Abstract

Multibeam bathymetry and high-resolution seismic profiles depict in detail the characteristics of submarine gullies present in the upper continental slope offshore of the mouths of the Tiber and Volturno Rivers in the Tyrrhenian Sea and the Simeto River in the Ionian Sea. Upper slope gullies are interpreted as depositional features, growing because of faster aggradation on intergully areas with respect to their axes. The dispersal of river-flood sediment through plunging of hyperpycnal flows has been interpreted to be the limiting factor in sedimentation in gully axes. However, the generation of hyperpycnal flows requires a sediment concentration of several tens of kilograms per cubic meter in the river flows to overcome the higher density of seawater. This threshold is difficult to reach in medium-sized rivers (i.e., catchment basin of 500–5000 km2), such as the Tiber and Volturno. Two alternative scenarios of enhanced sediment availability for these rivers have been proposed. The first scenario is related to sea-level fall and lowstand stages, when the climate conditions may have been different and a huge amount of unconsolidated, fine-grained sediment deposited during the previous highstand stage may have been eroded from the river valleys, coastal plain valleys and shelf delta, as suggested in pioneering work of one of the authors (Chiocci) and Bill Normark. This scenario might explain the formation of some of the gullies offshore of the Tiber River mouth, which developed during the last glacial maximum; however, it is unable to explain the gullies within deposits related to transgressive and highstand system tracts, such as offshore of the Simeto and Volturno River mouths. An alternative scenario is thus proposed on the basis of the relationship observed between the studied rivers and the presence of large volcanic edifices in their catchment basins. Explosive volcanic activity could have drastically modified their catchment basins and caused a sudden and large supply of loose tephras, increasing the sediment load in watercourses and consequently favoring the development of gully-forming hyperpycnal flows. The timing of volcanic activity on the three study areas matches, in fact, the presence or absence of gullies in upper slope depositional sequences.

INTRODUCTION

Submarine gullies are small-scale, straight and shallow channelized features mainly developed on the upper continental slope (Fedele and Garcia, 2009, and references therein). They are thought to transport a volumetrically significant amount of sediment from the shelf down to the basins (Dott and Bird, 1979; Field and Clarke, 1979), and contribute to the architecture of petroleum reservoirs (Hewlett and Jordan, 1993). They were little appreciated until the recent development of multibeam seafloor imaging became widely available for scientific purposes (Hughes-Clarke and Mayer, 1996). Their small size and low morphological relief hinder their study both in ancient uplifted sequences and on the modern seafloor (Spinelli and Field, 2001; Surpless et al., 2009, among others). Before the widespread use of multibeam bathymetry in the 1990s, work based on high-resolution seismic-reflection data depicted and characterized upper slope gullies in the framework of the Quaternary evolution of continental margins (i.e., Chiocci and Normark, 1992; Blum and Okamura, 1992). Pioneering work (Chiocci and Normark, 1992) analyzed single-channel sparker profiles (0.3–1 kJ sound source with a frequency of 100–1500 Hz) in the upper slope area offshore of the Tiber River mouth. This work characterized the morphology of gullies at a fine scale similar to that achievable with multibeam bathymetry (cf. Figs. 1, 2, and 3). With the advancement of multibeam techniques, gullies have been increasingly recognized along different continental margins (i.e., Field et al., 1999; Gardner et al., 2003; Posamentier and Basden, 2000; Spinelli and Field, 2001; Rebesco et al., 2009).

Here we focus on regularly spaced (tens to thousands of meters) gullies of the outer continental shelf to the upper slope offshore western Italy. Their regular spacing is similar to that observed for slope gullies developed on the northern California margin (Field et al., 1999; Spinelli and Field, 2001) and on the southwestern Japan forearc (Blum and Okamura, 1992).

