canyon formed within a main landslide scar, which does not reach the shelf edge. The related narrow shelf-edge indentation is caused by retrogression, exclusively affecting a 150-m-long stretch of the original landslide crown. (B) Depth profile crossing the Torremuzza western landslide, highlighting different scarps and terraces. (C) Zoom of slope shader with contours (every 2.5 m) of a portion with rugose seafloor. (D) Zoom of slope shader with contours (every 2.5 m) of shelf-edge indentations in the Torremuzza eastern landslide. The third-order indentation is bowl-shaped.

The Torremuzza eastern landslide (Figures 4A, 5A) forms a depression with a maximum along-slope width of approximately 5 km. Its main scarp has a semi-circular form and is responsible for a first-order shelf-edge indentation, with an along-slope width of 4 km. In its eastern proximal part, relatively gentle areas, bounded by steep scarps, are interpreted as rotated blocks (Figure 5A). In the central part of the landslide body, preserved within the depletion zone, two linear scarps bound a channel and connect upslope to a bowl-shaped scarp, which causes a second-order shelf-edge indentation with an along-slope width of 500 m (Figure 5D). The bowl-shaped area is interpreted as the landslide’s source region, whereas the distal linear scarps, which limit a channel, hint at debris-flow processes downslope from the landslide's main source area.

Other features that can also be interpreted as relatively large landslides, with dimensions similar to those of the Torremuzza ones, are the Tusa landslide (Figure 4A) and the Rais Bergi landslide (Figure 4B). They have their main scarps in the upper slope, representing blind landslides. Minor scarps and terraces at the base of the main scarp of the Tusa blind landslide are evidence of rotated blocks (Figures 4A, 6A). In the distal part of the displaced material, two channels follow the lateral flank of the landslide (Figure 4A). Similar channels along the lateral flanks of the Rais Bergi landslide connect upslope to bowl-shaped depressions bounded by downslope-convex scarps (Figures 4B, 7). Combining the observations from the two landslides, we interpret the channels as being formed by debris-flow processes that have reworked the sediment previously displaced by the main landslide movement and have their source region in the bowl-shaped depressions. Successive failures resulting in sediment displacement through different processes, including debris flows, are an integral part of submarine landslides (Bull et al., 2009). Along the northern Sicilian Margin, a similar example of a debris flow occurring at the expense of a previously deposited landslide has been shown in the Gioia Basin (Gamberi et al., 2011).

Figure 6

Slope map featuring canyons and landslides labeled Tusa 1 to 5. Contours indicate depths with features like rugose seafloor and rotated blocks. Various lines mark landslide scarps, debris-flow channels, and duck-foot features. Insets A and B provide detailed views of specific areas with color-coded indentations and channels. Scale bar included.

Figure 6. (A) Slope shader with bathymetric contours (every 5 m) of the Tusa Canyon System. The Tusa 2 canyon is the only worm canyon. Its proximal segment, which indents the shelf edge, is a duck-foot scoop. The other canyons are racket canyons. They form in the depletion zone of major landslides whose deposits display a rugose seafloor. Successive instability generates erosional debris flows that form channels and contribute to the excavation of the canyon axis. (B) Zoom of slope shader with contours (every 2.5 m) of the shelf-edge indentations in the Tusa 4 racket canyon.

Figure 7

Slope map of the Rais Bergi area showing features like landslide scarps, debris-flow channels, and axial channels marked with various colored lines. Labels indicate

Figure 7. Slope shader with bathymetric contours (every 5 m) of the Rais Bergi Canyon System. All the canyons are worm canyons. The two central canyons indent the shelf edge with their duck-foot scoop proximal parts. The eastern and western canyons are blind. At the base of the eastern lateral flank of the Rais Bergi landslide, a debris-flow channel forms downslope from its bowl-shaped source areas in correspondence with the re-entrances of the landslide’s main scarp.

4.3 Worm canyons

The main distinctive characteristics of the worm canyons are their restricted width, in general less than 500 m, and steep flanks, with an average dip of 20° but reaching values up to 30° (Figures 7–10).

