**Geomorphological Signature of Late Pleistocene Sea Level Oscillations in Torre Guaceto Marine Protected Area (Adriatic Sea, SE Italy)**

**Francesco De Giosa 1, Giovanni Scardino 2, Matteo Vacchi 3,4, Arcangelo Piscitelli 1, Maurilio Milella 1, Alessandro Ciccolella <sup>5</sup> and Giuseppe Mastronuzzi 2,\***


Received: 10 September 2019; Accepted: 12 November 2019; Published: 16 November 2019

**Abstract:** Morphostratigraphy is a useful tool to reconstruct the sequence of processes responsible for shaping the landscape. In marine and coastal areas, where landforms are only seldom directly recognizable given the difficulty to have eyewitness of sea-floor features, it is possible to correlate geomorphological data derived from indirect surveys (marine geophysics and remote sensing) with data obtained from direct ones performed on-land or by scuba divers. In this paper, remote sensing techniques and spectral images allowed high-resolution reconstruction of both morpho-topography and morpho-bathymetry of the Torre Guaceto Marine Protected Area (Italy). These data were used to infer the sequence of climatic phases and processes responsible for coastal and marine landscape shaping. Our data show a number of relict submerged surfaces corresponding to distinct phases of erosional/depositional processes triggered by the late-Quaternary interglacial–glacial cycles. In particular, we observed the presence of submerged marine terraces, likely formed during MIS 5–MIS 3 relative highstand phases. These geomorphic features, found at depths of ~26–30, ~34–38, and ~45–56 m, represent important evidence of past sea-level variations.

**Keywords:** morphostratigraphy; sea-level changes; marine terraces; river incisions; Adriatic Sea

#### **1. Introduction**

Ice-cores and marine sediment records indicate that the climate of the last 500 ky was characterized by ~100 ky warm–cold cyclicity [1,2] which led to repeated transitions between glacial and interglacial periods. These transitions triggered cycles of accretion and melting of the major ice-sheets with consequent major oscillations of sea-level position [3–5] and significant modifications of the on-land and sea-floor landscapes.

Morphostratigraphy applied to landscape evolution has allowed recognition of relict sequences of past morphogenetic processes associated with these glacial and interglacial climatic phases [6,7]. In particular, past landscapes have been often reconstructed through a combination of geomorphological and sedimentological analyses [8–10]. In the Mediterranean Sea, this multidisciplinary approach has proved to be very useful to understand genesis and evolution of particular landforms and sea-floor

features as well as to infer relative sea-level (RSL) changes and their influence on coastal landscape evolution (e.g., [11–14]).

However, many relict landforms are often difficult to observe and to analyze because they are located in submarine areas or covered by thick layers of more recent sediments. This is, for instance, the case of colder climatic phases when sea level was many m below the modern position, considering the tectonic contribution as well (e.g., [15,16]). This issue was progressively resolved thanks to recent technological advances that allow collecting high-resolution data to characterize the morpho-bathymetry and morpho-topography of a specific zone, even in underwater environments [17–19], using remote sensing techniques and spectral images.

In this paper, we investigated the coastal and off-shore zones of the Marine Protected Area (MPA) of Torre Guaceto (Adriatic Sea, SE Italy, Figure 1) through remote sensing techniques, morphological, and stratigraphic surveys in order to recognize and correlate chronologically subaerial and submerged landforms.

**Figure 1.** (**a**) Study area located in Apulia region; (**b**) location of Torre Guaceto surveyed area (Adriatic Sea, SE Italy); (**c**) surveys were performed from Punta Penna Grossa to Apani Islands.

Pleistocene marine terraces and Holocene deposits can provide an important insight into late-Quaternary rates of vertical displacements [20,21]. In south-eastern Apulia, evidence of Marine Isotope Stage (MIS) 5 are rare and poorly constrained in terms of chronology and elevation. The sole available indirect age comes from the south-eastern Murge (Torre Santa Sabina locality near Brindisi, Figure 1b). Here, a coastal deposit situated at ~3 m a.s.l. overlies a colluvial deposit bearing Late Paleolithic–Mousterian flints and, thus, it can be correlated to a generic MIS 5 [22,23]. In the northern part of Apulia, near Manfredonia (Figure 1a,b), MIS 5 deposits were found at depth of −22 m [24]; it implies significant subsidence rate (−0.17 mm/y) that are comparable to the areas placed in a very different geodynamic context (e.g., Trieste, Versilia, and Sarno plains [20]).

In this study, we identified the presence of relict marine terraces and other subaerial landforms presently lying along the Apulian continental shelf. We correlated them to the major late-Quaternary climatic phases coupling models with data available on land [20,21]. This analysis allowed reconstructing the coastal evolution of the Torre Guaceto Marine Protected Area (MPA) during the last 150 ky.

#### **2. Geological Settings**

Torre Guaceto is a tower of the XV century, placed on a promontory located north of the town of Brindisi, inside of the MPA of Torre Guaceto (Figure 1). The study area overlaps the Canale Reale river, which divides the Murge plateau from the Taranto-Brindisi plain [25,26]. Different lithological units crop out in this area, whose deposition is connected to past sea level stands and tectonic factors during Middle-Late Pleistocene [25,27]. The bedrock is represented by Mesozoic limestone, which is overlain by discontinuous marine deposits of Plio-Pleistocene age (up to 70 m), belonging to the Calcarenite di Gravina Fm, and the etheropic argille subappennine informal unit [28,29]. These units are covered by middle-upper Pleistocene biocalcarenitic beach and dune deposits [11,25,30]. Along the Murgia scarp they crop out as stepped terraces stretching from −400 m to a few m above the present mean sea-level. Although along the coastal area of Ionian Apulia the younger marine terraces are generally characterized by the presence of a senegalensis faunal assemblage that, in combination with U/Th ages, indicates its deposition during the MIS 5.5 (e.g., [23,31–33]), the Adriatic coastal area of Apulia does not permit any chronological attribution due to the lack of geochronological or paleontological markers. Every chronological attribution related to the upper/late Pleistocene derives from the use and applications of the morphostratigraphy principles [6]. All along the Adriatic coast stretching north to Brindisi and in particular between Torre Guaceto and Punta Penne, sea-level drop associated with the last glacial maximum (LGM, ~26 to ~21 ky BP, [34]), caused the incision of the basement, in correspondence on the current hydrographic network, showing features of sapping processes [11,23]. Seaward, sapping valleys cut the upper Pleistocene biocalcarenitic sedimentary cover. Surveys performed on the land allowed forming a hypothesis that the shaping of the paleo-beach-dune system occurred during MIS 5; the sapping valleys were shaped due to the increase of the relief energy caused by the lowering of sea level. Their maximum shaping would correspond to the lowermost sea-level stand (−120 m) at the LGM [11,25,27,30].

The following Holocene marine transgression (last 12 ky BP) and inter-strata dissolution produced a series of sub-circular inlets that host pocket beaches (Figure 2). On their border, a polyphasic dune ridge was recognized, formed by two aeolian sediment generations, the first at ~6.0 ky BP and the second at ~2.5 ky BP [11,25,30]. The south-eastern part of the foredune extends for ~500 m reaching a maximum elevation of −12 m msl. This is composed of brownish sand layers intercalated with brown soil rich in *Helix* spp. [30,35]. Geo-archaeological and bio-stratigraphic analysis indicate that RSL was 2.25 ± 0.2 m below the present one at about 3.5 ky BP and that the total RSL variation in the last ~2.0 ky BP was about of 0.9 m below the present mean sea-level (msl) [36,37].

**Figure 2.** The promontory of Torre Guaceto is shaped on a sequence of Calcarenite di Gravina Fm. (late Pliocene–Early Pleistocene) and of late Pleistocene–Tyrrhenian biocalcarenite [26].

The comparison of these data with the available GIA (glacial isostatic adjustment) models [38,39] indicates a low rate of tectonic subsidence of this coastal area at least during the last 125 ky. This is further corroborated by the absence of significant historical seismicity and by the GPS data that indicate zero to weakly negative on-going vertical movements [40]. Subsidence rate increases on the northern sector of Apulia, reaching values of −0.3 mm/y as indicated by tectonic structures observed in seismic profiles and boreholes off-shore the Gargano Promontory and Tremiti Islands [41,42]. In this tectonic framework the significant sea-level changes are added, in particular during the last 100 ky, when sea-level falls on MIS 5.4 and MIS 5.2 determined downward and seaward shifts of the shoreline and the decrease of sediment supply, possibly in response to the reduction in basin width that hampered lateral advection [43].

#### **3. Materials and Methods**

In this work, we merged high-resolution bathymetry derived by LIDAR remote sensing (with a penetration into water-column of ~50 m) techniques with detailed MIVIS spectral images (with a penetration into water-column of −7 m). We further corroborate these data by scuba diving surveys (e.g., [44,45]).

Remote sensing data consisted LIDAR data (Airborne Laser Terrain Mapper—ALTM Optech's Gemini 167 kHz, near infrared, Teledyne Optech, Toronto, ON, Canada) and Daedalus AA5000 MIVIS (Multispectral Infrared and Visible Imaging Spectrometer, Italian National Council Research CNR, Italy) hyperspectral images both for the on-shore and nearshore coastal zone (e.g., [17]). LIDAR Optech's Gemini 167 data are part of a wider acquisition of remote sensing data in the areas ascribed to the protected marine areas of Calabria, Campania, Apulia, and Sicily. This LIDAR system allows obtaining elevation data to an accuracy of 5 to 10 cm. This LIDAR system flies up to 4000 m to cover large coastal area where a high degree of accuracy and speed is necessary and where accessibility is difficult, as in back-dune zones or in steep coastal slope.

The surveys collected a point distribution, with each point consisting of x-y-z coordinates and associated reflectance value. Acquired points were processed in GIS environment to build digital surface models (DSMs) and digital terrain models (DTMs). In order to characterize the submerged coastal zone particularly for the near shore bathymetry up to −7 m, hyperspectral images have been used and acquired with MIVIS scanner. This instrument, property of the National Research Council of Italy, is a 102 channel scanner covering visible and near infrared (0.43–0.83 μm), middle infrared (1.15–1.55 and 1.98–2.50 μm) and thermal infrared (8.21–12.70 μm) regions of the electromagnetic spectrum, providing a wealth qualitative information of the surveyed area. MIVIS scanner has a geometrically correct scan line that, due to the movement of the aircraft, is displaced with the roll, pitch, yaw, and with changes in velocity and direction. The operational flight heights of the scanner can range from 1500 up to 5000 m above ground; at this height, the nadiral pixel dimension ranges from 3 up to 10 m, integrated with a GPS system and a gyro.

In GIS environment, LIDAR data and hyperspectral images have been combined to build digital terrain model (DTM) and digital surface model (DSM) with a grid cell width of 4 × 4 m, in order to define the major landforms occurring both along the coast and the shallow continental shelf up to a depth of −56 m (Figure 3).

Two scuba diving surveys across the incision immediately to the ESE of the promontory of Torre Guaceto (Figure 4) have been performed. The geomorphological surveys were traced up to a depth of about 18 m along both sides of the incision (Figure 5).

**Figure 3.** Morpho-topographic and morpho-bathymetric DTM (in blue) and DSM (in brown) of Torre Guaceto area with bathymetric profiles traces (in white) and direct scuba surveyed areas (in white dots).

**Figure 4.** The channel between Torre Guaceto promontory (reported in aerial view of Figure 3) and homonymous islands (in background); the sea-floor is cut by a sapping valley characterized by classic box profile, shaped in the Calcarenite di Gravina Fm. up to a depth of about 18 m.

**Figure 5.** Sapping notch and unstable block (**a**) and two large sub-horizontal steps (**b**) shaped on the southern slope of the submerged sapping valley between the islands and Torre Guaceto Promontory.

#### **4. Results**

The coastal tract comprised between Torre Guaceto and Punta Penne (Figure 1b), is characterized by a sequence of small calcarenitic islets (Apani Islands). They are made of cemented beach and dune sediments which likely represent the MIS 5.5 deposits according to the correlation with similar deposits outcropping all along the coast of Apulia. In addition, a further continuous dune belt can be found inland [23,25,31,32].

The composite survey campaign, performed in this study, allowed recognizing a number of significant submarine landforms; in particular, we observed the presence of near-flat surfaces and other morphological features likely related to submerged incisions. Bathymetric reconstruction revealed a staircase of near-flat surfaces which can be observed all over the different surveyed sectors (Figure 6).

**Figure 6.** Staircase geometry of near-flat surfaces and ravinement surface observed along the Torre Guaceto shelf.

These surfaces are characterized by gentle slopes which do not exceed 4–5 degrees (about 6%) occurring at four different depth ranges (represented in T1-T2-T3-T4 levels in Figure 7) in all the analyzed transects (A, B, C, and D).

**Figure 7.** Elevation profiles of Torre Guaceto MPA continental shelf. In blue, the upper limit of the surfaces (T1-T2-T3-T4) connected to the past sea-level stands are represented.

The T1 level occurs on shore; its inner (landward) and outer edge (seaward) are placed at 10 and 5 m, respectively.

The second surface (T2 level) occurs underwater. It develops between −26 (inner edge) and −32 m (outer edge). A third erosive surface (T3 level) occurs between −34 (inner edge) and −38 m (outer edge) while a fourth erosional surface (T4 level) ranges between −45 (inner edge) and −56 m (outer edge), with a great extent in correspondence of the profile AA' (Figure 7).

Our scuba surveys further documented a ravinement surface. This surface is gently sloping (2–3◦) seaward, developing between the present shoreline and −18 m (Figure 8). It is discontinuously covered by coarse and medium sands which, in some sheltered areas, define the present beaches [46].

**Figure 8.** Slope analysis of Torre Guaceto continental shelf;(**a**) slope changes in correspondence of the near-flat surfaces limits; (**b**) the mean morpho-bathymetry slope.

Both bathymetric data and scuba diving surveys showed the presence of submerged incisions orthogonal to surfaces boundaries and connected to the current land hydrographic network. These sea-floor features, recognizable along the whole investigated portion of the sea-bottom (e.g., from 0 to −56 m), cut all the submerged near-flat surfaces (T2 to T4, Figure 7).

#### **5. Discussion**

The new sets of topographic and bathymetric data reveal well preserved evidence of past sea-level stands that shaped the coastal and marine landscape near Torre Guaceto. Unfortunately, the geomorphological markers described in this study lack dateable material; therefore, the chronological constraint of their origin can be only speculated on the basis of bathymetric cross-correlations. In the first approximation, our geomorphological markers can be compared with modelled eustatic values [45,47]. The remains of the highest terrace (T1), most likely formed during the MIS 5.5 (~125 ky) when the RSL was −7 m above the present msl [20,23,37,48].

According to Rovere et al. [49] marine terraces can be shaped by marine erosion or can consist of shallow water to slightly emerged accumulations of materials redistributed by shore erosional and depositional processes (e.g., marine-built terraces) [50]. The width of marine terraces ranges from few hundreds of meters to up to 1–2 km and can stretch along many kilometers of coastline. The mapped submerged near-flat surfaces fit well with this morphological description. For this reason, we interpreted the surfaces as relicts of marine terraces. In the absence of any evidence of discontinuity in the sedimentary bodies, it is impossible to find any evidence of their depositional or erosive genesis. However, the hypothesis that these submerged surfaces could represent erosional marine terraces seems to be supported by seismic profiles performed in central Adriatic Sea, where a significant erosion of MIS 5.5–5.1 shelf progradational units during last 100 ky was observed [26,27].

The chronological frame of the submerged surfaces is complex. This is mainly because the presence of submerged features is very seldom reported in the Mediterranean [44]. The sea-level stands that shaped the reconstructed terraces must be located below both the present and the last interglacial ones. According to the available eustatic curves, different Marine Isotope Stages (3, 5.1, 5.3, 6.5, 7.1, 7.3, and 7.5) peaked below both MIS 5.5 and MIS 1 sea-levels. However, the evidence of river incisions shaped through sapping processes [11,23] is observed in all the submerged marine terraces. This incision reached maximum rate during LGM [6], when sea level was −120 m lower. For this reason, all surfaces must necessarily be older than MIS 1 and younger than MIS 5.5.

The submerged terrace found at depths −26 and −32 m (T2 level) can be tentatively attributed to MIS 5.3 (~101 ky BP) that peaked at −30 m on the sea-level curve by Grant et al., 2014 [5]. At MIS 5.1 (~81.5 ky), sea-level stand allowed the genesis of a new erosional marine terrace encountered at a depth variable between −34 and −38 m, corresponding to T3 level at −38 m. In the following phase, a sea-level drop was observed up to a new sea-level stand on MIS 3 at ~54.5 ky, with formation of marine terrace in correspondence of T4 level, at a depth variable between −45 and −56 m, peaked at −52 m on the sea level curve by Grant et al., 2014 [5].

The general tectonic framework of Torre Guaceto area reveals a low subsidence rate of 0.02 mm/y [15,22,23,51]. We corrected the current depth of terraces according to the subsidence rate in order to attribute the actual elevation for each boundary at the moment of their genesis (Table 1).

In the case of the MIS 5.5 surface, a displacement of 2.5 m in 122 ky has been calculated considering a tectonic rate of 0.02 mm/y [22,37,48]. This implies a corrected surface altitude ranging between 12 (inner edge) and 3 m (outer edge) at the moment of surface genesis.

