The Neogene tectonics of the central Mediterranean are related to the subduction and trench rollback of the Ionian basin under Eurasia, causing the opening of the Liguro-Provencal and Tyrrhenian back-arc basins and the formation of the Calabrian accretionary wedge. The Calabrian accretionary wedge is a partially submerged accretionary complex located in the Ionian offshore and laterally bounded by the Apulia and Malta escarpments. While the history of the back-arc extension process is fairly well defined, the structure and evolution of the wedge are still poorly known. We have analyzed and interpreted the available reflection multichannel and single-channel seismic profiles in the Ionian offshore and integrated them with other geological and geophysical data acquired in the last 40 years. Here we present unpublished seismic profiles to provide a new map of the tectonic structure of the Ionian offshore and to define the structure of the wedge and its evolution during the last 15 Myr. Our reconstruction points out that the Messinian salinity crisis represents an important break in the evolution of the wedge, as the basal décollement ramps up onto the Messinian salt deposits, producing a dramatic and fast forward propagation of the frontal thrust and resulting underplating of the underlying crustal Ionian sequence during progressive trench rollback. Our results provide new insight into the style of accretion in a weakly converging setting, which is typical for the Mediterranean region.
 The shape of an accretionary wedge is considered to be primarily controlled by the strength of its basal décollement and by the resistance of the accreted material [Davis et al., 1983; Dahlen et al., 1984]. Coulomb wedge theory furnishes the fundamental tools to predict wedge evolution, but other external parameters can produce important departures from the model prediction. Among others, erosion and sedimentation are well established parameters playing a fundamental role in the style of wedge growth [e.g., Storti and McClay, 1995]. In the Mediterranean region, several accretionary wedges, from Cyprus to Gibraltar, were at work during the Messinian. The Messinian salinity crisis caused not only an acceleration of the erosion rate of shelves and mountains but also the deposition of thick salt deposits [e.g., Ryan, 2009], which, because of their low strength, may have played an important role in the accretionary process [Davis and Engelder, 1987; Costa and Vendeville, 2002]. The deposition of salt, for example, strongly influenced the mode of extension for subsiding back-arc basins, such as the Gulf of Lyon, and the mode of accretion at convergent margins. A striking example arises from the Mediterranean Ridge, where the salt basal décollement produced a low-tapered structure and fast forward propagation of the frontal thrust [Ryan et al., 1982; Kastens et al., 1992; Chaumillon et al., 1996; Chaumillon and Mascle, 1997; Kukowski et al., 2002; Polonia et al., 2002; Reston et al., 2002a; Costa et al., 2004]. The impact of the Messinian salinity crisis on the Calabrian prism in the central Mediterranean (Ionian Sea) is far less studied, and basic questions concerning its evolution remain open.
 Here, we reconstruct the evolution of the Calabrian accretionary prism in the Ionian Sea, analyzing and interpreting the complete set of seismic data acquired in the area over the last 40 years. Our reconstruction, based on seismic line interpretations, emphasizes that the deposition of salt during the Messinian salinity crisis dramatically influenced the wedge growth, creating a low, almost flat-tapered accretionary complex and producing underplating of the underlying thick sedimentary pile.
 Apatite fission track analysis shows that little erosion occurred in the basement nappe during the last 20 Myr [Thomson, 1994], indicating that most of the present-day topography results from the recent uplift of the belt. Stratigraphic and structural relationships in the fore-arc basin indicate that uplift started during the lower Pliocene [Roda, 1964; Bonardi et al., 2001], accelerating during the Pleistocene as attested by a set of marine terraces sculpted into the basement rocks [Ferranti et al., 2006].
3. Seismic Data Set: Methods
Figure 2 shows the locations of the multichannel seismic reflection profiles used in this study (i.e., Italian Commercial (F and D zones), Mediterranean Sea survey (MS), Calabrian Arc survey (data supplied courtesy of Fugro), ION Streamers profiles, CROP seismic reflection profiles, PM01 of MCS Prismed survey and ARC of the Archimede academic cruise by the Ecole Normale Supérieure team) along with borehole logs, ESP refraction data, dredged and gravity or piston cores, the Ocean Drilling Program (ODP) and the Deep Sea Drilling Project (DSDP).
