Deep-water sand-fairway mapping as a tool for tectonic restoration: decoding Miocene central Mediterranean palaeogeography using the Numidian turbidites of southern Italy

This paper aims to illustrate how the sedimentology of sandy turbidites can inform palaeogeographical reconstructions in tectonically complex regions. The relative positions of the plate interiors and the major continental blocks are well constrained globally, certainly for the Mesozoic to present day (e.g. Müller et al. 2016). Interpretations of the tectonic evolution of these regions are commonly illustrated on time-series of palaeogeographical maps that display transient arrangements of continental fragments, oceans and sedimentary basins (e.g. Stampfli and Borel 2002; van Hinsbergen et al. 2020). These in turn can underpin 3D models of plate interactions and associated geodynamic processes such as rates of slab roll-back (e.g. Lucente et al. 2006). They are also used to erect models of past oceanographic circulation and as inputs to climate models. However, constraining the positions of smaller blocks and basins within complex areas of plate convergence, such as in SE Asia (e.g. von Hagke et al. 2016), the southern Caribbean (e.g. Meschede and Frische 1998) and in the western Tethyan regions (e.g. Le Breton et al. 2017), is far less certain. Testing the variety of different palaeogeographical reconstructions, and choosing between alternatives, necessarily involves adding new data and syntheses. Here we use a case study from the central Mediterranean during the early–middle Miocene, building upon exceptional studies of sandstone provenance (e.g. Thomas et al. 2010; Fornelli et al. 2015, 2019; Critelli et al. 2017; Critelli 2018). In doing so, we bridge the scale gap between outcrop and plate configurations by integrating observations and interpretations from several field studies. New insights arise from applying concepts developed in recent years on the deep-water sedimentology, especially concerning structurally confined turbidity currents.

As examples of subaqueous gravity flows, turbidity currents seek bathymetric lows. Therefore, tracking their pathways provides powerful constraints, not only on the relative bathymetry of their substrate but also on the bathymetric relief (structure) of the pathways they follow. In this paper, we use the Numidian deep-water sandstones and associated deposits that are preserved in Sicily and the southern part of peninsular Italy to understand relationships between the southern Apennines and eastern Maghrebian orogenic belts, which now host these strata. Our aim is not only to revise palaeogeographical restorations of the central Mediterranean during the Miocene but also to provide a rationale for the general application of stratigraphic and sedimentological methods applied to deep-water deposits in the study of orogens and their associated basins.

Syntectonic turbidites have been widely used to calibrate palaeogeographical reconstructions; for example, in dating collision between India and Asia (e.g. Rowley 1996; Hu et al. 2016). These studies have used classical approaches, treating the deposits as blankets that seal tectonostratigraphic units and establish the timing of their juxtaposition. Alternatively, turbidite provenance has been used to inform the proximity between land-masses at the time of deposition and thus date impending collision (e.g. Hu et al. 2016), including in the central Mediterranean (e.g. Critelli et al. 2017). However, substantial further information can be gleaned from the sedimentology of turbidites, and deductions of the deep-water sediment processes derived from these studies. Advances in understanding turbidites have accelerated in the past decade: by deducing sediment processes, we are able to infer the character of the pathways along which the causative turbidity currents flowed. These deductions can be used to identify the location, amplitude and continuity of structures in the syndepositional seabed and in turn, inform palaeogeographical models.

Significant insights on the scale of deepwater depositional systems, their depositional architectures and relationships to evolving seabed structures have come from modern systems, using both high-resolution bathymetric maps and 3D seismic volumes. In ancient, deformed basin systems that inform palaeogeographical reconstructions of tectonically complex regions such as the central Mediterranean, seismic-scale features are only rarely preserved or recognized. Consequently, it is the outcrop-scale sedimentology that yields the critical information, equivalent to utilizing well penetrations without seismic data for investigations in modern examples. Despite this limitation, we aim to show that significant insights are still possible.

This paper first outlines the key sedimentological elements of deep-water turbidites before introducing the tectonic setting of our case study in the central Mediterranean (Miocene). We then focus on the sedimentological and stratigraphic data of the Numidian turbidites of Sicily and southern Italy, building a depositional framework that informs discussion of palaeogeographical reconstructions in the region. This case study illustrates the general approach we take and that could be applied elsewhere.

Concepts of turbidite sedimentology have been developed over many decades (e.g. Mutti 1992; Meiburg and Kneller 2010, and references therein). Our challenge is to use turbidite deposits and their inferred transport processes to deduce the morphology of their host basins. In the following section (Fig. 1) we use the terminology and approach of Southern et al. (2015), who classified the shapes of basin morphology, their controls on turbidite systems and resultant facies distributions.

Traditional understanding of turbidite systems, as typically reported in textbooks and reviews (e.g. Mutti and Ricci Lucchi 1978; Reading and Richards 1994; Stow and Mayall 2000; Pickering and Hiscott 2016; amongst many others), uses concepts largely dating from what Shanmugam (2016) termed the ‘heydays of submarine fan models’ (1970s–1980s). In these models, sediment volumes build out onto open, laterally continuous basin plains. These unconfined turbidite systems (Fig. 1a) characterize some of the largest depositional bodies on Earth. Modern examples include the mega-fans that are building into ocean basins (e.g. Niger, Indus, Ganges–Brahmaputra, Amazon). Large flows in large basins generate deposits across which lateral facies variations occur over long distances. These systems can have locally auto-confined flows, within submarine channel systems that build levees. Classical descriptions argue that outside these channels, flows are free to expand and wane and the reduction in the capacity of flows to carry grains generates simple fining-upwards beds that are characteristic of the classical so-called Bouma sequence (e.g. Bouma 1962; Bouma and Ravenne 2004; Fig. 1a). Recent work has suggested that flow transformations can generate facies changes over short distances (e.g. Kane et al. 2017). Nevertheless, unconfined systems tend to build sediment bodies that broadly fine outwards, away from the fan apex. Fine-grained facies fringe the fans with coarser sands closer to the fan apex and its distributor fan-top channels (Fig. 1d).

A contrasting scenario exists where turbidity currents are restricted laterally by confining slopes so that they flow along structurally controlled corridors. These are confined systems (in the sense of Southern et al. 2015: Fig. 1b and c). It should be noted that the distinction between unconfined and confined turbidites relates the size of the causative flow to the size of the basin into which they flowed. It is probable that many ancient outcropping turbidite systems that have been studied, certainly in syn-orogenic and other active basin settings, are confined by the architecture of their host basin. Southern et al. (2015) divided confined systems into contained (ponded) or uncontained (Fig. 1b and c).

Mutti et al. (2009, p. 305) argued that, until their review, ‘studies have shown the importance of structurally induced submarine topography in controlling facies distribution patterns’, noting that investigating these interactions would require ‘close cooperation between stratigraphers, sedimentologists and structural geologists’. Although, at their time of writing, few such multidisciplinary studies had been attempted (indeed Mutti et al. were sceptical that such co-operation would ever happen), there has since been substantial work on confined turbidites. Much of this effort has been directed at forecasting sandstone distribution as possible hydrocarbon reservoirs in the subsurface, by using combinations of scaled analogue and numerical experiments (e.g. Albertão et al. 2015; de Leeuw et al. 2018), observations and measurements from active natural systems (e.g. Gamberi and Rovere 2011; Stevenson et al. 2013) and studies of outcrops of ancient deposits (e.g. Southern et al. 2015; Liu et al. 2018). In uncontained confined systems, turbidity currents tend to flush down sinuous corridors (Fig. 1e). In this regard, flows tend to overrun significant parts of their confining conduit without leaving deposits, a process generally termed flow bypass.