Figure 8

Slope map depicting Sant’Ambrogio canyons with key features. Contour lines indicate depths. Main and minor landslide scarps, debris-flow areas, and channels are color-coded. Labels identify Sant'Ambrogio western blind, central, and eastern worm canyons.

Figure 8. Slope shader with bathymetric contours (every 5 m) of the Sant’Ambrogio Canyon System. The two main canyons are worm canyons that indent the shelf edge with their duck-foot scoops. Three channels develop in the duck-foot scoops and converge downslope in the canyon axial channel. Note the ubiquitous arcuate scarps that make up the margins of the worm canyons. The axial channels display abrupt variation in direction coinciding with the limits between the single arcuate scarps.

Figure 9

Slope map of underwater canyons with contour lines indicating depth. Main and minor landslide scarps are highlighted in red and pink, respectively. Features such as

Figure 9. (A) Slope shader with bathymetric contours (every 5 m) of the two Pollina worm-type western canyons. They are very narrow, with a segmented course and scalloped flanks. The Pollina Western 1 canyon shows two terraces and scarps, which can be indicative of rotated blocks in the proximal part of the canyon. Both canyons form first-order shelf-edge indentations approximately 250 m wide. In both canyons, second-order indentations correspond to the bowl-shaped heads of very narrow channels, interpreted as the source area of debris flows responsible for the channel excavation further downslope. (B) Zoom of slope shader with contours (every 2.5 m) of the shelf-edge indentation in the Pollina western canyons. The second-order indentations are bowl-shaped.

Figure 10

Slope map illustrating the Pollina landslide area with various features labeled: main and minor landslide scarps, worm canyons, racket canyon, and indentations. Color-coded lines represent channels and scarps, with distinct shades for each type. An inset labeled

Figure 10. (A) Slope shader with bathymetric contours (every 5 m) of the Pollina Central and Eastern Canyons. The eastern canyons are blind worm canyons. They are flanked northward by the Pollina landslides. The central Pollina Canyon is a racket canyon and is the largest in the study area. It also forms a large first-order indentation stretching a 1500-m wide margin sector. Its head presents numerous channels connected upslope to bowl-shaped depressions interpreted as the source regions of debris flows responsible for the channel excavation and third-order shelf-edge indentations. (B) Zoom of slope shader with contours (every 2.5 m) of the shelf-edge indentations of the Pollina Central racket canyon. First-order indentations are bowl-shaped.

“Worm canyons have scalloped margins, indicating that they consist of separate segments. The single segments have a planform with predominantly rounded margins with concavity toward the canyon axis and result in enlargements and restrictions to the canyon’s course, which resemble the body of a segmented worm (Figures 7–10A). Often, abrupt changes in the direction and gradient of the canyon’s axis occur at the junction between the single canyon segments (Figures 7–10). The flanks of the worm canyons often have a rough, rugose seafloor (Figures 7, 8, 9A), which can be an indication of landslide deposits. Steeper and less inclined portions in the proximal part of the Pollina western canyon 1 can be evidence of scarps bounding rotated landslide blocks (Figure 9A).

All the above observations suggest that worm canyons form through the joining of elemental features that have a circular or oval planform and can be interpreted as corresponding to the depletion zone of rotational spoon-shaped or elongated landslides forming a train of retrogressive failures focused along a narrow corridor. In some cases, however, landslide scars trending perpendicular or at a high angle to the canyon axis are evident in the flanks of the worm canyons, further up from the linear narrow canyon axis. These can be observed in Rais Bergi 1 (Figure 7), the Sant’Ambrogio central worm canyon (Figure 8), and the Pollina western 1 and 2 worm canyons (Figure 9A). They serve as evidence that the lateral collapses of the canyon flanks can also contribute to the enlargement of the worm canyons. In some cases, where the landslide material has not been completely removed, the canyon flanks preserve chaotic deposits above the landslide basal rupture surface and consequently have a rugose morphology (Figures 7, 8, 9A). More often, however, the canyon flanks have a smooth seafloor, suggesting they are not floored by landslide deposits. We interpret this latter setting as resulting from the complete evacuation of the source area of the canyon-forming landslides through flow-type movement, which can be erosional, such as a debris flow. In this case, the seafloor of the canyon flanks corresponds with the rupture surface along which the failed sediment was removed.