Figure 3 shows the acoustic stratigraphy of four sections along a NW-SE profile crossing the Calabrian wedge, from the Crotone-Spartivento basin to the foreland basin. Five main seismostratigraphic units (PQ, ME, PM, MC and acoustic basement) separated by regional reflectors (A, B, K and S) were detected in the foreland basin. These reflectors, which are well known in the Mediterranean region [Finetti and Morelli, 1973; Ryan and Cita, 1978; Casero et al., 1984; Reston et al., 2002a; Polonia et al., 2002; Hieke et al., 2003], are clearly recognizable in the undisturbed Ionian foreland region, where they are calibrated by ESP refraction data [de Voogd et al., 1992] and by DSDP (Site 374, Messina) drilling. The PQ unit consists of subparallel, high-frequency, low-amplitude reflectors, consistent with terrigenous and turbidite deposits of Plio-Quaternary age, penetrated by DSDP Site 374 and oil wells near the coast. The unit is characterized by an average thickness in the seismic reflection profiles of 200–400 ms TWT (about 400 m) in the entire Ionian offshore, while in the Crotone-Spartivento basin, it reaches a maximum thickness of 2 s TWT. The PQ unit, which is almost undeformed in the foreland basin, is folded and faulted in the internal part of the wedge. In the fore-arc basin in particular, the PQ unit is gently folded and affected by normal faults (synsedimentary or postsedimentary) and involved in salt tectonics, as documented by diapiric structures. The base of the PQ unit is marked by a strong high-amplitude reflector, named the A or M reflector [Ryan, 1969], which is almost continuous throughout the basin and easily recognizable. The A or M reflector represents either the top of the Messinian evaporite and clastic sediments when present (ME unit) or an erosional surface, marked by a continuous and strong reflector often delineating a prominent angular unconformity. Therefore, because the M reflector could be a time-transgressive marker and does not always represent the top of the Messinian, it can be more properly defined as the base of the PQ unit. The ME unit, bounded by the A (or M) and B reflectors, is present in both the inner portion of the wedge, as indicated by diapiric structures, and in the foreland basin, where it reaches an average thickness of 0.4 s TWT. Here, the ME unit consists of an upper high-velocity layer related to the upper evaporitic layer, and a lower less reflective one related to the Messinian clastic sequence. The base of the ME unit is marked by the B reflector, a planar, continuous and high-amplitude reflector often representing the first flat reflector below the diapiric structures [Finetti and Morelli, 1973]. Unlike the Plio-Quaternary sequences and the top of the Messinian evaporites (DSDP Site 374), no direct wells were drilled through the pre-Messinian sequences. Geological correlation, previous works in Mediterranean regions, and seismic velocities were used to interpret these sequences. The pre-Messinian units are composed of the PM unit, a poorly reflective and well-layered sequence, consistent with deep water clastics and marls (Tertiary terrigenous sediments [Ryan et al., 1973; Hsü et al., 1978]). This unit is deformed under the Ionian basin, and its thickness varies from 1.5 to 0.4 s TWT. Moving landward, the sediments are progressively incorporated into the Calabrian wedge. Its lower boundary, the K horizon, represents the top of the Mesozoic carbonate series (MC unit), characterized by strong, relatively continuous, and low-frequency reflections with an average thickness of 1.5 s TWT. Below this unit, the S reflector is commonly described as the top of the acoustic basement; it lies at about 8 s TWT in the foreland and has been interpreted as the upper layer of oceanic crust (layer 2a) [de Voogd et al., 1992]. This poorly defined basement is characterized by a reflective sequence that shows increasing velocity with depth (ESP data).
 Following Rossi and Sartori , the wedge can be divided in four main tectonic domains, marked by different stratigraphies and styles and ages of deformation (Figures 4 and 5).
 1. The Crotone-Spartivento basin is a subsiding, relatively undeformed, basin filled by Plio-Quaternary sediments. The basin deposits lie on top of either the crystalline basement or the inner accretionary wedge. The southern boundary of the basin is represented by the fore-arc outer thrust.
 2. The inner accretionary wedge represents the inner portion of the wedge, composed of pre-Messinian sediments, accreted mainly during pre-Messinian time. It is separated from the outer accretionary wedge by the main external ramp. A major tectonic feature, the “Crotone-Spartivento slope,” divides the inner wedge into two portions characterized by different topographic slopes.
 3. The outer accretionary wedge represents the frontal part of the wedge, composed of deformed Messinian to Quaternary deposits.
 4. Foreland basin represents the foreland basin of both the Calabrian Arc and the Mediterranean Ridge. The foreland basin shows compressive pre-Messinian structures.
4.1. Crotone-Spartivento Fore-Arc Basin
 The Crotone-Spartivento basin is a NE-SW oriented, 220 km long and 40 km wide fore-arc basin extending from Crotone to Spartivento both off- and onshore Calabria (Figures 4 and 6). It has so far been interpreted as a “fore-arc” basin, as it is composed of relatively undeformed middle-upper Miocene to Quaternary deposits unconformably covering the Calabria-Peloritani Arc nappe stack [Amodio Morelli et al., 1976; Cavazza and De Celles, 1998]. The basin is bounded seaward by a post-Messinian SE verging fore-arc outer thrust. This reverse fault represents a relatively continuous element across the mapped area (Figure 4).
 The basin stratigraphy is characterized from bottom to top by Serravallian (or slightly older Aquitanian) to Tortonian clastic sequences (conglomerate, sandstone and mudstone), cropping out in the Crotone area (San Nicola and Ponda formations [Roda, 1964]). These are followed by the Messinian sequence, which displays a variable thickness and a regional internal angular unconformity. On top of the Messinian unconformity, a thin layer of Messinian salt is present (Figure 7). The last and thicker (0.4 s TWT to more than 2 s TWT) sequence is composed of a Pliocene to Quaternary subhorizontal and well-stratified unit. On shore and in the shallow water area, this unit is characterized by an overall fining upward, from conglomerates and sandstones to marine shales.
 Several unconformities show a more complex tectonosedimentary history of the basin, defined by an overall progressive seaward migration of the depocenters. Relationships between Miocene and Pliocene deposits onshore in the Crotone basin suggest that the progressive migration of the basin depocenter is locally related to the uplift of the belt, which probably started at the late Miocene-Pliocene boundary, during a new thrusting phase. The clearest unconformities in the seismic sections are the infra-Messinian and infra-Pliocene angular unconformities, which rests on top of quasi-transparent lower Pliocene Trubi Formation (Figure 7a). These unconformities are probably related to the activity of several landward and seaward dipping thrusts and back thrusts, clearly active during Messinian-Pliocene time (Figure 7).
