Evidence for stratigraphic changes across the northern sub-basin margin has been addressed by Hodgson and Haughton (2004). Phase 3 sheets abut directly against the El Cautivo Fault Zone (ECFZ) which juxtaposes the ponded turbidite sheets to the south against thick marls to the north. There is no evidence for significant thinning of the turbidite sheets before they are cut by the fault zone. Only a thin interval of Phase 3 sheets (30 m thick) occurs directly north of the ECFZ just beneath the transition to Phase 4 fill. These cannot be tied at a bed level to sheets within the sub-basin. The sandier Phase 4 succession steps across the fault with only minor (m scale) later displacement.

Unlike the sheet turbidites, the underlying Alfaro Member calcarenites do not extend off the crest of Alfaro hill northwards as far as the ECFZ. Nested channel surfaces step up to the north and the calcarenites pinch out approximately 1 km south of the fault trace. Calcarenitic beds are however caught up in the thick gouge zone associated with the EFCZ. Calcarenitic beds also locally cap hills on the north side of the fault zone and are interpreted to represent the erosional remnants of another Phase 2 channel system, this one filled with sand with a mixed carbonate-siliciclastic provenance. At El Cautivo (Figure 2), this channel appears to have been truncated by the ECFZ to which it is oblique at a low angle in plan view.

Interpretation: Phase 4 sandstones overstep the ECFZ yet the Phase 3 sheet turbidites are truncated by the fault, implying the fault was active during or at least towards the end of Phase 3 deposition and then became inactive or buried by an increase in the rate of sediment accumulation. The expansion of the Phase 3 stratigraphy across the fault implies either an element of growth and/or lateral displacement of a thicker against a thinner sheet turbidite succession. The Alfaro calcarenite-filled channels step towards the fault zone but are separated from it suggesting the fault did not directly control the channel location. The northward stepping of channels may reflect early tilting towards the fault. The obliquity between channel orientations and the fault trend also suggests the fault was largely post channel formation, as do the large blocks of calcarenite resembling channel sediment (some cemented) that float within the fault gouge zone. The blocks were presumably entrained where the faults propagated to displace the channel fills cutting across the fault trace.

The Phase 3 ponded sheet succession thickens northwards away from the sub-basin, reaching 100 m thick c.2 km north of the El Cautivo Fault before the sandstones pinch out laterally again into marl in the Arroyo del Verdelecho (Figure 2). The same sheets can also be traced eastwards 2.5 km in the Rambla de Reinelo where the thickest bed reaches >5 m thick. In general, however, the event beds are thinner than their equivalents in the expanded sub-basin succession and thick mudstone caps are less well developed. The character of the sheets outside the sub-basin is well illustrated by a large exposure in the Rambla de Tabernas (just upstream of its junction with the Arroyo del Verdelecho; Figure 6). Here a sandstone-mudstone couplet is characterised by a parallel-laminated sandstone with local convolute folding of the lamination upstream of a thinned part of the bed. The sandstone component of the event bed varies from 80 to 200 cm due to an irregular base as it rests on up to 10 m of remobilised marl. The bed is displaced by a series of normal listric faults orientated N115E and dipping to the south. The faults root out at the base of the muddy slumped interval but do not extend to the upper part of the outcrop to cut the next sandstone sheet. The fault planes have locally been intruded by marl.

Interpretation: Gradual expansion and then thinning of the Phase 3 sheet succession northwards away from the El Cautivo Fault Zone suggests that ponding of turbidity took place in gentle depressions up to 2 km across and > 3 km long, as well as in the deeper fault-controlled bathymetry of the sub-basin. The thickest sheets encountered outside of the sub-basin are in the Rambla de Reinelo and coincide with the hinge of a syncline. It is possible therefore that the basin floor outside of the sub-basin was warped by contemporary folding, with turbidites deposited along developing synclinal axes. Some of the sandstones directly overlie cohesive mudflow deposits suggesting that they were emplaced on mobile mud beds; turbidity currents may have either triggered local substrate flows or they may have interacted with mudflows released by the same external trigger (see below). The strongly layered nature of the sheet succession with sands encased in marls may have encouraged packet sliding off low gradient slopes, with gravity deformation continuing by brittle deformation and shallow faulting after emplacement of the sandstone and its associated mudstone cap.