In our case studies, gullies are located offshore river mouths (Tiber, Simeto, and Volturno Rivers), and we therefore initially interpreted their development to be related to fluvial processes. The dispersal of flood sediment as a result of plunging of hyperpycnal flows from rivers (Mulder and Syvitski, 1995; Piper and Normark, 2009, and references therein) might be a primary mechanism for gully formation. However, the generation of these flows requires a high suspended particle concentration in the river freshwater (∼40 kg/m3), so many medium and large rivers are not able to generate hyperpycnal flows (Mulder and Syvitski, 1995). On the contrary, this phenomenon may easily occur during unusually extreme conditions, such as joukulhaups, lahars, and dam breaking, where a large concentration of sediment is present, or in so-called “dirty” rivers, characterized by a relatively small but steep drainage basin and by a torrential regime (i.e., the stream bed might be dry for months or seasons, whereas water supply is short and intense during abrupt flash floods; examples of such regimes are the El Coyote River [Nava-Sanchez et al., 1999] and the fjords of British Columbia [Prior and Bornhold, 1990]). Similar conditions exist in small and steep rivers draining the central and southern Apennines in Italy, where hyperpycnal flows can occur in a 100 yr period (Milliman and Syvitski, 1992; Syvitski and Kettner, 2007). However, the occurrence of gullies in front of medium-sized Italian rivers, such as the Tiber and Volturno, is counterintuitive and may potentially give new hints for the generation of hyperpycnal flows from river floods.

The aim of this paper is to characterize the morphology and seismic-reflection facies of upper slope gullies developed in front of the Tiber, Simeto, and Volturno Rivers. These characteristics are then discussed in the context of literature on submarine gullies found in other continental margins to elucidate the factors involved in their development. Specific attention is devoted to the mechanisms responsible for the generation of hyperpycnal flows. Two hypothetical scenarios of enhanced sediment delivery from the river catchment basins to the studies gullies are proposed: (1) the dismantling of highstand deposits during glacial sea-level lowstands (as hypothesized by Chiocci and Normark, 1992; Field et al., 1999, among others); and (2) sudden, large-magnitude sedimentary inputs from explosive volcanic activity.

GEOLOGICAL SETTING

Tiber River

The Tiber River source is Mount Fumaiolo in the Northern Apennine Chain. The catchment basin is the largest in central Italy, crossing different geological units: terrigenous facies and flysch deposits, the carbonate Apennine Chain, and the volcanic district of Mounts Sabatini and Albani Hills in the river's lower reach (MSVD and AHVD in Fig. 2A). The lower Tiber River course is actually constrained by the presence of these two volcanic districts. Volcanic activity at Sabatini spanned between 800 and 250 ka, whereas Colli Albani activity was between ca. 600 and 25 ka (De Rita et al., 1995; Funiciello et al., 2003); the larger explosive eruptions occurred from 600 to 350 ka (Marra et al., 2003). When the river debouches into the sea, sandy bedload bidirectionally feeds the beaches of the delta (Bellotti et al., 1989), while muddy suspended load feeds the prodelta region across the width of the inner/middle shelf. This mudbelt is mainly transported northward as a result of the interaction of the hypopycnal plume and geostrophic shelf currents (Chiocci and Normark, 1992).

The continental shelf offshore the Tiber River is 15–30 km wide, with a shelf break at 120–150 m (Table 1). Pleistocene shelf stratigraphy is mainly characterized by siliciclastic lowstand forced regressive deposits (Chiocci, 2000) and post–6 ka deltaic highstand deposits, which overlie a transgressive erosional surface (Bellotti et al., 1989).

Volturno River

The Volturno River is sourced from Mount Azzone in the Southern Apennine Chain. The catchment basin crosses sedimentary and volcanic units, such as Roccamonfina units in its upper reach (RVC in Fig. 2B, 550–150 ka; Rouchon et al., 2008). To the north the Volturno lower reach bounds the Phlegrean Fields, the volcanites of which were emplaced in the past 40 k.y. (PF in Fig. 2B; last eruption in A.D. 1588). The Phlegrean Fields volcanic history has been dominated by explosive and hydromagmatic activity; two major eruptions occurred 37 ka and 12 ka (Campanian Ignimbrite and Neapolitan Yellow Tuff; Florio et al., 1999). The Volturno catchment basin also recorded the distal part of explosive eruptions at Mount Somma-Vesuvius (So-Ve in Fig. 2). Subsurface stratigraphy of the Volturno Plain shows a Pliocene–Quaternary succession formed by alluvial deposits, marine and transitional sediment, and intercalations of lavas and volcaniclastic deposits (Romano et al., 1994).