Some of the worm canyons indent the shelf edge. The Pollina western canyons (Figures 9A,B) form two 500-m-long, first-order indentations, containing the heads of narrow channels that cause second-order indentations less than 100 m wide (Figure 9B). The latter appear to be connected to localized failures of small, bowl-shaped source regions connected to channels downslope.

Blind worm canyons are also present in the Sant’Ambrogio (Figure 8), Rais Bergi (Figure 7), and Pollina (Figure 10A) canyon systems and are tributaries to the main canyons.

4.4 Racket canyons

A planform resembling a tennis racket, with a proximal circular, oval-shaped, or elongate depression that connects downslope to a narrow linear channel along the axis of the canyon, is the main distinctive characteristic of racket canyons. The proximal part generally has an along-slope width of 1 km but can reach a width of 5 km. It has gentle flanks, dipping on average at approximately 15° but up to 20° (Figures 5A, 6A, 10A). The distal linear channel, which resembles the handle of the racket, occupies only the thalweg of the canyon, has an average width of approximately 100 m proximally, and enlarges downslope.

The Pollina Central Canyon (Figure 10A) is a representative example of a racket canyon indenting the shelf edge. It has a landward part with a maximum width of 1.4 km, and lateral margins defined by relatively linear crests that converge downslope toward the canyon axis (Figure 10A). The eastern part of the canyon depression has a rough seafloor with both arcuate and elongated scarps, relatively flat terraces, and channels. To the west, the more landward part of the canyon depression has an uneven seafloor with an often-rugose pattern whereby narrow and shallow channels with a dendritic pattern converge downslope in the canyon axis. The channels are narrow, very straight features, with a width of approximately 100 m, a depth of 10 m, and a slightly larger, semi-circular head. The surficial morphology of the canyon floor is therefore in agreement with the presence of sediment with variable degrees of disruption, from intact blocks forming terraces to completely chaotic units elsewhere on the rugose seafloor. It is the evidence of landslide deposits, which, in our interpretation, are above a large, spoon-shaped, rotational rupture surface within a depletion area not completely evacuated. In this interpretation, the channels can be the evidence of successive failures, most likely of the debris-flow type, that affected the main landslide body. Within the canyon depression, which forms a first-order shelf-edge indentation, single scallops, likely resulting from minor successive landslides, form second- or third-order shelf-edge indentations (Figures 10A,B). In particular, the heads of the channels in the main landslide deposits, due to debris-flow processes, form third-order shelf-edge indentations (Figure 10B). Other racket canyons in the Tusa system (Figure 6A) and the Torremuzza canyon (Figure 5A) show a single or fewer channels scouring the deposits of the main landslide. These channels always converge downslope into a distal linear channel, which is the main axial trunk of the racket canyons.

In some of the racket canyons, such as in the Tusa 5 (Figure 6A), a channel develops exclusively downslope from the deepest scarp affecting the landslide body, which is only slightly deformed, as shown by the prevalence of scarps and terraces, evidence of rotated blocks. We interpret this setting as resulting from debris-flow channelization of only the distal part of the material involved in the main slides. The heads of some of the racket canyons, such as the Tusa 4 and the Torremuzza canyons, do not indent the shelf edge for the majority of their extent but have smaller-scale re-entrances that cause local indentations (Figures 5, 6B). This setting can be indicative of retrogressive failure processes, whereby the original, main large landslide scarp does not reach the shelf edge, which is indented only by the scarps from successive smaller-scale failures.

4.5 Duck-foot scoops

The proximal part of the majority of worm canyons consists of a “duck-foot scoop” (Figures 6A, 7, 8): a downslope-narrowing triangular depression with lateral flanks, which, although initially very subtle, gradually increases in relief downslope. Narrow, low-relief downslope-converging channels are the main features of the majority of the duck-foot scoop canyons (Figures 6A, 7, 8). The duck-foot scoops form first-order shelf-edge indentations, with an average width of 500 m. They contain second-order indentations that correspond to the heads of the internal channels (Figures 7, 8) or to smaller-scale landslide scarps (Figures 6, 8). The second-order indentations, which correspond to the channel heads, are approximately 100 m wide. The duck-foot scoops are here interpreted as the result of repeated landslide processes. We suggest that an initial shallow failure empties a part of the upper slope, forming the overall depressed structure of the duck-foot scoops. Successive landslides are then responsible for further sediment removal and for the formation of the channels, explained as due to debris flows with the source in the proximal part of the duck-foot scoops. We envisage that these debris flows are recurrent, can progressively erode and deepen the duck-foot scoops at the canyon head, and most likely cause erosion also downslope.