The Alfaro Member calcarenites and overlying Phase 3 ponded turbidites can be followed southwards off the Alfaro ridge to the area north of Rambla de Indalecio where a series of exposures constrain the stratigraphic and structural geometry of the southern margin of the sub-basin. When traced down the dip slope to the southwest, the Alfaro Member calcarenites and basal part of the ponded sheets are overstepped by the Phase 4 basin fill. In the Barranco de los Baños, the Alfaro calcarenites are overlain by the basal part of the ponded sheet succession which is discordantly overlain by upper Phase 3 sheets. The result is that only a thin and incomplete interval (c. 50 m thick) of the Phase 3 fill is developed here (compared to >300 m to the north). Further south again, Alfaro calcarenites that floor the sub-basin to the north are progressively rotated to near vertical and juxtaposed against pale grey scaley marls by a decametre-wide vertical fault zone trending N090E (Figure 7). Ductile deformation of the adjacent, now cemented and brittle, calcarenites suggest that they were cohesive but not fully lithified at the time of deformation (Figure 7B). Mode I vuggy calcite veins occur in the fault zone and these have an average strike of N155E. Some sub-vertical dextral E-W shear planes are also developed in this fault zone and are occasionally utilised by fibrous calcite veins. When the fault is traced 400 m to the east, calcarenites flanking the fault are incorporated into the fault zone and deformed to produced folds with axes orientated N070E which were then cut by minor N120E-trending dextral faults (Figure 7C). To the south of the fault, Alfaro calcarenites are not present, and Phase 2 marls are directly overlain by Phase 4 sheet turbidites. The contact is a sharp surface, generally planar but with local steps, against which the younger succession systematically onlaps in a south-easterly direction. The onlapping succession also contains evidence for significant gravity-induced deformation directed away from the slope (discussed in more detail below). The mapped distribution of the Phase 4 turbidites implies they extend without displacement across the shaley gouge zone that truncates the Alfaro calcarenites.

Interpretation: The scaley and calcite-veined gouge resembles a thinner version of the anomalously thick gouge zone bounding the northern side of the sub-basin. It is similarly interpreted as a syn-depositional fault, the key evidence being the overstep of the younger Phase 4 stratigraphy and the Gordo megabed embedded in it across the fault trace without displacement. The southern fault zone is orientated ENE-WSW and projects eastwards into a dextral step in the basement of c. 1 km. Overall the two bounding faults are slightly oblique to one another in map view and they converge to the east. As late uplift and subsequent erosion have produced an oblique slice, the geometry is interpreted as upward splaying of the principal displacement zone with the sub-basin sitting between two active strands of the same linked fault system. Small scale internal deformation features and vein orientations indicate the fault bounding the sub-basin to the south was a dextral fault zone (like the El Cautivo Fault Zone) but in this case with a component of downthrow to the north (Figure 7D). Deformation flanking the fault zone to the south and/or longer wavelength uplift linked to the rise of the basement blocks may have tilted the muddy Phase 2 succession down to the north to provide the relatively smooth slope against which the younger Phase 4 stratigraphy laps out. The change from Phase 3 to Phase 4 turbidites coincides with overfilling of the sub-basin and a switch to deposition over a wider area that draped the then-inactive faults and onlapped an unstable confining slope to the south.

The sub-basin lying between the two fault strands is c.2 km wide and dominated by a succession of laterally extensive Phase 3 ponded turbidite sheets (Haughton, 2000; Hodgson and Haughton, 2004) up to 300 m thick (Figure 5). The younger of these sheets extend across the full width of the sub-basin. They are dominantly siliciclastic sandstones, but include at least 5 sheets with a carbonate provenance (in all cases debrite-turbidite couplets with the sand fraction dominated by bioclastic grains). A distinctive bundle of closely spaced beds c.140 m below the top of the sub-basin fill (highlighted on Figure 5) can be traced laterally between sections. Three stacked siliciclastic turbidite sheets (referred to as beds 1, 2 and 3) and an overlying calcarenite sheet (marker bed 4) have a relatively constant thickness when traced north to south across the sub-basin floor for 800 m (Figure 8B), but thin as they approach the fault corridor at the south side of the sub-basin where the calcarentic sheet pinches out completely, implying it thins laterally at a faster rate than siliciclastic sheets of similar thickness. Complete pinch out of the siliciclastic sheets cannot be demonstrated due to lack of outcrop in the main rambla bed. N-S section correlations of relatively thin sheets at a deeper stratigraphic level demonstrate subtle but systematic southward thinning of the succession, mainly involving thinning of the bioturbated sandy marls between the sheet sandstones (Figure 8A). Beneath the sheet succession, a marl interval occurs in the central sub-basin, resting on top of the Alfaro Member calcarentites. This marl is absent to the south where sheet sandstones higher in the section directly overlie the Alfaro Member. Just north of the Barranco de los Baños large calcarenite sandstone blocks (up to 70 m across) occur locally interleaved with the Phase 3 stratigraphy.