For the MIS 5.3 surface, a displacement of 2.02 m has been calculated in 101 ky. It implies a depth range between −23.98 (inner edge) and −28.98 m (outer edge), while for the MIS 5.1 surface a displacement of 1.63 m has been calculated in 81.5 ky attributing a corrected depth range between −32.37 (inner edge) and −36.37 m (outer edge). Finally, for the MIS 3 surface, a displacement of 1.09 m has been calculated in 54.5 ky which implies a depth range between −43.91 (inner edge) and −54.91 m (outer edge) at the moment of surface genesis.


**Table 1.** Depth of surfaces detected in each profile corrected for the tectonic displacements.

These terraces can be referred to different relative past sea-level stands occurring during MIS 5-3, corresponding to peaks which can be observed in different model curves (e.g., [5,52–55] Figure 9).

**Figure 9.** Sea-level changes from 150 ky to the present derived from Red Sea records (modified after Grant et al., 2014). Light blue bands show erosional marine terraces depth ranges surveyed in the Torre Guaceto area.

The position of each marine terrace fits well with evidence of MIS 5 deposits already described by Mastronuzzi et al., 2011; 2018 [25,26]. This correlation was made under the assumption of minimal tectonic movements of this area [52] that show significantly different neotectonics pattern with respect to the northern part of the Apulia [21,24,34–36].

Morphobathymetry suggests that a valley network developed coeval with the sea-level stand through sapping processes [11,23]. In fact, since the evolution of the aquifer and the seawater/fresh water interface is strictly linked to the sea-level, the development of each sapping valley was largely influenced by the Late-Quaternary sea-level changes and the consequent shifts of the coastline [11,30]. Each sea-level (high) stand induced the development of a marine terrace, a shoreline and a number of short valleys. These valleys developed orthogonally to the coastline since the sapping processes was conditioned by structural alignment and/or by general geometry of the local basement [11]. Maximum incision occurred during MIS 2, when sea level was 120 m lower than present.

During the Holocene sea-level rise, incisions were flooded and filled by sediments, until the slowing down of sea-level rising rates (7 ky BP, [38,56], Figure 10), causing the widening of the ravinement surface between −18 m and the present sea level.

**Figure 10.** Relative sea-level prediction for Egnatia site [41]; the red dashed line indicates the slowing down of the sea-level rising rate at 7 ky.

Applying the morphostratigraphy principles on the near-flat surfaces depth—both subaerial and submerged—and to the sea-level trend recognizable during the last warming time, it is possible to reconstruct the following morpho-evolutionary steps:


#### **6. Conclusions**

Morphostratigraphic approach carried out in Torre Guaceto area allowed recognizing and chronologically correlating a variety of sea-floor features shaped by different sea-level stands in the Late-Quaternary. Last interglacial phases were already described in the literature [25,26,48] while, for the submerged features, new technologies have been useful to individuate the morphodynamic phases following the MIS 5.5 sea-level highstand.

Remote sensing and spectral images allowed detecting the underwater landforms which were used to perform a high-resolution mapping of submarine environments at large spatial scale. In particular, the use of LIDAR and MIVIS instruments allows surveying up to a depth of 55 m, highlighting the main underwater landforms as near-flat surfaces and submerged fluvial incisions.

Morphostratigraphic analysis of the sea-floor features allowed reconstructing the different shaping phases that were correlated to the chronological constrains deriving from the study of the shallow water and subaerial landforms. This analysis clearly indicated that the underwater marine terraces of Torre Guaceto were older than MIS 2; we tentatively attributed their formation in correspondence of the highstand peaks of MIS 5.5–MIS 5.3–MIS 5.1–MIS 3 (Figure 11).

**Figure 11.** Phases on Torre Guaceto area: (**a**) First Marine Isotope Stage 5.5 phase–(**b**) second Marine Isotope Stage 5.5 phase–(**c**) last glacial maximum (LGM)–(**d**) bronze Age–(**e**) present.

The morphostratigraphic approach presented in this paper represents a fundamental first step to investigate the submerged landforms; further investigation, possibly corroborated by additional coring campaign may provide more precise insights into the genesis of the submerged landforms and into the climatic phases that shaped them.

**Author Contributions:** Conceptualization: F.D.G., G.S. and M.V.; Data curation: F.D.G. and M.V.; Formal analysis: M.V.; Funding acquisition: G.M.; Investigation: A.P.; Methodology: M.M.; Project administration: G.M.; Resources: A.C.; Visualization: A.C.; Writing—original draft: F.D.G. and G.S.; Writing—review & editing: A.P., M.M. and G.M.

**Funding:** This research has been realized by a research agreement between the Consortium of Torre Guaceto and the Department of Earth and Geo-environmental Sciences of the University of Bari, approved in the Department Council on 1 March 2019.

**Acknowledgments:** This paper is the result of studies performed in the framework of the agreement between the Torre Guaceto Marine Protected Area and the Department of Earth and Geo-environmental Sciences of the University of Bari. We thank all collaborators for their logistic and technic support in every phase of this work. We are thankful to the reviewers for their revisions that allow us to improve the quality of the paper. MV is funded by the Rita Levi Montalcini programme of the Italian Ministry of University and Research (MIUR).This work has been carried out under the umbrella of the IGCP Project n. 639 "Sea-level change from minutes to millennia" (Project Leaders: S. Engelhart, G. Hoffmann, F. Yu and A. Rosentau). We extend our gratitude to the MOPP-Medflood (INQUA CMP 1603P) project for fruitful discussions during the workshops.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Sound Velocity in a Thin Shallowly Submerged Terrestrial-Marine Quaternary Succession (Northern Adriatic Sea)**

#### **Ana Novak 1,2,\*, Andrej Šmuc 2, Sašo Poglajen 3, Bogomir Celarc <sup>1</sup> and Marko Vrabec <sup>2</sup>**


Received: 30 December 2019; Accepted: 15 February 2020; Published: 18 February 2020

**Abstract:** Estimating sound velocity in seabed sediment of shallow near-shore areas submerged after the Last Glacial Maximum is often difficult due to the heterogeneous sedimentary composition resulting from sea-level changes affecting the sedimentary environments. The complex sedimentary architecture and heterogeneity greatly impact lateral and horizontal velocity variations. Existing sound velocity studies are mainly focused on the surficial parts of the seabed sediments, whereas the deeper and often more heterogeneous sections are usually neglected. We present an example of a submerged alluvial plain in the northern Adriatic where we were able to investigate the entire Quaternary sedimentary succession from the seafloor down to the sediment base on the bedrock. We used an extensive dataset of vintage borehole litho-sedimentological descriptions covering the entire thickness of the Quaternary sedimentary succession. We correlated the dataset with sub-bottom sonar profiles in order to determine the average sound velocities through various sediment types. The sound velocities of clay-dominated successions average around 1530 m/s, while the values of silt-dominated successions extend between 1550 and 1590 m/s. The maximum sound velocity of approximately 1730 m/s was determined at a location containing sandy sediment, while the minimum sound velocity of approximately 1250 m/s was calculated for gas-charged sediments. We show that, in shallow areas with thin Quaternary successions, the main factor influencing average sound velocity is the predominant sediment type (i.e. grain size), whereas the overburden influence is negligible. Where present in the sedimentary column, gas substantially reduces sound velocity. Our work provides a reference for sound velocities in submerged, thin (less than 20 m thick), terrestrial-marine Quaternary successions located in shallow (a few tens of meters deep) near-shore settings, which represent a large part of the present-day coastal environments.

**Keywords:** sound velocity; Quaternary sediment; submerged alluvial plain

#### **1. Introduction**

In geophysical (acoustic/seismic) investigations of the subsurface, velocity modeling is essential for converting two-way travel time of the observed reflections into the depth domain. Velocity data are routinely extracted from multi-channel seismic data [1]; however, in many circumstances, particularly in shallow near-shore settings, obtaining offshore multi-channel data is not feasible. Restricted navigation, legal constraints, busy marine traffic, relatively low resolution of the acquired data and the surveying cost itself often make acquisition and maneuvering with streamers and seismic sources impractical or even impossible. In such settings, high-resolution single-channel seismic and acoustic surveys provide a common alternative, but the velocity data must then be obtained by other means, such as in situ

measurements, laboratory core logging, and geo-acoustic modeling [2–20]. Due to costly offshore core drilling, these approaches are mostly focused on the upper few meters of the seafloor sediment. Consequently, the acoustic properties of surficial seafloor sediments have been well known for some time [4,5,11,15,19], but sound velocity in thicker (more than 10 m) sedimentary sequences has rarely been investigated (e.g., [17]). Therefore, when velocity data for depth conversion of single-channel seismic or acoustic data are not acquired during surveying, a velocity value corresponding to the surficial sediment grain size (e.g., [19]) or a previously published value from a nearby location is usually used. Whereas this approach is sufficient for geophysical surveys of uniform sedimentary layers, it produces significant uncertainties when dealing with pronounced lateral and vertical variability in sediment composition and architecture.

Typical example of such complex settings are shallow continental shelf areas drowned during the global sea-level rise that followed the Last Glacial Maximum (LGM) lowstand when the global mean sea level was approximately at −130 m [21,22]. During transgression, earlier terrestrial depositional environments (e.g., alluvial plains) were overlain by terrestrial and marine-derived sediments deposited in fluvial, estuarine, and open marine settings, resulting in complex sedimentary architecture and highly variable sedimentary types [22–24], which markedly affect the sub-bottom propagation of acoustic waves [1]. Precise mapping of the 2D and 3D geometries of the sedimentary bodies in these near-shore shallow environments and their appropriate time-to-depth conversion are essential for reliable interpretation of high-resolution acoustic and seismic surveys, not only for unraveling their depth and depositional history, but also for geotechnical site assessments in various engineering projects.

During the LGM lowstand, vast areas of the presently submerged continental shelves were exposed (e.g., Figure 1a) and amounted to approximately 40% of additional landmass in Europe and 5% globally [25,26]. Therefore, the shallow-most near-shore parts of presently submerged continental shelves extend over a considerable area globally and represent an important and often poorly studied geological environment. Due to the steadily increasing interest of the geological and archaeological research communities in shallow, presently submerged, and often buried landscapes [27–41], accurate depth conversion is crucial for future geological studies and paleoenvironmental reconstructions in such settings. We investigate an example of a transgressed and submerged alluvial plain in the Bay of Koper (Gulf of Trieste, northern Adriatic Sea; Figure 1) to provide sound velocity values for thin (up to 20 m thick) Quaternary sediments deposited in terrestrial-marine sedimentary environments located in shallow near-shore environments a few tens of meters deep. Our work ranks among the few studies that are not limited only to the surficial seafloor sediments, but also include the entire sedimentary succession from the seafloor to the base of the sediment on the bedrock.

#### *Setting*

The post-LGM sea-level rise induced significant changes in the sedimentary environments of the northern Adriatic Sea with terrestrial environments transitioning in paralic and later shallow marine environments [22,42–56]. In the Gulf of Trieste, where our study area is located (Figure 1), the Late Pleistocene alluvial plain transitioned into a paralic environment until open marine conditions finally prevailed approximately 10,000 years ago [22,42,44,45,55–64].

The Bay of Koper is located in the southeastern part of the Gulf of Trieste, which represents the northeasternmost extension of the Adriatic Sea (Figure 1). The seabed morphology of the Bay is smooth with depths ranging up to 20 m in the open part (Figure 1b). The main fluvial source draining into the Bay is the Rižana river with its mouth located in the reclaimed eastern part of the Bay within the Port of Koper complex. The smaller Badaševica stream is located west of the city of Koper.

The hinterland of the Bay of Koper is composed of Eocene turbidites (flysch) comprising interbedded sandstones and marlstones, which are overlain by Quaternary alluvial and paralic sediments in the valleys of Rižana and Badaševica (Figure 1b; [65,66]). Offshore the Eocene succession is unconformably overlain by Quaternary terrestrial and paralic sediments topped by Holocene marine sediments [55,56,59,62,67–71]. The Quaternary succession in the Bay of Koper, which was recognized as a submerged fluvial valley of the Rižana river (Figure 1b, [68–71]), is up to a few tens of meters thick and is composed of a lower alluvial part and an upper paralic part; however, alternations of terrestrial and paralic sedimentary environments have also been observed [68,72]. The alluvial sediments are generally composed of fine-grained clastic sediments with occasional gravelly and sandy horizons, whereas the paralic sediments are mostly composed of silty clay [68–70]. The Holocene marine cover comprises fine-grained bioclastic sediment with the surficial sediments showing a clear zonation: sandy silt near the coastline, clayey silt in the central part, and silt in the outer part of the Bay [62,68–71]. Holocene sedimentation rates in the Gulf of Trieste are relatively low and amount to a few millimeters per year [44,57,69,73]. Repeated multibeam bathymetric surveys in the Bay of Koper do not show significant changes in the seafloor morphology [74] and therefore imply a low-energy sedimentary environment.

**Figure 1.** Geographical location of the study area. (**a**) Regional map of the northern Adriatic, which was entirely subaerially exposed during the Last Glacial Maximum (LGM). Bathymetry data from [75]. (**b**) Geological map of the hinterland of the Bay of Koper. Two red rectangles in the Bay of Koper mark the areas investigated in this study. The red rectangle in the Bay of Muggia marks the study area of [76]. Bathymetric data are simplified after [77] and [78]. Geological data are simplified after [65,66,79–82].

#### **2. Materials and Methods**

We used archive geotechnical reports from the borehole database of the Geological Survey of Slovenia. Boreholes were mainly located in the NE part of the Bay (Figure 2) and were drilled in the late 1980s and early 1990s for geotechnical investigations supporting various infrastructure projects of the Port of Koper. Borehole metadata and descriptions are provided in Table 1 and Figure 5. The boreholes were drilled with rotary drilling and were cored. Sediment core samples were used for geomechanical testing and were not preserved. We therefore reconstructed the borehole sedimentary logs (Figure 5) from borehole descriptions contained in the geotechnical reports.

Sub-bottom sonar profiles were acquired in June 2016 on board vessel Lyra with the Innomar SES-2000 Compact sub-bottom sonar. Profile transects were designed to directly cross the borehole locations. Navigation north of the second pier was obstructed by a containment boom and very shallow water depths near the coastline. For this reason, some of the sub-bottom profiles are located at some distance from the borehole locations (Table 1 and Figure 2b). We used a transmitter frequency of 8 kHz. A total of 7 sub-bottom sonar profiles were acquired (Figure 2). At shallow depths, the seafloor was also observed visually to distinguish between sedimentary, rocky, and seagrass-covered seabed. The sub-bottom profiles were visualized and interpreted in the IHS Markit Kingdom software (Version

2018, IHS Markit, London, UK). Two-way travel times (TWTs; in milliseconds) of the seafloor (sfTWT) and the top of the weathered/compact bedrock (dTWT) were determined at borehole locations (Table 2).

**Figure 2.** Study area. (**a**) Bay of Koper with borehole (white circles) and sub-bottom sonar profile locations (thick black lines). (**b**) and (**c**) close-ups of the studied areas with bathymetry indicated by thin grey lines. Red lines show the locations of sub-bottom sonar profiles presented in Figures 3 and 4. White circles indicate borehole locations. For clarity the "A-III-" prefix is not shown in (**b**).


**Table 1.** Boreholes used in this study (GeoZS: Geological Survey of Slovenia; IGGG: Institute for Geology, Geotechnics and Geophysics Ljubljana).

**Figure 3.** Sub-bottom sonar profiles (with superimposed boreholes) where the top of the Eocene bedrock is expressed as a single undulating medium-to-high amplitude reflection ((**a**), (**b**) and (**c**)). For profile locations, see Figure 2. Blue overlay marks the Quaternary sediment and orange overlay marks the bedrock. Peat layers are indicated with brown arrows. Red overlay within the Quaternary section marks isolated diffraction hyperbolas. Red overlay in the water column marks reflection events above the seafloor. Green overlay marks the extent of seagrass meadows on the seafloor. Dredged areas are marked with grey arrows. Multiples are indicated by thin black arrows.

**Figure 4.** Sub-bottom sonar profiles (with superimposed boreholes) where the top of the Eocene succession is expressed as a medium-to-high amplitude reflection unit with a well-definable top from which downward shallow-dipping reflections emerge ((**a**–**c**)). For profile locations, see Figure 2. For an explanation of the color overlays, the reader is referred to Figure 3.


**Table 2.** Thickness of Quaternary sediments at borehole locations from borehole logs (thb) and sonar profiles (thTWT), the depth of the seafloor (sfTWT), and the top of the bedrock (dTWT) from the sonar profiles and the calculated average sound velocity in Quaternary sediments at the borehole location.