Building on pioneering studies such as that by Kneller and McCaffrey (2003), Stevenson et al. (2015) described the critical sedimentological observations needed to establish flow bypass, the transit of turbidity currents after the partial deposition of some of its sediment content. These include abrupt grain-size breaks in vertical sections that imply flows only dropping coarse parts of their sediment load, with the remaining finer grained fractions continuing down-system (Fig. 1c). Other deposit characteristics include coarse-grained lags of granule- and pebble-grade clasts at the base of beds. These form by the reworking of bed-loads by multiple flow-events without being buried by fallout from the overriding suspension cloud. Isolated pebbles in sandstones are here interpreted as clasts that were stripped out from lags, entrained as saltating outsized grains towards the base of the turbidity current (including within plugs of coarse clasts constituting traction carpets, in the sense of Mutti 1992; Sohn 1997) and then dropped out downstream onto aggrading sand left by a weakly waning flow. These features are generally common in confined turbidites where individual flows can experience complex velocity variations and interaction with submarine structures (created by not only submarine channels but also the margins of basins). The various combinations of different spatial and temporal accelerations produce markedly different vertical and lateral variations in the resulting turbidite deposit.

Experiments, with reference to channel–levee complexes (de Leeuw et al. 2018), show that confined turbulent flows develop vertical fractionation of grain sizes. The lower part of the flow is represented by a fast-moving, high-concentration component with increasingly dilute, finer-grained and slower-moving components above. Deposition from these flow components creates coarser, sand tracts in the channel base and builds finer-grained levees on the flanks. Levees aggrade by flows overtopping them and waning away onto the unconfined slopes beyond. For flows that are fully confined by structured bathymetry, the grain sizes equivalent to levees will accumulate up the confining slopes, or be flushed through the system. The coarse sand components will accumulate along the axis of the conduit. Finer-grained deposits, falling out from higher in the turbidity current, tend to accumulate higher on the flanks of the confining bathymetry (Fig. 1f). The result is that turbidite facies can vary over short distances (hundreds of metres) when compared laterally, across the flow direction.

Turbidite sand fairways are the geological record of preferential pathways of confined flows. Their deposits are characterized by clean, coarse-grained sandstones (e.g. Joseph and Lomas 2004) with common parallel lamination indicative of aggradation as the residual flow continues to pass. Abrupt bed tops with grain-size breaks are indicative of bypassed fine-grained fractions (e.g. Kneller and McCaffrey 2003). Here we consider the grain size of sandstones containing primary bed forms such as parallel lamination to be representative of the maximum available grain size that could be carried by turbulent suspension at that point in the causative flow. Larger clasts were therefore presumably carried as dense bed-load and transported at a slower velocity than that of the turbulent suspension cloud, as a traction carpet (in the sense of Mutti 1992; Sohn 1997). The presence of coarser sediment (very coarse sand, granules, pebbles, etc.) in deposits far down-system presumably requires that they have been carried by multiple flows. Most critically, the occurrence of pebbles originating from the original source area in distal deposits requires the upstream regions across which the turbidity currents passed to have been confined so that their capacity to carry coarse sediment was retained. The high-concentration components of the flows should hug bathymetry, moving continuously downslope along the basin fairways. Therefore, it is the distribution of coarse sandstone fractions across a depositional system that is the most informative of the relative bathymetric variations within a basin.

Contained (ponded) systems retain the entire grain-size range carried by the turbidity current (Fig. 1b) including the mud, silt and very fine sand along with coarser fractions. Individual flows therefore leave beds with muddy and silty caps as the suspension cloud is trapped into the basin (e.g. Patacci et al. 2015, 2020). Southern et al. (2015) noted that this tendency increases the opportunities for subsequent flows to entrain mud. This can change flow dynamics, increasing the propensity of forming so-called hybrid beds (Haughton et al. 2009; Baas et al. 2011) where clean sands, deposited from turbulent suspension clouds, are interleaved with muddy debrites that record cohesive flow mechanisms.

A challenge in understanding ancient turbidite systems is to track clast provenance. When energetic enough, turbidity currents can erode the substrate across which they transit. The entrained clasts contaminate the resultant deposit and can confuse interpretation of sediment source and basin morphology. The problem is avoided if the turbidites of interest are composed of sands rather distinct from the substrate; for example, quartz-rich turbidity currents routed and deposited upon substrates exclusively composed of carbonates. In orogenic systems, it may be that such turbidites will be the first to enter an otherwise sediment-starved deep-water basin system just as that basin is beginning to be deformed by the orogen. This is the ideal scenario in which to use turbidites to decode the basin geometry.

To illustrate how the concepts from sedimentary geology outlined above can inform tectonic studies, we use the Numidian (Miocene) turbidite system of Italy. These rocks are found through Sicily and the southern Apennines (Wezel 1970; Critelli 1999, 2018; Guerrera et al. 2005, 2012; Fig. 2), incorporated into the eastern Maghrebian chain, part of the Alpine orogeny, which was developed by convergence between the African and Eurasian continents from the late Mesozoic into the Cenozoic (e.g. Elter et al. 2003). The region is a complex collage of continental blocks and former ocean basins of various ages that are caught between the continental interiors of Africa and Europe. Although the relative motion between these two bounding continents is well established within a globally compatible plate-tectonic reference frame (e.g. Muller et al. 2016), there are many competing models for the evolution of blocks and basins between them (e.g. Meulenkamp and Sissingh 2003; Handy et al. 2010; Zarcone et al. 2010; Guerrera et al. 2012; Puglisi 2014; van Hinsbergen et al. 2020; and many more).

The early Miocene is generally considered to be a pivotal time for the regional tectonic framework of the ancestral Mediterranean, with significant reorganization of continental blocks (e.g. Corsica and Sardinia), opening of oceanic basins (e.g. Gulf of Lion, Balearic Sea) within the broad area of plate convergence between Africa and Europe, and the initiation of the modern Apennine chain (e.g. Lucente et al. 2006; Le Breton et al. 2017; and references therein). Different models imply different linkages of ocean basins through the region, a key element of which is the westward extent of oceanic lithosphere now represented by the floor of the deep basin holding the Ionian Sea (Speranza et al. 2012). Many palaeogeographical models consider this to be a remnant of a vestigial arm of the Tethyan ocean that once entirely separated north Africa from Apulia and various other micro-continental blocks (e.g. Fig. 2a, modified after Thomas et al. 2010; and references therein). For Guerrera et al. (2012; Fig. 2b), this inferred corridor became occluded by convergence between a composite continental block (the so-called Mid-Mediterranean ‘microplate’) and the north African margin, with detritus shed from these blocks forming the fill to a ‘flysch basin’.

Turbidites of the Numidian system, part of the flysch basin fill, are composed of generally well-sorted medium to coarse, very mature quartz sandstones. Historically, their provenance has been contested (see Parize et al. 1986; Thomas et al. 2010). However, petrological studies, including zircon compositions, now clearly indicate that they have been derived from cratonic northern Africa (Fornelli et al. 2015, 2019; Critelli et al. 2017; Critelli 2018). As such, the Numidian system is a prime example of a craton-derived sand system, and its deposits are found widely around the margins of the SW Mediterranean.

For Guerrera et al. (2012), the Numidian turbidites were fed by a variety of fans that were small compared with the size of the basin (Fig. 2b). For Critelli et al. (2017) and Fornelli et al. (2019), the Numidian sandstones were deposited in elongate strips (fairways) fed from a narrow entry point (broadly along what is now northern Tunisia; Fig. 2c). In both scenarios illustrated in Figure 2b and c, the ocean basin also received sediment from the fledgling orogen to the north. Neither provides explanations for how deposits from these systems remain distinct from each other, an issue we address below.