A major remobilised unit (referred to here as the Alfaro Slump; Figure 9) is intercalated with sheet turbidites within the sub-basin fill (c. 75 m below the top of Phase 3), but only in the northern part of the sub-basin. This is well exposed in high cliffs on the western margin of the Rambla de Tabernas (Figure 9A) immediately south of the El Cautivo Fault Zone. The slump is 45 m thick and is composed of chaotic sheared muddy matrix (more than 90%) with scattered deformed sandstone blocks (up to 1 m across) as well as poorly defined marl blocks which are recognised by their bedding-discordant fissility. It has a sheet-like geometry with no convergence of turbidite beds above and below. The slump has a planar base with the underlying marls and it sits stratigraphically above a distinctive calcarenite debrite-turbidite couplet (marker bed 4; see above). When followed to the north, the slump can be traced to within 10 s m of the El Cautivo Fault although recent erosion means the actual contact is no longer preserved. Sheet turbidites beneath the slump extend to and are truncated by the fault zone and a similar relationship is inferred for the slump unit. No correlative of the slump can be found in the Phase 3 succession north of the El Cautivo Fault Zone. The slump can be followed 300 m south, along the Rambla de Tabernas to a sharp bend in the river where another fault with a strike of N120E truncates both the slump and underlying turbidite beds. The fault plane is sharp with a stepped geometry; a sub-vertical segment truncates the upper part of the slump, but the fault flattens close to ramble level. Significantly, the slump cannot be identified in the exposures south of the fault (contrast with Hodgson and Haughton, 2004). The geometry and timing of fault displacements have been documented by carefully logging the succession straddling the fault (Figure 9B) and by tracing beds using detailed photomontages. Marker bed 4 is displaced vertically c. 60 m by the fault. On the downthrown side (to the north), there are two bundles of thin beds in the 25 m thick section below the slump. Only the lower of these can be identified south of the fault in a similar position relative to bed 4. The slump cannot be correlated across the fault, but the turbidites above it forming the upper half of the cliff do not appear to be affected by the fault. Indeed the bed located at 17 m on the log south of the fault oversteps the fault and corresponds to the second thick bed above the slump (at 79 m on the northern log). The topmost beds on the cliff are composed of the lowermost beds of the Phase 4 (post sub-basin) fill. The boundary is placed at a step increase in the frequency of thin beds.

There is no bedding discordance at the top of sub-basin fill in the northern part of the sub-basin. However, the transition from Phase 3 isolated ponded sandstone sheets to Phase 4 fill is reworked by a major mass transport deposit in the southern part of the sub-basin (MTD-1 described in detail below). When the sub-basin fill dips are flattened by restoring the mean Phase 4 fill dips to horizontal, the residual dip is to the north implying tilting towards the El Cautivo Fault.

Interpretation: Filling of the sub-basin took place mainly by suspension settling of muds interspersed with sporadic turbidity currents, the larger volume of which became ponded and evolved to higher concentrations as sustained flows continued to drain into what must have been a relatively small bathymetric deep. Although the margins of the sub-basin are faulted, it was probably not greatly wider than the 2 km currently preserved. The strike extent of the sub-basin is speculative as it dips beneath younger stratigraphy to the west, and its eastern extent has been removed by erosion. Evidence for a thicker succession in the northern part of the sub-basin, thinning of individual sheets as the southern margin is approached, residual dip to the north and local angular discordances in the south suggest a wedge-shaped profile with greatest subsidence in the north. Tectonically-induced facies asymmetry has been noted in other deep-water basins, notably the Peïra Cava Basin, SE France (Cunha et al., 2017) and the northern Apeninnes, Italy (Tinterri et al., 2017). If the Alfaro Member calcarentites fill channels at the base of the sub-basin as is suspected, their distribution (thicker to the south and lapping out to the north) implies initial subsidence may have been focussed further south and then migrated to adjacent to the main fault strand in the north when this propagated to the sea floor. Blocks of calcarenite within the sheet turbidites close to the southern margin imply tilting may have led to uplift and reworking of the Alfaro Member next to the southern boundary fault.

Younger Phase 3 sheets (and the marls between them) vary little in thickness across the floor of the sub-basin implying this was sub-horizontal by this stage. Thinning adjacent to the southern margin of the sub-basin reveals the presence of a fault-controlled, bounding slope. More rapid pinching out of the calcarenitic compared to the siliciclastic flows suggests a difference in the flow characteristics. The former may have been thinner with more concentrated lower portions as they tend to be coarser grained with less silty and micaeous components. The ponded suspensions may have inflated to differing extents, explaining the difference in pinchout geometry.