5 Discussion

5.1 Out-of-grade margin and landsliding

Our seismic data do not allow the reconstruction of the southward continuation of the Finale Basin’s tectonic setting on the continental shelf. On land, high-angle, dextral, strike-slip faults with NW-SE orientation are responsible for the most recent deformation; antithetic, sinistral, NE-SO-oriented faults also contribute to the recent deformation style (Giunta and Giorgianni, 2010). They are part of the South Tyrrhenian system, and the analysis of the elevation of relict alluvial plains substantiates that these faults may be active. Further tectonic structures are N–S-trending faults, part of the medial Tyrrhenian system (Barreca and Carbone, 2008).

Along the coast, discontinuous marine terraces have been raised to elevations of up to 450 m. They are not older than 450 Ky and indicate a recent rapid and conspicuous uplift of the emerged area to the south of the Finale Basin (Giunta and Giorgianni, 2010). The shelf edge (Figures 2, 4) is deeper than the other areas of the northern Sicilian Margin and the expected regional depth. Hence, subsidence of the external part of the continental shelf accompanies the uplift of the onshore region. This setting may be due to faults that separate the inner shelf from the outer shelf, as suggested for the offshore to the east of the study area (Sulli et al., 2013). Another possibility is that the margin is subject to an important and ongoing northward tilting, as proposed for the northwestern Sicilian offshore (Lo Iacono et al., 2014). In any case, a steepening of the continental margin’s progradational depositional package is to be expected, with the development of out-of-grade conditions. Furthermore, the seismic line parallel to the margin shows that the troughs and highs perpendicular to the margin are formed by faults with an NNW–SSE trend (Figure 3A). Therefore, local steepening and tilting of structural elements occur along the margin, contributing to the widespread instability conditions in the study area.

Along out-of-grade continental margins, slope degradation is essential in order to re-establish a graded margin profile (Ross et al., 1994; Pirmez et al., 1998). Slope degradation is generally achieved through erosional processes that are best displayed by an abundance of canyons and landslides, particularly along the slope sector of continental margins. In the study area, canyons and landslides are pervasive and affect the entire continental slope (Figures 2, 4). Tilting of sedimentary successions is one of the factors that predispose sediment instability along continental margins (Lee et al., 2007). We thus conclude that it can also be one of the main preconditioning factors underlying the widespread unstable conditions leading to the frequency of landslides and canyons in the study area. Within this intrinsically unstable setting, the frequent earthquakes recorded in the study area and its surroundings (Figure 1B) can serve as recurring triggers for landslide initiation. Therefore, the Finale Basin demonstrates the importance of long-term tectonic processes of margin deformation in fostering landslide susceptibility and of earthquakes in triggering sediment failure along active continental margins.

In the seismic profiles of Figure 3, the vertical distribution of seismic facies makes it apparent that the erosional and depositional processes associated with canyons and landslides affect only the most recent infill of the basin. Therefore, our analysis shows that the disequilibrium originated by the tilting of the northern Sicilian Margin must relate to a recent, abrupt change in the tectonics of the northern Sicilian Margin. In the absence of any age dating, we cannot speculate on its timing. This change should coincide with the onset of the uplift of the adjacent land areas, for which conclusive data are not available.