Severe sea storms occur annually during the winter, most coming from the west and southwest, with associated powerful longshore currents that redistribute part of the sediments supplied from the river (Budillon et al., 2005, 2006, and references therein).

The continental shelf offshore the Volturno River is <10 km wide, with a shelf break at ∼120–125 m (Table 1). Its late Pleistocene–Holocene stratigraphic architecture is characterized by highstand deltaic systems fed by the Garigliano and Volturno Rivers (Bartole et al., 1984; Marani et al., 1986; Aiello et al., 2000, and references therein).

Simeto River

The Simeto River is the second longest river in Sicily, with a length of 114 km (Table 1). The upper part of the Simeto River drainage basin is on the Apennines-Maghrebian Chain terrains, overlain by postorogenic Miocene and Pliocene units (Lentini et al., 1996). The river flows around the western and southern flanks of Mount Etna before crossing the low-lying Plain of Catania to enter the lonian Sea (Fig. 2C). The subaerial Catania Plain and offshore area represents a segment of the foredeep-foreland system of the Apennines-Maghreb Chain, which is characterized by a strong subsidence after late Pliocene time, creating as much as 600 m of accommodation space for sediment accumulation (Torelli et al., 1998).

The morphology and sedimentation of the Simeto River have been strongly influenced by the activity of Mount Etna, evidenced by the recognition of a series of discontinuous fluvial terraces along its northeast bank, capped by lava flows emitted from the volcano over the past 500 k.y. (Chester and Duncan, 1979, 1982). Mount Etna has been particularly active since 100 ka; Plinian activity occurred between 15.5 and 15 ka, and at 122 B.C. (Coltelli et al., 2000). The explosive activity profoundly affected the sedimentation of the subaerial part of the delta, evidenced by the abundance of fall-out debris that produced alternating dark volcanic ash and light terrigenous siliciclastic layers in its late Pleistocene–Holocene coastal plain (Longhitano and Colella, 2007).

Currently, the Simeto River is building its delta into the Ionian Sea, characterized by powerful longshore currents and storm waves; the storm waves remove most of the bedload from the river mouth off the shelf. The continental shelf offshore the Simeto River mouth shows an average width of 6 km and a shelf break at ∼100 m (Table 1). The shelf disappears north, toward Etna volcano, where a deep and wide canyon system is present.

DATA AND METHODS

The data used for this study were collected during several oceanographic cruises from the Sapienza University of Rome and the Istituto di Geologia Ambientale e Geoingegneria–Consiglio Nazionale delle Ricerche onboard the research vessels Urania, Universitatis, and Odin Finder.

For the sector offshore the Tiber River mouth, multibeam data were acquired during a 2003 cruise by means of a 30 kHz multibeam system having an angular coverage sector of 135 beams per ping at 1° (for seismic profiles, see details in Chiocci and Normark, 1992).

For the sector offshore the Volturno River mouth, multibeam data were acquired during a 2003 cruise by means of a 50 kHz multibeam system with an angular coverage sector of 125 beams per ping at 1.5°. High-resolution seismic-reflection profiles were acquired with a Datasonics Chirp profiler during the same cruise.

For the sector offshore the Simeto River mouth, multibeam data were acquired during 2007–2009 by means of two multibeam systems, the first one at 455 kHz with an angular coverage sector of 240 beams per ping at 0.5° for shallow-water sectors (40–120 m below sea level, bsl), and the second at 50 kHz for deep water (depths >120 mbsl). The seismic profiles were acquired by means of 4.5 kJ sparker during a 2009 cruise.