The thick interbedded slump has been attributed to local collapse of part of the El Cautivo Fault scar (Haughton, 2000; Hodgson, 2002). This is consistent with the muddy composition of the slump and its position adjacent to the El Cautivo Fault Zone. The analysis of the intra sub-basin fault limiting the slump to the south demonstrates that it was also active during deposition, initiating after the first bundle of thin beds above the bed 4 marker and ending shortly after slump deposition. The fault temporally partitioned the floor of the sub-basin providing deeper accommodation between it and the El Cautivo Fault. Occasional turbidity currents arriving from the west were preferentially trapped and confined on the downthrown block between the two faults (Figure 9C). Superficially the fault to the south appears extensional, but this would require an unrealistically high (>50 m) scarp to totally confine the slump. The major cross-sectional bend (step) on this fault requires a strike-slip component of displacement; furthermore bed geometries of sequences adjacent to this curved fault are incompatible with significant dip-slip displacement (Figure 9A). The preferred interpretation is therefore that the fault has laterally displaced the slump so that any overstep of the fault by the slump is out of the plane of section (Figure 9C); i.e. now below the ground for a dextral offset. The proximity and the angle between the faults bounding the slump suggest the small fault to the south formed a synthetic fault splaying from the dextral El Cautivo Fault.

The conformable contact between the Phase 3 sub-basin fill and the overstepping Phase 4 succession implies the fault-induced wedging and lateral displacement of the sub-basin had finished before Phase 4 deposition when the linked faults became inactive.

MTD-1 differs from the younger failures in that it remobilised both Phase 4 and Phase 3 deposits. It is very well exposed in the Rambla de Tabernas and in nearby road cuts giving a particularly good 3D insight into its internal structure and relationship to the undisplaced stratigraphy (Figure 11A). MTD-1 is up to 160 m thick centrally and about 50 m thick close to its toe. It can be followed laterally for up to 600 m, and extends about 350 m across the top of the sub-basin which it partly reworks. Phase 3 and 4 turbidite beds incorporated in MTD-1 are largely intact albeit complexly deformed, so much of the remobilisation can be classified as a slump (sensu Martinsen, 1994). In low cliffs on the east side of the Rambla de Tabernas, the basal detachment to the MTD is seen. This is a planar, bedding parallel bundle of clay-lined surfaces that lies stratigraphically about 10 m above marker bed 4 in the Phase 3 sub-basin fill (see earlier). The surfaces separate lensoid bodies of imbricated sandstone beds (duplexes) ranging from a few centimetres to a few tens of centimetres high. The slump transport direction deduced from the bed imbrication is from east (N100E) to west (Figure 11C).

The toe of MTD-1 is seen in a north-facing cliff (20 m high and 100 m long) at a prominent bend in the Rambla de Tabernas (Figure 11B). The basal detachment here is at rambla level and it has ramped up to within the overstepping Phase 4 turbidites. The ramp between here and where the basal detachment occurs on the east side of the rambla is estimated to be 40 m high. Both Phase 3 and 4 sheet turbidites are rafted intact above the detachment but roll into a large recumbent fold that is inclined to the west and corresponds to the toe of the failure. The toe is draped by in-situ Phase 4 turbidites in which there are many small dip discordances (Figure 11B).

Approximately 200 m to the north, MTD-1 truncates a 5 m thick sandstone sheet which can be mapped northwards away from the slump as part of an undisturbed succession of Phase 4 sheet turbidites. The thick sandstone is cut vertically by the lateral (northern) edge of the MTD, but a thick mudstone cap overlying the sandstone appears to overstep the junction between sandstone and the remobilised deposits, thinning and lensing out southwards over the MTD (Figure 11E).

MTD-1 is also seen beside the old N340 road in a high and unstable cut face north of Barranco de los Baños (a 300 m long and 30 m high exposure; Figure 11D). The northern end of the cliff preserves Phase 3 turbidite sheets (c. 50 m below marker bed 4), these are flexed down into MTD-1 which dominates the southern end of the exposure. Soft sediment deformation here has thus cut deeper still into the sub-basin fill. The eastern extension and headwall of the failure cannot be observed as this part of the MTD has been removed by recent erosion.

Interpretation: Exceptional exposure conditions encountered in the Tabernas Basin permits the reconstruction of the 3D geometry of MTD-1 (Figure 11F). An important feature of MTD-1 is that it is composed of local basin floor deposits, including translated ponded turbidites with thick mud caps that can only have formed on a relatively flat and deep part of the palaeo-sea floor. Phase 3 sandstone sheets preserved beneath the failure show evidence for thinning to the south suggesting proximity to a slope. Phase 4 turbidites that overstep the faulted southern margin of the sub-basin are also involved in the failure. A possible explanation is that gravity instability took place by sliding of the initial onlap wedge off the slope flanking the south-side of the sub-basin. This slide then interacted with and drove deformation of the earlier sub-basin fill. However, kinematic evidence from the stepped basal detachment, forced folds within the sliding mass and the position of the lateral limit of the failure suggests dominantly westward transport, parallel to the sub-basin margin. This suggests that failure may have been influenced by down to the west rotation of the sub-basin and its flanking slope (the Sierra Alhamilla to the east was uplifted to became emergent around this time). The large frontal fold may relate to a ramp on the underlying detachment, and to buttressing at the toe of the MTD. Minimal internal disaggregation is seen, at least in the section provided by the outcrop, consistent with limited transport and a low drop height due to the early stage in the onlap; MTD-1 involved slumping and sliding, but no transformation to a debris flow.