The Numidian deposits have been swept up within tectonic units of the Maghrebian chain and southern Apennines and carried onto the orogenic forelands of Apulia and the Hyblean plateau of SE mainland Italy and Sicily respectively (Fig. 3a). These translations involved substantial tectonic rotations, of up to c. 100° (e.g. Speranza et al. 2003; Monaco and De Guidi 2006; Barreca and Monaco 2013). The Sicilian outcrops experienced clockwise rotations whereas those in the southern Apennines have experienced an anti-clockwise rotation (Fig. 3b), essentially corresponding to ‘double saloon doors’ (in the sense of Speranza et al. 2003; Martin 2006). Restoring the Numidian outcrops by applying counter-rotations reveals that they define a broad SSW–NNE-trending tract (Fig. 3c), similar to part of that proposed by Critelli et al. (2017; Fig. 2c).

In the eastern Maghrebian and southern Apennine orogens described here, the Numidian system was deposited upon Mesozoic and early Cenozoic substrata exclusively comprising carbonates, marls and mud-rocks. These attributes, as introduced above, make the Numidian sandstones ideally suited for use as tectonic tracers: we can deduce that all quartz clasts must have transited the basin system between their source area and their ultimate site of deposition without contamination from other clastic sources. Therefore, we can use the deposits to infer flow processes and the nature of the pathways down which the causative flows were routed.

The most widely studied parts of Numidian system in the Central Mediterranean are those in Sicily where they are incorporated into the thrust belt of the eastern Maghrebian chain (Fig. 4). The thrust belt is largely composed of Mesozoic and Paleogene carbonates, mudstones and marls generally inferred to have formed parts of the rifted continental margin of Africa into the Tethyan ocean. These pre-orogenic strata, apparently from distinct palaeogeographical domains from the margin, are now stacked in a series of thrust sheets that juxtapose Mesozoic platform and basin systems (Butler et al. 2019, and references therein). In most reconstructions these juxtapositions are assumed to reflect displacements during thrust sheet emplacement (e.g. Guerrera et al. 2012). However, deep-water Paleogene strata (the Argille Varicolori Formation; e.g. Ogniben 1960) locally overlie all the various Mesozoic palaeogeographical domains, indicating a period of restructuring before the Maghrebian thrust systems developed (Butler et al. 2019). The Numidian successions were deposited upon these complex substrates and represent the first influx of quartz sand into this part of the Mediterranean basin for over 200 myr. Regional palaeoflow was from west to east, in the modern reference frame (Pinter et al. 2016, 2018).

The lower to middle Miocene strata show significant variations across Sicily, illustrated in a series of schematic composite stratigraphic columns (Fig. 5a) and tied in a chronostratigraphic chart (Fig. 5b). Until recently, the age of the Numidian succession has been significantly misinterpreted, largely because of the inclusion in published fossil assemblages of microfauna reworked from older strata (a common source of contamination in turbidites) and the lack of relative stratigraphic control of sample sites. To correct this, in our precursor studies we collected microfossil assemblages from logged sections where simple stratigraphic superimposition provided tests of relative age. Microfossil assemblages were screened for reworked material. The resultant age patterns (Fig. 5b) are reproducible and resolve the Numidian of the Nebrodi basin of northern Sicily (Fig. 4) to be Aquitanian to late Burdigalian in age (Pinter et al. 2016). In central–east Sicily, the Numidian is slightly younger with ages of up to late Langhian (Pinter et al. 2018).

In the Nebrodi Mountains, northern Sicily (Fig. 4), the Numidian successions (Pinter et al. 2016) chiefly lie unconformably upon Cretaceous platform carbonates of the Panormide palaeo-tectonic domain (e.g. Dewever et al. 2010; Fig. 5a, column A). The Numidian is capped by a distinctive series of turbidites, termed the Reitano Flysch (Grasso et al. 1999). Outliers of Reitano Flysch unconformably overlie Numidian turbidites that are folded into their substrate of Argille Varicolori Formation (Fig. 5a, column B). These Numidian rocks include Langhian fauna (Pinter et al. 2018), as does the Reitano Flysch above (de Capoa et al. 2000, 2004), indicating that significant deformation and erosion happened during this short stage.

In central Sicily (column B in Fig. 5b), many parts of the Numidian succession build up from brown claystones, which we infer to represent deposition from turbidity currents that carried their coarser grain-size fractions elsewhere (Pinter et al. 2018). The first influx of medium to coarse sand (Burdigalian) represents switching of the main pathways of turbidity currents into these parts of the basin. In central Sicily, higher in parts of the Numidian successions the units are capped by siliceous marlstones, dated as Langhian (Pinter et al. 2018). These deposits we interpret as recording the deviation of the main turbidity currents to other parts of the basin, leaving the marlstone areas relatively sediment-starved.

In eastern Sicily, around the thrust culmination of Monte Judica (MJ in Fig. 4), Burdigalian strata are characterized by thick series of brown claystones, inferred to be part of the Numidian succession (Fig. 5a, columns C, D and E; e.g. Lentini & Carbone 2014), presumably deposited off the main conduits for sand flux. Here, significant sand bodies are restricted to Langhian-aged parts of the series (Pinter et al. 2018). They pass up into further mudstone and thin-bedded fine sandstones of Serravallian to Tortonian age.

The frontal (currently southern) parts of the thrust belt, represented by the so-called Gela Nappe (location shown in Fig. 4), contain a rhythmically banded marl with hard-grounds, the Licata Formation (Fig. 5, column F; Grasso et al. 1997). This deep-water unit essentially charts very limited deposition and is interpreted to have lain laterally far from and above the routes followed by the Numidian turbidity currents. It lies on multi-coloured mud-rocks (the Argille Varicolori Formation) of Oligocene and older age. Any lateral transition between the Licata Formation and the brown clays is unrecognized, potentially hidden by subsequent thrusting. The foreland area of the Hyblean Plateau saw shallow-water carbonate deposition (Pedley 1981; Fig. 5a, column G).

Northern Sicily contains a variety of Miocene turbidites, including the Capo d'Orlando ‘flysch’ (Bonardi et al. 1980). These lie unconformably upon Variscan metamorphic basement and metamorphosed Mesozoic cover, generally inferred to represent part of the Calabrian orogenic belt (e.g. Lentini 1982), preserved within the Peloritani Mountains of Sicily (Fig. 4). The Capo d'Orlando turbidite succession includes pebbles of crystalline basement but is generally characterized by thick beds of medium-grained immature sandstone with lithic fragments and detrital mica derived from the Calabrian basement rocks. The Capo d'Orlando turbidites are locally unconformably overlain by late Burdigalian to early Langhian ‘Calcareniti di Floresta’ (Aldega et al. 2011). These carbonate sandstones are reworked from earlier carbonate formations preserved within the Peloritani Mountains. Collectively the Floresta and Capo d'Orlando units are broadly the syntectonic cover to the Calabrian thrust sheets, deposited in thrust-top basins.

Calabria-derived turbidites that overlie the Numidian system or its immediate substrate are termed ‘Reitano flysch’; Fig. 5a, A and B; Grasso et al. 1999). These unconformable units include thin medium- to fine-grained volcaniclastic sandstones (e.g. Balogh et al. 2001) generally referred to as the ‘Troina–Tusa flysch’ (‘Tufiti di Tusa’ of Ogniben 1960). De Capoa et al. (2002) showed that at least some material was derived from Miocene volcanic rocks similar to those in Sardinia (essentially part of the Calabrian orogen).