**Figure 5.** Borehole logs determined from geotechnical reports (see Section 2).

The thickness of Quaternary sediments in the borehole (thb, in meters) was derived from the geotechnical reports and was calculated as the depth from the top of the core to the top of the bedrock represented by weathered or compact Eocene turbidites (see Figure 5). The thickness of Quaternary sediments from the sub-bottom profiles (thTWT, in miliseconds) was obtained by subtracting the TWTs from the top of the bedrock (dTWT, in miliseconds) and the seafloor (sfTWT, in miliseconds) at the borehole location (see Figures 3 and 4; Table 2). When the profiles did not directly overlie the borehole (Figure 2b and Table 1), the part of the profile closest to the borehole was used to determine thTWT

(Figure 4a,d). The average sound velocity in Quaternary sediments at the borehole location (v, in meters per second; Table 2) was calculated using the following formula:

$$v = \frac{2 \times 10^3 \times th^b}{th^{TWT}}.\tag{1}$$

#### **3. Results**

#### *3.1. Boreholes*

Borehole logs are provided in Figure 5. The Quaternary succession (including the Holocene marine sediment) is composed of fine-grained clastics with occasional gravelly horizons. Only core A-III-8/90 contains sandy horizons. Soil-rich and peat horizons are present in some of the boreholes. Horizons with bivalves and/or gastropods occur in all boreholes; however, no remarks on the species or their environment are provided in the borehole geotechnical reports. Due to the lack of detailed descriptions of the boreholes, we did not attempt to interpret the sedimentary environments. However, it is clear that the Quaternary sediments comprise terrestrial-marine deposits. The bottom parts of all the boreholes consist of weathered and/or compact bedrock built of Eocene interbedded sandstones and marlstones.

#### *3.2. Sub-Bottom Sonar Profiles*

The sub-bottom profiles with superimposed boreholes are shown in Figures 3 and 4. Boreholes more than 5 m away from the nearest sub-bottom profile were projected orthogonally to the profile (Table 1, Figure 4a,d). The seafloor is marked by the first strong sub-horizontal reflection and is indicated by a blue arrow. The Quaternary sequence is indicated by a light blue overlay. Quaternary sediments are seen as (1) acoustically transparent units (Figures 3 and 4), (2) units containing onlapping or concordant reflection geometries (Figures 3 and 4), and (3) units with sigmoidal (prograding) reflection configurations (Figures 3a and 4b,c). Eocene bedrock is indicated by a light orange overlay. The often undulating unconformity at the top of the bedrock is expressed as (1) a medium-to-high amplitude reflection under which deeper reflections are not observed (Figure 3) or (2) an up to 5 ms TWT thick medium-to-high amplitude reflection unit with a well-definable top from which downward short shallow-dipping reflections emerge (Figure 4). The acoustic record does not discriminate between weathered or compact bedrock.

In addition to the two described units, the sub-bottom profiles contain several other features. Columnar-shaped reflections (gas flares) within the seawater column are located above the seafloor and are indicated by a light red overlay in Figure 3b,c and Figure 4c. Rough seafloor morphologies with plentiful diffraction hyperbolas are observed above the dredged areas (Figure 2a,b, Figures 3c and 4a,c), which were excavated to accommodate ships with larger drafts in the Port of Koper. Slightly rougher seafloor morphologies are also produced by seagrass (most commonly *Posidonia* sp.) meadows (marked by a light green overlay in Figures 3a and 4d), which were visually recognized during sub-bottom sonar acquisition. Significant, yet variable reflection degradation is seen directly beneath the areas covered by seagrass. Within the Quaternary sedimentary column, peat layers produce medium-to-high amplitude, sub-horizontal, 1–2 ms TWT thick reflections (Figures 3b and 4b). All sonar profiles show many small diffraction hyperbolas scattered between 25 and 15 ms TWTs within the sedimentary column (Figures 3 and 4). Seafloor multiples appear on the majority of the profiles (Figure 3a,b and Figure 4a–c) due to the shallow acquisition depths. Figure 3a also contains a multiple of the top of the bedrock.

#### *3.3. Average Sound Velocity in Quaternary Sediments*

Thicknesses of Quaternary sediments from drilling reports and geophysical data along with the calculated average sound velocities at borehole locations are provided in Table 2. The mean, median, and standard deviation for the whole dataset are 1523.6 m/s, 1539.7 m/s, and 144.0, respectively.

#### **4. Discussion**

#### *4.1. Sound Velocity Variation*

The sound velocity variation in the presented dataset is significant (Section 3.3). Some of the calculated velocity values can be considered less reliable due to significant distance between the respective borehole and its closest sub-bottom sonar profile (Table 2). If we omit the two boreholes that are separated by more than 20 m from their nearest sub-bottom profile (A-III-4/88 and A-III-10/90), the mean, median, and standard deviation for the dataset become 1556.4 m/s, 1569.2 m/s, and 143.1, respectively. Since the standard deviation remains relatively high, we discuss the principal influences on the sound velocity scattering below.

#### 4.1.1. Influence of Overburden

Sound velocity in sediments strongly depends on porosity, which is in turn influenced by overburden and compaction [1,4]. Here, the overburden comprises the combined weight of the water and sedimentary columns at the borehole locations. In Figure 6, we plot the calculated velocity against the thickness of Quaternary sediments taken from borehole logs and sonar profiles (a proxy for the weight of the sedimentary column), the depth of the seafloor taken from sonar profiles (a proxy for the weight of the water column), and the depth of the top of the bedrock taken from sonar profiles (a proxy for the weight of the water and sedimentary column). A strong scattering of plotted data points demonstrates that there is no relation between the determined sound velocity and these parameters. Additionally, in the correlation plots, the *x*-axis values of the minimum (red) and maximum sound velocities (green) are often quite similar (Figure 6a,c,d), again suggesting that velocities are uncorrelated to the overburden thickness. This shows that, in thin sedimentary successions located in shallow water depths, overburden does not significantly influence the sound velocity.

**Figure 6.** Plots of sound velocity versus (**a**) the thickness of Quaternary sediments (taken from borehole logs), (**b**) the depth of the seafloor (taken from sonar profiles), (**c**) the depth of the top of the bedrock (taken from sonar profiles), and (**d**) the thickness of Quaternary sediments (taken from sonar profiles). For clarity, A-III- and V- prefixes are removed from borehole labels. Data points in orange mark less reliable velocities (boreholes more than 5 m away from the nearest sub-bottom sonar profile; Table 2 and Figure 2b). Data points in green and red mark the maximum and minimum calculated velocities (Table 2).

#### 4.1.2. Influence of Grain Size

The influence of mean grain size on sound velocity in surficial marine sediments is well known [4,5,19,83,84]. Although granulometric analyses of the cored sediments used in our study were unavailable, general grain size classes could still be determined from borehole geotechnical logs (see Section 2). Therefore, an estimation of the influence of grain size on the sound velocity in our dataset is possible. In Figure 7, the calculated sound velocities at our study site were added to the plot, correlating sound velocity with the mean grain size of surficial sediments of continental shelves from [19]. Clearly, our calculated velocities (at boreholes located close to the acquired profiles) correspond well with the expected sound velocity range of the predominant grain size class determined from the borehole logs (Figures 5 and 7), even though the dataset of [19] is based on surficial sediment samples. Boreholes penetrating exclusively clay (A-III-6/88 and V5/95 Istrska) are in the lower sound velocity spectra, whereas boreholes mainly encountering silt (A-III-5/88, A-III-9/90 and A-III-11/90) are in the middle velocity spectra, corresponding to silt mean grain size velocities. Maximum sound velocity was calculated for Borehole A-III-8/90, which is the only core containing sandy sediment. Although clay dominates in this borehole, the amount of sandy sediment seems to be sufficient enough to significantly increase the average velocity in the Quaternary succession. A special case is presented by Borehole A-III-13/90, which penetrated clay; however, its calculated velocity corresponds to the mean grain sizes of silt. This discrepancy can be attributed to (1) an inadequate geotechnical description of the core, (2) an abrupt increase in sound velocity within the gravelly horizons (Figure 5), or (3) the presence of an overconsolidated layer within the sequence. In conclusion, the generally good agreement between the calculated velocities and expected velocities for the predominant grain size classes indicates that the composition of the stratal succession is a major factor influencing velocity variations in our study area.

**Figure 7.** The relation between sound velocity and mean grain size for continental shelf sediments (after [19]) with added velocities from our study. Our estimated sound velocities and the velocity value from [76] for a similar setting in the Gulf of Trieste (see Section 4.3) are shown by white horizontal overlays. Asterisk at Borehole A-III-11/90 indicates that this borehole is separated by more than 5 m from the nearest sub-bottom profile. Pie charts display the proportions of sediment types in each borehole core (for details, see Figure 5).

#### 4.1.3. Influence of Gas Presence

Abundant diffraction hyperbolas are present within the Quaternary succession (Figures 3 and 4; Section 3.2). They could be produced by reflections from gravel horizons; however, their occurrence does not correlate with gravel layers determined in the boreholes (Figures 3 and 4). The diffraction hyperbolas more likely indicate low concentrations of gas in the sedimentary column, commonly encountered in high-resolution geophysical profiles [85–87]. Reflective features in the water column (Figure 3b,c and Figure 4c; Section 3.2), which commonly result from gas-bubble plumes emitting from the seafloor [88–90], further indicate gas occurrence. The lowest sound velocity in our dataset was

calculated for Borehole A-III-7/88, which penetrates a hyperbola-dense zone clearly visible on the corresponding sub-bottom sonar profile (Table 2 and Figure 3c). Since even minor gas concentrations (1–2%) dramatically reduce sound velocity in sediments [83,84,91–93], we attribute this significantly lower velocity value to gas in the Quaternary sediment. In the northern Adriatic seabed, gas seeps are commonly observed and are attributed to both deep and shallow sources [94–98]. Since the diffraction hyperbolas in our sonar profiles are constrained only to a narrow zone in the uppermost part of the Quaternary sequence (Figures 3 and 4), we propose that the gas (probably methane) originates from a degradation of organic matter contained in the Holocene paralic and marine sediments and/or Late Pleistocene terrestrial sequences [62,97,99]. Gas production related to biological processes in seagrass meadows can also greatly hinder the propagation of acoustic signals [100–102]; however, the contribution of this effect is difficult to determine from our dataset since only a single borehole (V-5/95) is located within a meadow (Figure 3a). Nevertheless, significant signal attenuation below the meadows can be observed in the geophysical data (Section 3.2, Figures 3a and 4d).

#### *4.2. Are Average Sound Velocities an Oversimplification?*

Using average sound velocities for depth conversion of a highly heterogeneous Quaternary succession can be considered a gross oversimplification as the velocity strongly varies with grain size [4,5,19,83,84]. However, when comparing the sub-bottom sonar profiles and the borehole logs used in our study (Figure 5,Figure 3, and Figure 4), a good alignment between the reflections and the main sedimentological boundaries from the borehole logs is apparent (Figure 3b,c and Figure 4a,b). Especially peat layers prove to be very effective reflectors, which has already been noted in the northern Adriatic Sea by other authors [43,47,50–52,76,97,103,104]. This demonstrates that average velocity can be quite effectively used for robust depth conversion of sonar profiles in thin and shallow Quaternary successions.

#### *4.3. Comparison with the Sound Velocity from the Bay of Muggia*

An earlier study in a similar geological setting [76] reported sound velocity in Late Quaternary sediments from the neighboring Bay of Muggia (Figure 1b), which comprise Rosandra river deposits submerged in the Holocene transgression. There, the Quaternary succession is between 20 and 30 m thick and is composed of clay and silt with occasional sand and peat horizons. The water depth extends between 18 and 21 m. The Bay of Muggia site is therefore quite similar to our study site both in sediment thickness and composition. Using P-wave seismic refraction [76] led to a sound velocity value of 1595 m/s, which fits within the range of velocities estimated in our study (Figure 7). This implies that, also in the Bay of Muggia, the sound velocity in the Quaternary sediment is largely controlled by the sediment type and further corroborates sedimentary type as the major factor influencing sound velocity in shallow, thin, terrestrial-marine Quaternary sedimentary environment successions.

#### *4.4. Choosing the Appropriate Velocity for the Depth Conversion of Geophysical Data*

The results of our study show that sound velocity in thin (up to 20 m thick) submerged terrestrial-marine Quaternary successions located in near-shore areas few tens of meters deep is mostly controlled by the predominant grain size class of the succession (Section 4.1.2). The sound velocity for depth conversion in these settings can be chosen based on the predominant grain size class. We show that the sound velocity vs. grain size relationships previously documented in surficial sediments [19] are also valid for buried and submerged Quaternary successions (Section 4.1.2). Therefore, published values for sound velocity of surficial sediments can be utilized for depth conversion of shallow offshore high-resolution geophysical data, as long as the selected grain size corresponds to the predominant grain size class of the Quaternary succession.

Terrestrial-marine Quaternary successions often contain significant amounts of degrading organic matter; consequently, locally present gas further influences sound velocity in these settings (Section 4.1.3). Different gas indicators can easily be recognized from high-resolution geophysical data [85–93],

facilitating the mapping of low-velocity areas. As the velocity decrease associated with the presence of gas is often quite variable e.g., [93], we suggest avoiding detailed velocity analysis in gas-rich areas.

#### **5. Conclusions**

We used geophysical and borehole data to determine sound velocities through the Quaternary fill of a submerged alluvial plain containing terrestrial, paralic, and marine sediments. Our study shows that an average sound velocity through the Quaternary sedimentary column is sufficient for depth conversion of high-resolution geophysical profiles acquired in thin (up to 20 m thick) Quaternary successions in shallow (up to 20 m) water depths. We find that, in these settings, the main factor influencing sound velocity is the sediment type (i.e. mean grain size) contained within the studied sedimentary column, whereas overburden effects do not show any influence. However, where gas is present in the sedimentary column, it reduces sound velocity by a few hundred meters per second and becomes the dominant factor influencing sound velocity.

We found that, for a good approximation of the average sound velocity at a borehole, the velocity typical for the most represented sediment type in the borehole column can be employed. Nevertheless, in highly heterogeneous sedimentary settings, such as the Bay of Koper investigated in this study, significant lateral variations in average velocity will occur within a small area, necessitating a careful selection of multiple, most representative values, if relying on velocities published in the literature.

Using our study area in the northern Adriatic, we provided reference values for sound velocity in thin, mud-dominated Quaternary sedimentary successions in shallow coastal areas. Velocity values determined in our study correlate well with the sound velocity vs. grain size relationships previously documented in surficial sediments [19], showing that these published values can also be used for shallow sub-bottom sedimentary sequences.

**Author Contributions:** All authors have read and agree to the published version of the manuscript. Conceptualization: A.N. and M.V.; methodology: A.N.; validation: A.N.; formal analysis: A.N.; investigation: A.N., S.P., and B.C.; resources: S.P. and B.C.; writing—original draft preparation: A.N.; writing—review and editing: A.N., A.Š., S.P., B.C., and M.V.; visualization: A.N.; supervision: M.V.; project administration: A.N. and M.V.; funding acquisition: M.V.

**Funding:** This research was funded by: the Slovenian Research Agency, Young Researcher grant number 38136; the Slovenian Research Agency and Harpha Sea d.o.o., grant number L1-5452; the Slovenian Research Agency, grant number J1-1712; the Slovenian Research Agency, research programme P1-0195. The APC was funded by the Slovenian Research Agency, research programme P1-0195.

**Acknowledgments:** This work was supported by the Slovenian Research Agency (Young Researcher grant Nr. 38136), by joint funding of the Slovenian Research Agency and Harpha Sea d.o.o. within the project L1-5452 (Application of sonar in research of active tectonics and paleoseismology in low-strain environments) and by funding of the Slovenian Research Agency within the project J1-1712 (Record of environmental change and human impact in Holocene sediments, Gulf of Trieste) and within the research programme P1-0195 (Geoenvironment and Geomaterials). We would like to acknowledge IHS Markit Kingdom and their University Education Grant, which provided us with IHS Markit Kingdom software licences. The crew of the vessel Lyra (Iztok Rant and Rok Soczka Mandac) is acknowledged for their assistance and hospitality during sub-bottom acquisition. We would also like to thank Karoly Nemeth and an anonymous reviewer for their comments, which allowed us to improve the manuscript.

**Conflicts of Interest:** The authors declare that there is no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Submarine Geomorphology of the Southwestern Sardinian Continental Shelf (Mediterranean Sea): Insights into the Last Glacial Maximum Sea-Level Changes and Related Environments**

**Giacomo Deiana 1,3, Luciano Lecca 1, Rita Teresa Melis 1, Mauro Soldati 2, Valentino Demurtas 1,\* and Paolo Emanuele Orrù 1,3**


**Abstract:** During the lowstand sea-level phase of the Last Glacial Maximum (LGM), a large part of the current Mediterranean continental shelf emerged. Erosional and depositional processes shaped the coastal strips, while inland areas were affected by aeolian and fluvial processes. Evidence of both the lowstand phase and the subsequent phases of eustatic sea level rise can be observed on the continental shelf of Sardinia (Italy), including submerged palaeo-shorelines and landforms, and indicators of relict coastal palaeo-environments. This paper shows the results of a high-resolution survey on the continental shelf off San Pietro Island (southwestern Sardinia). Multisensor and multiscale data—obtained by means of seismic sparker, sub-bottom profiler chirp, multibeam, side scan sonar, diving, and uncrewed aerial vehicles—made it possible to reconstruct the morphological features shaped during the LGM at depths between 125 and 135 m. In particular, tectonic controlled palaeo-cliffs affected by landslides, the mouth of a deep palaeo-valley fossilized by marine sediments and a palaeo-lagoon containing a peri-littoral thanatocenosis (18,983 ± 268 cal BP) were detected. The Younger Dryas palaeo-shorelines were reconstructed, highlighted by a very well preserved beachrock. The coastal paleo-landscape with lagoon-barrier systems and retro-littoral dunes frequented by the Mesolithic populations was reconstructed.