The trigger for the failure is uncertain, but the relationship with a large sandstone-mudstone couplet in the laterally equivalent undisturbed section to the north may be significant. The slump appears to cut the sandstone component of this bed, but to be onlapped by the mud produced by suspension fallout. This suggests MTD-1 may have been contemporary with a large-volume turbidity current, with failure occurring after the sand had fallen from suspension, but before the mud cloud rained out. Such a relationship could reflect a common trigger (seismic shaking) or a role for the large-volume (sloshing?) turbidity current in destabilising or loading the slope it impinged on. The former (common trigger) is perhaps less likely as the local slope failure might be expected to precede the arrival of the turbidity current.

A second more extensive slope failure associated with the southern onlap slope occurs approximately 20–80 m above the Gordo megabed although locally it cuts down to the level of the Gordo megabed. MTD-2 can be traced between a series of outcrops for a distance of c. 3.5 km parallel to the SE to NW transport direction (Figure 10) using the underlying Gordo event bed as a marker. Its width is unclear due to outcrop limitations. MTD-2 blanketed the southern part of the underlying sub-basin, and was sourced higher on the southern onlap surface. It is up to 130 m thick and characterised by a variable degree of internal disaggregation, locally forming blocks dispersed in a well-mixed matrix. The style of deformation also varies laterally along the failure. MTD-2 translated Phase 4 basin floor turbidites downslope, with no evidence for input of external slope-derived sediment. It does, however, locally contain pebbles and cobbles of Nevado-Filabride schist.

Close to the onlap surface (approximately 200 m from the projected onlap, and 100 m above the Gordo event bed), the inferred head region of the failure is seen. Here, a 6 m thick package of discordant inclined beds is overlain by a sandstone/mudstone couplet with an irregular base. When traced c.100 m to the NW, the basal deformed zone becomes more complex with several internal detachment levels (Figure 12D). There is also an increase in the deformation intensity with local folding beneath an irregular contact with the overlying sandstone-mudstone couplet.

A farther 1.8 km downslope, a large fresh road cut through the failure (200 m long and 15 m high) just south of Barranco de los Baños shows the deformation cuts down to within 20 m of the Gordo megabed (Figure 12A). The MTD here comprises upper and lower packages of soft-sediment deformed but coherent turbidites sandwiching a highly sheared and remoulded central part. The latter includes sheath folds (with SE-NW orientated stretching) and large (6 m across) folded and locally overturned rafts of Phase 4 turbidites floating in a poorly sorted, mud-rich matrix (Figures 12A,B). Significantly, extrabasinal clasts (mainly black schist and quartz vein pebbles up to 10 cm across) also float within the matrix. Overall, the deformed zone is estimated to be 130 m thick.

MTD-2 is also seen on the west side of the Rambla de Tabernas in an east-facing cliff up to 20 m high and 250 m long. The upper part of the MTC has been removed by recent erosion at this site. The Gordo event bed is also missing in this outcrop and deformation associated with MTD-2 merges downwards with MTD-1 deformation, making the base of MTD-2 hard to pick. MTD-2 deformation at this site is characterised mainly by multiple slide surfaces and discordances, particularly low in the body, with an increase in strain towards the central part of the MTD where tight folds and complex internal disruption are observed in rafts floating within a homogenised matrix.

The toe of the failure can be reconstructed from outcrops in the area close to the junction of the Rambla Seca with the Rambla la Gitana. Here a large overturned fold (at least 30 m wavelength) of Phase 4 turbidites is developed. Further to the NW, the post-Gordo succession is largely intact but traversed by several soft-sediment thrust and reverse faults and associated forced folds (Figure 12C). Restoration of the fault geometry indicates compression from the south-east.

Interpretation: MTD-2 differs from MTD-1 in that it preserves evidence for internal disaggregation and matrix formation. This suggests that this failure initiated as a slide but locally broke up internally to become a debris flow. Transformation of just the central section of the MTD may reflect a combination of partitioning of deformation in the lower part to the basal shear surface and reduced drainage of water from the central part, increasing pore fluid pressure. Alternatively, the failure may have occurred in two phases. MTD-2 transported Phase 4 turbidites blocks and is interpreted to have formed by collapse of part of the onlap-wedge. The toe of the failure was associated with a large fold and it is interpreted to have been frontally-confined like MTD-1. Compressional deformation spread out down slope from the confined front, deforming the basin floor that buttressed the failure. Transport was to the NW on the basis of internal and external shear indicators and the geometry of the inferred frontal fold.