The Calabrian-derived flows (feeding the Troina–Tusa, Reitano and Capo d'Orlando turbidites) and those from north Africa (feeding the Numidian turbidites) remained distinct. That these systems do not mix indicates that the broad basin area between the fledgling Calabrian orogen and the foreland was structured. At least some of this structuring must have happened during the Langhian (Fig. 5b) so that Numidian sand deposition could continue in the Mont Judica area (Fig. 5a, column C) whereas Reitano turbidites were restricted to the north (Fig. 5a, columns A and B). Coeval strata now preserved in thrust sheets to the south in Sicily are dominated by mudrocks and marlstones (and carbonates on the foreland) indicating that the sandy turbidity currents did not transit these areas of the basin. There are no purely African-derived sandstones younger than Langhian that have been identified in Sicily. Where preserved, the Numidian successions pass up into mudstones. Younger sandstones in Sicily (Serravallian and younger) exclusively rework previous deposits from the thrust wedge or are derived from the Calabrian orogen.

Several existing facies models have been erected for the Numidian that use variations in the proportion of sand to silt/mud, thicknesses and stacking pattern of beds, and these schemes have in turn been tied to models of unconfined turbidite fans (e.g. Guerrera et al. 2012). This approach has assumed that different facies were deposited in distinctly separated parts of submarine fans and their current proximity has resulted from tectonic juxtaposition of later, far-travelled thrust sheets (e.g. Bianchi et al. 1989). Thus, Numidian stratigraphy has directly influenced cross-section-scale interpretations of structural geometry (Butler et al. 2019). However, recent stratigraphic and sedimentological field research on the Numidian of Sicily, allied to geological mapping (Pinter et al. 2016, 2018), has revealed continuous, transitional stratigraphic sections and lateral connectivity between different facies previously interpreted as having been far removed. There are mappable lateral pinch-outs of sandstone bed-sets and onlap onto substrate. These short-range (3–5 km) facies changes and substrate relationships are not consistent with unconfined submarine fan models.

Full detailed descriptions and interpretations have been provided by Pinter et al. (2016, 2018) and only a brief summary is provided here. As noted above, the Numidian turbidites in Sicily are characterized by exceptionally mature quartz sandstone (Fig. 6a). There are significant thicknesses of coarser material with quartz granules and pebbles up to 5 cm in diameter (e.g. Thomas and Bodin 2013; Fig. 6b). Classically described as structureless or massive (e.g. Johansson et al. 1998, and references therein), this outcrop character reflects the exceptional sorting and locally extensive dewatering and local liquification of the deposits (Fig. 6c). When not dewatered, the majority of the sandstones contain parallel lamination (e.g. Fig. 6d). In some locations, the coarse facies contain clasts of fine-grained carbonates and mudstones (Fig. 6e) that can be readily correlated with substrate lithologies from the basin floor (Pinter et al. 2018). There are also rare, well-rounded clasts of Numidian sandstone. We interpret these clasts as recording erosion by the causative turbidity currents both of basin-floor substrate and of slightly older Numidian sediment.

The Numidian facies contrast markedly with the Capo d'Orlando (Fig. 6f) and Reitano turbidites (Fig. 6g), which are characterized by diverse sand compositions with lithic fragments that include metamorphic and granitic material indicative of a source from the fledgling Calabrian orogen. These too have coarser-grained components including thick conglomerates. As with the background sandstone, conglomerate clast types are highly variable (Fig. 6h).

Typically, sandstones of the Numidian system form units several tens of metres thick that can commonly be shown to be amalgamated (Fig. 7a), with individual depositional units generally between 50 and 200 cm thick. The amalgamated bed-sets show various stacking patterns. In places, especially within northern Sicily, the bed-sets combine to create units several hundred metres thick (e.g. Fig. 7b). More commonly, the sandstone bed-sets are separated by finer-grained, thinner-bedded units (Fig. 7c). Individual bed-sets can be traced for several kilometres, through continuous Numidian outcrop. The outcrop of Numidian strata includes finer-grained, more thinly bedded sandstones (Fig. 7d), siltstones and mudstones (e.g. Fig. 7e). Early studies (referenced by Guerrera et al. 2012) interpreted the various facies of the Numidian to originate from widely separated parts of unconfined submarine fans and they have been assigned to distinct stratigraphic formations and interpreted to lie in different thrust sheets. Our mapping in northern Sicily (Pinter et al. 2016) demonstrates that the amalgamated sandstones (specifically the outcrops in Fig. 7b) pass laterally into the thin-bedded facies over 3–5 km, approaching an unconformity with Mesozoic substrate. We therefore conclude that the various facies represent lateral changes in the behaviour of causative flows, with the amalgamated sandstones lying in the main flow path and the thin beds being marginal to the flow pathway.

Stratigraphic sections of Numidian sandstone vary in thickness from up to 1500 m to less than 200 m (Fig. 8). Full detailed logs have been provided by Pinter et al. (2016, 2018). Beds invariably show abrupt grain-size breaks with coarse sand grade material passing directly into very fine sand and silt. Even in thin sandstone facies (e.g. Fig. 7d), medium to fine sand grade passes abruptly up into silt. Collectively these relationships, as laid out by Stevenson et al. (2015), imply substantial sediment bypass. The main amalgamated sandstone bed-sets (e.g. Fig. 7a–c), especially where they contain granule- to cobble-sized grain suites, represent the main conduits for the causative turbidity currents. They represent parts of sand fairways that, prior to later tectonic disruption, would have formed continuous ribbons across the basin. Several locations contain thick, marly intervals that imply transient shut-downs in sand supply. This is interpreted as reflecting temporary re-routing of causative turbidity currents and clastic starvation in parts of the basin. A general migration of sand fairways from, in current orientation, north to SE across eastern Sicily is inferred from the diachroneity of thick sandstones established from the biostratigraphy of the associated mudstones (Pinter et al. 2018). Collectively, we deduce that the Numidian system in Sicily was confined but uncontained, with the causative flows directed along sinuous corridors across the basin.

The sedimentology of the Reitano turbidites is significantly different from that of the Numidian (Fig. 9a). As noted above, the Reitano is preserved in distinct stratigraphic outliers, in places unconformably overlying a deformed substrate of Numidian turbidites and its own substrate of Argille Varicolori Formation (e.g. Fig. 9b). As a system (Cassola et al. 1992, 1995), it too shows significant variations in grain size, with coarse fractions up to cobbles (20–30 cm diameter clasts). Sandstone beds can be amalgamated into thick bed-sets (Fig. 9a). However, where bed tops are not eroded by younger units they commonly grade upwards from coarse sand through finer sand grades into silty bed caps. Generally, the sandstones are dewatered but primary parallel lamination is common. Finer sand-fractions towards bed tops commonly show convolute lamination. Away from amalgamated bed-sets, individual sandstone beds commonly have with mud caps locally attaining thicknesses of several metres (Fig. 9c). ‘Sandwich units’, where individual beds have interiors of muddy debrites and remobilized laminated sands contained between bed tops and bases of well-sorted sandstone, are common (e.g. Fig. 9d). These hybrid beds (in the sense of Haughton et al. 2009) are common constituents of ponded turbidites, as might be expected in confined–contained systems (Southern et al. 2015). The presence of thick mud-caps and the associated propensity for hybrid beds is consistent with our deduction that the causative flows of the Reitano turbidites were ponded within mini-basins developed above a deforming thrust wedge that had incorporated earlier Numidian turbidites.