5.2 Landslides and canyon initiation

Incised valleys on the shelf and river mouths at the shelf’s edge, particularly during low sea-level stands, can contribute to the formation and the successive evolution of canyons on the upper slope. River mouths were almost surely present at the shelf’s edge during the low sea-level stands in the study area, as shown by incised valleys elsewhere along the Sicilian northern margin (Sulli et al., 2012; Gamberi et al., 2014). In the absence of seismic data control on the shelf, we thus cannot rule out that during past low sea-level stands, rivers were also present at the shelf edge in the study area. In fact, a channel on the slope to the east of the Torremuzza eastern landslide reaches the shelf edge. It can provide evidence of past geomorphic elements developed on the slope in connection with river processes. On the other hand, our analysis shows that some of the canyons, with their heads in the lower part of the slope, at depths of up to 300 m, are blind and thus were unequivocally never directly connected with rivers (Figures 4, 6, 7, 10). However, the contribution from longshore currents, storm action, and distal hyperpycnal flows during past low stand periods to canyon excavation for these features, which have their heads at a distance of only approximately 1 km from the shelf edge, cannot be excluded. Incidentally, however, the blind canyons are always associated with landslides and thus show a clear connection with slope instability and failure. In addition, relatively large landslides, such as the Tusa (Figure 4A), the Rais Bergi (Figure 4B), and the Pollina (Figure 10A) landslides, independent of canyons, are also present in the study area. The Torremuzza landslides indent the shelf edge in the eastern part of the basin (Figures 4A, 5A). They show that the instability processes in the upper slope and at the shelf edge can result in the excavation of depressions where successive sediment failures can eventually lead to canyon formation. We therefore conclude that, although a contribution from fluvial processes and other oceanographic shelf processes cannot be excluded, particularly during past low sea-level stands, the morphology and the evolution of the canyons in the study area are primarily controlled by landslide processes and the formation of depressed depletion zones with different volumes and in different locations along the slope.

According to our interpretation, two types of landslides, distinguished by their dimensions and positions, lead to the formation of worm canyons and racket canyons (Figure 11). Worm canyons result from the coalescence of single, relatively small-sized rotational slides (approximately 1 km along-slope width), with a concave-up rupture surface, that develop sequentially along a narrow corridor (Figure 11A). Initially, a small failure causes a slide in the lower slope (stage I, Figure 11A). Successive retrogradation of instability causes further landsliding landward, resulting in a narrow incised corridor. Such a stage is exemplified in the study area by the Sant’Ambrogio western blind worm canyon (Figure 8). Concomitant flow-type processes, such as debris flows, cause extensive seafloor erosion downslope, leading to canyon initiation as in many other examples worldwide (stage II, Figure 11A). Subsequently, continued upslope landslide retrogradation and related erosional flows contribute to the landward shift of the canyon head, which, however, still has its head in the slope and remains blind. Erosional flows also contribute to the downslope widening and deepening of the canyon axis, which can be further enlarged by lateral collapses along its flanks. Such a situation (stage III, Figure 11A) is exemplified by the Pollina eastern 1 and 2 blind worm canyons (Figure 10A), the Rais Bergi eastern and western blind worm canyons (Figure 7), and the Sant’Ambrogio western blind worm canyon (Figure 8). Shelf-edge indentation is the last step in the intrinsic evolution of worm canyons. The upslope migration of the canyon head, through the successive activation of relatively small-scale landslides, eventually leads to the failure of the shelf edge (stage IV, Figure 11A), as shown by the Sant’Ambrogio central and eastern worm canyons (Figure 8), the Rais Bergi 1 and 2 central worm canyons (Figure 7), the Pollina western 1 and 2 worm canyons (Figure 9A), and the Tusa 2 worm canyon (Figure 6A).

Figure 11

Illustration showing the evolution of worm and racket canyons over time. Panel A depicts four stages of worm canyon evolution, including initial slope failure and the formation of duck-foot scoops. Panel B demonstrates two cases of racket canyon evolution with large retrogressive failures. The right side features a legend with lines indicating main and minor landslide scarps, debris-flow channels, axial channels, and indentations in various colors.

Figure 11. Sketch of the landslide-formed canyons in the Finale Basin. (A) Worm canyons are narrow features resulting from laterally confined retrogressive landslides. Often, duck-web-shaped areas are present in their proximal part and indent the shelf edge. (B) Racket canyons form following large landslides. In both canyon types, debris flows, with both retrogradational and progressive styles of movement, follow the main landslide. The resultant debris-flow channels are the main causes of canyon excavation. A full description of the canyon evolution is provided in the text.