 The upper Pliocene-Quaternary sediments are also deformed mostly by synsedimentary normal faulting, forming tilted blocks and rollovers, by extensional diapirs in the inner part of the basin [Barbieri et al., 1982; Zecchin et al., 2004], and by shallow compressive structures in the outer sector. These deformations, associated with other sliding and slumping features [Barbieri et al., 1982], can probably be related to gravity sliding processes above a Messinian salt layer (Figure 7b). One of the largest gravitational structures is in the Crotone peninsula (Figure 4 and Figure 7b, left), where a thick Pliocene and Pleistocene sequence slides seaward over the Messinian salt layer, producing large convex listric normal faults onshore. Overall, we cannot exclude that the massive gravitational sliding was triggered by the growth at depth of compressional structures.
Figure 7 also shows the presence, below the upper Miocene sequence (Ponda and San Nicola formations), of south dipping reflectors probably representing the top of the Calabrian basement unit, cropping out in the Sila massif, and overthrust onto a stratified sequence made up of sedimentary sequences that are probably carbonate or Eocene-Oligocene flysch deposits.
4.2. Inner Accretionary Wedge
 The inner accretionary prism is an arcuate, 130–140 km long and 280–300 km wide belt extending from the fore-arc outer thrust to the main external ramp, which separates the highly deformed older portion of the wedge from the outer accretionary wedge and foreland basin (Figure 4). We thus consider the main external ramp to be the outermost frontal ramp of the wedge, where the basal décollement cuts up through the sedimentary sequence and is located at the base of the ME unit (Figures 4 and 5). In map view (Figure 4), this element has a convex shape running from the eastern margin of the Hyblean plateau toward the Apulian plateau. In the western sector, this ramp is reactivated with normal/strike-slip motion, as attested by the growth of an extensional basin that is probably Pleistocene in age (Figures 4 and 11).
 The wedge itself consists of an extremely deformed nappe stack of pre-Messinian units scraped off from the downgoing plate. It is characterized by several NE-SW oriented and southeast verging thrusts, locally reactivated in post-Messinian time, deforming the M reflectors. On top of the M reflector, a thin layer of Plio-Quaternary sediments is also present and is locally involved in the deformation. In the southeastern part of the inner accretionary prism, for example, we observe a spectacular post-Messinian positive inversion structure (Figure 8), as it involves a thick basin sequence characterized by high amplitude and continuous reflectors. In map view, this inverted structure closes toward the southwest (Figure 4).
 The inner wedge can be divided into two domains characterized by different topographic gradients (from 0.6° in the inner portion to 0.14°). The divide is marked by a regional thrust front marked by a clear 0.6 s TWT (∼450 m) morphological step (Crotone-Spartivento slope [Rossi and Sartori, 1981]) (Figures 4 and 5). This structure has been previously interpreted either as the superficial expression of the subduction zone [Sioni, 1996], or it is considered as one of the main thrusts of the Calabrian Arc [Rossi and Sartori, 1981]. As discussed below, we consider this structure to be one of the most recently activated thrusts, with a surface expression of deep underthrusting.
4.3. Outer Accretionary Wedge
 The outer accretionary wedge extends from the main external ramp to the foreland basin where the Plio-Quaternary unit is undeformed. On the western side, it pinches out against the Malta Escarpment, while on the eastern side it collides with the Mediterranean Ridge (Figure 4). The total length of the outer wedge is about 100 km toward the abyssal plain, decreasing to ∼40 km toward the Malta Escarpment (Figure 9). The outer accretionary wedge displays a thickness decreasing outward from 1.8 s TWT at the main external ramp, to ∼0.3 s TWT in the abyssal plain. The topographic slope of the wedge (α) is about 0.36°. The wedge itself is seismically quasi-transparent without internal reflection, and is bounded by the B and A reflectors, representing the base and top of the Messinian unit (ME), respectively. While the A reflector has been directly reached by drilling (Site 374) as part of the Deep Sea Drilling Project, which penetrated the Plio-Quaternary sequence and over 80 m into the Messinian evaporite formation [Hsü et al., 1978], the nature of the B reflector is only inferred [Finetti and Morelli, 1973]. The presence of the evaporitic layer in the outer wedge has been also confirmed by the ESP 4–5 [de Voogd et al., 1992] running in the foreland basin and farther upslope onto the Calabrian wedge. Below the outer wedge, it is possible to observe the subducting sediments characterized by subhorizontal parallel reflectors, locally truncated by the basal décollement, which could be interpreted as Mesozoic and Tertiary clastic sequences and by Ionian crust, as suggested by the related refraction velocity [de Voogd et al., 1992]. The wedge is covered by a thin sequence of Plio-Quaternary turbidite deposits [Müller et al., 1978; Hieke, 2000]. The PQ unit thickens toward the Ionian abyssal plain (i.e., 450 m in ESP 5). The Plio-Quaternary unit shows small folds, creating the typical cobblestone topography [Hersey, 1965], interpreted as megaripples [Emery et al., 1966], karst structures [Ryan et al., 1973; Hinz, 1974], or related to gravitational viscous flows in the frontal thrust area [Rossi and Sartori, 1981; Finetti, 1982].