**Keywords:** submarine geomorphology; morphostratigraphy; sea-level changes; Last Glacial Maximum; Sardinia; Italy

#### **1. Introduction**

Sea-level variations connected to climatic oscillations [1] cause changes in the landscape of coastal areas and continental shelves [2]. The comparative geomorphological analysis of emerged and submerged areas is particularly effective for revealing the landand seascape changes [3,4]. Landscape evolutionary phases can be reconstructed considering morphostructural and morphostratigraphic settings and using geomorphological, seismic, sedimentological, palaeontological, and isotopic data. The detailed reconstruction of the submerged coastal palaeo-landscape is useful to understanding the dynamics of the human population during the Last Glacial Maximum (LGM) [5,6]. As such, marine and continental geomorphological analyses are crucial for better representing and understanding the Pleistocene landscape evolution [3,7–11].

This study aims to obtain new insights into the palaeo-geographic evolution of the San Pietro continental shelf of southwestern Sardinia (Figure 1) during the last cold stage

**Citation:** Deiana, G.; Lecca, L.; Melis, R.T.; Soldati, M.; Demurtas, V.; Orrù, P.E. Submarine Geomorphology of the Southwestern Sardinian Continental Shelf (Mediterranean Sea): Insights into the Last Glacial Maximum Sea-Level Changes and Related Environments. *Water* **2021**, *13*, 155. https://doi.org/10.3390/ w13020155

Received: 27 November 2020 Accepted: 4 January 2021 Published: 11 January 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

(MIS 2) by analysing erosional and depositional landforms formed during the LGM sealevel lowstand, as well as the palaeo-geographic coastal evolution connected to the LGM sea-levels. Several studies used different methodological approaches and analysed various palaeo-sea-level indicators (e.g., palaeo-cliffs, lowstand depositional terraces, beachrocks, fossiliferous deposits) to evaluate the post-glacial sea levels in the Mediterranean Sea in the past 20 ka. Previous studies also successfully applied the glacial-hydro-isostatic adjustment (GIA) models [12–17].

**Figure 1.** Geographic location and structural setting of the study area: (**a**) Sardinia Island within the Mediterranean Sea; (**b**) San Pietro Island on the SW side of Sardinia; (**c**) structural sketch map of the Mediterranean area. Red lines mark thrust fronts; white line the Sardinian-Corse block translation 30 Myr BP; yellow line Sardinian-Corse block translation 25 Myr BP (mod. after Carminati and Doglioni, 2008 [18]); Black line the Sardinian-Corse block translation 14 Myr (Gattacceca et al., 2007 [19]); Light blue color line, isobath of −130 m represents the coastline during the Last Glacial Maximum (LGM).

> LGM shorelines are known from other areas of the Mediterranean Sea, including the Adriatic continental shelf [17,20–22], southern Tyrrhenian margin offshore Sicily [22–26], Calabria [27], and Malta [2,11,28,29].

> For example, morphobathymetric data acquisition (i.e., high-resolution multi-beam and seismic data) integrated with direct survey methods (i.e., remotely operated vehicle (ROV) and diving) allowed scientists to obtain a particularly rich database of the southwestern Sardinian continental shelf [30–35].

> Herein, we analysed the structural and volcanic geological settings linked to the Oligo-Miocene rifting of the western Mediterranean and Sardinian-Corsican blocks to highlight the geomorphological features of the continental shelf surrounding San Pietro Island. These data contribute to the knowledge of the coastal palaeo-landscape and its evolution from LGM to the Holocene (Figure 1a,b). In particular, submerged high rocky

coast morphotypes, a large palaeo-valley, a palaeo-lagoon and the successive phases of post-glacial sea level stationing were analyzed.

#### **2. Geological and Structural Settings**

The southwestern continental margin of Sardinia is characterised by normal faults that define intrashelf and intraslope basins [36]. This part of the Sardinian continental margin has been explored using geophysical surveys and deep drills, defining the order and geometry of the depositional sequences [36–41] (Figure 1c). High-angle normal fault systems characterised the western Sardinia continental margin setting between the Middle-Upper Oligocene and Miocene when, owing to the Apennine-Maghrebian chain orogeny, the intra-back arc basins opening caused the formation of an extensive system of rifts [42–45]. The genesis of the margin was clarified based on the ECORS-CROP Programme seismic data by examining the extensional tectonic inversion of a compressive structure of the Pyrenean western branch (Figure 1c) [39].

The margin formed as the transition between the western Mediterranean rift and the western branch of the Sardinian rift system and later assumed the structural and evolutionary characteristics of a divergent margin [36] (Figure 2). The kinematic analysis of the central Mediterranean shows that the Sardinia-Corsica block rotated until about 15 Ma later it became almost stable [18,19]. However, in the western part of the base of the margin, a significant earthquake (38.21◦–08.21◦; 5.4 Mw) was recorded on in August 1977 [46]. Furthermore, the INGV (Istituto Nazionale di Geofisica e Vulcanologia) earthquake catalogue, which contains the seismic records for the past 25 years, shows three other major earthquakes in southern Sardinia: one earthquake with a magnitude of 5.5 in August 1988 along the Sant'Antioco active fault, from Toro Island to Quirino Seamount, and two earthquakes with a magnitude of 4.5 in March 2006 at the sea prolongation of a major fault NW–SE Campidano graben that marks the western edge of this Plio-Quaternary graben. Therefore, slight fault movements that produce an occasional seismicity are still present and affect the margin.

**Figure 2.** Sectioned block diagram of the Sardinian southwestern continental shelf off the San Pietro Island. (**1**) Acoustic basement (volcanic complex—Lower-Middle Miocene); (**2**) lower sedimentary sequence (Middle-Upper Miocene); MS) Messinian erosional surface; (**3**) sedimentary sequence poorly or not stratified in the lower part, with undulating stratification in the upper part (Lower-Middle Pliocene); (**4**) upper sedimentary sequence, prograding complex of the external platform, superficial deposits in the proximal platform (Upper Pliocene—Quaternary) (after Lecca, 2000 modified [36]).

The sedimentary units preceding the Oligo-Miocene Sardinian rifting stage are represented by the Palaeozoic basement, marine clastic Eocene series, and fluvial sandstones and claystones of the Cixerri Formation (Upper Eocene to Lower Oligocene). The initiation of the Oligo-Miocene rifting was accompanied by the andesitic volcanism (Upper Oligocene to Aquitanian) of the Sulcis block. The subsiding basin was filled by the fluvial sediments of the Ussana Formation and the marine marly-arenaceous and carbonate sediments of the Lower Miocene [42].

The continental margin off San Pietro Island is characterised by a steep slope, which extends to the Sardinian-Balearic abyssal plain to a depth of approximately 2800 m [41]. The inner and intermediate continental shelf is characterised by the extensive outcrops of volcanic rocks, consisting of ignimbrites (comendites) and pyroclasts [35]. From a geochemical point of view, the rhyolites predominate, while the dacites characterise the basal volcanic formations [43,47]. Explosive volcanic eruptions occurred on San Pietro Island during the Burdigalian, Miocene (15–17 Ma). From a morphostructural point of view, the ignimbrite outcrops are characterised by wide mega-cuestas, calderas, necks, and dikes (Figure 2) [35].

The presence of the Oligo-Miocene volcanites at the tectonic block boundaries is marked by magnetic anomalies on the inner continental shelf [43] and was documented by analysing the rock samples from the lower margin of the Seamount Quirino [36] (Figure 3). On the distal shelf, the Miocene volcano-sedimentary and sedimentary strata rest on the volcanic substrate. The Miocene sedimentary sequence, up to the pre-evaporitic Tortonian marls, tends to be characterised by an erosional surface tied up to the Messinian eustatic fall [48]. In the lower part of the Miocene sedimentary sequence, the clinoform reflections are spaced wider, and the ages close to the Burdigalian are suggested [36] (Figure 4).

#### **3. Geomorphological Setting**

The major factor controlling the evolution of submarine canyons in the Mediterranean basin is the Messinian salinity crisis, which induced a significant forced sea-level fall of approximately 2000 m from the present-day sea level [42]. The consequent emergence of the continental margin led to intense erosion [37,48,49]. The following Pliocene flooding event deposited a thick mud drape over the entire continental shelf [7].

The shelf break is located at depths of 190–220 m and hosts the Plio-Quaternary prograding sedimentary wedge [35]. Both the shelf break and the Upper continental slope are eroded by the canyon heads formed via the retrogressive erosion processes. Intrachannel landslides are observed in the canyon sidewalls, while the Upper continental slope is distinguished by creeping areas and complex landslides often associated with pockmark fields due to fluid emissions [34,50].

The Messinian eustatic sea-level fall has been recognised on the Sparker seismic tracks acquired during the MAGIC (Marine Geohazard Along Italian Coasts) Sardinia Channel 2009 survey off Cala Fico. That study identified a palaeo-valley with polycyclic evolution that engraved both the volcanic substrate and the lower sedimentary sequence. Convoluted and plane parallel reflectors seem to characterise the Quaternary sequence.

Arenaceous beachrocks are represented by two extensive outcrops located to the north of La Punta and Piana Island at depths of 45–50 m. The outcrops display prominent erosion features both on the top surface and at their edges [51] (Figures 3 and 4).

**Figure 3.** Geolithological sketch map of study area from the Geological Map of Italy. Scale 1:50,000—Sheet 563 "Isola di San Pietro" (Rizzo et al., 2015 [35]).

**Figure 4.** Geological section from seismic data of: Oceanography and Seabed—Mineral Resources—PLACERS Project CNR—Profilo 2-78/1, Fix 115/127—Sparker 1KJ: (1) Volcanites (Lower-Middle Miocene; (2) emission chimney; (3) dikes; (4) marine sedimentary sequence with inclined reflectors (Middle-Upper Miocene; (5) Messinian erosional surface; (6) marine sedimentary sequence with undulated reflectors. (Pliocene-Pleistocene); (7) Holocene-current drape; (8) fault.

The beachrocks are slightly tilted seaward, presenting a typical character of beach sand bodies, with the sedimentary structures (e.g., parallel lamination and wedge-shaped, sigmoidal, and inclined stratification) common for coastal environments [52].

Considering the outcrop depth, these beachrocks are attributed to the end of the Younger Dryas event and are interpreted to be formed when the eustatic sea level dropped during the Pleistocene–Holocene marine transgression.

The actual and subactual sediments on the distal continental shelf off San Pietro Island are represented by pelitic sands and sandy pelites. These deposits contain variable bioclastic fractions, composed of foraminifera and the degradation products of algal bioherms. These algal bioherms colonise the rocky substrate and can be found both outcropping and suboutcropping [35] (Figure 2).

The Middle continental shelf has medium-grained, slightly pelitic sands, which bioclastic component increases towards the lower limit of *Posidonia oceanica* meadows. The areas farther offshore are dominated by biogenic gravels consisting of red algae (mäerl). These gravels form patches and hydraulic dunes. The near-continental shelf and the peri-littoral area are dominated by deposits linked to the retreat of high rocky coasts. In particular, base cliff deposits consist of sub-rounded heterometric blocks of volcanic lithology and landslide deposits with isolated sub-angular mega-blocks. The submerged beaches are characterised by medium- to coarse-grained sands with a predominantly quartz composition, whereas medium- and fine-grained sands are present in the bays of the southeastern sector. The sandy deposits with a predominantly quartz composition are located near the shoreline and Upper limit of the *Posidonia oceanica* prairie and have an important carbonate bioclastic fraction.

The first studies published on the LGM palaeo-shoreline of the western Sardinian shelf were conducted northward of our study area and indicated the existence of both erosional landforms and sedimentary sequences, in distal continental shelf, at depths between −120 and −140 m [53–56].

#### **4. Materials and Methods**

#### *4.1. Seismic Data*

The dataset used herein includes the seismic analogic data (Sparker 0.8 KJ) and by a high-resolution 3.5 kHz seismic sub-bottom profiler. These data were purchased from R/V Bannock (CNR) and collected during the oceanographic cruise "Placers 78/1" as part of the "Oceanografia e Fondi Marini" project. These data allowed the reconstruction of the Upper continental margin geological structure [36].

In order to reconstruct the palaeo-geomorphological setting, in particular, the intermediate continental shelf palaeo-hydrography, digital seismic surveys were carried out by R/V "Universitatis" during the oceanographic cruise "Canale di Sardegna 2009" in the frame of the MAGIC Project. The seismic surveys used a seismic energy source (Sparker 100/1000 J, Applied Acoustic CSP 20200, Great Yarmouth, United Kingdom), while the subbottom surveys aimed to reveal the structure of the surface deposits and were carried out using a geoacoustic source (Geochirp II-CP931, GeoAcoustics–Kongsberg, Great Yarmouth, United Kingdom) (Figure 5).

**Figure 5.** Data locations: spatial coverage multi-beam echosounder (MBES), side scan sonar (SSS) and uncrewed aerial vehicle (UAV); seismic profiles, scuba-dive and ROV stations; dredging sampling points.

#### *4.2. Multibeam, Singlebeam, and Side-Scan Sonar Data*

Morphobathymetric data were acquired during the oceanographic cruises "Canale di Sardegna 2009" and "Sardegna 2010" using R/V "Universitatis CoNISMa" as part of the Marine geohazard along Italian coasts (MAGIC) Project. The 50 kHz multi-beam echosounder (MBES, RESON SEABAT 8160) was calibrated with continuous sound velocity detection lines and vertical profiles. Onboard R/V Universitatis, the integrated system contained a motion sensor and gyro (IXSEA OCTANS) and a satellite differential GPS (Global Positioning System). The geocentric datum WGS84 and the UTM projection were chosen for navigation and display. The data collected during the survey were integrated with the Official National Italian Geological Cartography (CARG) project data.

Side-scan sonar data acquisition was performed on the proximal continental shelf with depths ranging from 10 to 50 m as part of the "Mapping of *Posidonia oceanica* meadows along the coasts of Sardinia" project funded by the Italian Ministry for the Environment on R/V "Copernaut Franca". A 100–500 kHz, dual-frequency sensor was used with a towfish (Model 272/T, EG&G Marine Instruments, Massachusetts, USA) connected to the Triton Elics system (Triton Elics International, Portland, OR, USA) with ISIS software (Triton) for geo-referenced acquisition and Delf Map for the construction and correction of the mosaic. The correct positioning of the acquired data was ensured by a GPS receiver with differential correction (Trimble 5007, Sunnyvale, CA, USA).

In the coastal areas with depths of 5–20 m, single-beam echosounder and lateral sonar data were acquired using a towed sensor (1 MHz, Starfish 990, Tritech, Aberdeenshire, Scotland) and sonar (200/800 kHz, Lowrance Elite 12, Tulsa, Oklahoma) with a hull transducer (Lowrance Simrad Active, Tulsa, Oklahoma) (Figure 5).

#### *4.3. Direct Seabed Observations*

On the internal shelf, diving surveys and sampling were carried out during the surveys "San Pietro Sub 2006" and "San Pietro 2010" as part of the "Official National Italian Geological Cartography" project. Fifteen underwater survey stations were set down to a depth of 50 m. Direct observations aimed to elaborate the interpretative keys for the geophysical data. Two teams of four geologists were engaged in the underwater surveys. The first geomorphological survey data were reported on tablets equipped with a depth gauge.During the underwater survey, sediment sampling was carried out with a vacuum core, whereas rocks were sampled with a chisel and heavy hammer. Six cores of unconsolidated sediments, five sedimentary rock samples (beachrocks and eolianites), and 20 samples of acid volcanites were collected. The data were synthesised using special survey cards (Figure 6). Direct seabed observations in the distal shelf areas and, in particular, the exploration of the palaeo-cliff walls at depths of 85–140 m were conducted using ROV *Polluce* III R/V Astrea (ISPRA, Istituto Superiore Protezione e Ricerca Ambientale). These surveys were carried out as part of the "CORALLIUM RUBRUM" and "MARINES-TRATEGY" projects, being sponsored by Italian Environmental Ministry—Autonomous Region of Sardinia. High-definition ROV images supported the habitat mapping of deep rocky bottoms dominated by red algal coralligenous assemblages and coral settlements (*Corallium rubrum* and *Leiopathes glaberrima*) [31,32]). These images allowed scientists to calibrate the geomorphological interpretation of palaeo-cliffs, especially regarding gravityinduced processes (Figure 5).

#### *4.4. Dredging and Shell Sampling for Radiocarbon Analysis*

The "SULCIS dredging survey" (2011) was conducted on the distal continental shelf, onboard R/V Gisella, using a classic submerged cylindrical dredger and two-cylinder experimental dredger. The dredging route was planned upon the analysis of morphobathymetric data and seismic profiles. The coordinates for the core sampling sites were determined using a differential GPS onboard the ship. The seabed depth at each core sampling point was acquired from the digital terrain models (DTM) processed using sonar multi-beam data. The volcanic rocks were not sampled because massive coralligenous bioconstructions with thicknesses greater than 50 cm covered them (Figure 5).

**Figure 6.** (**a**) Diver engaged in underwater geomorphological survey, using tablet with compass, clinometer, depth gauge and collimator at −15; (**b**) Fault mirror exhumed by erosion in Cala Fico at 10 m; (**c**) Foot cliff deposit with subspheroidal blocks at −18 m; (**d**) Lamination of pyroclastic lavas (Comenditi) in Cala Vinagra at 13 m.