Racket canyons, in contrast, form within the depletion zones of larger rotational slides (Figure 11B), with an along-slope width up to 5 km. The large landslides, which precede canyon development, can have their main scarp in the upper slope, such as the Torremuzza eastern landslide (Figure 5A; case I; Figure 11B), or indent the shelf edge, such as the Torremuzza western landslide (Figure 5A; case II; Figure 11B). Successive canyon development occurs within the depletion zones of such relatively large slides. Examples of canyons that developed in landslides that reach the shelf edge are the Pollina Central racket canyon (Figure 10A) and the Tusa 1 racket canyon (Figure 6A). In contrast, the Tusa blind racket canyon is an example of a canyon formed within a slide that did not reach the shelf edge (Figure 6A). In both cases, depletion zones are not completely evacuated and contain part of the slide deposits, with both rotated blocks and more disrupted units, as shown by the seismic line of Figure 3C. Eventually, landslide processes progress upslope, and indentations of the shelf edge can also occur in cases where the initial landslide was not shelf-indenting, such as in the case of the Tusa 3 and 4 racket canyons (Figure 6A) and the Torremuzza racket canyon (Figure 5A). In both cases, at least part of the sediment involved in the main landslide undergoes a flow-type movement. The resulting debris flows erode the seafloor downslope from the depletion zone of the main landslide and form the axial channel of the racket canyons. Successive instability can cause both smaller-scale failures of the already-emplaced landslide deposits and retrogradational failure of the main landslide scarp. These successive landslides can form further flow-type movements that can enlarge the axial channel of the canyon (Figure 11B).

Within the general association between landslides and canyons, our reconstructions show that two different failure processes operate in the study area to form the observed canyon types. The evolution of the worm canyons, triggered by a relatively small initial failure in the lower slope, fits the upslope-erosion model of canyon development. The latter is the most frequently identified trend of canyon evolution in many margins worldwide (Green et al., 2007; Green and Uken, 2008; Puga-Bernabéu et al., 2011; He et al., 2014; Wu et al., 2022; Harishidayat et al., 2024; and references therein). In contrast, racket canyons form in connection with relatively large landslides that develop at or very close to the shelf edge. Similar locations of canyon-initiating landslides have been described in the tectonically active areas of the deformed South Caribbean belt (Naranjo-Vesga et al., 2022). Here, canyon-initiating landslides affecting the shelf edge have been related to fault activity and steep slope (3°). Our study area is also affected by active tectonics and tilting, and it has a very steep continental slope (between 10° and 15°). We thus conclude that the active tectonic setting of the northern Sicilian Margin favors the instability of the upper slope and at the shelf edge and triggers landslides that drive the formation of racket canyons.

5.3 Canyon and debris flows

The evolution of both worm and racket canyons depends heavily on the occurrence of debris flows associated with seafloor failures. Debris flows can be one type of mass movement occurring in sequence during the movement of landslides with a complex style of failure (Varnes, 1978; Cruden and Varnes, 1996). As an example, a debris flow can be generated when a landslide body is mostly disintegrated (Mulder et al., 1997; Gee et al., 1999; Piper et al., 1999; Talling et al., 2007; Strachan, 2008; Laberg et al., 2017; Brothers et al., 2019; Cardona et al., 2020; de Lima Rodrigues et al., 2020; Cukur et al., 2021). In a submarine environment, debris flows can also be generated when landslide deposits are affected by long-term instability long after their initial emplacement (Jing et al., 2024). For example, secondary remobilization of relatively large landslides has been suggested on the surface of debris avalanches in the Poverty Margin (Mountjoy and Micallef, 2012), the Adriatic Basin (Dalla Valle et al., 2015), and the Gioia Basin in the southern Tyrrhenian Sea (Gamberi et al., 2011; Rovere et al., 2014; Gamberi et al., 2020).