4.4. Foreland Basin
 The foreland basin is represented by a triangular flat basin within the Ionian abyssal plain. It is bounded by the Medina Mounts, a complex tectonic ridge structure [Finetti, 1982], by the frontal thrust of the Mediterranean Ridge and by the toe of the Calabrian accretionary complex, marked by a gradual onset of folding in the PQ unit. The Medina Mounts and their extension (Victor Hensen structure (VHStr) or Marconi [Finetti, 1982], Nathalie structure (NStr) [Hieke and Wanninger, 1985] and Valdivia structure (VaStr) [Hieke et al., 2003]) form a belt located between the Sirte basin and the Ionian abyssal plain and show a sharp NE-SW escarpment. A vertical displacement of about 0.1 s TWT of the M reflector above NStr indicates post-Messinian movements [Hieke et al., 2003]. In some places, the Medina Mounts fault crosscuts the seafloor, suggesting recent extensional tectonics.
 The seismic stratigraphy of the Ionian basin has been calibrated using DSDP Site 374 [Hsü et al., 1978] and ESP 5 refraction data [de Voogd et al., 1992]. The PQ unit is undeformed and lies horizontally above the ME unit. Moving toward the prism, PQ shows little deformation with small wavelength folds (cobblestone topography). Below the B reflector, a well-stratified sequence of thick Mesozoic carbonates and Tertiary clastic sediments is present. The minimum total thickness of these sedimentary sequences should be of 4–5 km [de Voogd et al., 1992], but could be higher depending on the interpretation of the nature of the underlying crustal layer.
 In the central abyssal plain, the Ionian crust is deformed by a system of NE-SW striking reverse faults that mainly dip to the southeast (Figures 4 and 10). The overall structure resembles a ∼50 km wide pop-up. Deformation occurred during deposition of the pre-Messinian PM unit, with southeast dipping reflectors. The increase in thickness of the MC unit inside the deformed area also indicates that the compressional structure reactivated a preexisting extensional basin. Since the ME unit is not involved in the compressive deformation, this event was probably Tortonian in age [Sioni, 1996; Chamot-Rooke et al., 2005]. The total amount of shortening is about 5–6 km [Chamot-Rooke et al., 2005]. This event was probably responsible for the rise of the Medina Mounts, even though in some places, faulting along the escarpment seems to still be active [Hieke et al., 2006].
4.5. Lateral Margins
 The two lateral margins of the Calabrian accretionary wedge are quite different (Figures 4, 11, and 12). The western margin is located east of the Malta Escarpment. Figure 11 displays an E-W striking seismic profile perpendicular to the escarpment that extends from the Hyblean plateau to the Calabrian accretionary prism. The profile shows the contact between the outer accretionary wedge and the foreland basin to the west, composed of the Hyblean plateau down-faulted toward the Ionian Sea, with the severely deformed inner accretionary wedge to the east. The thickness of the outer accretionary wedge varies from 1.2 s TWT near the contact with the inner accretionary wedge to 0.2–0.3 s TWT toward the Malta Escarpment. The thinning of the outer wedge is not progressive but decreases abruptly along a fault zone with probable strike-slip movement. Salt diapiric ascents are associated with this fault (Figure 11). In map view, this fault zone is composed of different segments running parallel to the Malta Escarpment for ∼100 km (Figure 4). Toward the east, the boundary between the inner and outer wedges is marked by the main external ramp, interpreted as a strike-slip structure [Consiglio Nazionale delle Ricerche, Progetto Finalizzato Geodinamica, 1983; Streamers Profiles Working Groups et al., 1996]. However, the presence of a basin filled by growth strata of the PQ unit indicates that the structure was recently reactivated with a normal component of slip. This extensional basin, with an average thickness of 0.1–0.2 s TWT, runs parallel to the Malta Escarpment and then turns E-W, following the main external ramp (Figure 4). The strike-slip fault system and the main external ramp join together farther north into a Pliocene to recent strike-slip/normal fault system well described along the eastern side of Sicily [Hirn et al., 1997; Monaco et al. 1997; Bianca et al., 1999; Adam et al., 2000; Argnani and Bonazzi, 2005].
 The eastern margin of the Calabrian accretionary prism is defined by the southern Apennines' frontal thrust, by the Apulian Escarpment and, to the south, by the contact with the Mediterranean Ridge. Seismic profiles in Figure 12 show the flexure of the Apulian plateau below the Southern Apennines to the west and below the Hellenides to the east. The Apulian foreland divides two foredeep basins related to these fold and thrust belts. The profiles show a progressive closure of the Bradanic foredeep basins to the south, where the Apulian foreland is no longer flexed but abruptly truncated by the Apulian erosional escarpment (Figure 12). As shown in Figure 4, this NW-SE trending geomorphic feature places the southern sector of the Apulian platform in contact with the deformed sequence of the inner accretionary wedge.
 The closure of the Bradanic foredeep and the termination of the flexure of the Apulian plateau indicate that the Plio-Pleistocene Southern Apennines' frontal thrust cannot be linked with the outer wedge of the Calabrian Arc. This could mean that these two structures evolved independently in recent times.
