INTERACTIONS BETWEEN DEEP-WATER GRAVITY FLOWS AND ACTIVE SALT TECTONICS

A BSTRACT : Behavior of sediment gravity ﬂows can be inﬂuenced by seaﬂoor topography associated with salt structures; this can modify the depositional architecture of deep-water sedimentary systems. Typically, salt-inﬂuenced deep-water successions are poorly imaged in seismic reﬂection data, and exhumed systems are rare, hence the detailed sedimentology and stratigraphic architecture of these systems remains poorly understood. The exhumed Triassic (Keuper) Bakio and Guernica salt bodies in the Basque–Cantabrian Basin, Spain, were active during deep-water sedimentation. The salt diapirs grew reactively, then passively, during the Aptian–Albian, and are ﬂanked by deep-water carbonate (Aptian–earliest Albian Urgonian Group) and siliciclastic (middle Albian– Cenomanian Black Flysch Group) successions. The study compares the depositional systems in two salt-inﬂuenced minibasins, conﬁned (Sollube basin) and partially conﬁned (Jata basin) by actively growing salt diapirs, comparable to salt-inﬂuenced minibasins in the subsurface. The presence of a well-exposed halokinetic sequence, with progressive rotation of bedding, beds that pinch out towards topography, soft-sediment deformation, variable paleocurrents, and intercalated debrites indicate that salt grew during deposition. Overall, the Black Flysch Group coarsens and thickens upwards in response to regional axial progradation, which is modulated by laterally derived debrites from halokinetic slopes. The variation in type and number of debrites in the Sollube and Jata basins indicates that the basins had different tectonostratigraphic histories despite their proximity. In the Sollube basin, the routing systems were conﬁned between the two salt structures, eventually depositing amalgamated sandstones in the basin axis. Different facies and architectures are observed in the Jata basin due to partial conﬁnement. seismic imaging

Subsurface studies have shown that salt structures deforming the seafloor can exert substantial control on the location, pathway, and architecture of lobe, channel-fill, levee, and mass-transport deposits ( Fig.  1; e.g., Mayall et al. 2006Mayall et al. , 2010Jones et al. 2012;Wu et al. 2020;Howlett et al., in press). Turbidity currents that were ponded, diverted, deflected, and confined by salt structures (Fig. 1) are well documented in the eastern reflection data suggest that the basal salt layer is present at depth connecting the Bakio and the Guernica structures ( Fig. 4; Robles et al. 1988;Poprawski and Basile 2018). The structures were close to the seafloor during the middle Albian creating highs that were capped by isolated carbonate platforms and influenced the deposition of the BFG Robles 1991a, 1991b;Pujalte et al. 1986; Cámara 2017; Roca et al. 2020). Slope apron facies, deposited at the platform edge, and subsequent stratigraphy formed tapered halokinetic sequences against the west of the Bakio diapir ( Fig. 4; García-Mondéjar and Robador 1987;Soto et al. 2017;Roca et al. 2020).

Bakio Stratigraphy
Studies of magnetic anisotropy in the Bakio-Guernica area demonstrate a minimal Pyrenean compressional overprint to the stratigraphy (Soto et al. 2017), as the tholeiitic ophitic inclusions in the diapir acted as buttresses forming shadow areas protected from the compression. Hence the area is used to study syn-halokinetic deposition without a regional tectonicdeformation overprint.
The Aptian-middle Albian Urgonian stratigraphy (middle Albian Sequence 2: H. dentatus Zone of Agirrezabala and López-Horgue 2017) FIG. 3.-A) Geological map located in Figure 2 and B) stratigraphic column for the study area. A) Compiled from Espejo and Pastor (1973), Espejo (1973), Garrote-Ruiz et al. (1991, 1992, 1993a, 1993b, Pujalte et al. (1986), García-Mondéjar and Robador (1987), Robles et al. (1988Robles et al. ( , 1989, Vicente Robles (1991a,1991b), Poprawski et al. (2014Poprawski et al. ( , 2016, Ábalos (2016), and fieldwork observations. Lateral facies changes in carbonates around the salt outcrops at Guernica are modified from García-Mondéjar and Robador (1987). The Guernica structure has been weathered away and forms a present-day estuary. Orange lines show locations of stratigraphic logs shown in succeeding figures, dashed lines indicate missing section, and numbers refer to subsequent figures where logs are presented. B) Abbreviations for stratigraphic units are shown in ( ) and formation names of Poprawski et al. (2014Poprawski et al. ( , 2016, where they differ from those used in this study, are shown in [ ]. Numbers adjacent to the stratigraphy refer to regional sequences of Agirrezabala and López-Horgue (2017), based on biostratigraphy. Line of section is shown for Figure 4; for full extent see Figure 2.