All data were positioned by means of differential global positioning system; the multibeam data were processed with CARIS Hips and Sips software, where all sensor data were merged and corrected for the effects of tide, attitude sensor (roll, pitch, and yaw), and sound velocity variations that occurred during the survey. Manual editing and automated filters were used to clean erroneous soundings in order to obtain digital terrain models with a cell size of 20 m.

The main hydrometric parameters of the Tiber, Volturno, and Simeto catchment basins were obtained through measurements collected at the Ripetta station of Rome, the Amorosi station, and the Giarretta station, respectively (Table 1; Adragna, 1964; Pagano, 1969; Dall'Oglio, 1977). Data for the suspended load of these rivers are limited to 1933–1946 and 1949–1969 for the Tiber River; 1956–1959 for the Volturno River; 1936–1942 and 1957–1962 for the Simeto River. Suspended load data are merely indicative as they cannot be considered representative of the strongly changing climate of the late Quaternary. Moreover, the measure of suspended load is intrinsically underestimated by daily flow statistics that do not reveal the magnitude of peak events. In southern California, small- and medium-sized rivers can contribute as much as 90% of their load in a single flood lasting a few days (Warrick and Milliman, 2003; Warrick et al., 2008).

RESULTS

Upper Slope Gullies Offshore the Tiber Delta Mouth

New multibeam data acquired in 2001 (Figs. 3A, 3B) depict in detail the morphology of 15 submarine gullies (imaged in Chiocci and Normark, 1992) on the basis of high-resolution seismic profiles (Fig. 1). The gullies are V-shaped linear or slightly sinuous features that trend toward the west-southwest, perpendicular to the isobaths; they extend from the shelf break at ∼150 mbsl to 450 mbsl (Fig. 3A). Gully lengths vary between ∼1.5 and 5.5 km and appear to be controlled by the presence of a north-northwest–south-southeast–trending, wide channelized feature, where some gullies abruptly terminate (Figs. 3A, 3B). This feature was interpreted as the superficial expression of a growth fault affecting the continental slope (Chiocci and Normark, 1992). Gullies are 120–430 m wide and their depths are between 2 and 15 m; gully spacing ranges from 300 m to 1300 m (Table 2).

On high-resolution seismic-reflection profiles, gullies are characterized by concordant reflectors with little evidence of erosive truncation along their wall (Figs. 3C and 3D; Chiocci and Normark, 1992). Their initial shape is commonly a smooth negative relief with gently sloping walls, whereas their relief increases upward as a consequence of the higher aggradation in intergully areas with respect to the gully axes (Chiocci and Normark, 1992; Fig. 3D). Gullies are present only below the Last Glacial Maximum (LGM) unconformity and correlative conformity, that have been dated as ca. 27 ka through the integration of seismic-reflection data and accelerator mass spectrometry radiocarbon dating on foraminifera sampled in a core collected on the outer shelf facing the Tiber River mouth (P. Bellotti, 2010, personal commun.).

Gullies were recognized within at least four different shelf-slope prograding units (Fig. 3C) interpreted as fourth-order depositional sequences (Chiocci, 2000). Each of these sequences reflects one glacioeustatic cycle of ∼100 k.y.; the amplitudes of the sea-level fluctuation during these cycles were ∼100–120 m (Hays et al., 1976). The largest gullies tend to persist through successive depositional sequences, while others are abruptly buried by the following sequences.

Upper Slope Gullies Offshore the Volturno River Mouth

Multibeam data show a network of 10 gullies in the upper slope (gradients of 8°–10°) facing the Volturno River mouth (Figs. 4A, 4B). The gullies are commonly V-shaped and have a linear or slightly sinuous pattern, extending from the shelf break at ∼160 mbsl down to 440 mbsl, with lengths of 1–4 km. Gully widths are ∼100–300 m, the depths of the intergully areas are 2–12 m, and intergully spacing is 130–850 m (Table 2). Some gullies terminate on an oblique channel and/or larger gully owing to the complex canyon system affecting the slope.