The sedimentology and map geometry of Numidian deposits in Sicily indicate that they are part of a confined but uncontained turbidite system. The present-day outcrop has been strongly modified by later thrusting but the overall disposition of units from north to south across the island has not been (Butler et al. 2019). Pinter et al. (2016) showed that the Numidian sand system in Sicily was deposited in structurally confined conduits, apparently controlled by embryonic thrust structures that deformed the basin floor. The development of thick sandy bed-sets without significant incision is consistent with these being deposited from confined–uncontained turbidity currents (e.g. Liu et al. 2018). These sand fairways are locally controlled by active thrusting. The chronostratigraphy of the Numidian system (Fig. 5b) shows a southward migration of the principal sand fairways through time (from Burdigalian to Langhian) and the northern parts of the system are overlain by the orogen-derived Reitano turbidites. That the sedimentology of the Reitano is consistent with deposition by confined–contained flows indicates that thrust-top basin morphology remained during subsequent deformation. Active thrusting therefore served to keep the orogen-derived (Reitano) and African-derived (Numidian) sand systems distinct at least within the preserved outcrops of Sicily. The effect of active thrusting was to confine the causative turbidity currents of the Numidian so that suspension clouds could carry medium to lower coarse sand grains through the conduits and, by entraining bed-load, deliver pebbles and granules to the furthest parts of the system exposed in eastern Sicily. The Numidian of western Sicily (back upstream) remained a deep-water deposit. Any coastline in the early Miocene must have lain yet further upstream. The present minimum separation between the downstream Numidian outcrops of eastern Sicily and the closest possible coastline exceeds 200 km. Therefore, we deduce that the causative turbidity currents flowed at least this far into the ancestral Mediterranean. However, more outcrops of Numidian strata lie in the southern Apennines (Fig. 3c) and these provide further insights on the extent of the system and the efficiency of its causative flows.

The outcrops of Numidian in the southern Apennines are found in the eastern part of the Campania–Basilicata region, where they occur as a series of ridges from Monteverde to Valsinni (Fig. 3c). That these sandstones form part of the same depositional system as found in Sicily is confirmed by the African sand provenance (e.g. Fornelli et al. 2015, 2019; Critelli 2018). The Numidian in the southern Apennines lies upon a Cretaceous–Oligocene succession of mudstones and marlstones with thin interbedded carbonates, the so-called Flysch Rosso (e.g. Zuppetta et al. 2004). Clay chemistry indicates a weathered Archean basement source, presumably the African craton (Mongelli 2004). This succession is part of the Lagonegro–Molise basin and it is time- and facies-equivalent to the basinal mudstones of the Argille Varicolori Formation described for the central–east Sicilian Numidian basin. Collectively the Numidian and its Mesozoic substrate evolved into a major thrust sheet, the Lagonegro allochthon, that was emplaced onto the Apulian foreland by counter-clockwise rotational overthrusting, chiefly in the late Miocene and Plio-Quaternary (e.g. Mazzoli et al. 2006). These results concord with the Numidian sandstones of the southern Apennines having been deposited during the Langhian (Patacca et al. 1992; D'Errico et al. 2014; Critelli et al. 2017). These age constraints are supported by the age of successor deposits in southern Italy (e.g. Zuppetta et al. 2004; Critelli 2018).

The data collected from the Numidian sections of the southern Apennines comprise three sedimentary logs, which represent key parts of the system. We use these data together with sedimentary observations to evaluate facies and then depositional processes. The selected sections are described from the northernmost part of outcropping succession (Monteverde) to the southernmost part (Valsinni section).

The Numidian succession in Monteverde (location X in Fig. 3c) chiefly comprises a thick interval of fine-grained sandstones and siltstones at the base that coarsens upwards. This locality has proven important for establishing a north African provenance for the Numidian of the southern Apennines (Fornelli et al. 2015). Notwithstanding rather sparse outcrop, a stratigraphic thickness is estimated to exceed 100 m.

The lower parts of the Monteverde section are characterized by medium- to thick-bedded sandstones of 0.1–1 m thick, composed by clean quartz grains of medium to fine grain sizes (Fornelli et al. 2015). The sandstones are generally ungraded or weakly graded and present a typical grain-size break in the top contact (in general, from medium sands to silts). The basal contact is generally concordant or slightly erosive (c. 1 cm). The sandstones are interbedded with finer-grained intervals of siltstones, with rare thin-bedded sandstones of maximum 2 cm thickness. The siltstone intervals are laminated or massive, with common debritic aspect. Rare quartz granules are found encased in the debritic intervals. These interbedded fine-grained intervals are concordant with the sandstone bedding, which overall forms a tabular geometry for the deposits.

The stratigraphic top of the Monteverde section and passage to younger formations is unknown. The upper part of the preserved section is characterized by more amalgamated coarse sandstones with pebbly intervals. The short section, at the Elephant House in Monteverde town (Fig. 10a), is representative of this facies. Amalgamated bed-sets form a 7 m thick package of sandstone. Depositional banding defined by grain-size variations occurs in this amalgamated package and individual sandstone beds are a maximum of 1 m thick. The sandstones are poorly graded to ungraded, composed of well-sorted, very coarse to granular sands. Although beds appear to be massive and unstructured, careful observation reveals weak parallel lamination and banding (Fig. 10b), which is otherwise obscured by diffuse pipes, dish and disaggregation textures indicative of dewatering processes. Although the sands are otherwise well sorted, there are many outsized granules and isolated pebbles (up to 2 cm in diameter) dispersed through the sandstone (e.g. Fig. 10c). These coarse fractions are also present as lags at the base of the sandstone beds, with shallow erosional relief of a maximum of 5 cm. Otherwise, bed bases are flat.

The Numidian succession in the Accettura area (location Y in Fig. 3c; Selli 1962; Boenzi et al. 1968) is steeply dipping, with a stratigraphic thickness exceeding 450 m (D'Errico et al. 2014), and generally coarsens upwards. A representative log for the upper part of the section is provided here (Fig. 11a). It is characterized by alternations of thick-bedded sandstones and thin-bedded fine-grained sandstones and siltstones. The lower part of the section is characterized by amalgamated bed-sets of sandstones up to 10 m thick, with individual sandstone beds of at most 2 m thickness. Further amalgamation of stacked sandstone beds characterizes the preserved top of the section (e.g. Fig. 11b). The sandstones are ungraded or slightly graded, well sorted and medium grained, with parallel lamination and rare convolute lamination towards bed tops. However, most bed tops are sharp, with distinct grain-size breaks from medium sand passing abruptly into siltstones (e.g. Fig. 11c). These sandstones are characterized by weakly normally graded coarse sands with granules and small pebbles on bed bases. Outsized quartz granules are distributed through beds that otherwise show parallel lamination defined by weak alignment of coarser grains. Bed bases are slightly erosive (c. 5 cm). The thick sandstone beds are interbedded with thin-bedded fine to medium sandstones with siltstones up to 1 m thick. The thin-bedded sandstones are ungraded or slightly graded medium to fine grained with tabular, flat bed bases and tops. Where exposed, the siltstones are grey, laminated or massive.

The Numidian of the Salandrella section passes upwards into the arenaceous–calcareous deposits of the Serra Palazzo Formation. The Serra Palazzo sandstones comprise texturally and compositional diverse clasts (metamorphic, granite together with quartz, including angular pebble-grade material). They are interpreted to represent the first significant input of coarse clastic material into this part of the basin from the developing orogen.