METHODS AND DATA
The dataset comprises 28 sedimentary logs (totaling 821 m of stratigraphy) collected along the Bakio-Guernica coastline. The logs were collected at a 1:25 scale, with 1:10 scale used locally to capture additional detail. Halokinesis during BFG deposition (García-Mondéjar 1996;Poprawski et al. 2014Poprawski et al. , 2016Roca et al. 2020) generated syndepositional basin-floor relief and led to the development of multiple discrete depocenters (Vicente Bravo and Robles 1991a, 1991bAgirrezabala 1996). As such, correlating stratal surfaces in and between depocenters is difficult and un-crewed-aerial-vehicle (UAV) photography was used to aid stratigraphic correlations (Hodgetts 2013). Paleocurrent, bedding, and structural data were collected to determine the influence of syndepositional basin-floor relief and to quality-control the pre-existing geological map of Poprawski et al. (2014Poprawski et al. ( , 2016. Paleocurrent readings were taken where sedimentary structures were clear enough to permit unambiguous data collection. Sparse biostratigraphic data (Agirrezabala and López-Horgue 2017) hinders correlation across the structures; hence we refer to the Lower (LBF) and Upper (UBF) Black Flysch group only, to avoid further subdivisions based on geographic location (e.g., Robles et al. 1988;Vincente-Bravo and Robles 1991a, 1991bPoprawski et al. 2014).

Basin Subdivision
To aid comparison, the study area has been divided into two depocenters: the Jata and Sollube basins (Figs. 3,4). These are analogous to subsurface minibasins, defined as relatively small (5-30 km) synkinematic depocenters subsiding into thick salt (Hudec and Jackson 2007;Jackson and Hudec 2017). The Jata basin is confined to the east by the Bakio diapir. The Sollube basin is confined on both its western and eastern sides by the Bakio and Guernica structures, respectively (Fig. 3), and hence is more confined than the Jata basin.

LITHOFACIES
This study focuses primarily on the facies variability in the siliciclastic BFG. A description and discussion of the carbonate facies of the BMF and BBF is provided in Poprawski et al. (2016). Here, the carbonate facies are tabulated for reference in Table 1 (and Supplemental Fig. 1) and their depositional process interpreted. The BFG lithofacies presented in Table 2 and in Figure 5 represent ''event beds'' and are classified based on outcrop observations. ''Mud'' is used here as a general term, for mixtures of clay, silt, and organic fragments. Where individual facies are heterogeneous, multiple photographs are shown to illustrate this lithological and sedimentological variability (Fig. 5). FIG. 4.-Schematic structural-stratigraphic cross section through the Bakio and Guernica diapirs. Full extent is located using B-B 0 in Figure 2, partial extent is also shown in Figure 3. The section combines Poprawski and Basile (2018), Robles et al. (1988), field observations, and publicly available vintage onshore seismic lines from IGME. Facies are indicated where known or inferred from the literature but are left blank where they cannot be inferred. Two times vertical exaggeration for clarity.

INTERPRETATION OF DEPOSITIONAL ELEMENTS
Facies associations (  (Prélat et al. 2009;Spychala et al. 2017) and channel-levee (Kane and Hodgson 2011;Hubbard et al. 2014) nomenclature that best fit field observations. Facies associations and geometries are described separately, because architecture alone is not diagnostic of the depositional sub-environment, and multiple facies associations can form a similar architecture.

STRATIGRAPHIC EVOLUTION
Extensive exposures permit detailed lithostratigraphic analysis (Figs. 3,8,9,10), allowing investigation of the role of salt-induced relief on depositional patterns (Figs. 11,12,13). The following sections describe the exposures from oldest to youngest, first focusing on the flanks of the Sollube and Jata basins (Gaztelugatxe Island and Bakio West Bay, respectively) and then the axis of the Sollube basin (Cabo Matxitxako).

Gaztelugatxe Island
The cliff sections to the south of Gaztelugatxe Island (located in Fig. 3) provide a semi-continuous section through 120 meters of the BFG, 2 km northeast of the exposed Bakio diapir, and c. 9 km northwest of the Guernica structure. The stratigraphy is subdivided into five lithostratigraphic units (GX1-5).
Description.-GX1 is 8 m thick and consists of bioturbated mudstones, calci-debrites, calci-turbidites, thin-bedded turbidites, and mud-rich debrites. It shows an overall coarsening-and thickening-upwards from centimeter-to meter-scale interbeds of each facies. GX2 is 10 m thick and dominated by carbonate-clastic debrites, with angular clasts of the Gaztelugatxe Limestone, up to 1 m in diameter. GX2 pinches out downslope, forming a triangular geometry (Table 4, Fig. 7E). GX3, has a minimum thickness of 42 m, onlaps GX2 and is recognized as the first purely siliciclastic succession; comprising thin-to medium-bedded turbidites, debrites, hybrid beds, and bioturbated mudstones. GX4 has a minimum thickness of 9 m, and its base is marked by a meter-thick slump,  Table of carbonate facies detailing the major observations of the six facies which constitute the early Albian Bakio Marls and early middle Albian Bakio Breccias formations. For further discussion see Poprawski et al. (2014Poprawski et al. ( , 2016. Supporting facies photographs are provided in supplementary material.