 The overall geometry of the Calabrian accretionary prism, as deduced from our interpretation of seismic reflection profiles, is shown in Figure 13. As described before, the wedge can be divided into an inner domain, probably made up of an Upper Cretaceous to Messinian sedimentary sequence [Barone et al., 1982] and an outer domain formed exclusively of Messinian and post-Messinian deposits. The boundary between these two domains is marked by the main external ramp, a ramp cutting down through the Tertiary and Mesozoic sequences and probably joining the deep reflectors of the basal décollement (14 s TWT) recognized in the inner part of the wedge in ION-Streamer profiles [Streamers Profiles Working Groups et al., 1996] (Figure 13).
 The nature of the outer wedge, due to the absence of internal reflections within it, remains a point of debate and is subject to various interpretations. Three alternative models can be put forward to explain the structure of the outer wedge. In the first model, the outer wedge represents an allochthonous gravitational “olistostrome” terrane [Rossi and Sartori, 1981]. In the second model, conversely, the outer wedge represents a sedimentary flexural wedge filled by Messinian deposits [Casero and Roure, 1994; Holton, 1999]; in this model, the wedge is bounded by reflectors A and B, the base of unit PQ and the top of the Tortonian sequence, respectively. The third model considers the outer wedge as a tectonic wedge made up of deformed Messinian deposits [Polonia et al., 2008]. Hence, in this model, the B reflector, representing the basal décollement of the wedge, ramps into the Messinian deposits and then flattens to run along the base of the evaporites. The first hypothesis can be rule out as the outer wedge preserves lateral continuity of its internal structure and shape along the entire margin, as shown in Figure 9. Moreover, there are no evidences of niche structures and/or sliding detachment expected for large-scale deep gravity process.
 The sedimentary wedge hypothesis imposes that the Messinian outer wedge is a flexural accommodating depositional setting. This second hypothesis can be hardly applied to our setting as (1) the outer wedge lacks of the expected internal stratification, typical of foredeep basins, and is conversely characterized by a chaotic seismic facies in ME unit and by a presence of intense folding within PQ unit, and (2) the size of the wedge (100 km long and 300 km wide) would have been far larger than other flexural basins within the Mediterranean domain [Casero and Roure, 1994]. In this scenario, in fact, the frontal thrust should have attained its present-day position already 5–6 Myr ago. But, the opening of the Tyrrhenian back-arc basin imposes a large (∼300 km) post-Messinian drifting event of the Calabrian fore arc. Therefore, either the accretionary complex was unrealistically longer than the present-day one, or the whole Messinian fossil Calabrian wedge was rigidly translated southeastward accommodating the shortening over the African margin, where there is no trace of such deformation.
 Several lines of evidence support the third model.
 1. The velocity model of de Voogd et al.  indicates that the outer wedge attains a mean velocity of 4.3 km/s, compatible with a wedge made up of Messinian sediments.
 2. The flat base of the evaporites (B reflector) is discordant with the underlying reflector. Therefore, the B reflector could represent the basal décollement rather than a stratigraphic horizon. In this model, the well-stratified folded cover of unit PQ, with its peculiar hummocky morphology, originated from an array of compressive structures within the outer wedge.
 3. From de Voogd et al.'s  velocity model (1500 m/s water, 1800 m/s unit PQ, and 4300 m/s as average value for unit ME), we estimated a basal décollement dip β of about 0.96°, and a topographic slope α of 0.36°. This very low tapered wedge is expected for a very weak basal décollement, and it falls into the salt orogenic wedge field [Davis and Engelder, 1987].
 5. The deformational style of the outer wedge could be imagined as dominated by a bulk coaxial thickening, typical of a viscous salt-like rheology. Restoring the present-day wedge shape to the thickness of the Messinian sequence in the foreland basin and assuming no volume change, we estimate that more than 200 km of coaxial shortening may have been accommodated inside the outer wedge. This model could then correctly account for the amount of shortening and subduction expected after the Messinian for the Calabrian subduction zone [Faccenna et al., 2001].
 All of these points suggest that the synaccretionary deposition of the Messinian evaporite-bearing deposits on top of the accretionary wedge and in the frontal undeformed foreland region could have played an important role.
 As observed in the Mediterranean Ridge [Polonia et al., 2002; Reston et al., 2002b], this event could have dramatically influenced the overall geometry and deformation style of the wedge. Unlike the Mediterranean Ridge, the Calabrian wedge shows post-Messinian deformation in the inner domain as well, probably reactivating preexisting structures. This indicates a back-stepping of the deformation front, suggesting a synchronous movement in a different part of the wedge. This process can be explained in terms of critical wedge theory, as the fast frontal propagation above the salt layer would have anomalously lowered the wedge profile. The inner deformation could then be interpreted as necessary to reestablish a critical taper profile. This could also explain the onset of late Messinian-Pliocene compressional uplift of the Sila-Serre-Aspromonte belt, concurrent with the onset of subsidence in the Crotone-Spartivento and Paola basins.
 Eventually, the upward post-Messinian jump of the basal décollement inside the shallow layer of the outer wedge is expected to produce large-scale underthrusting of the Meso-Cenozoic sequences below the wedge itself. Because of their inherent buoyancy, these layers are likely underplated beneath the inner wedge producing progressive crustal thickening.