The Numidian succession at Colobraro in Valsinni (location Z in Fig. 3c; Lentini et al. 2002; Zuppetta et al. 2004) crops out in a ridge that forms a NNW–SSE monocline structure and the section reaches 800 m in thickness (Carbone et al. 1987). The site was important for establishing a mid- to late Langhian age of Numidian sand deposition in the southern Apennines (D'Errico et al. 2014). A representative part of this succession is shown in Figure 12. It is characterized by amalgamated bed-sets of sandstones of up to 15 m thick, with individual sandstone beds of a maximum of 1 m thick separated by thin intervals of fine-grained facies (a few centimetres thick). The sandstones are generally composed of ungraded well-sorted medium sand, with coarse sand intervals restricted to bed bases and lags. Crude parallel lamination is observable in most beds, but is otherwise obscured by dewatering pipes. The fine-grained facies are characterized by thin-bedded sandstones of a maximum of 3–4 cm thick, interbedded with laminated siltstones. The thin beds are ungraded medium-grained sandstones with rare convolute lamination.

The Numidian of the Colobraro–Valsinni section passes upwards into a thick sequence composed of marlstones and arkosic sandstones called the Serra Cortina Formation (Lentini et al. 2002). We interpret these strata as being derived from the fledging Calabrian–Apennine orogen, essentially equivalent to the Serra Palazzo Formation in the Salandrella section. However, these orogen-derived successions need not be fed from the same submarine fan system. Understanding their depositional systems alongside the tectonic controls would be an interesting study but is one that lies beyond the scope of our paper.

The Albidona flysch includes upper Burdigalian to Langhian turbidites (Selli 1962; Cesarano et al. 2002) and so is broadly time-equivalent to the Numidian. It was sourced principally from the Calabrian arc (Cavuoto et al. 2004, 2007) and lies unconformably on the obducted Ligurian subduction–accretion complex preserved along the Tyrrhenian coast of the southern Italian mainland. The Albidona flysch changes in character up-section, from chaotic immature siliciclastic deposits at the base to turbiditic arenaceous–clayey material with thick intervals of marlstones towards the top (Finetti et al. 2005).

Further west in the orogen, the Burdigalian–Langhian turbidites of the Cilento group unconformably overlie strongly deformed and weakly metamorphosed deep-water successions ascribed to the Ligurian accretionary prism (Cammarosano et al. 2004). Cavuoto et al. (2007) described the Cilento Group as being orogen-derived. The system contains thick sandstones that are locally amalgamated. Strikingly, the Cilento system also contains rare calciturbidites with primary algal clasts and exceptionally thick (tens of metres) mud-caps. These megabeds are analogous to similar deposits in the Marnoso–Arenacea basin of the northern Apennines (e.g. the Contessa megabed; Gandolfi et al. 1983) and may have a similar provenance (a proto-Abruzzo carbonate platform in what is now the central Apennines). Their thick mud-caps indicate that these carbonate-rich flows were entirely contained (Fig. 1b), and therefore that the Cilento basin was isolated from the basin system that hosted the Numidian system.

Both the Albidona and the Cilento turbidites may represent a structurally equivalent unit to the Reitano and Capo d'Orlando turbidites of Sicily. As with the Sicilian examples, the Albidona–Cilento and Numidian systems do not appear to have mixed. Therefore, a similar explanation is proposed. The orogen-derived turbidites of the southern Apennines formed systems ponded in thrust-top basins. Their causative turbidity currents were contained by these basins and did not contaminate the Numidian sand fairway.

As noted above, in the southern Apennines, the Numidian is overlain by the Serra Cortina and Serra Palazzo formations, together with further orogen-derived sandstones of the Gorgoglione Formation (Critelli et al. 2017; and references therein). This indicates that by the Serravallian the Numidian sand system was not able to reach the southern Apennines.

Finally, mature quartz sandstones forming the Bifurto Formation unconformably overlie parts of the shallow-water limestones of the ‘Apennine platform’ (Selli 1957). Traditionally these sandstones are considered to be a distinct unit. However, zircon compositions reported by Fornelli et al. (2019) show the Bifurto and Numidian sandstones to have the same, African provenance, uncontaminated by clasts from the fledgling Calabrian orogen. For our purposes, the Bifurto can be considered to simply be part of the Numidian sand fairway. The variations in the Mesozoic geology of its substrate are similar to those discussed above for Sicily and imply substantial restructuring of the platform and basin morphology that was initiated during the Mesozoic. Consequently, the distributions of these Mesozoic rocks are unreliable guides to Miocene palaeogeography (Butler et al. 2019).

The Numidian outcrops of the southern Apennines are characterized by thick, tabular sand-bodies that can be traced laterally for at least several kilometres, this extent being limited by outcrop quality and later deformation. Erosional features are limited to a few centimetres at the base of individual beds; there are no major incisional features into the underlying deposits. The sandstone-rich intervals are characterized by intense amalgamation of typically ungraded coarse to medium sandstones (e.g. Fig. 11b). These intervals pass abruptly into cogenetic very fine sandstones and siltstones, without showing a transition through intermediate grain sizes (medium and fine sands). In many cases, internal structures such as primary banding and parallel lamination are evident, suggesting that the beds were formed by progressive aggradation. There are no obvious associations with stacked finer sands as associated constructional levees. D'Errico et al. (2014) reported a general grain-size decrease from south to north in the southern Apennines. However, the major sand bodies are remarkably homogeneous across the region. Indeed, medium to coarse sands still represent the predominant grain sizes even in relatively distal areas, which also include pebbles (e.g. Fig. 10c from the Monteverde section; X in Fig. 3c).

Collectively the sedimentology, as for the Numidian sandstones of Sicily, suggests that the turbidity currents continued to largely bypass the basin floor. We deduce that the causative turbidity currents were strongly confined so that they maintained their sediment-carrying capacity. Although outcrop in the southern Apennines is not sufficient to demonstrate the lateral facies changes and relationships with substrate that we have been able to show in Sicily (Pinter et al. 2016, 2018), the apparent absence of significant incisional relationships or associated constructional levee facies suggests that confining bathymetry was provided by structures on the basin floor.

Through the Langhian, the Numidian sand fairway remained uncontaminated by orogen-derived sediments. Therefore, orogenic detritus (Albidona and Cilento turbidites) was ponded, presumably within enclosed thrust-top basins located higher on the westward slope of a seaway that lay east of the ancestral Calabrian mountain belt. The Numidian system lay towards the base of this seaway. However, this does not constitute a simple foredeep basin (see Guerrera et al. 2012). Rather, the basin floor was structured, presumably by fledgling thrust systems that went on to develop into the imbricate systems now found within the Lagonegro allochthon (e.g. Zuppetta et al. 2004).

The sedimentology of the Numidian sand system is not compatible with the depiction of Numidian fans as small and unconfined, with multiple input points both east and west of modern Sicily (Guerrera et al. 2012). Nor is it compatible with the single unconfined fan depiction of Thomas et al. (2010); Critelli et al. (2017; Fig. 2c) showed the Numidian system as forming elongate sand ribbons across a deep ocean basin then running up onto the Mesozoic Apulian platform (Fig. 2c). Fornelli et al. (2019) modified this model by showing the Numidian sandstone ribbons running across rift-related relict topography ahead of the Apennine–Maghrebian thrust front. We concur with Critelli et al. (2017) and Fornelli et al. (2019) that the Numidian system was fed axially, from the SW of the Sicilian thrust belt. As the foreland area, SE of modern Sicily, was essentially marine we deduce that there was no significant sediment influx from this direction. However, both of these other studies imply that turbidites lie in narrow pathways across an open marine basin (e.g. as illustrated in Fig. 2c) without discussing the basin structure necessary to generate these fairways. We now examine the relationship between the confined Numidian turbidites and the inferred basin structure.