Strongly erosional debris flows have been responsible for channel excavation in the modern seafloor (Tripsanas et al., 2008) and outcrop examples of submarine deposits (Butler and McCaffrey, 2010; Dakin et al., 2013; Ogata et al., 2014; Hodgson and Brooks, 2018; Cardona et al., 2020). In the study area, the erosional behavior of secondary debris flows is best shown by the debris-flow channels linked upslope to bowl-shaped source areas (Figures 5D, 9B). The idea that debris flows are a frequent process in the study area is supported by the abundance of deposits with a blocky surface, as shown by the bathymetry downslope from the canyon head in the basin ramp (Figure 2). Here, a blocky seafloor outlines discrete areas, often with an axial, linear, narrow corridor with a smoother seafloor. This morphology is typical of lobes with blocky deposits fed by debris-flow channels in the submarine environment, as described elsewhere.

Debris flows erode the main landslide deposits and imply successive instability (Figure 5). It is therefore reasonable that debris flows evolving from the main landslides, responsible for the initial depressions of both the worm and the racket canyons, can cause the erosion and substrate entrainment necessary for the incision of the axial channels further downslope in the system. We thus suggest that the lower parts of the canyons, downslope from the depletion area of the main landslide, are due to excavation by repeated debris flows, arising both from flow transformation during the movement of the main slide and from successive failures that rework the main slide body (Figure 11). In general, we stress the importance of debris-flow type-landslides in the initial stages of canyon evolution.

5.4 Duck-foot scoops and flow slides

One particular trait of the worm canyons in the study area is their proximal part, consisting of a duck-foot scoop (Figures 7, 8). The channelized character of the duck-foot scoop could hint at some control by cross-shelf flows connected to shelf-edge oceanographic processes. It could also be a remnant of past low-stand periods, when river-connected flows could have directly impacted the upper slope. However, in at least some cases, a clear connection of the channel head with a landslide scarp is apparent in the study area (Figures 6, 8). The duck-foot scoops resemble the triangular-shaped canyon headwalls that form a concave tripartite head, cutting landward into the shelf of the Otway Basin (Wu et al., 2022). These features are interpreted as shelf-edge landslide scarps, representing the proximal expression of multistage retrogressive failure, which played a major role in canyon initiation. Features similar to the duck-foot scoops characterize the proximal landslide-dominated domains of some canyons in the Atlantic and Pacific Margins of the United States (Brothers et al., 2013; 2019). Duck-foot scoops are also similar to canyon heads reported in the South China Sea, where multiple dendritic submarine gullies deepened and widened canyons through recurrent seafloor erosion and successive sediment flows (Li et al., 2022). They also resemble the 700–1,500-m-wide slope failures abundant at the head of Tributary A in the Patia Canyon (Ratzov et al., 2012). All of the above examples and our observations suggest that recurrent retrogressive instability at the heads of worm canyons plays an important role in the formation of duck-foot scoops. In this scenario, mainly governed by sediment instability, it is also possible that part of the morphology of the duck-foot scoops reflects the influence of past processes, linked to rivers or more recent oceanographic constraints.

The duck-foot scoops in the study area, and the above examples of comparable features form the upper reaches of canyons, close to the shelf edge. In the study area, the downslope apex of the duck-foot scoop shows a narrow depth range, between 260 m and 350 m. Studies of the Sicilian continental margin have shown that wedges of presumably coarse-grained sediment formed at the shelf edge during the low-stand periods of the Quaternary (Gamberi et al., 2017; 2020). These clinoform wedges taper downslope, terminate at an abrupt gradient reduction in the slope, and accumulate under conditions of a high sedimentation rate. The seismic lines presented in this article confirm that such low-stand clinoform wedges are also present in the study area (Figures 3B–D). We thus suppose that the peculiar character of the duck-foot scoops at the heads of the Finale Basin canyons results from the specific failure behavior of the shelf-edge wedges. Tripsanas et al. (2008) noted that retrogression reaching the upper slope involved granular sediments that formed cohesionless debris flows. Similarly, we suggest that the duck-foot scoops are the result of processes of instability involving young, presumably coarse-grained material with a low degree of cohesiveness close to the shelf edge. In particular, we propose that repeated flow slides involving the loose granular sediments of the shelf-edge low-stand wedges lead to the progressive excavation of the duck-foot scoops.