Facies Name Description Interpretation
Thin-bedded calcareous sandstone 0.01-0.1-m-thick beds of bioclastic (corals and shell fragments) very fine-fine grained sandstones. Commonly normally graded with flat tops and flat bases. Weak planar, ripple, and convolute lamination.
Medium-bedded calcareous sandstone 0.1-0.3-m-thick very fine to medium-grained normally graded sandstones, with flat bases and flat tops. Planar, ripple, and convolute lamination observed. Mud clasts and intense dewatering also present.
Medium to high-density turbidites: Presence of tractional structures suggests deposition from a dilute turbidity current. High mud-clast percentage could suggest imminent flow transformation (Barker et al. 2008). Limestone breccia 10þ-m-thick beds of matrix-or clast-supported limestone breccia, with erosional bases and undulating tops. Poorly sorted beds consisting of subangular to angular limestone megaclasts, which can be normally, inversely or non-graded. Megaclasts commonly contain entire rudists and fragmented corals.

Mass-transport deposit:
Poorly sorted clasts suggest deposition from ''flow freezing'' of a flow with yield strength (Iverson et al. 2010). Limestone clasts are similar in composition to the Gaztelugatxe Limestone, suggesting that it is their source (Poprawski et al. 2014;. Fossiliferous poorly sorted carbonate mudstone 0.03-0.2-m-thick poorly sorted, non-graded carbonate mudstone with fossil fragments. Beds are laterally discontinuous, with undulose, gradational bases and tops. Cm-to dm-size bioclasts of urchins, brachiopods, bryozoans, bivalves, corals, crinoid stems, and rarer mollusk shell fragments.

Debris flow:
Fragmented bioclasts, poor sorting, and undulose contacts suggest deposition from a laminar flow (Nardin et al. 1979). Fossils are fragmented, indicating reworking, but are not lithified, indicating direct reworking from an active platform or reef.
Remobilized carbonates 5þ-m-thick packages consisting of a combination of the above facies that have been slightly remobilized but maintain bedding planes. Contacts are erosional, scalloped, or smooth, and underlying mudstone units often appear sheared. Convolute lamination and soft-sediment deformation are present.

Slide deposits:
The remobilization but maintenance of individual bedding planes and sheared basal contacts indicates that these are slide deposits. Lack of internal deformation suggest these deposits have been remobilized post lithification, conceivably due to halokinetic movements (Ferrer et al. 2014;Poprawski et al. 2016). Clast-rich poorly sorted carbonate mudstone 0.1-1þ m thick poorly sorted beds with angular limestone clasts in a mud-silt matrix. Clasts can be cm to m scale, generally 1-12 cm, and are rich in mollusk fragments. Rare lithics and organics are observed. Undulose tops reflect clast topography and, bases are flat, weaklyerosional or undulose. Weak normal grading and rarer reverse grading are observed.
Carbonate-clast-rich debrites: Poor-sorting and large clast size indicates en-masse deposition from a laminar flow (Nardin et al. 1979;Iverson 1997;Sohn 2000) Weak normal grading suggests that some turbulence was influencing the flow. Clast angularity suggests proximity to source area. Lack of unconsolidated fossil debris suggests that lithification has occurred before reworking into the flow.

High-density turbidites:
The coarse grain size, thick beds, and amalgamation surfaces suggest deposition from a highly concentrated turbulent flow, indicating that these beds are turbidites. Weak stratification indicates traction-carpet deposition (Lowe 1982), suggesting high-density turbidites. Thick-bedded sandstones 0.5-1þ-m-thick beds of very-fine-to coarse-grained normally graded sandstones, which lack primary depositional structures and are commonly dewatered (Fig. 5C). Bases can be sharp, erosional, stepped, or amalgamated, commonly along a mudstone amalgamation surface with a subtle grain-size break (Fig. 5C), and tops are often flat. Plane-parallel lamination, mud clasts, and soft-sediment deformation are occasionally observed.
High-density turbidites: The general massive structuration of these deposits suggests that they represent rapid aggradation beneath a highly concentrated flow (Lowe 1982).