A thick debrite-turbidite or megabed occurs stratigraphically 80 m above the Gordo megabed in the southern Tabernas Basin away from the onlap slope. Extrabasinal pebbles incorporated in MTD-2 resemble those in northerly-derived, large-volume event beds and are unlikely to have come off the southern slope. The thick sandstone–mudstone couplet also blankets headwall slides and slumps close to the onlap. The association between slumping and sliding and the emplacement of a large megabed is again significant and may suggest a link.

The youngest failure, MTD-3, occurs approximately 110 m above the Gordo megabed and is more difficult to link to the onlap surface and turbidite wedge, partly because it crops out remote from the onlap to the south and west of the junction of the Rambla la Gitana with Rambla Seca, beyond the confined toe of the underlying MTD-2. Although it is displaced on a series of NE-SW trending faults, it can be mapped over a strike length of at least 1.5 km. It is up to 65 m thick and dominated by white marl in which are dispersed sparse deformed clasts and rafts (up 10 s m long) of distinctive pebble-bearing calcarenites (locally micro-conglomerates). The marl unit thins northwards to c. 15 m over 1 km. The rafts tend to be folded. Rounded pebbles as clasts and within the rafts are dominantly basement dolomite, the surfaces of which are extensively bored. The calcarenites contain extensive shell fragments, including corals. MTD-3 is directly overlain by a thick sandstone-mudstone couplet, so the overall appearance of the unit is of a debrite-turbidite couplet; however, the capping sandstone is not a calcarentite. It is a siliciclastic sand of metamorphic provenance similar to the bulk of the basin floor turbidites making up the Phase 4 basin fill.

Interpretation: MTD-3 was not derived from the onlap wedge, but from higher on the southern slope. The pebble types and calcarenitic composition of the rafts in the lower muddy unit suggest emplacement as a mudflow dispersed from a southern or southwestern source (possibly the Sierra de Gádor). Abundant borings and coral fragments suggest a partly shallow water source, not the deep water reworking of basin-floor turbidites seen in the older MTDs. Whereas MTDs 1 and 2 were slope-attached and frontally-confined, MTD-3 appears to have more passively filled deep bathymetry without disturbing the onlap wedge. It was directly followed by a contained turbidite, not a trailing plume from the mudflow but a separate event dispersed from a different (siliciclastic) source. The sharp switch in composition implies that this is unlikely to a case of textural and compositional fractionation in a single flow. This again indicates a connection between different failure events, in this case possibly sharing the same trigger as there is little mixing between the two deposits. The stacking of event beds one on top of the other suggests little mounding on the top of the lower debrite, despite its thickness.

Turbidity currents are commonly confined by lateral and/or distal slopes against which their deposits thin and pinch out. Outcrop examples of turbidites that can be traced laterally to onlap against slopes have been described from a number of basins (e.g. Sinclair 2000; McCaffrey and Kneller, 2001; Amy et al., 2004; Smith and Joseph 2004; Marini et al., 2015; Spychala et al., 2015). Most examples involve simple aggradation of confined turbidites against stable slopes that are either bare of sediment or draped by mud. Less is known about onlap settings where the onlap is inherently unstable and the slope/base of slope is modified by episodic slope failures, although this situation is likely to be common in basins on active substrates, such as slope mini-basins in the Gulf of Mexico and elsewhere (Madof et al., 2009; Beaubouef and Abreu, 2010) or thrust-top basins (Tinterri and Tagliaferri, 2015). Onlaps associated with the Alfaro sub-basin are instructive on a number counts:

Firstly, they show that the bed style can reflect the degree of confinement (see also Hodgson and Haughton 2004). Confined turbidites within the small sub-basin have a different vertical structure to those filling the enlarged but still confined basin once the sub-basin topography was healed. This is best seen in the thinner beds which in the Phase 3 fill comprise sandstone beds capped by structureless sandstone divisions interpreted as fallout from sandy suspensions ponded in local basin floor depressions. Once the depressions were filled, thinner turbidite beds show well laminated sandstone divisions throughout with evidence for flow reversals from ripple laminations. The flows reflected off the confining slopes but dissipated before sandy suspension clouds could form (a function of weaker containment).

Secondly, flows of different provenance (siliciclastic vs calcareous) can have different onlap geometries against the same slope. Calcarenitic sandstones pinch out laterally on the south side of the sub-basin before siliciclastic sandstones of similar thickness are lost. The former tend to be coarser grained and may have formed thinner flows and suspension clouds that did not inundate the slopes as much (McCaffrey and Kneller, 2001). The finer grained and polymodal siliciclastic flows produced thicker suspensions that extended further up slope.