The Numidian system conforms to models of confined turbidites, where causative turbidity currents are preferentially routed along elongate, structurally controlled conduits. This structural confinement of the causative turbidity currents was most plausibly provided by active thrust anticlines that developed within a deep-marine seaway. In our model (Fig. 13) the routing of turbidity currents effectively fractionates coarse sand and larger clasts from the finer-grained fractions. Sand was preferentially deposited along the main pathways taken by turbidity currents, forming fairways. Finer fractions, as well as being flushed through the system, accumulated on the flanks of the fairways and over-spilt into adjacent parts of the basin floor. Where fold amplification continues, sediment routing can evolve, delivering coarse sand to previously largely depleted or bypassed parts of the basin. Abandoned parts of previously active flow paths can become starved of significant detrital input. This evolution is broadly supported by the chronostratigraphy of the Numidian system in Sicily (Fig. 5b).

Not only do folds and the resultant seafloor relief control the behaviour and routing of the main Numidian turbidity currents, they can also serve to hold back detritus shed from the fledgling Calabrian orogen (Fig. 13). In the Sicilian part of our study, these orogen-derived materials include the Reitano turbidites. In the southern Apennines the Cilento turbidites, which also unconformably overlie deformation structures, represent a tectonostratigraphic unit equivalent to the Reitano of Sicily. Both systems are interpreted here to have been restricted to distinct thrust-top basins. There are no indications that significant sand components from these orogen-derived systems entered the flow pathways for the Numidian. Again, sediment type was fractionated by the structure of the basin.

As noted in our previous work in Sicily (Pinter et al. 2016, 2018), the Numidian sandstones overlie a variety of strata that originally were deposited under significantly different palaeobathymetries. Traditional accounts of Italian geology emphasize the importance of these distinct, pre-Numidian successions in defining tectonostratigraphic domains, and assume particular arrangements for their palaeogeographical disposition during Numidian deposition and in reconstructing tectonic displacements in the southern Apennines and Sicily. For example, the designation in Sicily of ‘internal Numidian’ and ‘external Numidian’ by Guerrera et al. (2012) and many others relies exclusively on characterizing their immediate substrata. As noted elsewhere, deposition of Numidian turbidites upon rocks deposited originally on the carbonate platform (the so-called Panormide units) of Sicily requires this former platform to have experienced substantial subsidence prior to Miocene times. Thus, the definition of palaeogeographical domains such as platforms and basins, defined by Mesozoic strata, is an unreliable guide to basin geometry in the Miocene (Butler et al. 2019). The same deduction arises from accounts of the southern Apennines. Not only do Numidian turbidites overlie Flysch Rosso of the Lagonegro basin but, by including the Bifurto Formation within the Numidian system, also shallow-water Mesozoic carbonates (e.g. Fornelli et al. 2019). These are generally conflated into the contiguous Alburno–Cervati–Pollino platform (e.g. Iannace et al. 2005, and references therein; otherwise termed the Apennine Platform). However, for the causative flows of these far-transported Numidian sands to reach these substrates of the southern Apennines they must have crossed the Sicilian system together with the intervening basin. The southern Apennine Numidian would have deposited in deeper water than these up-system locations. Clearly then, the Apennine ‘platform’ was not a platform during the Langhian; indeed, it lay at greater bathymetries than the Argille Varicolori of central Sicily. This implies that, notwithstanding Miocene thrusting, the Mesozoic array of platforms and basins of this part of Tethys were significantly restructured, with new palaeogeographical juxtapositions presumably at some time in the early Tertiary.

The confined nature of the Numidian turbidites can be traced across both Sicily and the southern Apennines. In neither case did the turbidity currents, as recorded by the deposits discussed here, enter an unstructured foredeep or a broader deep-marine basin. We infer therefore that thrust systems provided structural continuity between the Maghrebian system of north Africa and Sicily and the southern Apennines (Fig. 14). The Numidian turbidity currents that reached the southern Apennines must have passed along the fairways in Sicily. Therefore, bathymetry increased from SW to NE around the thrust belt. During the Langhian the strata of the southern Apennines lay under deeper water than the thrust system of Sicily.

A challenge remains in defining the required confining slope to the SE of this thrust belt, such that turbidity currents from the Numidian system did not break out into the Ionian Sea basin (hence the question mark in Fig. 14). This inferred confining feature lies within what we refer to here as the ‘Calabrian Gap’. Most existing palaeogeographical reconstructions depict the Calabrian Gap as containing an arm of the Tethyan ocean floor projecting westwards from the site of the modern Ionian Sea (e.g. Fig. 2a), separating the two orogenic foreland blocks of Hyblea and Apulia (Fig. 3b and c). Notwithstanding its widespread adoption in palaeogeographical reconstructions for the central Mediterranean, it seems unlikely that such a continuous seaway existed, certainly during the Langhian, or presumably the Numidian flows would have navigated a pathway through the structured seabed to this bathymetric low.

A variety of palaeogeographical reconstructions can satisfy the requirement for a confining slope to the SE of the Numidian sand fairway to fill the Calabrian Gap between Sicily and the southern Apennines. Le Breton et al. (2017) suggested that Hyblea and Apulia were in close proximity at 20 Ma and that the two blocks have separated as Apulia experienced a counter-clockwise rotation and convergence with the eastern side of the Adriatic Sea (Dinarides). Restoration of this displacement closes the Calabrian Gap. The confining slope to the Numidian system is provided by a near continuous platform and associated NW-facing slope. It is the subsequent rotation of Apulia and its divergence from Hyblea that opened the Calabrian Gap for the Calabria arc to migrate southeastwards into the Ionian basin. However, this model requires substantial right-lateral displacements to cut the Ionian basin and would be expected to offset the escarpment that now defines the SW edge of the Apulian platform and Adriatic Sea. No such structure has been recognized (e.g. Catalano et al. 2001). It seems most likely that Hyblea and Apulia have remained in the same relative position, at least since the Mesozoic.

Rather than displace Apulia relative to Hyblea, the Calabrian Gap may instead have been filled by the continuation of continental crust from these two blocks, much of which now lies buried and partially telescoped by the southern Apennine and Sicilian thrust systems. In this model the western limit of Ionian Tethys coincided with the modern Malta escarpment. That there has been continuity between Apulia and the Mesozoic carbonate platforms of the southern Apennines and continental north Africa is strongly indicated by the dispersal of terrestrial megafauna during the late Cretaceous (e.g. Zarcone et al. 2010).

A third alternative, and our preferred model, for filling the Calabrian Gap is to invoke a submarine thrust belt largely occluding the deep basin lying on Ionian oceanic crust (Fig. 14). The challenge with this model is to create a scenario where Numidian turbidites are entirely restricted to lie on the thrust belt. Their flows cannot have breached the thrust belt and accessed the deep ocean basin that would have lain ahead of it.

One remaining challenge facing our model for a structurally confined Numidian turbidite system lies in accounting for the volumes of finer-grained sediment that is presumed to have been carried over the coarser grain fractions through flow bypass. Earlier flows that deposited their coarse sand in Sicily could of course have carried their finer sand and mud into what is now the southern Apennines. However, the significant bypassed sediment inferred here to have been associated with the coarser fractions of the southern Apennines has no obvious down-system continuation within which it might have accumulated. One possibility is that the distal flows broke out of their thrust-top confinement to enter a true foredeep. In Sicily, for example, the more forelandward stratigraphic sections are dominated by mud and very fine sands (Fig. 5, column E). Although some of this material may represent lateral overspill from confined flows on the thrust belt, it is possible that some may also represent distal deposits from older flows.