Mediumbedded sandstones
0.1-0.5-m-thick beds of very fine-medium grained, normally graded sandstones. Beds are rich in tractional structures, particularly plane parallel laminations (Fig. 5D). Ripple laminations is observed in bed tops, and beds are more frequently structureless towards bases. Bed bases are flat with tool marks or loaded, and tops are flat or convolute and often rich in mud clasts (Fig. 5D). Occasionally amalgamated.

Medium-density turbidites:
Based on their tractional structures and normal grading, beds of this lithofacies are interpreted as deposition from a dilute turbidity current. These beds are interpreted as mediumdensity turbidites due to their bed thickness and common lack of structures in the lower part of the bed.

Low-density turbidites:
Tractional structures and normal grading indicate deposition from a dilute turbidity current and are therefore interpreted as low-density turbidites. Common banding may reflect some periodic suppression of turbulence associated with flow deceleration or increased concentration (Lowe and Guy 2000;Barker et al. 2008). Ripples with opposing paleoflow suggests topographic interference.

Siltstone and very thinbedded sandstones
Packages of 0.1 m composed of individual fine siltstone to fine sandstone events less than 0.01 m. Beds form discontinuous drapes within mudstone (Fig. 5G), with flat bases and flat tops. Parallel and ripple lamination and diagenetic phosphate nodules are observed (Fig. 5G).

Background sedimentation:
Fine grain size indicates low-energy conditions, representative of background sedimentation via suspension fallout. Discontinuous siltstones suggest that lamination may be present below the scale visible in outcrop, representing deposition from a dilute turbidity current (Boulesteix et al. 2019). Poorly-sorted mudstone 0.1-1þ-m-thick siltstone to fine-sandstone rich mudstones (Fig. 5I). Poorly sorted, matrix-supported, clast-rich deposit with starry-night texture. Granules, organic fragments, mud clasts and rare shelly fragments present, often with subtle alignment. Bases are flat or undulose, tops flat or loaded (Fig. 5I).

Mud-rich debrites:
The poorly sorted matrix and clast-rich nature indicates enmasse deposition from a laminar flow (Nardin et al. 1979).
Poorly-sorted muddy sandstone 0.1-1þ-m-thick, mud-rich poorly sorted matrix-supported, fine to medium grained sandstones with starry-night texture (Fig. 5J). Organized mudstone clasts and sporadic granules to pebbles are observed. Flat-undulose tops and flat-graded base are common (Fig. 5J). Rare normal grading and grain-size segregation and infrequent sheared layers present.

Sand-rich debrites:
En-masse deposition from a laminar flow (Nardin et al. 1979;Iverson 1997;Sohn 2000). Weak normal-grading suggests that some turbulence was influencing the flow and therefore deposition from a transitional flow regime is interpreted (Baas et al. 2009(Baas et al. , 2013Sumner et al. 2013). Chaotic clastrich matrix supported deposit 0.5-3-m-thick, poorly sorted deposit with a poorly sorted matrix of mudstone to fine sandstone. Clasts include: cm-to m-scale sandstone balls (Fig. 5K), showing internal lamination and soft-sediment deformation, dm-m scale sandstone and heterolithic subangular rafts, deformed siderite nodules, limestone clasts, gastropod and sponge fragments, mud clasts, and phosphate nodules. Beds are flat-topped, and bases are weakly loaded (Fig. 5K).

Mega-debrites:
The poorly sorted matrix and large clast size are suggestive of ''flow freezing'' indicating deposition in a debris-flow regime (Iverson et al. 2010). These deposits are interpreted as mega-debrites due to their large clast size (rafts), suggesting they are derived from localized mass failure. Bipartite or tripartite beds 0.1-1.5-m-thick beds that contain multiple parts (Fig. 5L, M). Typically consisting of a lower fine-medium sandstone (division 1) overlain by a poorly sorted, muddy siltstone-sandstone (division 2) with a flat to slightly undulose base (Fig. 5L, M). Division 3 is sometimes present, consisting of cleaner siltstone or fine-grained sandstone loaded into division 2 (Fig. 5L). Division 1 can contain planar lamination and weak cross-stratification ( Fig. 5L) but is often massive with sporadic-slightly organized mud clasts (Fig. 5M). Division 2 is organic-rich, highly deformed, and can contain sporadic granules or pebbles (Fig. 5L, M). Starry-night texture is observed in this division. Division 3 is more frequently planar laminated than division 1 but can be highly chaotic (Fig. 5L).