Thirdly, when confinement is associated with an active fault, fault zone widening can trim the onlapping sandstones as has happened along the northern margin of the sub-basin. Ponded sheets show little thinning towards the fault due to the basin asymmetry, and are locally entrained as blocks in the fault zone.

Lastly, once the sub-basin was healed, the slope marking the southern limit of the basin was progressively onlapped but was subject to episodic collapse of the onlapping sediment wedge. Small scale slides involving thin bedded turbidites are common in the onlap area. Correlations using marker beds suggest a possible link between these slides and the thicker MTDs observed downslope on the floor of the basin. The slide complexes may record the instability leading to initiation of MTDs within the onlap wedge. MTDs 1 and 2 are composed mainly of wedge and basin floor turbidites and represent locally detached and frontally confined mass transport deposits with basal shear surface ramps, rafted blocks of remobilised turbidites, and well developed recumbent folds associated with frontal thrust propagation (Frey-Martínez et al., 2006; Moscardelli and Wood, 2008; Bull et al., 2009). The slump toes appear to have formed positive topography on the sea floor against which later turbidites onlap. MTD-3 was the furthest travelled and comprises exotic calcarenitic sandstone blocks, abundant borings, shells and coral fragments floating in a muddy background. The overall shape is also different with a more sheet-like geometry filling the deep bathymetry of the basin and draped by a thick ponded megabed (Figure 10).

Each of the local slope-sourced MTDs is associated with a thick, basin-wide megabed either lateral to it (MTD-1,-2) or directly on top of it (MTD-3). McCaffrey and Kneller (2001) drew attention to the possibility of debris flows triggered by turbidity currents impinging on distal slopes. Puigdefàbregas et al. (2004) describe outcrops of the Grés d’Annot onlap, SE France, showing evidence for deformation induced by flow reflection against the basin confining slope. Overpressure in the slope muds associated with the sudden arrival of a large volume turbulent flow was thought to trigger injection, delamination and en-masse downslope movement. This sort of mechanism may explain the origin of MTD-1 and 2, particularly as large volume flows impinged on a slope that had been tectonically rotated and was being rapidly onlapped and loaded. One curious aspect of the slope stability is that the largest megabed event to impinge on the slope (Gordo megabed which occurred between MTD-1 and -2) did not trigger a failure. Instead, each of the three megabed elements (basal breccia or debrite, central graded sandstone and upper mudstone cap) lap out directly against the slope (Figure 13). It is possible the unusually thick basal debrite which contains abundant large intraclasts blanketed the slope protecting it from turbulent pressure fluctuations in the overlying expanded cloud. Alternatively, MTD-1 had already released the potential energy stored in the onlap wedge which was then relatively stable at this time.

This sort of distal failure triggering mechanism cannot apply for MTD-3 where the failure must have taken place high on the slope and did not involve remobilisation of the onlap wedge; the slope failure also preceded arrival of the megabed from a different source area. It is possible that in this case the separate slope failures had a similar trigger (a large earthquake?).

Neogene basins in SE Spain have variously been interpreted as an extensional (Platt and Vissers, 1989; Vissers et al., 1995; Augier et al., 2005), thrust-top (Frizon De Lamotte et al., 1995) or strike-slip basins (Montenat et al., 1987a; Montenat et al., 1987b; Kleverlaan, 1989b; Haughton, 2000). The uncertainty partly reflects the difficulty of linking faults to basin formation in areas where the stress field has evolved rapidly, where many of the faults are post-depositional, and where Neogene normal, thrust and strike-slip faults are all developed but are not necessarily directly related to basin formation. In addition, later rotation of earlier low-angle extensional structures to vertical can result in their misinterpretation as strike-slip faults (Valetti et al., 2019), and strike-faults can act as transfer structures in areas that are fundamentally undergoing heterogeneous extension (Martínez-Martínez, 2006; Giaconia et al., 2014).

Stratigraphic evidence discussed above (and in Hodgson and Haughton 2004) from the southern part of the Tabernas Basin strongly favours a strike-slip origin for the subsidence that opened up deep-water accommodation in that strike-slip faults are identified that were clearly active at the time of deposition. A wide range of different types of strike-slip basin have been described (e.g. Wood et al., 1994; Aksu et al., 2000; Lazar et al., 2006) ranging from fault-bend to stepover basins to basins formed between diverging or splaying faults. Other strike-slip basins form above relatively straight but divergent sectors of strike slip faults as negative flower structures (Bozkurt and Koçyiçğit, 1996; Aksu et al., 2000; Barnes et al., 2001) or due to transform-normal extension (Ben-Avraham and Zoback, 1992).