#### *4.5. Aerial and Uncrewed Aerial Vehicle Inland Remote Sensing*

To obtain high-resolution aerial photos and topography suitable for the mapping of the onshore Sardinian coastal sector, we analysed the available topographic data produced by LiDAR (light detection and ranging) surveys. These aerial photogrammetric surveys were carried out by the Autonomous Region of Sardinia in 2008. The high-resolution aerial photos allowed us to analyse the coastal sector with high precision, down to a depth of 15 m. A cell size of 1 m and a mean vertical resolution DTM of approximately 30 cm were extracted.

In the most important sectors, such as the Capo Altano landslide, the surveys were performed with uncrewed aerial vehicles (UAVs, DJI Matrice 200, Shenzhen, Guangdong) equipped with a megapixel camera (ZENMUSE X5S 20.8). The survey was conducted by the UAVs flying at altitudes of 40–80 m above the ground level and maintaining a stable speed of 2.5 m/s. The acquired images were analysed and processed using the photogrammetric PhotoScan software (Agisoft, St. Petersburg, Russia). Being constrained by 12 ground control points, the resulting orthorectified mosaic and digital elevation model (WGS 84 datum and UTM 32N projection) had a cell size of 5 cm/pixel and were deemed precise enough to be used for geomorphological analysis (Figure 5).

#### *4.6. Data Processing and Cartography*

The MBES data covered 500 km2, with track lines parallel to the coast. The multi-beam data cleaning and filtering were performed using the PDS2000 software package, while the Global Mapper software was used to construct the bathymetric map of the UTM (WGS84) Zone 32 N projection. The bathymetry was plotted on a grid at 5 m node spacing as a contour plot to display detailed bathymetric information. It was also plotted as a slope

value and an illuminated 3-D perspective view to visualise prominent features within the investigated area.

Sub-bottom profiler (SBP) data processing was performed using the Triton Elics Information suite software package. The navigation data were plotted in a geographic information system (GIS) application (DelphMap). We exported the processed seismic data in the GeoTIFF format. The side scan sonar (SSS) data processing provided the georeferenced grey-tone acoustic images of the seafloor at a resolution of 1 m.

Bathymetry was investigated by analysing the acquired multibeam data, while for the Sardinia emerged coastal sector, DTM with a 5 × 5 m cell size of the Autonomous Region of Sardinia was used.

High-resolution multibeam bathymetry was combined with the echo-types of chirp sonar data, documenting the high morphological complexity of the study area. The geomorphological map of the study area was created using ArcGIS by analysing and interpreting data at a scale of 1:5000 to obtain a highly detailed and accurate final map.

#### **5. Results**

The San Pietro Island continental shelf morphology presents a strong structural control, in accordance with the tectonic style of the passive continental margin of southwestern Sardinia. A system of normal faults, including low-angle faults, predominates and likely led to the evolution of both intraplatform and intraslope basins (Figures 2 and 4). However, on the proximal shelf, structural morphologies predominate and are often linked to volcanic processes. The distal shelf transition is abrupt and is represented by a normal fault system trending 40◦ N in the northern sector and N-S in the central-southern sector at depths of 80–140 m. The fault walls show the morphological evidence of a polycyclic evolution in the marine, coastal, and continental environments. The continuity of the rocky outcrops is interrupted by extensive areas with very low slopes, where the surface deposits are represented by medium-grained sands with bioclastic components and mäerl biogenic gravels. These deposits are affected by hydraulic dunes, and their granulometric features are highlighted in the backscatter side-scan sonar images. *Posidonia oceanica* is nearly absent in the western and northern coastal strips and is limited to small discontinuous areas, where seagrasses are visible on the rocks at 10–25 m depth. Starting from Punta delle Colonne and towards the east and northeast, a large *Posidonia oceanica* prairie almost completely colonizes the San Pietro Channel [57].

High rocky coasts dominate on San Pietro Island, and the highest cliffs characterize the coast exposed to the NW waves. During extreme marine events, the waves in this area reach a height of 10 m and a length of over 200 m. In the western sector (Sandalo Cape), plunging cliffs consisting compact lava rocks prevail, while pseudo-stratified pyroclastic volcanites and cliffs with abrasion platforms often masked by large subangular rockfall deposits or cones with subspheroidal boulders characterize the northern and southern coasts.

Off the fault walls, at depths of 150–170 m, a small intraplatform basin is filled with the onlapping Miocene sedimentary strata and features an isolated outcrop of volcanites. The distal platform has a very low slope located at depths of 170–200 m and composed of fine-grained sands. The sand pelitic component increases towards the open sea up to the net topographic convexity of the shelf edge located at an average depth of 220 m.

In a context dominated by volcanic and tectonic-controlled morphologies, we detected several sea-level and climate-change indicators dating back to the Upper Pleistocene and Holocene. As such, morphometric data refers to the palaeo-stages when a basal platform was located at depths of 125–135 m; morphometric and palaeontological data indicate the existence of a palaeo-lagoon at depths of 120–127 m; seismic data allows the identification of a buried palaeo-valley with a base level at a depth of 130 m; and side-scan sonar and petrographic data reveal the presence of beachrocks at depths of 45–48 m (Figure 7).

**Figure 7.** Location of sites depicted in the following figures.

#### *5.1. Structural Landforms*

Tectonic control on morphology is evident both in the coastline area and on the continental shelf. A fault-controlled slope affecting the Cala Vinagra comendites was recognised at the base of Punta di Cala Fico promontory, with an edge located at a depth 5 m and a base at depths of 15–25 m (Figure 6b). Two fault-controlled slopes with the same lithology were found 800 m off Sandalo Cape and Punta Becco, with an edge at 10 m depth and a base at 40 m depth. In the western sector, fault wall alignment is controlled by the tectonic lineaments trending N and 345◦ N, long between 1 and 3 km, revealing an organised subparallel pattern. In the northern sector, the fault walls follow a 60◦ N line for approximately 5 km in the same direction as the tectonic lines that control the present-day high coastline from Capo Altano to Porto Paglia (Figure 8a).

Structural surfaces, linked to the submerged ignimbrite bedrock, characterize the entire intermediate continental shelf. They are irregular and are covered by superficial sediments up to 12 km off the coast of Punta Spalmatore. The open-sea limit is represented by the edge of the palaeo-cliffs controlled by the N–S trending faults at 90 m depth. These surfaces are interrupted using their relief due to differential erosion, necks, and dikes. The structural surfaces that are slightly inclined towards SW characterise monoclinal "cuesta" reliefs found off the coast of Cala Lunga (Island of Sant'Antioco). Some sectors (e.g., off the coast of Cala Fico) present fault control. The distal platform at depths of 150–190 m presents the Miocene sedimentary sequence outcrop (Figure 3).

**Figure 8.** (**a**) Digital terrain model (DTM) from MBES data showing the drowned volcanic landforms in the continental shelf of San Pietro Island. The great volcanic edifice off Corno Island and the main fault sistems are highlighted. (**b**) Aerial photo of NW sector of San Pietro Island; the lava flow structures are evident from Becco Nasca source area (white arrows) and the lava structures from the volcanic edifice off Corno Island, currently submerged (yellow arrows). (**c**) DTM shadow relief and morphometric sections of a tabular volcanic structure (neck?). (**d**) DTM shadow relief and morphometric sections of mega-dikes, in relief due to differential erosion. These morphologies rise up to 12 m above the basal erosion surface, with longitudinal development up to 10 km. (**e**) DTM shadow relief and morphometric sections of a volcanic crater, showing a double collapse-rim structure. (**f**) Crater 3D model, the lowered internal flank and the central depression are evident.

#### 5.1.1. Volcanic Landforms

The proximal portion of the continental shelf is dominated by the medium-Upper Miocene outcrops of acid volcanites. From a morphological point of view, numerous volcanic landforms were noted, including primary (e.g., craters or calderic depressions) and secondary landforms highlighted by differential erosion processes. The most important volcanic edifice on San Pietro Island was found on the continental shelf and occupies 37 km2, expanding up to 6.5 km off Capo Sandalo (Figure 8a). The emission centre corresponds to the Islet of Corno, where the ignimbrite lavas (Cala Lunga Group, Middle Miocene) of Cala Vinagra were sampled. Marine erosion processes partially eroded summit members, including the Becco Nasca's comendites (Figure 8b). The residual deposit is represented by the hills of Capo Sandalo and Monte della Borrona, where the undulated morphologies

of rope lavas attest that the lavas flowed towards SE (from the sea towards the interior of the island).

A larger emission centre was found 8 km from the Gulf of Mezzaluna. It presents a sub-circular and tabular mesa morphology with a diameter of approximately 1 km (Figure 8c). The relief basis is located at a depth of 100 m, and the top is found at 82 m depth. We interpret this landform as a volcanic neck; however, it should have been a huge volcanic chimney, the largest in Sardinia. This landform might also be interpreted as a volcanic plateau, similar to those recently recognised in the western Sardinian continental shelf [58], approximately 60 km north of the study area. These landforms are typical of basic volcanites and, therefore, appear unsuitable for the volcanic context of San Pietro Island. Regardless, this hypothesis requires further investigation. We attempted sampling the rocks of the volcanic neck by dredging. However, red algae bioconstructions completely covered the rocks and prevented us from sampling. A system of eight emission centres, showing neck morphologies, is distributed along a strip extending for 5 km to the SSE of the main volcanic edifice. A crater was found about 10 km off the coast of Punta Geniò, where only the eastern half-rim is preserved. It rises from the seabed at depths of 84–100 m. This volcanic edifice indicates two phases of activity (Figure 8e), and its morphology is similar to that of the volcanic features found on the seabed in front of the Phlegraean Fields, with their unlithified light grey pumiceous cinerite [59,60] (Figure 8f). A depression with a sub-circular perimeter and a diameter of 1200 m was found 1 km to the east the crater off the coast of Punta Geniò. The depression starts at a depth of 105 m and reaches 135 m. A depression with a similar morphology was classified as a caldera [61]. The entire group of emission centres following a tangential trend is crossed by a system of mega-dikes affected by differential erosion for more than 10 m [62]. The mega-dikes have a slightly sinuous form and can extend for up to 5 km without interruption, trending from 5◦ N to 350◦ N. The only exception is represented by a dike of considerable thickness, which follows a tangential trend near the main emission centre and is oriented NW-SE. This dike was likely emplaced subsequently to the N-S-trending dike system (Figure 8d). Such extensive mega-dikes are either contemporary or were formed immediately after the rifting phase [63]. The basement rocks are often draped by thin layers of mobile sediments, which partially cover the erosional landforms engraved in the volcanic substrate.

#### 5.1.2. Palaeo-Cliffs and Related Landforms

The morphobathymetric DTM of the studied continental shelf shows a clear discontinuity between the proximal and distal shelves. The discontinuity follows the offshore limit of the volcanites. This limit is represented by the alignments of rocky walls up to 50 m high with evident tectonic control and a prevalent orientation of 340◦ N.

The fault walls were subjected to polycyclic processes due to variations in the eustatic sea level in the cliff environment. Judging by the depth of the basal platforms (125–140 m), the last phase of subaerial erosion can be attributed to the LGM sea-level fall. These palaeocliffs are set in the volcanites, with their base locally reaching a depth of 145 m, and form plunging cliffs, similar to the modern cliffs along the Sandalo Cape coast [64].

The palaeo-cliff base is predominantly located at the shore platforms and is often affected by iso-oriented shallow erosive channels in line with the main tectonic lineaments. The basal abrasion platform has an irregular shape and is frequently covered by large subangular blocks of multi-decametric dimensions or by rockfall deposits. Some large blocks were found hundreds of metres from the detachment areas and recalled the evolutionary model of the block slides diagnosed in other submerged areas, such as the continental sector of the Gulf of Cagliari [34], the southern Apulian margin off the coast of Santa Maria di Leuca [65], and the Malta continental shelf [3,29,66,67] (Figure 9).The palaeo-cliff surfaces are often sub-vertical and are affected by sub-vertical fracture systems, which run parallel to the main tectonic lineaments. In some areas, sub-orthogonal joints are present and probably represent columnar cooling fracturing, similar to that in the southern coastal sector of Sardinia. The cliff summit edges are developed at depths of 80–90 m, exhibiting a

palaeo-cliff system 30–50 m high, on average. In some palaeo-cliff sectors, double ridges can be observed, pinpointing to the extensional trenches with counter-slope flanks. They were interpreted as distensional landforms and correlated with mass movement involving rotational kinematics (Figures 9b and 10).

**Figure 9.** (**a**) DTM from MBES data showing the Last Glacial Maximum (LGM) palaeo-cliffs. (**b**) DTM 3D from MBES data showing some landslides affecting the submerged palaeo-cliffs: 1—main scarp; 2—distensional trench; 3—landslide bodies; LGM palaeo-sea-level (blue line). (**c**) detail of tectonic controlled palaeo-cliffs with localization of morphometric profiles. (**d**) Morphometric sections and hypothesized sliding surfaces (red lines).

The interpretation of the kinematics of these drowned-landslides was based on the geomorphological surveys of similar palaeo-landslides located along the coast (between Capo Altano and Porto Paglia. In this sector, large landslides with rotational kinematics were systematically observed [68]. The first landslide is located 500 m north of Altano Cape, while the second landslide has been recently found to the south of Porto Paglia (Figure 10A). Both palaeo-landslides have their foot fossilised by regression eolianites (MIS 4, MIS 3). Therefore, their movement likely occurred at a high sea-level stand during the last interglacial period (MIS 5) [9] (Figure 10D). From a morphological point of view, the first palaeo-landslide is distinguished by a complex detachment niche and is organized in two scarps. A wide trench and a counter-slope terrace are considerably lowered and are partially covered by colluvial deposits. This landslide shows the evidence of recent reactivation. The second palaeo-landslide has a detachment niche with a single scarp, a counter-slope terrace at the base of the niche, and a trench partially buried by collapsed blocks (Figure 10C,D).

In order to correlate the shapes of the submerged palaeo-cliffs with the modern (subaerial) cliffs and relate them to the landforms associated with rock falling and toppling, we carried out proximity remote sensing surveys by UAVs on some modern cliffs engraved in the same volcanic lithologies on Sant'Antioco Island and Altano Cape.

**Figure 10.** (**A**) Location of the studied coastal palaeo-landslides; (**B**) Excerpt of Geological Map of Italy—scale 1:50,000 Sheet 555 "Iglesias" (Pasci et al., 2015 [33]), with landslides location; (**C**) Aerial photo, photo 3D view and Lidar DTM 3D of landslide 1 (**C1**) and landslide 2 (**C2**); (**D**) Palaeo-landslide 1 3D interpretive models (**D1**) and Palaeo-landslide 2 (**D2**), showing rotational kinematics, probably due to the basal erosion during the high-stand MIS 5.5 and the subsequent foot fossilization by continental deposits of the Upper Pleistocene (MIS 4,3,2). Geolithological legend: (1) sandstones and conglomerates, Cixerri Formation—CIX (Eocene-Oligocene); (2) ignimbrites, tuffs—AQC (Middle Miocene); (3) ignimbrites, lavas—SRC (Middle Miocene); (4) eolianites (Upper Pleistocene—MIS 4-3?); (5) eolianites and colluvia—PVM (Upper Pleistocene—MIS2). Morphological legend: (a) detachment niche; (b) trench; (c) counter-slope terrace; (d) rotational sliding surface; (f) cliff engraved in the Pleistocene aeolian deposits that fossilize the landslide foot.

The ROV surveys allowed us to explore the palaeo-cliff morphology, particularly, the extensional trenches. We observed that erosional channels interrupted the continuity of the cliff, whereas the niches and hollows, formed by differential erosion, created the environments protected from the coelenterate colonies of *Corallium rubrum* (SDC—the biocoenosis of semi-dark caves). By contrast, the top surfaces of the cliffs were almost completely colonised by incrusting algae *Pseudolitophillum expansum* (coralligenous biocoenosis).

5.1.3. Fossil Palaeo-Valleys

The LGM palaeo-hydrographic network has only been partially recognised because the valley incisions are buried by very coarse-grained and gravelly bioclastic sands, especially the mäerl facies, which inhibit the penetration of the chirp elastic signal. The only buried palaeo-riverbed that was completely identified starts from the Ria di Cala Fico and is demarcated by a fault wall that continues for approximately one km offshore trending 280◦ N. The palaeo-riverbed top is located at depths of 5–10 m, and the base is demarcated at depths of 20–35 m (Figure 6b). Offshore, the palaeo-drainage system is deflected by an orthogonal fault system (Figure 11).

**Figure 11.** Seismic data of LGM palaeo-valley.

Beyond the intermediate platform with depths of 80–90 m, the morphobathymetric DTM shows a significant incision with steep rocky slopes, which is partially filled with sediments. This surface morphology possibly masks a buried palaeo-valley. To investigate this incision further, a Sparker 0.1–1 kJ seismic survey was planned and carried out. The survey proved the existence of a palaeo-valley, whose incision started in the Middle Miocene, immediately after the emplacement of the lava volcanites (Comenditi Cala Lunga Group). Subsequently, the incised valley was filled by marine sediments with inclined stratification and partially interbedded with the predominantly pyroclastic volcanites of the Middle-Upper Miocene (ignimbrites of the Cala Lunga Group) (Figure 11). The next erosive event was probably related to the Messinian low eustatic sea level [69]. In Plio-Quaternary, the palaeo-valley was completely filled by shallow-marine deposits with wavy laminations. The last identified incision down to a depth of 115 m occurred during the LGM sea-level low stand. The valley was filled by sediments originating from an environment with low wave energy, where low lighting caused a decrease in bioclastic productivity. These deep palaeo-valleys were discovered in a lower to Upper offshore environment due to the fast post-glacial to Holocene sea-level rise. In seismic images, the valley infill is represented by semi-transparent sandy mud alternating with more reflective sands of episodic storm nature (Figure 11—Section 2).