5.5 Canyon and shelf indentation

In the study area, shelf-edge indentation is associated with the landslides and the canyons. In Tusa 3 and 4 (Figure 6A) and the Torremuzza racket canyon (Figure 5A), the first-order shelf-edge indentations correspond exclusively to a local landward step of the canyon headwall scar. This setting can result from the original complexity of the main scarp of the canyon-forming landslide, or it can be the result of its successive retrogradation through smaller-scale landslides. As an example, successive debris flows can be responsible for the second- and third-order shelf-edge indentations within the first-order notch created by the main landslide scarp (Figure 6B). In the landslides on the eastern basin side (Torremuzza landslides, Figure 5A) and in the Pollina Central Canyon (Figure 10A), the first-order indentation corresponds to the main landslide scarp. Here also, however, further landslide headwalls or channel heads form second- and third-order indentations (Figures 5D, 10B). Minor landward notches, corresponding to the source region of single debris-flow channels, also form the second-order indentations within the duck-foot scoops (Figures 7, 8). All the above observations imply that the hierarchical nature of the shelf-edge indentations results from the progressive local concentration of instability within the original larger landslide headwall. In the case of the racket canyons, the position of the original large landslide crown leads to differences in the character of the indentations. When it is below the shelf edge, indentation occurs only through the failure of the successive smaller-scale retrogradational landslides (case I, Figure 11B). Alternatively, the crown of the main landslide can by itself indent the shelf edge, and the subsequent retrogradation of instability processes adds further localized landward migration of the previously formed shelf-edge notch (case II, Figure 11B). In general, thus, our analysis infers that the shelf indentations in the study area are accomplished through landslides with a hierarchical arrangement (Figure 11). In particular, we suggest that a diminishing landslide (Cruden and Varnes, 1996), where the volume of displaced material decreases in time, can explain the hierarchical arrangement of the shelf-edge indentations in the study area. In general, our reconstruction can enhance the understanding of the processes connected to continental margin degradation in tectonically active and failure-prone continental margins.

6 Conclusion

Our analysis shows that repeated landslides are the main process leading to canyon formation in a tectonically active area of the northern Sicilian continental margin. Initial landslides can have different locations, dimensions, and characteristics, leading to two different types of canyons: worm and racket canyons. The presence of these two types of canyons in an area dominated by a common geodynamic setting implies that local variability of geological factors along the margin must play an important role in canyon evolution. The particular flow type of many of the landslides, which exhibit characteristics pointing to frequent debris-flow processes, is an important aspect that regulates canyon evolution in the study area. Canyons and landslides show different degrees of retrogradation and can indent the shelf edge. In particular, erosional features that indent the shelf have a hierarchical organization, with a small bowl-shaped debris-flow source area at the lower side of the spectrum. More generally, the results of our analysis of shelf-edge indentations highlight that the effects of events with largely variable magnitude must be evaluated when reconstructing geohazards in connection with seafloor erosion and degradation associated with canyons.

Data availability statement

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

Author contributions

FG: Formal Analysis, Writing – review and editing, Investigation, Visualization, Project administration, Validation, Funding acquisition, Resources, Supervision, Methodology, Conceptualization, Writing – original draft. ES: Funding acquisition, Visualization, Software, Methodology, Writing – review and editing, Data curation, Investigation. VF: Formal Analysis, Methodology, Data curation, Software, Writing – review and editing.

Funding

The authors declare that financial support was received for the research and/or publication of this article. The financial support for the research was recived in the frame of the MaGIC Project funded by the Italian Civil Protection Department.

Acknowledgements

The authors thank all the researchers and technicians who contributed to the acquisition and processing of the data. They are particularly grateful to Alessandra Mercorella and Elisa Leidi, who performed the majority of the multibeam data elaboration and built the final DTM.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The authors declare that no Generative AI was used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

Antonioli, F., Ferranti, L., Fontana, A., Amorosi, A., Bondesan, A., Braitenberg, C., et al. (2009). Holocene relative sea-level changes and vertical movements along the Italian and istrian coastlines. Quat. Int. 206, 102–133. doi:10.1016/j.quaint.2008.11.008