Hybrid beds:
Tractional structures in division 1 and 3 suggest that these deposits formed under turbulent flows. Starry-night texture, poor sorting, and mud content suggest that division 2 was deposited under a transitional-laminar flow regime (Haughton et al. 2009). Flow transformation from turbulent to laminar can occur through flow decelerations (Barker et al. 2008;Patacci et al. 2014)  Mudstone with occasional, rare drapes of siltstone. I) Poorly sorted mudstone, foundered into by a thick-bedded sandstone. J) Poorly sorted muddy sandstone, containing sporadic granules and raft blocks. K) Chaotic clast-rich matrix-supported deposit encased between units of thin-to-medium-bedded sandstones. L) Tripartite bed consisting of lower medium-bedded sandstone with weak cross-lamination, middle poorly sorted mudstone and upper poorly sorted sandstone. M) Bipartite bed consisting of lower thickbedded sandstone which becomes mud-clast rich upwards overlain by a poorly sorted mudstone above.

SALT-INFLUENCED DEEP-WATER SUCCESSIONS
J S R 41 J S R overlain by interbeds (centimeter to decimeter scale) of turbidites, debrites, slumps, slides, hybrid beds, and mudstones.
GX5 is identified on the western side of Cabo Matxitxako (Fig. 8), having a minimum thickness of 30 m. There is approximately 400 meters of missing stratigraphy between GX4 and 5 (Robles et al. 1988), but GX5 is projected to lie stratigraphically above GX4. GX5 comprises predominantly amalgamated medium-to high-density turbidites showing evidence of soft-sediment deformation.
Thick-bedded nature suggests deposition from high concentration turbidity currents with relatively high rates of aggradation preventing the development of tractional sedimentary structures (Kneller and Bramney 1995;Talling et al. 2012). Common amalgamation and entrainment of mudstones clasts in thickbedded sandstones indicate that the parent flows were highly energetic and capable of eroding, entraining, and bypassing sediment during the passage of flow (Lowe 1982;Mutti 1992;Stevenson et al. 2015). Similar deposits elsewhere have been interpreted as lobe-axis deposition (Walker 1978;Prélat et al. 2009;Kane et al. 2017). Thin-bedded granular-cobbly sandstones are associated with overlying and adjacent amalgamated thick-bedded sandstones and are thought to represent a mostly bypassing equivalent of the depositional thick-bedded sandstones within the lobe axes (Kane et al. 2009).
Pinching out upslope ( Lobe off-axis Composed principally of normally graded structured to structureless medium-bedded sandstones ( Fig. 5D) with less common thin-bedded ( Fig. 5F) and thick-bedded sandstones ( Fig. 5C). Ripples at the tops of beds commonly show paleoflow directions opposite to those that are measured from flutes and grooves on bed bases. Mudstones, poorly sorted mudstones, sand-rich mudstones, and rarer chaotic clast-rich matrix-supported deposits are periodically or randomly interspersed in this facies association (Fig. 6).
A medium-density turbidite interpretation is given for these units based on the preservation of both structured and structureless sandstones. Similar preservation of both deposit types has been interpreted as off-axis lobe environments, deposited by decelerating turbidity currents (Prélat et al. 2009;Soutter et al. 2019). Opposing paleocurrent directions in event beds is characteristic of topographically influenced flows (Kneller et al. 1991;Bakke et al. 2013). Periodic deposition of mudstones suggests episodic system shutdown.
Poorly sorted mudstones, sand-rich mudstones, and chaotic clast-rich matrix supported deposit occurrence indicate periodic laminar flows which could indicate nearby active topography (Kneller et al. 1991;Mayall et al. 2010).
Pinching out upslope ( Siltstone and very-thin-bedded sandstones and mediumbedded-sandstones ( Fig. 6) are infrequently observed. Poorly sorted mudstones, sand-rich mudstones, and rarer chaotic clast-rich matrix-supported deposits are periodically or randomly interspersed in the otherwise organized thin-bedded sandstones and bipartite or tripartite beds (Fig. 6).
Thin-bedded, structured sandstones are interpreted to be deposited from low-concentration turbidity currents (Mutti 1992;Talling et al. 2012). Bipartite and tripartite event beds are interpreted as hybrid beds (Haughton et al. 2009). The transformation of flows in hybrid beds observed here document a change in flow process from high-medium-concentration turbulent to laminar or transitional, to low-concentration turbulent (Remacha et al. 2005;Baas et al. 2011). Thin-bedded sandstones and hybrid beds underlie lobe and lobe-axis facies associations and are therefore interpreted to be deposited adjacent to such deposits. Abundant hybrid beds and thin beds indicate lobe-fringe deposition elsewhere (