The terminal section of the palaeo-valley is enclosed within a narrow incision with walls approximately 10 m high, where we interpreted a submerged palaeo-delta at depths of 130–140 m.

#### 5.1.4. Palaeo-Lagoon

Approximately 7 km off the coast of Punta Spalmatore, the palaeo-cliff is interrupted by a deep incision that is connected shorewards to a large depressed area at depths of 120–130 m. During the LGM sea-level low stand, this depressed area could have formed a Ria with a head bay lagoon. The palaeo-lagoon is asymmetrical, being characterised by a southern arched bank with a low slope and a deeper northern rectilinear bank (Figure 12a).

To verify the existence of this lagoon, we sampled the relevant deposits by dredging. We used a two-cylinder dredger and started from a depth of 125 m (lat 4329430.040 N, long 426914.014 E) to a depth of 129.8 m. The base was located close to the rocky wall (lat 4329755.209 N, long 426723.042 E), following a 200 m long cross-shaped church (Figure 12a,b).

The dredger sampled compact greenish-grey sandy silt, which, when washed, revealed a significant fossil content with both intact and fragmented lamellibranchs, gastropods, and serpulids (Figure 12b), marking the transition from the meso-littoral to the infralittoral planes (Figure 12c) [30]. The sampled fauna comprised species common for lagoon and meso-littoral environments, such as the Bivalvia (e.g., *Mytilus* cfr. *Edulis, Mytilus* cfr. *Galloprovincialis, Glycymeris* sp., *Parvicardium exiguum, Pitar* cfr. *Rudis, Venus* cfr. *Casina*), Gastropoda (e.g., *Tectura virginea*, *Calliostoma laugieri*) (Figure 12d) [70], and Annelida (e.g., *Serpula vermicularis*).

Several samples were subjected to AMS (accelerator mass spectometry) 14C radiocarbon analysis at the Beta Analytic laboratories (Florida, USA). The radiocarbon analysis results confirmed that the sampled rocks were deposited during MIS 2, at the beginning of the deglaciation period (Table 1) [71]. The dating of *Tectura Virginia* was the closest to that of LGM, and this species is still present in the Mediterranean Sea and some areas of the Aegean and North Adriatic Seas. The most consistent populations of *Tectura virginea* are currently present in the eastern Atlantic (e.g., Scotland, Iceland) and the North Sea (e.g., Norway, Svalbard).

**Figure 12.** (**a**) Morphobatimetric DTM shadow relief showing the depression that hosted the palaeo-lagoon in the LGM; the arrow shows dredging (DR3); (**b**) sampled thanatocenosis. Bivalvia: (1) *Mytilus* cfr. *Edulis*, (2) *Mytilus* cfr. *Galloprovincialis*, (3) *Glycymeris* sp., (4) *Parvicardium exiguum*, (5) *Pitar* cfr. *Rudis*, (6) *Venus* cfr. *Casina*; Gasteropoda: (7) *Tectura virginea* (López Correa et al., 2010 [70] for the distribution of the *Tectura virginea* species). (**c**, **c1**) Ascending two-cylinder experimental dredger; the arrow indicates part of the dark grey sandy silt deposit; (**d**) enlarged image of *Tectura virginea*.



#### 5.1.5. Beachrocks

Limited cemented conglomerate and sandstone outcrops were noted in the northern sector of the shelf at depths varying between 45 and 49 m (Figure 13a). These outcrops are represented by polygenic and heterometric conglomerates alternating with arenaceous microconglomerates with a feldspar-quartz matrix. The fossiliferous content is high, predominantly including lamellibranchs and gastropods, with the evidence of radiolarians and echinoids. Carbonate cementation was determined to be polyphasic, with the initial formation of magnesian calcite cement precipitated from seawater in the form of acicular crystals and followed by the cryptocrystalline globules deposited via bio-precipitation. After partial dissolution, the cementation was completed by the idiomorphic crystals of calcite deposited from freshwater. The dynamics of cementation are linked to the sea-level oscillations during an overall sea-level rise (Figure 13d). These deposits show depositional characteristics and cementation typical of a beachrock and, consequently, can be referred to as palaeo-submerged shorelines. They define the exact position of an intertidal zone. The main beachrock deposits were observed and mapped at the same depth, approximately

3 km NW of the La Punta promontory. The side-scan sonar images showed weakly inclined banks dipping seaward. Owing to the gradual basal undermining of these beachrock outcrops by strong traction currents, extensive plains are covered by unconsolidated sediments (sand) distributed in patches (Figure 13a). The top outcrop surface is denoted by a typical sub-orthogonal fracture system linked to diagenesis, favouring the occurrence of landslides at the edges (Figure 13b,c).

**Figure 13.** (**a**) The 100 kHz side scan sonar image of the beachrock a depth of 45 m: (1) Outcrop of sandstones and micro-conglomerates with carbonate cement; (2) sandpatches sedimentary structures (mäerl); (3) medium and fine grained bioclastic sands. (**b**) Seabed picture a depth of 45 m, underwater survey of beachrock fracture system. (**c**) Detail of the beachrock outcrops affected by a sub-orthogonal fracture system and by block toppling due to basal erosion: 1—35◦ N fracture system; 2—300◦ N fracture system; 3—detachment niche; (**c1**) block diagram by diving survey. (**d**) Petrographic thin section of the beachrock—45 m: (1) acicular magnesian calcite coating: (2) micritic globular filling; (3) dissolution cavity with secondary clastic-micritic filling; (4) accretion of idiomorphic calcite in dissolution cavity. N.I. × 20.

#### **6. Discussion**

#### *6.1. Last Glacial Maximum (LGM) Coastal Palaeo-Landscape*

The collected data allowed us to identify the geomorphological evidence of a drowned palaeo-landscape attributable to a base level at approximately 125–145 m depth below the present-day sea level.

Continental landforms (e.g., river valleys and coastal plains), transitional environments (e.g., palaeo-lagoons and coastal landforms represented by cliffs), and associated depositional features were recognized. Different marine indicators observed at various depths helped identify a palaeo-sea-level, suggesting that the drowned palaeo-landscape formed between the LGM and early deglaciation stages [22,24]. These observations testify to the particular mobility of this continental shelf to hydro-isostatic rebounds [12], similar to those reported from the northern sector of San Pietro Island, from Fontanammare Bay to Capo Frasca (Figure 14) [54]. In our study area, the MIS 5 palaeo-sea-level indicators show relative tectonic stability [72,73], while those modelled by GIA on the southwestern Sardinian continental shelf display lower vertical displacement rates, with glacio-hydroisostasy constituting 0.62 mm/yr off the coast of Oristano and 0.60 mm/yr in the Gulf of Palmas [12]. Therefore, the theoretical sea-level drop attained during LGM (−120 m) could be extended by approximately 10 m (−130 m). Some research carried out on the continental shelf between Capo Pecora and Oristano indicates that the LGM palaeo-shoreline was at depths of 125–140 m [54] (Figure 14).

**Figure 14.** Geomorphological sketch of the San Pietro continental shelf. Submerged palaeo-landscape since LGM (20 ka) to 9 ka.

The LGM coastal palaeo-landscape of the study area can be subdivided into three sectors. The western sector is dominated by cliffs affected by landslides, where several large isolated blocks represent the islets and a deep Ria interrupts the cliffs' continuity and incises the head bay lagoon. The northwestern sector represents a high rocky coast with a gentle slope, probably coinciding with the lava flow fronts originating from the emission centre of Corno Island. An extensive basal wave-cut platform is situated at the base of these rocky coasts and affected by a dense system of iso-oriented erosional channels, with a prevalent direction of 330◦ N. The northern sector of Altano Cape is characterized by a low rocky coast with a series of beachrocks at depths of 115–120 m. It is located adjacent to two back-littoral areas with palaeo-lagoons. These beachrocks lie in a sub-parallel way and have a morphological response identical to MBES, in contrast to the beachrocks sampled at a depth of 45 m in the same sector (Figure 14).

#### *6.2. Post-Glacial Palaeo-Landscape Evolution*

Rapid sea-level rise from 130 m below the present-day sea level during the deglaciation period [22,65] likely contributed to the destabilisation of the palaeo-cliffs [74], with their geomechanical characteristics being worsened by periglacial processes. Such destabilisation could explain many massive rockfall deposits that currently cover the basal abrasion platforms.

The comparison between the geomorphological features of the subaerial coastal palaeolandslides (Figure 10) and the identified submerged palaeo-landslides (Figure 9) has shown that both types of landslides were affected by rotational kinematics. However, the submerged palaeo-landslides have low-angle sliding surfaces, and their landslide bodies are stacked (Figure 9c), dissimilar to the present-day landslides. These differences can be linked to instability after drowning, in wave energy conditions probably very different from the present ones [75,76]. To better understand the coastal landslides kinematics in this area, it is important underline that the waves currently interacting on the coasts of western Sardinia are the most energetic in the entire Mediterranean basin. In fact, the waves measured in Alghero (northwestern Sardinia) during extreme meteorological events have a maximum height of over 10 m [77,78].

#### *6.3. Younger Dryas Coastal Landscape*

At depths of 45–48 m, we diagnosed a littoral spit in the beachrock facies. It is well preserved and extends for approximately 15 km in the northern part of the study area (from La Punta to offshore of Capo Altano). The outcrops are represented by polygenic and heterometric conglomerates with a sandy matrix and carbonate cement. The fossil content varies, and fully bioclastic levels were observed. Rock cementation indicates that a palaeo-depositional environment changed from intertidal to supratidal [52]. The outcrops are characterised by apparent erosional landforms both on the top surface and the edges (Figure 13a).

The strata are tilted slightly seaward, representing an arrangement typical of the beach sedimentary body [51]. The sedimentary structures (e.g., parallel lamination and wedgeshaped, sigmoidal, and inclined stratification) are also common in coastal environments (Figure 13b).

In Sardinia, beachrocks are found at different bathymetric levels. The deepest beachrocks are located at 95–110 m depth off La Maddalena Island, whereas the shallowest beachrocks are found at a depth of 1 m in the Gulf of Palmas and incorporate Roman pottery [30]. The latter type of outcrop constitutes thin (1–2 m) and discontinuous strata.

The beachrocks found at depths of 45–50 m are particularly thick (4–5 m) and continuous. They start in southern Sardinia, pass along eastern Sardinia, and reach the Aléria Platform in central-eastern Corsica. These littoral spits are often associated with retrolittoral areas (i.e., palaeo-lagoons and palaeo-dunes). The same beachrocks were sampled at a depth of 45 m in the Gulf of Palmas, Gulf of Cagliari, Island of Serpentara, and Gulf of Orosei. The 14C analysis indicated that the beachrock ages range between 11 and 9.5 ky cal BP. In particular, the Cagliari beachrock revealed a date of 10,835 ± 170 ky cal BP [51]. As such, these palaeo-shorelines are attributed to the Younger Dryas cycle of eustatic oscillations. Off the Porto Paglia coast, the beachrock is interrupted, and the area behind it is distinguished by a sub-elliptical depression, interpreted as a palaeo-lagoon approximately 1 km in diameter (Figure 14).

#### *6.4. Lower Holocene Coastal Palaeo-Landscape*

We reconstructed the LGM coastal palaeo-landscape corresponding to the Upper Palaeolithic period. During this period, the evidence of human presence is still rare in Sardinia [79,80] (Figure 14), despite the short distance from the continent and the continuity with Corsica due to the lowered sea level during the LGM. [56]. The first evidence of human settlement in Sardinia is attributed to the Holocene [81]. The discovery of the Mesolithic site of S'Omu e S'Orku (SOMK) along the southwestern Sardinian coast, about 40 km from San Pietro Island, is of particular interest, being one of the few Mesolithic coastal sites in the western Mediterranean (Figure 15). By examining the sea-level rise curve [15,16] (Figure 16), we placed the ancient Holocene shoreline at 20 m depth. In the Holocene, the palaeo-landscape represented a vast coastal plain that extended between the present-day islands of San Pietro and Sant'Antioco and the mainland. High rocky coast was interspersed with extensive beaches with coastal dunes, as evidenced by a strip of coastal desert near Capo Altano [9]. A narrow bay was bordered by two rocky promontories between the islands of San Pietro and Capo Altano, while instead of the San Pietro canal, there existed a large bay and river mouth (Figure 14). The coastal morphological context of this part of Sardinia probably influenced the movements of the last Mesolithic groups. The human remains discovered in SOMK were dated to approximately 9 ky cal BP. The site is completely covered by red ocher and jasper artefacts (Figure 15b), the outcrops of which are found on San Pietro Island [81]. These artefacts testify to the movement of Mesolithic groups along wide beach areas close to the reliefs. Those Mesolithic groups benefited from the emerged land area between the mainland and the two small islands, where coastal lagoons favoured not only the mining of jasper and ocher but also offered food resources from sea and lagoon, especially in a period, when relatively scarce terrestrial fauna existed. The terrestrial fauna was mainly represented by the now-extinct genus *Prolagus* [82].

**Figure 15.** Mesolithic man in Sardinia west coast: (**a**) Location of S'Omu e S'Orku (SOMK) site and study area, palaeocoastline of Mesolithic period (9 kyr BP) at 20 m. (**b**) The heavily ochre-stained skeletal remains (after Melis and Mussi, 2016 [81]). (**c**) View of SOMK Mesolithic site. (**d**) General view of the present-day position of Mesolithic SOMK site along the Sardinian west coast.

#### **7. Conclusions**

The integrated study of new geomorphological, seismic, MBS ultrasound, direct and remote-sensing UAV data allowed the evolution of the coastal palaeo-landscape of the continental shelf off San Pietro Island (southwestern Sardinia) since the LGM to be reconstructed. We found robust evidence that during LGM, the sea level was approximately 130 m depth below the present-day level, in agreement with GIA model indications (Figure 17) [12].

The morphostratigraphic investigation carried out around San Pietro Island allowed recognizing sea-floor features and landforms related to different sea-level stands during the last 22 kyr. In particular, the research highlighted the following (Figure 14):


**Figure 16.** Relative sea-level prediction curve after Lambeck et al., 2011 [12] and Lambeck et al., 2014 [15], associated with palaeo-sea-levels indicators of San Pietro continental shelf: lagunar shells, (C1) *Acmea virginea* (gasteropoda)—18,982 ± 338; (C2) *Mytilus galloprovincialis* (Bivalvia)—15,350 ± 338 yr cal BP; BR) beachrock—10,835 ± 170 ky cal BP (De Muro and Orrù, 1998 [51]); SOMK) Mesolithic site (Melis and Mussi, 2016 [81]).

In literature, LGM palaeo-shorelines are normally investigated by focusing on single research aspects, such as seismic stratigraphy [20,21,27], lowstand depositional terraces [23] and high rocky coast evolutionary models [22]. Within our research, an effort was made to integrate a series of datasets including geological data and geomorphological evidence from both emerged and submerged areas of southwestern Sardinia. This made it possible to reconstruct the drowned palaeo-landscape in its complexity providing the means to infer evolutionary phases from the deglaciation to the present. With reference to the continental margin, the research allowed for the first time the description of geomorphological features and palaeo-landscapes associated with the LGM shoreline. Furthermore, chronological constraints for the development of peculiar landforms were achieved, thanks to the dating of correlative fossiliferous deposits.

**Figure 17.** Reconstruction of the palaeo-landscape evolution of the San Pietro continental shelf since the LGM to present: (**a**) LGM; (**b**) Younger Dryas; (**c**) Holocene, Mesolithic; (**d**) Present.

**Author Contributions:** Conceptualization, G.D., L.L., M.S. and P.E.O.; methodology, G.D. and V.D.; validation, R.T.M.; formal analysis, V.D.; investigation, G.D., P.E.O. and V.D.; resources: G.D. and P.E.O.; writing—original draft preparation, G.D., L.L., R.T.M., M.S. and P.E.O.; writing—review and editing, L.L., M.S., V.D. and P.E.O.; visualization, G.D., P.E.O. and V.D.; supervision, P.E.O. and M.S.; project administration, P.E.O. and M.S.; funding acquisition, P.E.O., M.S. and R.T.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** The study was carried out in the frame of the Project "MAGIC Marine Geohazard along Italian Coasts" funded by Italian National Civil Protection (Resp.: F.L. Chiocci—Resp. CoNISMa Unit: P.E. Orrù), of the CARG Project Geological Map of Italy, Scale 1:50,000 (Marine area)—Sheet 563 "Isola di San Pietro" and Sheet 555 "Iglesias" (Resp.: P.E. Orrù), of the SOMK Parco Geominerario 150-29.12.2017 and FdS grant number F74I19000960007 "Geogenic and anthropogenic sources of minerals and elements: fate and persistency over space and time in sediments" (Resp.: Rita Teresa Melis). The research is also part of the Project "Coastal risk assessment and mapping" funded by the EUR-OPA Major Hazards Agreement of the Council of Europe (2020–2021). Grant Number: GA/2020/06 n◦ 654503 (Unimore Unit Resp.: Mauro Soldati).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** We are thankful to Margherita Mussi for her precious suggestions about human heritage and georcheological implications. We are also grateful to Soprintendenza Archeologica di Cagliari for their valued support. The precious contribution of the three anonymous reviewers and of the journal editors is acknowledged.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Terraced Landforms Onshore and Offshore the Cilento Promontory (South-Eastern Tyrrhenian Margin) and Their Significance as Quaternary Records of Sea Level Changes**

**Alessandra Savini 1,2, Valentina Alice Bracchi 1, Antonella Cammarosano 3, Micla Pennetta 3,\* and Filippo Russo <sup>4</sup>**


**Abstract:** Climate change and tectonic uplift are the dominant forcing mechanisms responsible for the formation of long and narrow terraced landforms in a variety of geomorphic settings; and marine terraces are largely used to reconstruct the Quaternary glacial and interglacial climates. Along the Mediterranean coast, a considerable number of popular scientific articles have acknowledged a range of marine terraces in the form of low-relief surfaces resulting from the combined effects of tectonic uplift and eustatic sea-level fluctuations, as relevant geomorphological indicators of past sea-level high-stands. With the exception of a few recent studies on the significance of submarine depositional terraces (SDT), submerged terraced landforms have been less investigated. By integrating different marine and terrestrial datasets, our work brings together and re-examines numerous terraced landforms that typify the Cilento Promontory and its offshore region. In this area, studies since the 1960s have allowed the recognition of well-defined Middle to Upper Pleistocene marine terraces on land, while only a few studies have investigated the occurrences of late Pleistocene SDT. Furthermore, to date, no studies have consistently integrated findings. For our work, we correlated major evidence of emerged and submarine terraced landforms in order to support an improved understanding of the tectono-geomorphological evolution of the Cilento Promontory and to further clarify the geomorphological significance of submerged terraces.