 Unfortunately the deep structure of the wedge is poorly illuminated by few available deep seismic profiles. Receiver function analysis, however, reveals the presence of 9 km of “underplated sedimentary sequence” beneath the upper plate Moho at ∼20 km, and on top of a 5 km of subducting Ionian crust [Piana Agostinetti et al., 2009]. Hence, the rather thick (∼35–38 km) crustal thickness observed beneath Calabria [Piana Agostinetti and Amato, 2009; Piana Agostinetti et al., 2009] is probably related to the conspicuous underplating of shallow crustal layers scraped off from the subducting plate. The timing of this event is constrained by the erosion and deposition history of the Sila-Aspromonte Massif and Crotone-Spartivento basin. Fission track data [Thomson, 1994, 1998] show little or no erosion during the last 15 Myr, indicating that the present-day elevation of the Sila-Aspromonte Massif was only recently achieved. On the same line, the geometry of the Crotone-Spartivento marine sequences onlapping over the eastern flank of the Sila Massif shows an overall seaward migration of the basin depocenters [Rossi and Sartori, 1981] and thrusting during late Miocene-Pliocene time [Roveri et al., 1992]. These evidences indicate that the uplift of the inner portion of the wedge started in the late Messinian-early Pliocene time [Bonardi et al., 2001; Molin et al., 2004]. This process can be reasonably related to deep underplating and can be linked to the dramatic lowering of the taper profile achieved after the rapid frontward propagation of the basal dècollement within the salt layer.
 If this scenario is correct, we may reinterpreted the deep signals detected in the ION Streamer profiles (previously interpreted as lower crust reflectors by Streamers Profiles Working Groups et al. ) as the basal and roof thrusts of a prominent large-scale underplated body. This structure, indeed, is positioned in correspondence of the underplated sedimentary sequence imaged by receiver function. In Figure 13 is reported the lateral extension of these reflectors, projected beneath the inner wedge at a depth of 10–16 s TWT. The reconstructed geometry resembles strongly to a crustal duplex. It is worth noting that in correspondence of this structure the morphology of the wedge is remarkably different from the outer wedge portion forming a gentle, large-scale bulge (Figure 13b). In addition, the outer frontward tip of the deep structure is marked by a active morphological scarp (“Crotone-Spartivento” slope) marking the sharp change in the wedge taper.
 Conversely, the external boundary of the outer wedge is morphologically less evident, and only locally marked by a gradual onset of folding in the PQ unit [Gutscher et al., 2006]. Further high-resolution seismic data will be necessary to better establish the present wedge activity.
 Another interesting element is the presence in the foreland basin of a compressional structure composed of 10 km spaced N30°–40° striking southeast and northwest dipping thrusts accommodating a total displacement of about 10 km. The deformation involved growth deposits up until the upper Miocene (B reflector), so it probably developed between 13 and 8 Ma [Sioni, 1996; Chamot-Rooke et al., 2005]. The pattern of deformation and seismic facies can be clearly related to the inversion of a preexisting extensional basin. This type of intraplate deformation could be related to a pause of the subduction process. During this time the wedge was then unable to absorb deformation and convergence.
6. Evolution of the Accretionary Prism
 In the central Mediterranean, subduction is primarily related to intermittent trench retreat producing back-arc extension at an average rate of a few centimeters per year, and to a lesser degree to the slow convergence between Africa and Eurasia. The latter usually did not exceed 1 cm/yr [Jolivet and Faccenna, 2000; Faccenna et al., 2001]. Neogene shortening in Calabria is attested to by the nappe stacking, producing overthrusting of the Calabrian basement units over the flysch sequence, and is considered to have been completed in the Burdigalian [Roveri et al., 1992; Bonardi et al., 2005]. Extensional deformation accompanied the growth of the accretionary wedge in the back-arc region. The arcuate shape of the Calabrian Arc, in particular, formed mostly between the Serravallian, after the end of the Liguro-Provençal oceanic spreading phase [Faccenna et al., 2004, 2005; Cifelli et al., 2007; Mattei et al., 2007; Chiarabba et al., 2008] and the late Miocene and the Pleistocene, during the opening of the Tyrrhenian sea. Between 15 and 10 Ma, the retreat drastically decelerated [Faccenna et al., 2001], but a high spreading rate was reestablished during the opening of the Vavilov basin (6–4 Ma) and the Marsili basin (2–1 Ma). Nicolosi et al.  analyzed magnetic anomalies and concluded that the opening of the Marsili basin occurred through a succession of ultrarapid (2.1–1.6 Myr at 19 cm/yr) and slow spreading events (post-0.78 Myr, magmatic inflation of the Marsili Seamount) separated by periods of relative quiescence (about 1 Myr). Although the subduction zone below Calabria is seismically active down to a depth of 450 km [Selvaggi and Chiarabba, 1995], the subduction rate is presently only a few millimeters per year, and trench retreat has stopped its retrograde motion, as suggested by geodetic [Hollenstein et al., 2003; D'Agostino and Selvaggi, 2004] and paleomagnetic data [Mattei et al., 2007]. This suggests that subduction in the central Mediterranean is decaying.
 According to Faccenna et al. , the total amount of subduction for the Tyrrhenian system should be derived by adding the amount of convergence [Dewey et al., 1989] to the amount of back-arc extension [Patacca et al., 1990]. The value obtained is, since 10 Ma on, about 60 km of convergence and 380 km of retreat for a total amount of subduction of 440 km. This is the shortening that should have been accommodated in the Calabrian accretionary wedge. Our data indicate that the fold-and-thrust belt, formed during the Neogene, is composed of a system of NNE-SSW and NE-SW directed thrusts with a southeastward sense of transport.