A modern analogue for the palaeogeography proposed here for the central Mediterranean during early Miocene times lies in the SW Caribbean. Drainage from South America enters the Caribbean at the Gulf of Uraba and encounters a seabed structured by folds of the North Panama Thrust Belt (e.g. Silver et al. 1990). Thrust-top basins (e.g. the San Bias basin) are trapping at least part of the detritus being shed from the rising Panama orogen. Other examples could include the NW Arafura Sea, where the eastern part of the Banda arc impinges on western Papua, Indonesia.

The central tenets of our paper are that turbidites, as the products of subaqueous gravity flows, can map out bathymetric lows, and that their sedimentology may be used to infer the shape of the basins within which they are contained. Our work is consistent with other studies indicating that structurally confined but uncontained turbidite systems leave sand-rich fairways comprising stacked bed-sets (e.g. Liu et al. 2018; Casciano et al. 2019). These turbidite sand fairways reflect not only the connectivity between arrays of basins but also the relative elevation of parts of the basin floor. These are first-order elements that can be used to infer palaeogeography, especially charting parts of the history of vertical movement, in our case study, of the complex array of blocks and basins that have become incorporated into the Maghrebian–Apennine orogenic system. That coarse sand has been carried a long distance across the basins requires the causative turbidity currents to maintain their capacity to carry sediment, a deduction reinforced by the presence of large grains in the more distal deposits together with distinct grain-size breaks throughout (indicative of flow-stripping and bypass; Stevenson et al. 2015). This behaviour is characteristic of deposits from flows that were confined laterally. In the absence of autogenic channel–levee complexes, turbidites with these characteristics presumably reflect confinement by basin structures.

A key assumption in our analysis has been that the sand and coarse grain-size fractions within the Numidian turbidites have been derived from north Africa. This could be invalidated if the causative flows entrained sand from older successions that originally lay in their paths. However, for the Numidian system in Sicily and the southern Apennines, its substrate is represented by carbonates with their associated mudstones. There are no significant coarse siliciclastic successions from which quartz sand might have been entrained. Therefore, the Numidian sediment effectively acts as a tracer or dye, uncontaminated from its north African source, that tracks turbidity currents down-system. This is an important constraint. For our approach to be applied elsewhere, turbidite systems should be chosen that represent an early influx of siliciclastic material into an array of basins that otherwise, and previously, were starved of such sediment compositions. In orogenic systems, especially those of the western Tethys, where carbonate deposition dominated much of the Mesozoic, it is the earlier synorogenic strata that may prove the most amenable to our approach.

Coarse-grained quartz sandstones of the Numidian turbidite system (Burdigalian–Langhian in age) are found in Sicily and the southern Apennines of mainland Italy. These outcrops have been carried on tectonic allochthons and partially dismembered by rotational thrusting through the later Miocene and Pliocene. When these displacements are reconstructed, the Numidian sandstone defines a composite fairway that can be traced for over 300 km.

Both in Sicily and the southern Apennines, the Numidian sandstones have abrupt bed tops that show distinct grain-size breaks. The deposits include quartz pebbles and lags. This suggests not only that the capacity of the causative turbidity currents to carry coarse sand and bed-load was maintained for many hundreds of kilometres down-system but also that much of the flow bypassed the seabed. Flow bypass on this scale strongly suggests that the turbidity currents were confined (e.g. Stevenson et al. 2015). Our previous work in Sicily (Pinter et al. 2016, 2018) indicates that confinement was provided by active folding associated with the Maghrebian thrust belt.

We conclude that the Numidian turbidites were deposited in structurally controlled tortuous corridors, developed along synforms associated with a submarine thrust system, and that the Numidian turbidity currents flowed across an evolving tectonic allochthon (see also Butler et al. 2019). Structural evolution will have influenced the path taken by the flows, an inference consistent with the system's biostratigraphy for Sicily (Pinter et al. 2016, 2018). This shows diachronous migration of the main sand fairways through the Burdigalian and Langhian.

The Numidian turbidity currents were fed axially, derived from the SW corner of the thrust belt. The southern Apennine outcrops of Numidian sandstone show the same sedimentary characteristics as their Sicilian counterparts. Therefore, the thrust system and its bathymetric relief was continuous, from the main Maghrebian chains of northern Africa into the Apennine chain. Presumably the Langhian-aged sandstones of the southern Apennines were deposited by flows that bypassed through the sand fairways of that age in Sicily. Bathymetry increased anticlockwise around the thrust system. It is unlikely that there was deep bathymetric connection between this arcuate thrust system and the Ionian Sea basin to the SE. Certainly, by Burdigalian and Langhian times, any residual arm of the Tethyan ocean through this region had effectively closed.

The Numidian sandstones stratigraphically overlie various substrata representative of different Mesozoic palaeogeographical units (Butler et al. 2019, and references therein). For example, Numidian turbidites locally overlie Mesozoic rocks in platform facies in the southern Apennines but the causative turbidity currents from which these sandstones were deposited must have transited through the sand fairways in Sicily, and these lie on apparently deep-marine strata (Argille Varicolori Formation). This implies significant restructuring of the Mesozoic palaeogeographical framework. The Mesozoic units are not, of themselves, indicative of the palaeogeography of blocks and basins in the central Mediterranean, certainly during the Miocene and probably through the early Tertiary.

Stratigraphy has long been used to inform palaeogeographical reconstructions. The Numidian case study developed here illustrates the utility of using the sedimentology of turbidites to gain understanding of basin structure that can inform palaeogeographical models on the scale of hundreds of kilometres. Our study benefits from using strata with distinctive sediment compositions, in this case hyper-mature quartz sand derived from north Africa that entered a seaway floored by carbonates, claystones, mudstones and marls. The distinct provenance reduces possible confusion with other, orogen-derived, more locally sourced turbidites. The approach is therefore most applicable to understanding systems using the first siliciclastic inputs into otherwise sediment-restricted, and carbonate or mud-rock floored, marine basins.

Key elements of our work were presented at the conference on ‘Deep-water depositional systems: advances and applications’ held at the Geological Society in 2017. We thank participants for discussions and apologize for the tardy preparation of this work. We are also indebted to two deceased colleagues at the University of Catania who supported our research endeavours over the years and generously discussed Sicilian and southern Apennine geology on numerous occasions: Mario Grasso and Fabio Lentini. M. Patacci and N. Lentsch are thanked for insightful reviews. However, of course, the authors are solely responsible for the interpretations presented here.

The research presented here was funded by BG-Shell in partnership with CNPq-Brazil (National Council for Scientific and Technological Development). R.M. acknowledges a Piano Triennale della Ricerca 2016-2018 grant awarded from the University of Catania.

RWHB: conceptualization (lead), formal analysis (equal), funding acquisition (equal), investigation (equal), methodology (equal), project administration (lead), supervision (lead), writing – original draft (lead), writing – review & editing (equal); PRP: formal analysis (equal), investigation (equal), methodology (equal), writing – review & editing (supporting); RM: formal analysis (equal), investigation (equal), methodology (equal), writing – review & editing (equal); AJH: conceptualization (supporting), formal analysis (supporting), funding acquisition (equal), investigation (supporting), methodology (supporting), supervision (equal), writing – review & editing (supporting)

Scientific editing by Linda Kirstein