Mudstones separating individual events are often slightly deformed or sheared and show drapes of discontinuous siltstone (Fig. 6).
The fine grain size, thin-bedded character, and low stratigraphic position of these beds is consistent with lobe-fringe deposition. The relative lack of hybrid beds in this facies association support a distal-lobe-fringe interpretation ( Poprawski et al. 2014Poprawski et al. , 2016. The carbonate deposits could have been remobilized from previous BBF deposits or remnant carbonate highs (Poprawski et al. 2014(Poprawski et al. , 2016. GX2 represents a period of increased mass failure, which is interpreted to be halokinetically driven due to the lentil-shape and diapir-centric distribution of these limestone breccias (Table 4: Fig. 7E; e.g., McBride et al. 1974;Hunnicutt 1998;Giles and Lawton 2002). The thin-bedded nature and presence of hybrid beds in GX3 suggests early BFG deposition in a proximal lobefringe environment Soutter et al. 2019). Thin-bedded debrites are interpreted to be delivered axially, based on their association with thin-bedded turbidites that show regional paleocurrents. Thickbedded, chaotic, clast-rich units are interpreted to be halokinetically driven based on variable slump-axis paleocurrent readings (Poprawski et al. 2014).
At the base of GX4, a meter-scale debrite overlies 30 m of missing section (Fig. 8), which given the low-lying geomorphology is likely mudrich. The overlying turbidites and debrites suggest deposition in a lobe offaxis setting (Prélat et al. 2009;Spychala et al. 2017) where the seafloor was, at least periodically, unstable (Fig. 12E). Meter-thick beds that stack into 30-meter-thick packages suggest that GX5 represents deposition in the
Erosionally based geobodies infilled with coarser clasts indicate active erosion and deposition. Common amalgamation and entrainment of mudstones clasts in thick-bedded sandstones indicate that the parent flows were highly energetic and capable of eroding, entraining, and bypassing sediment during the passage of flow (Lowe 1982;Mutti 1992;Stevenson et al. 2015), while weak cross-stratification, slight grading, and pebble imbrication are more typical of depositional conditions (Mutti and Normark 1987). This juxtaposition of depositional and erosional elements has been observed elsewhere in channel-lobe transition zones (Mutti and Normark 1987;Wynn et al. 2002;Pemberton et al. 2016;Brooks et al. 2018). The presence of cross-stratified gravels supports the facies association proposed by previous work (Vincente Bravo and Robles 1991a;1991b;. Tabular amalgamated beds (Fig. 7G) Channel-axis Thick-bedded sandstones (Fig. 5C), granular-cobbly sandstones (Fig. 5A), poorly sorted muddy sandstones (Fig. 5I), and chaotic clast-rich matrix-supported deposits (Fig. 5K). Thick-bedded sandstones typically gradationally overlie granular-cobbly sandstones, which are commonly grooved on the base, showing normal grading (Fig. 6). These successions are erosional into the underlying poorly sorted muddy sandstones or chaotic clast-rich matrix supported deposits, which exhibit some deformation and shearing (Fig. 6). Sandstone beds either erode into each other, are amalgamated, or less commonly are separated by thin beds of mudstone (Fig. 6). Low-angle crossstratification is observed (Fig. 5A). The sandy mudstones and chaotic units contain subangular to angular poorly sorted clasts of up to boulder size. The composition of these clasts includes limestone fragments, organics, siliciclastic fragments, slumped and reworked thin-bedded heterolithics, clasts of granite, deformed and reworked siderite, mud clasts, and fossil fragments (Fig. 6).
Common amalgamation, erosion, and entrainment of clasts in the sandstones indicate that the parent flows were highly energetic and capable of eroding, entraining, and bypassing sediment (Mutti 1992;Stevenson et al. 2015;Soutter et al. 2019). The coarse grain size and basal location of granular-cobbly sandstones suggests that these beds were deposited as a coarsegrained lag in a bypass-dominated regime (Hubbard et al. 2014). Erosionally based lenticular sandstones and their grading from cobbly-fine sandstone is consistent with deposition in a submarine channel described elsewhere (Hubbard et al. 2008;Romans et al. 2011;McArthur et al. 2020). Weak low-angle lamination in sandstone beds could indicate lateral accretion (Kane et al. 2010;Jobe et al. 2016). Poorly sorted muddy sandstones and chaotic units could represent channel collapse and margin failure (Flint and Hodgson 2005;Pringle et al. 2010;Jobe et al. 2017). The wide variation in clast composition, more diverse than that observed in any other facies association, indicates broader catchment area for these debris flows, which may indicate an extrabasinal provenance (Stevenson et al. 2015).