The Alfaro sub-basin formed a narrow trough delimited to north and south by oblique-slip faults that displaced the sea-floor. Subsidence was demonstrably asymmetric, thickening towards the fault with the more extensive gouge zone (the El Cautivo Fault) and with local unconformities and evidence for uplift and gravity sliding of earlier deposits along the southern margin (Figure 14). Splays of the main bounding fault also propagated through the fill, creating subtle irregularities on the sea floor, uplifting areas which modified the flow pathways and capturing a slump failure thought to relate to collapse of the master fault scarp. The syn-sedimentary oblique slip faults are interpreted to converge downwards and to link to a master fault. Outside and to the northeast of the sub-basin, a series of folds with fold axis trending at an angle of 30 anticlockwise to the E-W Alfaro fault zone also deformed the sea floor, creating depressions into which sheet turbidites thicken and highs that created local slope instability.

The convergence of dextral-oblique strike-slip faults at depth, evidence for displacements with important dip-slip components, the narrow trough-like sub-basin and the oblique orientation of folds suggest the sub-basin formed above a negative flower structure associated with a divergent strand of the fault that projects eastwards along the northern flank of the Sierra Alhamilla (Sanz De Galdeano, 1989; Figure 15). As bed dips in the vicinity of the Alfaro sub-basin are low, this cannot be an example of earlier extensional detachments that have been rotated to vertical to masquerade as strike-slip faults as has been argued elsewhere (Valetti et al., 2019). Evidence for folding and ponding of turbidites north-east of the sub-basin imply lateral changes between coeval transtension and transpression. In the Rambla de Reinelo area, the sea floor was gently warped, but moving further east, the earlier fill to the Sorbas Basin was already inverted in strong dextral transpression at this time (Haughton, 2001). The eastward change to transpressional deformation corresponds with the position of the western culmination of the Sierra Alhamilla. Northward translation of the Alhamilla as the regional folds amplified in the late Tortonian may account for this. The western end of the Alhamilla also marks a change in the style of turbidite deformation from strong ponding in deep depressions to the west, to weak ponding in shallow fold axes in the eastern Tabernas Basin. Hodgson (2002) also suggested that the Phase 2 sandy fan system exposed in the eastern part of the Tabernas Basin was deformed by growing transpressional folds during an earlier phase of bypass and basin floor fan growth. Close to Tabernas, early Phase 1 conglomerates were uplifted in the core of transpressional anticlines (Figure 2) against which the sandy fan onlapped. In the Sorbas area close to Lucainena (c.25 km to the east; Figure 1) Tortonian ponded sheets exposed just north of the Sierra Alhamilla closely resemble the Late Tortonian/early Messinian Alfaro sub-basin fill and may represent deposition within another but earlier transtensional depression later inverted in transpression.

Evidence for syn-depositional strike-slip deformation creating local accommodation on the sea floor in both the Tabernas and Sorbas basins is seemingly at variance with the recent focus on normal faults and regional extension in driving Late Miocene basin formation in the area (Andrić et al., 2018; Valetti et al., 2019). However, Giaconia et al. (2014) pointed to the importance of strike-slip transfer faults in areas undergoing regional heterogenous extension and showed that these bound Tortonian depocentes in the eastern Sorbas Basin. Similarly the Alpujarras Corridor (Sanz De Galdeano et al., 1985; Figure 1) to the west of the Tabernas Basin has been interpreted as a dextral strike-slip fault zone transferring deformation between areas undergoing different styles and amounts of Miocene to Recent extension (Martínez-Martínez, 2006). The Alfaro sub-basin lies on the eastward projection of the Alpujarras structure to link with the fault zone on the northern flank of the Alhamilla. The implication is that transtension on long transfer faults opened up local depocentres (sub-basins) distinct from the titled fault-block depocentres in the hanging wall of the main detachment faults.

Basins resembling the Alfaro sub-basin are seen today on the floor of the Alboran Sea south of the Tabernas area where they have been imaged on seismic profiles. The Alboran Basin is divided into several small sub-basins which have mainly been controlled by active right-lateral transpression due to oblique Africa-Iberia convergence since the late Tortonian (Bourgois et al., 1992; Comas et al., 1992; Woodside and Maldonado, 1992). Alvarez-Marrón (1999) interprets some of the sub-basins in the eastern Alboran Sea as negative flower structures. The Yusuf Basin is a 16 km long by 9 km wide depocentre bounded by a pair of sub-vertical faults. The dextral oblique-slip Yusuf Fault forms a master fault on the northern side of the basin. Subsidence in the basin was asymmetric wedging towards the Yusuf Fault (Figure 1). Subordinate faults on the south side of the basin step down into the basin. Moreover the downthrown/flexed basin margins are characterised by listric extensional faults caused by gravitational sliding.