**Keywords:** marine terraces; submarine geomorphology; coastal geomorphology; sea level oscillation; Tyrrhenian margin

#### **1. Introduction**

A terraced landform is any relatively flat horizontal or gently inclined surface bounded by a steeper ascending slope on one side and a steeper descending slope on the opposite side [1,2]. Terraces can be formed in many ways and in different geologic and environmental settings. In geomorphology, tectonic uplift and climate change are the dominant forcing mechanisms responsible for the formation of long and narrow terraced landforms. Resulting terraces can, therefore, be used for studying variations in tectonic, climate, and erosion, and for investigating how processes have interacted in the past and how they currently interact. The recognition of late Pleistocene uplifted coral platforms as indicators of past sea levels (i.e., reef terraces) was, for example, a significant finding in sea-level research [3]. Terrestrial, fluvial-glacial counterpart [4], coral reef terraces [3], and marine terraces [5,6] have been (and still are) largely used for reconstructing Quaternary glacial and interglacial climates. Where tectonic uplift considerably impacts coastal regions, sub-aerial marine terraces clearly document high-stands of sea level during interglacial stages [6], alternating with low levels during glacial stages. In temperate regions, Pirazzoli [7] noted

**Citation:** Savini, A.; Bracchi, V.A.; Cammarosano, A.; Pennetta, M.; Russo, F. Terraced Landforms Onshore and Offshore the Cilento Promontory (South-Eastern Tyrrhenian Margin) and Their Significance as Quaternary Records of Sea Level Changes. *Water* **2021**, *13*, 566. https://doi.org/10.3390/ w13040566

Academic Editor: Giorgio Anfuso

Received: 31 December 2020 Accepted: 17 February 2021 Published: 23 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

"marine-cut" terraces (or shore platforms) resulting from marine erosion and "marinebuilt" terraces formed by shallow-water and slightly emerged accumulations of materials removed by shore erosion. Along the Mediterranean coast, a range of sub-aerial marine terraces have been acknowledged to be relevant geomorphological indicators of past sealevel highstands (see [8] among other references). The study of Pirazzoli [7] has even allowed the definition of still popular marine stratotypes, outlining the first Quaternary chronostratigraphy (i.e., Calabrian, Emilian, Sicilian, Milazzian and Tyrrhenian for the Pleistocene, and Versilian for the Holocene). Although their work has been revised and refined, the chronostratigraphy is still used in gray literature and open discussions. The geomorphological significance of submerged terraced landforms [9–14], as evidence of Quaternary sea-level variations, has been less investigated and has only recently gained attention [15], thanks to advances in seafloor mapping techniques [16] and the resulting recognition of Submarine Depositional Terraces (SDT) (defined as sedimentary bodies with a clinostratified internal structure and a prograding growth towards the sea [17–21]). Minor studies have investigated the possible erosive nature of submerged terraced landforms formed on bedrock outcropping on the shelf and their relationship to sea-level oscillations, among them Bilbao-Lasa et al. [22]. Additionally, by taking into account both emerged and submerged terraces, a few investigations have integrated their findings to support an improved understanding of the tectono-geomorphological evolution of coastal areas and the physiography of the margin [15,23], as we have undertaken in our work for the Cilento Promontory.

In this region, systematic studies since the 1960s [24–34] have determined well-defined Middle to Upper Pleistocene marine terraces on land. The submarine sector was the subject of minor research during the 1990s. Among such studies, Trincardi and Field [35] investigated the origin and forming mechanisms of remnants of late Pleistocene prograded coastal deposits, locally preserved on the middle and outer portions of the shelf.

#### **2. The Cilento Promontory and Its Offshore: Geological Setting and Stratigraphic Framework**

The coastal area of the Cilento region (Southern Italy) (Figure 1), included between Agropoli and Pioppi, represents the western end of one of the most important peri-Tyrrhenian, morpho-structures belonging to the Campano-Lucano arch of the southern Apennine orogenic thrust system. Compressive tectonogenesis and structuring, initiated in the lower Miocene, appear to have ended in the Lower Pleistocene [29,36–40], through displacement of the Mesozoic-Tertiary bedrock of the Cilento coast during major Quaternary (Lower to Middle Pleistocene) tectonic activity [41–43]. The complex lithogenic history of the Cilento region has thus been shaped by numerous tectono-sedimentary events and orogenic shifts [44] that today allow us to distinguish different lithostratigraphic units outcropping along the coastline (Figure 1). Both siliciclastic and calcareous units outcrop on the Cilento Promontory. Siliciclastic units are primarily represented by the Cilento Flysch Unit [45,46] or Cilento Flysch Group "*Auct*" [41,42] that dominates in the north-western sector, and secondarily by the Ligurian and Northern Calabrese *Auct* tectonic units. Such units are often indicated as the "Internidi" [42,43] (Figure 1), the highest structural tectonic unit (thickness 1300 m) that emerges for a few hundred meters in the central-southern portion of the promontory. Calcareous bedrock outcrops on the south-eastern sector of the Cilento region within the Monte Bulgheria Unit [47]. The general structural setting is dominated by low-angle overthrust surfaces that are clipped and folded by subvertical transcurrent and extensional surfaces with a variable orientation from NW–SE to E–W. The Internidi have been described as tectonic overlap on calcareous units. However, such overlap is sometimes reworked and masked by recent tectonics that are responsible for major displacements that caused carbonate uplift and relief inversion. Indeed, all along the Cilento Promontory the highest peaks are formed by carbonate units, while the most erodible siliciclastic units are found in places only preserved along valleys and on morphological and structural lows [48].

**Figure 1.** The study area. The on-land portion of the Cilento Promontory is represented by a schematic geological map of the Cilento region (The map was adapted from a geological map containing thematic elements and underwater landscapes at a 1:110,000 scale) overlaid on a Digital Elevation Model (DEM) adapted from Campania Region Technical Cartography at a 1:5000 scale. Isobaths and offshore shaded relief were obtained from the EMODNET portal (https://www.emodnet-bathymetry. eu/, accessed on15 February 2021).

Quaternary deposits formed by marine, transitional, or alluvial sediments, where preserved, are found in angular unconformity on the described bedrock. Outcrops of coastal marine sediments (biocalcarenites, fossiliferous beaches and aeolian sands), often associated with typical forms of marine abrasion and bioerosion (shore platforms, fossil eroded notches, "Lithophaga" holes, etc.), are found in very small and discontinuous terraced strips (at various heights between 2 m and 300 m above sea level (a.s.l.)) along the entire Cilento coastal zone. Such deposits, sometimes alternating with "colluvial", pyroclastics, and paleosoils, have been the subject of careful and systematic studies since the 1960s [24–34], allowing chronological attribution to the Middle and Upper Pleistocene.

Offshore, the shelf of the eastern Tyrrhenian Margin lies between the uplifting Apennine chain on land and the Tyrrhenian offshore basin that has been subsiding at a rate of 1 mm/yr since the end of the Lower Pleistocene [49]. Due to the multifaceted tectonic history determined by the opening of the Tyrrhenian Sea, that is also associated with limited Plio-Quaternary sedimentation at places interrupted on the shelf by bedrock outcrops [48], the seabed topography is extremely complex. Seismic data has revealed a deformed acoustic basement, displaced by quaternary faults with very sharp and steep scarps similar to the ones detected on land. The continental shelf is wider in the northern portion, extending for almost 30 km to the north of Punta Licosa [50], and delimited seaward by an uncertain shelf break, from 180 to more than 200 m in depth. To the south, continental shelf width is reduced to less than 10 km, and in offshore Acciaroli the shelf break is sharper and is located at a water depth (w.d.) of roughly 130 m. In the southern sector (Figure 1), the shelf further narrows to 6 km and a transition between the shelf and the slope is evident, but located at variable depth, gradually decreasing from 140 m off Punta del Telegrafo to 130 m in .w.d. offshore of the south-eastern corner of the promontory, where the upper continental slope is much steeper and likely coincident with a tectonic escarpment [51–53]. Different sub-horizontal surfaces bounded by fairly continuous and slightly sinuous escarpments have also been

determined on the shelf. They have been interpreted as submarine terraces [48,54,55] that formed by local and prolonged low sea level stationing, occurring between the regression of the Last Glacial Maximum (LGM) and subsequent rapid post-glacial sea-level rise (i.e., the Flandrian Transgression). Only a few of the submarine terraced landforms located near sea level (8 m to 12 m below sea level (b.s.l.)) are ascribed to Marine oxygen Isotopic Stage (MIS) 5 [32,56,57], corresponding to the Last Interglacial period.

The continental slope is marked by depressions and topographic highs of variable dimension, down to a depth of 1600 m. Numerous escarpments document the existence of simple to complex landslide scars, testifying to the dominant role of mass-wasting phenomena in shaping the continental margin. A complex tectonic framework of bedrock is also still visible along the slope, where it has created local, intra-slope reliefs and marked tectonic lineaments [53–55].

#### **3. Data and Methods**

Our study was driven by the collection of major evidences of terraced landforms, both on land and in offshore areas of the Cilento Promontory coastal zone, recovered from scientific literature and detected on available Digital Elevation Models (DEM—i.e., Emodnet database—https://portal.emodnet-bathymetry.eu/ (accessed on 20 February 2021)—grid cell size 50 × 50 m and Magic project: http://dati.protezionecivile.it/geoportalDPC/rest/ document#MagicFoglio10/ (accessed on 20 February 2021)—grid cell size 50 × 50 m) for the offshore sector. Submarine terraced landforms were also manually and automatically detected by applying a geomorphometric analysis performed using Spatial Analysis Tool available in ArcGis®. All terraced landforms were then collected in a proper database by including information regarding dating, altitude or depth, and references (Tables 1 and 2). Landform spatial and temporal distributions were analysed in order to detect the role of associated bedrock and the structural framework in controlling distributions and geomorphological differentiation.

Terraces were grouped according to lithostratigraphic units of the corresponding bedrock (i.e., siliciclastic or calcareous), namely, from North to South: (1) Cilento Flysch and "Internidi" Units; and (2) the Carbonate Unit of Mount Bulgheria, with outcroppings regions located in the area surrounding Palinuro (Palinuro Cape, Mingardo river mouth and Camerota) on the southern coast of the Cilento Promontory (Figure 1).

Study of the offshore region was also supported by the availability of high resolution seismic data collected using a GeoAcoustic GEOCHIRPII (GeoAcoustic Limited, Shuttleworth Close, Doncaster, UK) Subbottom Profiler System (SBP) in 2003, between 10 and 130 m in w.d., as well as by results obtained from a sedimentological analysis performed on 16 gravity cores and 32 grab samples, as described in [58]. An interpretation of depositional and erosional processes, as detected from a seismo-stratigraphic analysis, was performed using the concepts of sequence stratigraphy [59,60].

#### *3.1. On Land Terraced Landforms*

To understand traces of described ancient marine deposits and sea-level markers, background knowledge of the study area was obtained from an extensive literature review [24,26,28,30–34,61–69], and a field survey. Table 1 provides on-land terraced landforms according to their altitude and dating (as ascribed in the scientific literature).

#### *3.2. Offshore Terraced Landforms*

Background knowledge for the submarine sector of the study area was obtained by collecting public bathymetry (EMODnet portal) and high-resolution seismic data, as described in Savini et al. [58], along with analogous remote data and evidence of direct observations as reported in the scientific literature [32,48,50,54–57]. To detect flat surfaces, basic geomorphometric analysis were performed in ArcGIS®. All areas with a slope value ≤ 1 and confined by a marked break of slope (according to [1]) were segmented and converted in polygons (Figure 2B).


**Table 1.** List, reported dating in the scientific literature, altitude, and referenced literature for on-land terraced landforms.

?: Uncertain value or not confirmed by consistent data.

**Table 2.** List, reported dating in the scientific literature, depth, and referenced literature for offshore terraced landforms. The table takes into account terraces cited in the scientific literature. The correspondence with terraces detected by geomorphometric analysis is reported in the last column on the right.


**Table 2.** *Cont.*


**Figure 2.** (**A**): A map of major terraced landforms detected on land and offshore in the Cilento Promontory, as reported in Tables 1 and 2, with the exception of the submarine depositional terraces (SDT) reported in [48,50]. (**B**): A slope map with red polygons indicating flat areas (slope ≤ 1) delimited by marked ascending and descending breaks in slope.

#### **4. Results**

#### *4.1. Terraced Landforms: Temporal and Spatial Distribution*

The presence of terraced landforms in the Cilento coastal area marked the emerged and submerged sectors (Tables 1 and 2). Scientific literature has documented at least seven orders of Pleistocene terraced surfaces on land (Table 1 and Figure 2), spanning from the Lower to Upper Pleistocene [32].

Offshore terraced landforms (Figure 2) were, instead, first detected using geomorphometric techniques (Figure 2B), then correlated with evidence in the scientific literature [32,48,50,54–58], and then grouped according to depth range of occurrence (Figure 2A). For terraced surfaces deeper than 50 m, we also refer to [50] and [48]. As discussed below, the distribution of both on-shore and off-shore terraced landforms was then resumed according to geological unit.

#### 4.1.1. Northern Cilento Group

On-land five terraced surfaces have been identified (Table 1 and Figure 2). Two were from the Middle Pleistocene (MIS 7 and 9) with clear evidence of uplift, apparently sealed by Upper Pleistocene deposits. The remaining three were from the Upper Pleistocene (MIS 5a, MIS 5c and MIS 5e), with no significant contribution from tectonics.

The submerged sector is typified by prolongation towards the sea of the "Punta Licosa" Promontory (Figures 1 and 2) that (1) provides an EW aligned spur formed by an outcrop of the acoustic basement that rests over more than 16 km2, between 25 and 80 m of w.d.; (2) likely originated from the Cilento Group synorogenic unit (or "Flysch del Cilento") and; (3) was bounded by direct faults. The spur rises from the surrounding seafloor through several sharp escarpments bounded by flat terraced surfaces (Figures 2 and 3).

According to the scientific literature, terraces are positioned at −8 m, −10/14 m, −17/27 m, and −43/50 m and reportedly range from MIS 5a or 5c up to MIS 3 [50,55,57,66]. A performed geomorphometric analysis distinctly outlined the marked stepped profile of the Licosa spur and several submarine terraces (slope ≤ 1), with a prevalence at −21/26 m, −47/52 m, and −76/86 m, having consistent lateral continuity (especially toward the south and for depth intervals of −21/26 m and −47/52 m) (Figures 3A and 4), were located. Small scale landforms resembling tension fractures at the crowning areas of modest landslides are frequent on the southern slope of the spur [55], an isolated group of terraced surfaces downward of small landslide scars was identified at −28/30 m (Figure 4). Ferraro et al. [48] detected additional terraced landforms at 86 m and even deeper at 107 m of w.d., both of an erosional origin, and indicated that the terraces formed at the outcrop of the acoustic basement. Further offshore, biogenic coarse sandy depositional bodies, bounded at their top by a ravinement surface, were described at 120 m and 160 m in w.d. Such bodies are developed over more than 20 km along-slope [48]. A small fragment of "*Arctica islandica*" (Linneo, 1767) was also recovered from a core sample ([48]; Pennetta pers. com.) allowing attribution of their formation to the last low-stand period (i.e., MIS 2). Trincardi and Field [50] also reported the occurrence of depositional bodies in the form of shelf-margin deposits, truncated at their tops by an outer-shelf ravinement surface at −150/160 m. The shelf-margin is reported to occur at −200 m in w.d. (Figures 1 and 2) (i.e., deeper than the sea level low-stand reported for the Last Glacial Maximum (namely 120 m in w.d.). Marani et al. [70] indicated that these sandy bodies appear to have formed in a shallow (<30 m deep) marine setting. This evidence, together with the detection of an outer limit for the ravinement surface generated by the post-Würmian Transgression at a deeper depth than the one reported for the eustatic minimum (i.e., 120 m), suggests that the outer continental shelf has been subject to important (tectonic) subsidence phenomena over the Holocene. Considering Mediterranean wave-base level in the order of 10/15 m [71], we speculate that the subsidence rate reported for the Tyrrhenian sea by Kastens et al. [49] (i.e., 1 mm/yr.) allowed a merger of the geodynamics of the outer portion of the continental shelf with the structural system that currently controls evolution of the Tyrrhenian basin.