On a very high resolution seismic-reflection profile, the gullies exhibit a concordant internal reflector geometry and little evidence of erosional truncation along their walls. Gullies are observed both above and below the LGM unconformity on the shelf and correlative conformity on the upper slope (Fig. 3C). The dating of this unconformity is speculative, although we are convinced of its validity. It is the youngest unconformity in the shelf stratal architecture that (1) has a regional extent; (2) truncates deposits of different age and nature but is always surmounted by transgressive and highstand systems tracts; (3) shows erosional characteristics down to ∼120 mbsl, i.e., the maximum depth reached by sea level during last glacial cycle. Where dating is available, the age of the deposit overlying it is always post-LGM. The unconformity is present in all the continental margins around Italy (Tortora et al., 2001) and in the Mediterranean (Chiocci et al., 1997); it is actually a polygenetic surface because after the subaerial exposure of the shelf during the LGM, it is often ravined by the rising sea level. In this case it is also diachronous, as it became younger coastward. In the outer shelf, however, where our study is, its age is usually between 18 and 10 ka.

Outer Shelf and Upper Slope Gullies in Front of the Simeto River

High-resolution multibeam bathymetry acquired in 2007–2009 offshore the Simeto River mouth exhibits tens of gullies on the outer continental shelf and upper continental slope, characterized by gradients of 2°–3° and 9°–12°, respectively (Figs. 5A, 5B). They start at ∼60 mbsl (this upper limit is obtained through the analysis of seismic-reflection profiles) and extend down to 300 mbsl, often merging downslope into larger channelized features. Gullies are ∼200 m to 3 km long, 30–240 m wide, and 2–20 m deep; intergully spacing ranges from 100 to 650 m (Table 2).

On high-resolution seismic-reflection profiles, gullies show an overall aggradation pattern in intergully areas, although erosional truncation is observed on gully walls (1–3 kJ sparker; Figs. 5C, 5D). Simeto gullies are developed only within postglacial deposits, i.e., above the erosive unconformity of the last glacial maximum (18 ka; LGM-U in Figs. 5C, 5D). The unconformity has been attributed to the shelf exposure during the last glacial maximum and ravinement during sea-level rise for the same reason expressed in the previous section. The unconformity truncates older canyon-fill deposits (PC in Figs. 4C, 4D); gullies are not present in these older deposits. These paleocanyons and their fill are located upslope of present-day canyons developed in the upper part of the continental slope (cf. buried and seafloor submarine canyons of the continental slope offshore New Jersey, USA; Pratson et al., 1994).

DISCUSSION

The acoustic characteristics of the gullies, i.e., concordant internal reflector geometry and little evidence of erosional truncation along their walls, lead us to interpret them as depositional features. Their growth seems to be related to a faster aggradation of intergully areas with respect to gullies axes, similar to features developed on the California margin facing the Eel River (Field et al., 1999) or larger submarine canyons developed offshore Brunei (Straub and Mohrig, 2009).

We believe that flood-generated hyperpycnal flows might be an important mechanism of axial gully sedimentation. It is important to note that cascading dense waters have been recently proposed as the source of flows able to carve submarine canyon and gullies on continental slopes (Canals et al., 2006; Micallef and Mountjoy, 2011). However, clear evidence of such dense water in the study area is lacking now; however, they could have been important during lowstand periods, when oceanographic conditions were probably different.

Although suspended sediment load data from the studied rivers are limited and could be underestimated with respect to the flood peak values (Warrick and Milliman, 2003), the maximum sediment concentrations in the Tiber and Volturno Rivers are always (at least for the monitored period) lower than the critical threshold of ∼40 kg/m3 required to overcome the density contrast between freshwater and seawater and to initiate a hyperpycnal flow (Table 1; Mulder and Syvitski, 1995). If hyperpycnal fluvial effluents are the primary mechanism of axial gully sedimentation, different scenarios of terrestrial sediment flux to the ocean are necessary to account for axial gully evolution. That is, scenarios and boundary conditions must have been extremely different relative to those recorded over historic times from stream gages. Two possible scenarios are proposed here: the first is related to sea-level fluctuations and the second is related to volcanic input.