Channelmargin
Thin-bedded sandstones (Fig. 5E, F) and poorly sorted mudstones with secondary medium-bedded sandstones and chaotic clast-rich matrix-supported deposits (Fig. 6). Thin-and medium-bedded sandstones are planar and ripple laminated (Fig. 5E). Poorly sorted mudstones and chaotic clast-rich matrix-supported deposits include angular to rounded clasts of limestone, siliciclastic fragments, and mud clasts. Medium-bedded sandstones erode into the tops of chaotic clast-rich matrixsupported deposits, and thin-bedded sandstones show loaded, flat and weakly erosional bases. This facies association appears beneath the channel-axis facies associations (Fig. 6).
The supercritical bedforms and thin-bedded nature of these deposits is similar to those described as channel-margin facies by others (Kane and Hogdson 2011;Hodgson et al. 2011;Hubbard et al. 2014;Jobe et al. 2017;McArthur et al. 2020). The location of this facies association beneath channel-axis deposits suggests that they were deposited adjacent to them, and this indicates they represent channel-margin facies association. The absence of debrites or slumps in GX5 (Fig. 8) suggests minimal halokinetic influence, either due to diapir inactivity or sediment accumulation, due to uplift and erosion from the Landes Massif (Agirrezabala 1996), outpacing the rate of seafloor deformation. The presence of amalgamated, laterally extensive medium-and high-density turbidites in GX5 supports deposition in a channel-lobe transition zone (Vincente Bravo and Robles 1991aRobles , 1991bRobles , 1995 or lobe-axis setting.

Bakio West Bay
The coastal cliff section at Bakio West Bay (Fig. 9) exposes c. 150 m of the BFG above its basal contact with the BBF (Robles et al. 1988), c. 1 km west of the Bakio diapir. This section is divided lithostratigraphically into three units (BW1-3), and is further divided by halokinetically driven unconformities into sub-units ( Fig. 9; Poprawski et al. 2014).
At the bases of BW3a, BW3b, and BW3c, 1-12-meter-thick debrites with variable thickness across the exposure overlie the angular unconformity (Fig. 9). Grain size varies from medium sand to boulders and clasts vary from rounded to angular. A more diverse range of clast rock types than elsewhere in the study area are observed, including limestone, sandstone, mudstone, organics, heterolithics, siderite, mafic material, granite, and siderite (Table 3). An undulose contact exists between the debrites and the 2-4-meter-thick amalgamated turbidites which overlie them (Fig. 9). These medium sand-to pebble-grade turbidites are lenticular in geometry, can be divided into meter-scale fining-upwards successions, and contain inclined stratification (Fig. 9).
Interpretation.-The angular unconformities are interpreted to be related to salt-diapir movements, and are interpreted as part of a tapered halokinetic sequence Poprawski et al. 2014). Unit  (Table 3). Peach highlights scale, either lens cap (52 mm), or indicated. Black arrow points to the north, and yellow shows the way up. A) Thick-bedded sandstones of the lobe axis. B) Interbedded sandstones and mudstones of the lobe off-axis. C) Thin-bedded sandstones interbedded with mudstones of the proximal fringe. D) Mudstones and very thin-bedded sandstones and siltstones of the distal fringe. E) Thick-bedded granular sandstones of the channel-lobe transition zone. F) Sandstones and poorly sorted mudstones of the channel axis. G) Thin-bedded sandstones interbedded with mudstones of the channel margin. Tabular bedded A package of stacked beds which show a continuous thickness laterally for 10s to 100s m, occasionally with some subtle thickness changes (Fig. 7A). Postdepositional faulting and tectonic and halokinetic tilt prevent these geometries from being traced on a 100s m-km-scale. Common in thinbedded (Fig. 7A) and medium-bedded sandstones (Fig. 5D). Tabular geometries are observed . 500 m away from diapiric influence.

J S R
This continuous, stacked geometry suggests constant depositional energy. Tabular architectures appear to be uninfluenced by topography, and are similar to unconfined settings (Prélat et al. 2009). Low-density turbidites are less affected by topography than more cohesive flow (Al-Ja'aidi 2000; Bakke et al. 2013) and therefore can run up topography for greater distances, without becoming ponded.
Concave upward (erosionally based) Curvilinear geobodies with variable thickness that are concave upward, consisting of a centroid and two margins, toward which the centroid thins, sometimes by up to 80%. (Fig.  7B). Erosionally based granular sandstones are present in the centroid of the geobody, often overlain by high-density turbidites, which become thinner-bedded towards the margins. The thickness of these geobodies is typically decimeter to meter scale, and thickness to width ratios can range between 1:10 and 1:50. The geobodies commonly erode and amalgamate with each other, and stack above the previous deposit.
Each geobody represents at least one event; the coarsegrained basal lag could represent a bypass event before the high-density turbulent flow which filled the geometry. These multipart geobodies, which are attributed to deep-water channels based on their geometries, stack on top of and erode into each other, suggesting increasing confinement (Mayall et al. 2010).