 Assuming the outer wedge as a tectonic complex, in the evolution of the accretionary complex, most of the post-Messinian shortening can be accommodated by growth of the outer wedge and consequent underplating of the Meso-Cenozoic Ionian sedimentary sequence.
 As discussed above, this option can also account for the first-order geometry of the wedge. Following this model, we can synthesize the evolution of the Calabrian wedge into five main stages (Figure 14).
Figure 14 shows the evolution, in five main steps, from the Serravallian until the present-day, of the Calabrian subduction system. The geometries of the deep structures, such as the Moho depth, are constrained by seismic data and receiver function analysis [Piana Agostinetti and Amato, 2009]. While the evolution of the wedge is only schematically sketched, the vertical and horizontal displacements follow the plate reconstruction [Faccenna et al., 2004; Patacca et al., 1990].
6.1. Phase 15–10 Ma: Intraplate Deformation and Subduction Zone Locking
 This time interval (Figure 14a) was characterized by a decrease in the subduction velocity as the rollback velocity decreased for a few million years between the end of the Liguro-Provençal spreading and the onset of the main phases of Tyrrhenian extension [Faccenna et al., 2001]. The net convergence velocity perpendicular to the trench was of the order of 1–2 cm/yr [Faccenna et al., 2001]. At that time, the Calabrian wedge was already built up, although its initial shape was poorly constrained. The stalling phase of rollback, exhibited by the decrease of back-arc extension and arc volcanism [Peccerillo et al., 2008; Lustrino et al., 2009], shows the poor efficiency of the subduction system. In the inner part of Calabria, this period is marked by extensional processes, with the growth of small basins developed mainly on the Tyrrhenian side [Mattei et al., 2004]. While few constraints are available to describe the geometry of the wedge, the foreland basin is fairly well illuminated by seismic images, showing the trace of intraplate compressive deformation in the Ionian abyssal plain about 300–400 km outside of the frontal ramp of the Calabrian wedge (R1). The intraplate compressional deformation reactivated a preexisting extensional basin, absorbing about 10 km of the plate convergence. The evidence for this significant intraplate shortening together with the reduction of the rollback velocity indicates that at this time, the subduction zone was mostly locked. This stasis phase in the accretion and wedge growth has been related either to deep slab deformation or to the entrance at the trench of a continental block [Faccenna et al., 2001; Gattacceca and Speranza, 2002].
6.2. Phase 10–6 Ma: Inner Wedge Growth
 During this time (Figure 14b), the amount of net convergence was about 40 km and the amount of retreat about 100 km [Faccenna et al., 2004]. The total subduction, about 140 km, was accommodated in the inner wedge, and the accretionary complex showed forward propagation with frontal accretion of the sedimentary sequences at its toe. The subduction was yet again efficient and the prism was attaining a new growth phase. Although the age of the single thrust fault and of the growth deposits cannot be inferred from the seismic lines, deformation was probably accommodated along several thrust faults.
6.3. Phase 6–5 Ma: Messinian Salinity Crisis
 The Messinian salinity crisis (Figure 14c) represents a crucial stage in the evolution of the Calabrian wedge. On one hand, the sea level drop related to this catastrophic climatic event could have produced isostatic rebound attested to by the subsequent, immediate discharge of a clastic sequence [Cavazza and De Celles, 1998]. However, it is probably more important that it resulted in the deposition of thick clastic and evaporite sequences. Messinian deposits crop out extensively in the Crotone basin region, where they reach thicknesses on the order of 500 m. In the foreland basin, the top of the ME sequence has been directly drilled, and the original total thickness is thought to be about 1 km. The signal of the Messinian salinity crisis is well marked over the whole Mediterranean area, either as an erosional or depositional surface. On shore and in the nearby Crotone basin soon after or just at the end of the Messinian salinity crisis, it is possible to recognize a renewal of the compressional episode with the presence of an intra-Messinian unconformity [Roveri et al., 1992; Cavazza and De Celles, 1998; Barone et al., 2008]. In our preferred model, the deposition of a thick evaporitic sequence in the abyssal plain could have also produced the upward and forward propagation of the wedge along a basal décollement within the weak salt layers of the outer wedge. This could have produced an overall lowering of the wedge profile due to fast forward growth for more than 100 km of the wedge.
6.4. Phase 5–3 Ma: Outer Wedge Growth
 During this phase (Figure 14d), the amount of subduction can be estimated at on the order of 190 km, including 10 km of convergence and about 180 km of rollback, related to the opening of the Vavilov back-arc basin. After the Messinian salinity crisis, the foreland propagation of the accretionary complex developed above a weak salt décollement layer localized at the base of the Messinian deposits. This implies a fast frontal propagation of the wedge above a weak salt décollement and progressive deformation of units ME and PQ, producing the wedge shape of the outer prism, characterized by internal shortening and vertical thickening and folding of unit PQ. The fast outward propagation reduced the wedge taper, lowering the topographic slope and increasing the total length of the wedge, for more than 200 km. This period of time is marked by the out-of-sequence reactivation of the inner portion of the wedge, accompanied by uplift of the belt, as attested, onshore Calabria, by the progressive onlapping unconformity of the Messinian-Pliocene unit onto the basement. At the same time, we register a strong subsidence phase both in the Paola (Tyrrhenian side) and in the Crotone-Spartivento (Ionian side) basins. The upward propagation of the basal décollement within the Messinian deposits produced underplating of the Ionian sedimentary cover, probably stacked below the inner portion of the wedge.