Sea-Level Fluctuations and Hyperpycnal Flows

In Chiocci and Normark (1992), it was first proposed that gullies may form during sea-level lowering and lowstands, when climate conditions may have been very different from those at present and when a huge amount of unconsolidated, fine-grained sediment deposited during the previous highstand may have been eroded from the river valley, coastal plain, and shelf delta (e.g., Cattaneo et al., 2003; Fernández-Salas et al., 2003; Ridente and Trincardi, 2005), dramatically increasing the sedimentary loads of watercourses. Collier et al. (2000) demonstrated that sediment discharge rates in the northern Mediterranean during the last glacial lowstand largely exceeded discharge rates during the preceding interglacial highstand interval, as a consequence of enhanced seasonality during the glacial period, with cool, dry summers and wet winters. The seasonality combined with the enhanced erosion and transport during cold and arid periods.

These conditions might have promoted significant deposition and progradation of the continental slope, which predominantly comprises interpreted forced regressive and lowstand depositional systems tracts on other continental margins (e.g., Piper and Aksu, 1992; Chiocci, 1994, 2000; Chiocci et al., 1997; Trincardi and Correggiari, 2000; Hernandez-Molina et al., 2000; McMurray and Gawthorpe, 2000; Kolla et al., 2000; Ridente and Trincardi, 2002).

Deposition on the slope during periods of falling and low sea level likely occurred along the continental margin, with ephemeral river systems incising across the subaerially exposed shelf and directly feeding canyon and gully heads at the shelf edge. When subaqueous density currents debouched from lowstand river mouth reach the slope break, centrifugal instability at the interfaces of gravity currents may have been responsible for erosion of the seafloor and gullies formation (Fedele and Garcia, 2009).

On the contrary, sediment starvation of the continental slopes might occur during the rise and highstand of sea level, when the river mouths are far from the slope and most bedload might be trapped in incised valleys or along the shoreline (Chiocci and Normark, 1992; Chiocci, 2000). These conditions might have fostered the escape of fine-grained, fluvial sediment only as nepheloid layers, as evidenced by the 25 × 106 t of fine-grained sediments delivered to the sea in few weeks during the January 2005 Eel River flood (i.e., Walsh and Nittrouer, 1999). Therefore, during sea-level rise and highstand, it is likely that regional dispersion of fine-grained sediment and settling from suspension was the dominant mode of slope deposition, producing stratigraphic units of uniform and limited thickness that may mimic the previous gully morphology if present.

Volcanic Activity and Hyperpycnal Flows

The previous scenario is reasonable for the Tiber gullies developed within the falling stage and lowstand deposits, but it is unable to explain the occurrence of gullies during sea-level rise and highstand conditions, as observed in the Volturno and Simeto cases.

An alternative scenario is proposed on the basis of the spatial relationship observed between the studied rivers and the large volcanic edifices surrounding their lower reaches (Fig. 2). Volcanic activity may dramatically increase the amount of sediment delivered by watercourses, causing drastic landscape modifications in an extremely short period, such as the diversion of the local drainage network or the damming of water courses. These phenomena were observed during and immediately after the most important explosive eruption of Mount Pinatubo (Philippines), when lahars descended from volcanic edifices and dammed surrounding streams. When the impounded water overtopped the volcanic debris dams, the flooding ensued, such as the extremely devastating events in 1991, 1992, and 1994 (Newhall and Punongbayan, 1996). In addition, the large and sudden tephra fall-out input in the river catchment basins could provide unconsolidated fine-grained volcaniclastic sediment to the system in the years following the event. In this regard, annual suspended-sediment yields following the catastrophic 1980 Mount St. Helens eruption were as much as 500 times greater than the typical background level, and they generally declined for more than a decade (Major et al., 2000). Note also that the widespread tephra deposition can destroy or severely damage the sediment-stabilizing vegetation in the catchment basin, so that erosive processes can be enhanced in the aftermath of explosive volcanic eruptions, especially on loose fine-grained tephras.