**Figure 3.** Histograms of depth values (depth range 0–120 m in w.d.) for the offshore areas of Northern Cilento group (**A**); Cilento Group and Internidi Units (**B**); and Southern Bulgheria Mount (**C**). The gray ellipses show the depth ranges that, on the map, are clearly delimited by sharp breaks in slope.

#### 4.1.2. Cilento Group and Internidi Units

The central continental coastal area of Cilento, between Acciaroli and Palinuro, is essentially composed by deformed units of Mesozoic-Tertiary Bedrock (members of the Cilento Group and Internidi Units) covered in angular discordance by Quaternary alluvial and coastal deposits. In places, coastal deposits represent the filling of localized morphotectonic depressions of actual alluvial (i.e., the plain of the Alento River) and coastal (i.e., plain of Casalvelino-Ascea—Figure 1) plains consisting of fluvial sediments, dune and beach-ridge deposits sometimes covered by continental colluviums, and slope debris. Marine beach-ridge deposits, emerging up to six m a.s.l. along the coast at Ogliastro [28] and Acciaroli [24], have been attributed to the Upper Pleistocene (Table 1 and Figure 2). In

this sector, no deposits or forms have been found that can be attributed, with certainty, to the Lower or Middle Pleistocene (Figure 2).

**Figure 4.** A 3D view of the seaward prolongation of the Punta Licosa Promontory, with polygons detected by geomorphometric analysis, representing terraced landforms according to [1] and distinguished in different colors according to depth range (yellow: 21/26 m; light blue: 47/52 m; green: 76/86 m). Small-scale landslide scars are also mapped (red lines) on the southern side of the submerged Punta Licosa spur. The Digital Terrain Model data products used to provide the 3D view have been derived from http://dati.protezionecivile.it/geoportalDPC/rest/document#MagicFoglio10 (accessed on 20 February 2021).

> Offshore, the morpho-structural depression, marked on land by the coastal and alluvial plain of Casal Velino-Ascea (Figure 1), is filled by sandy-silty deposits [58] over almost all of the central and southern continental shelf, in continuity with terrestrial physiography (Figure 2). The depression is confined to the north by the outcrop of the acoustic basement, that forms a southward-elongated ridge offshore Acciaroli, interrupted at the shelf break (Figures 1 and 2). The coupling of bathymetric and high-resolution seismic data clearly indicates marked terraced landforms along the ridge. Terraced landforms are particularly evident at a depth interval between 47 and 52 m (Figure 3B), in the form of an erosional surface (wave-cut platforms [58]) sculpted within the acoustic basement (i.e., bedrock, Figure 5A). As shown by the marine DEM and the associated slope value (Figure 2B), the terraced landforms are still in continuity with those detected northward at Punta Licosa, especially for the depth range 47/52 m. A north-south elongated depositional body (with a lenticular section formed by poorly defined sloping depositional units that resemble shoreface clinoforms [58]) rest at 55 m in w.d. in overlap above the southeastern edge of the acoustic basement that outcrops to the south of Acciaroli (Figure 5). According to sediment composition reported in [58,72], the depositional body likely formed in a shallow (<10 m deep) marine setting. Since an older origin would have resulted in aerial exposure due to sea level drop during the LGM, the absence of an obvious erosional surface at the top of the deposit and partial burial towards the sea due to the high-stand drape (Figure 5B), warrants ascription to the transgression that followed MIS 2.

#### 4.1.3. Southern Bulgheria Mount

On land, the best-preserved Quaternary landforms and deposits of the Cilento Promontory are found in Palinuro Cape (the Monte Bulgheria Carbonatic Unit), an area intensively studied and well described within the scientific literature [32]. The oldest evidence of flat eroded surfaces within the region are dated to the Upper and Lower Pliocene. Landforms are found at altitudes between 1200 and 400 m a.s.l. and have been associated with sub-aerial surfaces of fluvial-karst erosion, although some authors do not exclude marine abrasion as a potential origin [69].

**Figure 5.** High-resolution seismic profiles acquired offshore Cilento Group and Internidi Units. See Figure 1 for location. (**A**) A submerged terraced landform (wave-cut platform) is visible at 47–51 m in w.d. (corresponding to 63/68 ms). (**B**) A submerged terraced landform (wave-cut platform) is visible at roughly 51 m in w.d. (corresponding to 68 ms).

The oldest Pleistocene terrace outcrops around 300 m a.s.l. and corresponds to the Lower Pleistocene [64].

Overall, the area records several order of terraces positioned at variable altitudes that are not easily correlated to the Middle Pleistocene. Four order of terraces, at an altitude range between 20 m and 200 m a.s.l., have been ascribed to the Middle Pleistocene. Additionally, five orders of Middle Pleistocene marine abrasion terraces were carved along the coastal slopes at altitudes of 180/170 m, 140/130 m, 100 m, 75/65 m, and 50 m, for which important tectonization cannot be excluded.

Upper Pleistocene terraced surfaces outcrop continuously along the cost and they are located between 1.5 m and 10 m a.s.l. (Table 1). Two marine terraces located along a sea-cliff that marks the coastal area at 8/7 m and 3/2 m a.s.l., and beach-ridge deposits containing fragments of *Thetystrombus latus* Gmelin 1791 (=*Strombus bubonius*) are found at 3/2 m a.s.l. The location of these Upper Pleistocene landforms suggested a small (few meters) tectonic lowering of the area. In general, since good lateral continuity is preserved and since the terraces are quite well correlated, terraces created by Upper Pleistocene sea-level oscillations seem to document a relatively stable tectonic period [73].

As for the submarine portion between Palinuro Cape and Bulgheria Mount, four orders of submarine terraces, located at 7/8 m, 12/14 m, 18/24 m, and 44/46 m in w.d., have been extensively described within the scientific literature [32]. Since they show evidence of subaerial erosion associated with a regressive period, the first two terraces were ascribed to the Last Interglacial (MIS5), or to an earlier period. Evidence of former sea-level positions at 7/8 m in w.d. have been attributed to MIS5a with good confidence.

Terraces detected in the form of wave-cut platforms at 18/24 m in w.d. were, instead, ascribed to the last phase of MIS3, which seems to be characterized by long stationing that occurred during the post-Last Interglacial regression [56]. According to Antonioli et al. [32], good conservation of the deposit and the absence of subaerial erosion for terraces located

at 44/46 m in w.d. leads attribution to a lower standing period that occurred during the last, post glacial transgression (i.e., the Flandrian Transgression). Sparce terraces found at deeper depth were, in contrast, formed by depositional bodies and are ascribed to the last glacial low-stand period (MIS2). The geomorphometric analysis performed on the DEM detected quite large terraces at three main depth range: 55/63 m, 70/77 m, and 105/107 m. Seismic data well confirmed the erosional origin of mapped terraces located at 50/55 m in w.d. (Figure 6). Due to DEM resolution, which cannot resolve submarine terraces of small dimensions, many shallower terraces were likely difficult to detect.

#### **5. Discussion**

Abundant evidence of former Quaternary sea-level stationing, in the form of terraced landforms, occurs on the Cilento Promontory from north to south and on its offshore counterpart. One of the first observations obtained by grouping various terraces according to their lithological unit was consequent variation in the degree of conservation of erosive forms created by former sea level positions (i.e., marine terraces of dominant erosional origin). Such forms were, indeed, better preserved when associated with carbonate rocks of the Mount Bulgheria Unit. Bedrock of terrigenous origin (Cilento Flysch and Internidi Units) hinders the conservation of Quaternary erosive landforms [26,74], that were smaller and poorly represented. On the Mount Bulgheria Unit, a more widespread and continuous conservation of the Quaternary deposits was evident, allowing plentiful geological and geomorphological information to be obtained for reconstructing the age and alternation of former sea-level positions [25,28,29,31,32,34,61,62].

From analyzed data, it appears that older terraces can be detected up to 300 m a.s.l. Several Lower and Middle Pleistocene terraces have been cataloged. However, the complex tectonic history of the region makes it difficult to perform accurate correlations, although focused dating could improve our understanding of the post-orogenic tectonic differentiation that typified the variuos uplifting rates of the Cilento Units. Additionally, some sector of the promontory were tectonically displaced upwards by approximately 400 m during the late Lower and the Middle Pleistocene. The Upper Pleistocene experienced reactivations, but on a smaller scale and were differentiated from north to south.

A different Upper Pleistocene geodynamic behavior seems to characterize the offshore region where tectonic movements or relative stability, documented by the position of dated marine terraces on-land, are well correlated with the submarine sector for areas shallower than 120/130 m in w.d. In the offshore sector, the shallower orders of terraced landforms, likely generated as wave-cut shore-platforms, did not record relevant tectonic/vertical movements during the Upper Pleistocene. Since their position seems to be well correlated with former sea-level positions, as reported in the global mean sea-level curves (Figure 5), the result is in good agreement with documented research on land [75,76]. Seismic data additionally indicates that during the last period of sea level rise, a transgressive erosional surface (i.e., ravinement surface) formed in the area [50] and that its relationship to depositional bodies detected on the shelf critically improved constraints for ascribing the relative position of sea-level to detected submerged terraces. We gave considerable importance to the curve of isotopic stratigraphy in [77], and to other evidences reported for Holocene relative sea-level curves [78,79], where a short stasis is reported for the rapid Flandrian Transgression between 45 and 40 m in w.d.. This depth range is close to some of the mapped offshore terraces in the form of a wave-cut shore platform (Table 2, Figures 2 and 3) and displays strong lateral continuity all along the offshore sector of the Cilento Promontory (Figures 2B and 4). In contrast, as speculated on the basis of relative sea-level fluctuations documented by low-stand depositional bodies that formed SDT at 160 m in w.d. [50], the outer shelf appears to have experienced an important tectonic subsidence.

Using all collected evidence, we observed that submarine terraced landforms offshore of the Cilento Promontory can be distinguished as erosional and depositional, respectively representing paleo wave-cut shore platforms (see [80] for a comprehensive definition and differentiation from marine terraces) and SDTs (as described in [9] and references therein). In our study area, terraced landform distinction is marked by the depths at which they occur (Figure 5). On the outer zone of the continental shelf, and especially in areas deeper than 120 m in w.d., SDTs have been described by Ferraro et al. [48] and have been interpreted by Trincardi and Field [50] as shelf-margin deposits, with a different configuration according to physiographic shelf-break depth during the last sea-level low-stand (i.e., MIS2). Shelf margin deposits particularly occur offshore of the Cilento Promontory where the physiographic shelf-break is deeper than the position of the low-stand shoreline of the Last Glacial Maximum (MIS2). Trincardi and Field [50] highlighted the absence of such deposits, where the shelf-break was close to the shoreline during MIS2. The different configuration of SDTs, located on the outer shelf (as described in [50]) and the concurrent deepening and widening of the physiographic shelf break toward the north, warrants a distinction between the two main morpho-structural elements forming the shelf, as follows:


The two morpho-structural elements seem to represent the components of a regional fault system. The system is defined by NW-SE and NNE-SSW lineaments, marking the core area that separates the uplifting morpho-structural high forming the Cilento Promontory on the margin (interposed between the coastal depressions of the Sele Plain-Salerno Gulf to the north and of the Policastro Gulf to the south-east), and the Tyrrhenian basin offshore. The offshore tyrrhenian basin has been subsiding at a rate of 1 mm/yr since the end of the Lower Pleistocene [48] and from the Last Glacial Maximum until present could have been responsible for the lowering of the shelf break. The shelf s.s., represents a sector that experienced the same tectonic of the on-land system. Such a result is confirmed by a good correlation between the depth of marine terraces of erosive origin and eustatic sea level variations recorded for the last 200 ka (Figure 7) that are attributed, for the most part, to

shallower submarine terraces of the stationing of the Flandrian Transgression (40/46 m in w.d. —as reported for the offshore of the southern Bulgheria Mount [32]) and the stationing of MIS 3 (50/55 m and 70/76 m in w.d.), MIS 5c, and MIS5a (10/15 m and 18/24 m in w.d.). In the offshore the Cilento Flysch and Internidi Units, the strong lateral continuity that characterises the terraces located at 47/52 m in w.d. (Figures 2B and 3) suggests that they could also have an origin associated to the Flandrian Transgression.

**Figure 7.** A graph with eustatic sea level variations, expressed in meters, recorded during the last 200 ka cal (adapted from [76]). Thick horizontal lines indicate major depth intervals where submarine terraced landforms were located in the offshore region of the Cilento Promontory. Different colors refer to different regions as distinguished in Table 2 (light transparent blue for the offshore Cilento Group and Internidi Units, light transparent green for the offshore of the Bulgheria Mount); SDT: Submarine Depositional Terraces.

Sub-aerial marine terraces [7] have, indeed, been traditionally acknowledged to be relevant geomorphological indicators of past sea-level high-stands in regions subjected to tectonic uplift [8]. With time, uplift determines the formation of terraced coastlines, often with a step-like profile, where older terraces are higher and farther from today's coastline [7]. In this work, we focused on understanding the occurrence of paleo, wave-cut shore platforms (forming marine terraces) within the submarine domain [22]. Here, it is important to note that phases of relatively high sea-level or stationary phases during transgressive periods, in the end, determine the most favorable condition for wave-cut shore platform formations on rocky cliffs (because they cause marine processes to prevail over sub-aerial processes). A relative decrease in sea level would, instead, lead to a decrease in the efficiency of marine processes, with the formation of beaches or debris at the base of a cliff, preventing wave-cut shore platform formation. For this reason, wave-cut shore platforms are unlikely to form during low stands or regressive conditions. Therefore, for our study area, we conclude that the occurrence of marine terraces of erosive origin within the submarine domain resulted because shelf s.s. was subjected to the same geodynamics that impacted the Cilento Margin on-land; and because the area was relatively stable during the Upper Pleistocene and, therefore, during earlier high-stand (MIS5c and MIS5a) and stationary periods of the Flandrian Transgression. In contrast, the outer shelf has been lowering, at least since the end of the Lower Pleistocene, and has been involved in the same geodynamics that are altering the Tyrrhenian Basin, promoting the formation of SDTs.

Deeper submarine terraces of erosive origin, as described by [48], at 80/86 m and 100/107 m in w.d., are more difficult to interpret. Therefore, further investigation is required to confirm their actual association to bedrock outcrops.

A higher resolution of the DEM would also lead to a more effective morphometric analysis. The more accurate list of terraced surfaces that would result from the analysis, combined with an adequate reconstruction of late Quaternary environmental conditions

that controlled formation of wave-cut shore platforms (e.g., wave climatology, as performed in [22]), would provide more precise information to confirm the ascription of the terraces to a defined high-stand period.

#### **6. Conclusions**

An interesting finding of our study, obtained by coupling terrestrial and submarine terraced landforms [81], is the detection of two main types of submarine terraced landforms in the surveyed sector of the south-eastern Tyrrhenian Margin: (1) erosional terraces (wavecut or abrasion platforms) formed on outcropping bedrock and (2) depositional terraces (i.e., SDTs) generated by late Quaternary depositional sequences. A distinction between the two types of landforms actually depends on a set of parameters dominated by regional geologic settings (the type of bedrock, and geodynamic and sediment inputs) subject to sea-level oscillation. Submarine marine terraces that result from the generation of wave-cut shore platforms were predominantly generated during interglacial periods or during relevant stasis occurring in transgressive events. On the surveyed sector of the south-eastern Tyrrhenian Margin, they formed when bedrock outcrops were exposed on the shelf, making them vulnerable to substantial erosion due to marine processes. Bedrock outcrops also contributed to a disruption of late Quaternary sedimentation on the inner shelf. Based on this evidence, submarine terraces of a strictly erosional nature could not have been formed on a traditional passive margin subject to subsidence. In contrast, subsidence, provides accommodation for the formation of depositional bodies, that according to sediment availability, shelf morphology, and local sea level history provide a suitable condition for a variety of SDTs.

**Author Contributions:** Conceptualization, M.P., F.R. and A.S.; methodology, V.A.B., A.C. and A.S.; software, V.A.B. and A.S.; investigation, M.P., F.R. and A.C.; data curation, M.P., F.R., V.A.B. and A.S.; writing—original draft preparation, M.P. and F.R. writing—review and editing, A.S., F.R., M.P. and V.A.B.; visualization, V.A.B.; supervision, M.P. and A.S.; project administration, M.P. and A.S.; funding acquisition, A.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This article is an outcome of the Project MIUR—Dipartimenti di Eccellenza 2018–2022, Department of Earth and Environmental Sciences, University of Milano-Bicocca.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The bathymetric metadata and Digital Terrain Model data products have been derived from the EMODnet Bathymetry portal—http://www.emodnet-bathymetry.eu (accessed on 20 February 2021). and from http://dati.protezionecivile.it/geoportalDPC/rest/ document#MagicFoglio10 (accessed on 20 February 2021).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Water* Editorial Office E-mail: water@mdpi.com www.mdpi.com/journal/water

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18

www.mdpi.com ISBN 978-3-0365-1654-7