6.5. Phase 3 Ma to Today: Present-Day Setting
 This stage (Figure 14e) was also dominated by a localized and very fast episode of subduction and trench rollback, about 100 km, leading to the opening of the Marsili basin. During this time, the length of the slab was quickly reduced [Faccenna et al., 2005; Chiarabba et al., 2008]. The wedge evolution maintained the same characteristics of the earlier stage, characterized by out-of-sequence thrusting in the inner portion of the wedge, thickening of the outer wedge, uplift of the belt and subsidence localized in the Paola and Crotone-Spartivento basin. The ongoing stage of underplating could have reshaped the topography of the wedge. In this sense, the origin of the Crotone-Spartivento slope could be related to the more advanced slices of underplated unit.
 Overall, this model accounts for the total subduction of the last 4 Myr. Two hundred kilometers of internal shortening are considered necessary to restore the outer wedge to its original thickness, while the remaining 100 km were possibly accommodated in the inner wedge by the reworking of previous structures.
7. Trench Rollback and Crustal Thickening: Implications for the Style of Mountain Building
 The evolutionary scenario presented in Figure 14 predicts that the upward propagation of the basal décollement and the localization of the deformation in the outer wedge should have produced large-scale underplating of the Ionian upper crustal material. We estimate that more than 300 km of material should be underplated. Part of this material could have been directly eroded and subducted inside the subduction channel, which we expect was particularly open during the trench rollback episodes, but the rest should have been accreted at depth, probably forming an antiformal stack of upper crustal slices. The onset of deformation and uplift of the Calabrian belt could be interpreted as the surface expression of this process. The position of the Paola and Crotone Plio-Quaternary basins confirms this hypothesis, as it reveals that the Pliocene uplift of Calabria was accompanied by out-of-sequence thrusting and back thrusting. The origin of the Crotone-Spartivento slope morphological scarp, in the inner accretionary wedge, could also be related to a slice of crust that was underplated in a more advanced position with respect to the others already stacked in the hinge zone. Deep reflectors in seismic profiles [Streamers Profiles Working Groups et al., 1996] can be interpreted as the roof and sole thrusts of a crustal duplex. Receiver function and tomographic analyses [Neri et al., 2009; Di Stefano et al., 2009; Piana-Agostinetti and Amato, 2009] reveal that the crust beneath Calabria is up to 40 km thick in correspondence to the hinge zone (Figure 14e). This value is indeed elevated for a fore-arc slice on top of a retreating oceanic-style subduction zone.
 The evidence for underplating in the Calabrian accretionary wedge provides new insights into the style of mountain building. The Calabrian accretionary complex was generated during the rollback of the Ionian subducting slab, without substantial convergence. From a mechanical point of view, Calabrian accretion occurred without the presence of a backstop, as the fore-arc region is largely under extension flanking a back-arc spreading center. Therefore, crustal thickening can happen only if slices of subducted material are progressively incorporated and stacked in the wedge either at its toe or underplated at depth. The latter process can be activated if a roof and basal décollement isolate a slice of incoming crustal material from the preformed upper portion of the wedge and from its basement, respectively [e.g., Kukowski et al., 2002]. In Calabria, this mechanism finally led to the formation of a back thrust bounding the Paola basin [Barone et al., 1982] in a manner similar to that simulated by Fuller et al. . However, this newly formed structure does not reflect the formation of an indenting backstop, as classically imagined during collisional orogens [Schmid et al., 1996], but only accommodated uplift of the fore-arc during deep underplating. We speculate that the case of Calabria could be extended to other Mediterranean-type orogens [Brun and Faccenna, 2008], where convergence is almost negligible and where the paleogeographic setting generated favorable conditions for underplating. This process can then reconcile the apparent paradox of crustal thickening in nonconvergent settings.
 Our interpretation of a large data set of seismic reflection profiles in the Ionian offshore sheds new light on the structure and evolution of the Calabrian accretionary wedge, a key area for the geodynamics of the Mediterranean compressive margin. These new data allow us to divide the evolution of the wedge into two main phases: pre-Messinian and post-Messinian. The Messinian salinity crisis represents a crucial moment for the evolution of this accretionary complex. The pre-Messinian stage of wedge growth is characterized by frontal accretion. The post-Messinian stage, on the other hand, is dominated by fast forward propagation of the outer wedge above the salt décollement and consequential underplating of the Ionian subducting crust below the inner portion of the wedge. This underplating phase has a surface expression in the morphology of the accretionary complex.
 We thank TOTAL (Jean Loup Ligier, Jean-Claude Ringenbach, Remy Martin, and Paul Gibson) for promoting this work within a scientific collaboration with the University of Roma Tre. We would like to thank Fugro for giving us the permission to publish the seismic lines presented here. This paper resulted from the Ph.D. work of Liliana Minelli, under the supervision of Claudio Faccenna and Piero Casero. We thank Jean Françoise Roure, Alina Polonia, Alberto Malinverno, Nano Seeber, and Michael Steckler for interesting discussions. Financial support for this study was derived from the MURST program (Evolution of the Calabrian Arc, C. Faccenna).