We hypothesize that similar processes could have occurred during the past 18 k.y. (post-LGM) in the Simeto catchment basin, due to the intense explosive activity that characterized the most recent evolution of the Etna edifice (past 15 k.y., according to Coltelli et al., 2000; see also Fig. 6) as well as in the Volturno catchment basin due to the 12 ka Plinian eruption of the Phlegrean Fields (Florio et al., 1999). In this regard, a higher occurrence of flash-flood–generated hyperpycnal flows was observed for the small river draining the close Sorrentina Peninsula after the A.D. 79 Pompeii eruption of Mount Vesuvius, leading to the development of small fan-deltas offshore the mouths of these rivers (Sacchi et al., 2009).

Also in the Tiber case, the 800 ka volcanic activity of Mount Sabatini and Colli Albani Hill fits the occurrence of gullies in the late Pleistocene fourth-order sequences developed in the past 800 k.y. (Fig. 6; Chiocci and Normark, 1992). Conversely, gullies are not found in the last transgressive and highstand deposits, where a thin veneer of mud in the upper slope merely drapes them (Chiocci and Normark, 1992), consistent with the cessation of volcanic activity ca. 25 ka for Colli Albani district.

This volcanic supply hyperpycnal model is somehow similar to the sediment supply hyperpycnal model proposed for submarine gullies developed on other Mediterranean continental shelves and slopes facing dirty rivers, such as the Andarax and Guadalfeo Rivers (Garcia et al., 2006; Lobo et al., 2006), where gullies are interpreted to have developed as a consequence of flash-flood–generated hyperpycnal flows.

In the Etna case, there is also clear evidence of how the physiographic characteristics of the margin play an important role in gully formation. Two rivers flow at the base of the Etna Volcano and collect its volcaniclastic input: Simeto to the south and Alcantara to the north (Fig. 2). Gullies are present offshore the Simeto River mouth; however, they are lacking offshore the Alcantara River mouth, where the shelf is absent and a wide canyon system is present (Fig. 7).

CONCLUSIONS

The characterization of submarine gullies in the outer shelf and upper continental slope offshore three small- and medium-sized rivers in central and southern Italy provides an updated interpretation of these features (previously described in Chiocci and Normark, 1992).

The gullies are interpreted as depositional features related to a faster aggradation of intergully areas with respect to gullies axes. This process is interpreted to be related to flood-generated hyperpycnal flows that hinder or decrease sedimentation in the gully axis, based on the proximity of gully heads to river mouths. Nevertheless, generation of hyperpycnal flows during river flooding is connected to the watercourse sediment concentration, which is commonly low in medium-sized rivers. Hence, two scenarios of enhanced sediment availability in catchment basins have been proposed to explain the generation of gullies. The first scenario is related to sea-level fluctuations during the late Quaternary, taking into account greater sediment availability to shelf edge staging areas during the falling stages and lowstands, as suggested in Chiocci and Normark (1992) and Field et al. (1999). The second scenario is related to the emplacement of a large amount of volcaniclastic sediment during explosive eruptions in river catchment basins close to onshore volcanic edifices, increasing the sediment load of the rivers and leading to hyperpycnal flows.

We thank the crews of R/Vs Urania and Universitatis and Odin Finder and the people taking part in the surveys. P. Bellotti is gratefully acknowledged for the accelerator mass spectrometry dating on the core collected from offshore the Tiber River mouth. The manuscript greatly benefitted from the revisions of A. Cattaneo, R.B. Wynn, A. Fildani, and an anonymous reviewer. Chiocci expresses deep gratitude to the bright figure of Bill Normark, who is somehow the hidden author of this article, not only because it is based on his original ideas and observations, but because it would not have been possible without his teaching on curiosity-driven research, the need to rely on data rather than on model, and the capability to revisit the already expressed interpretations.