Pinching out upslope
Elements that change in thickness, but only in one direction (Fig. 7C). Commonly these geometries are amalgamated, with individual events displaying a convex-up geometry.
Thinning rate is approximately 10cm/m in Figure 7C.
Thinning of deposits indicates flow deceleration related to topography, which ultimately lowers flow concentration (Baas et al. 2011;Teles et al. 2016). The eventual pinch-out of the sandstone is due to the inability of the flow to run up the entirety of the topography.

Convex up
Packages are generally continuous in thicknesses on the scale of the outcrop, with beds thinning slightly to either side (Fig. 7E). The centroid is typically decimeters to meters thick. The upper surface of each deposit is commonly undulose with an overall, often subtle, convex-upward geometry (Fig. 7E). High-density turbidites dominate these architectures and are commonly stacked or amalgamated.
The upwards curvature and slight thinning of this geometry lead to their interpretation as lobate geometries (Prélat et al. 2009;Hodgson 2009;Sypchala et al. 2017). Shifting of the centroid of the lobe axis indicates that compensational stacking is influencing these deposits similar to that observed in unconfined settings (Prélat et al. 2009;Spychala et al. 2017).

Pinching out downslope
Packages are triangular in geometry and pinch out gradually. These architectures are very common at Gaztelugatxe Island (Fig. 3), where they consist of limestone breccia (Table 1) and have thinning rates of 6.7-10 cm/m downslope (Fig.  7E). Towards the top of Gaztelugatxe Island (closer to the contact with Gaztelugatxe Limestone) these architectures are amalgamated, whilst farther away from the limestone they are interspersed within tabular architectures. Successive thin-bedded, tabular deposits appear to onlap onto the topography formed by these downslope-thinning geobodies (Fig. 7E).
These deposits are interpreted as talus deposits, common around diapiric highs (Giles and Lawton 2002;Giles and Rowan 2012) and on fault scarps (Poprawski et al. 2014(Poprawski et al. , 2016. The similarity in facies and geometry to ''carbonate lentils'' described elsewhere (McBride et al. 1974;Hunnicutt 1998;Kernen et al. 2012Kernen et al. , 2018 and the likely close to the offshore Bakio diapir (Poprawski et al. 2016) suggest that these geometries are halokinetically driven. The source of this talus is interpreted to be the Gaztelugatxe Limestone due to its proximity and geometrical relationships. Onlap of successive deposits suggest that diapiric collapse was coeval with deep-marine deposition.

Undulose
Packages have an undulose, heterogeneous geometry (Fig.  7F). Individual beds vary in thickness and facies, and include thin beds, chaotic mud-rich debrites, and limestone breccias (Fig. 7F), but overall architecture maintains a broadly consistent thickness. The base of these architectures can be composed of limestone breccias (Fig. 7F, Table 1).
These remobilized units represent slump deposits. Ranging paleoflow directions, and both carbonate and siliciclastic inclusions, suggest they are derived from the diapir roof and flanks (Poprawski et al. 2014). The undulose geometries could overlie carbonate ''lentils'' or may reflect the reworking of ''lentils'' in these deposits (Fig.  7F).

Tabular amalgamated beds
Packages appear tabular and consist of beds which remain relatively consistent in thickness, with minor deviations related to previous topography (Fig. 7G). This architecture is principally composed of the channel-lobe transition zone facies association (Fig. 6). Concave depressions, which are spoon-shaped and meter scale in width, can be seen on bed tops and bed bases and are associated with undulations at bed scale (Fig. 7G). Overall the geometry is slightly concave up, with the center of each deposit thinning slightly on either side at the scale of the outcrop (Fig. 7G).
The dominance of channel-lobe-transition zone facies associations leads to an interpretation of a stacked, scoured, broad channel-lobe transition zone where erosional and depositional processes were active (Vicente Bravo and Robles 1991;Robles et al. 1995;Brooks et al. 2018). Spoon-shaped depressions are representative of megaflutes and scours (Robles et al. 1995). These cause a variable depositional topography which influenced subsequent flows, resulting in slight compensational stacking.  Figure 3. Transects of individual logs are separated by missing sections as highlighted, and are therefore not continuous. Similar sedimentary facies on either side of the fault suggest that GX3 continues on both sides of the structure, and therefore the impact of the structure is minor. Key for all the logs and pie charts shown. Thicknesses are in meters. GX# relate to stratigraphic units discussed in the text. Pie chart shows debrite data divided by predominant clast type and debrite type (Table 2), and relative proportions of all debrites at this section, and average thickness of each type is shown. BW1 marks the transition from BBF to BFG, and is interpreted as representing deposition at the base of slope of the carbonate platform, which was growing on the Bakio diapir ( Fig. 12D: Poprawski et al. 2014). Coarse-grained sandstones with lenticular geometries, scoured bases, and normal grading, such as those observed in BW2, indicate deposition in a channelized or scoured setting (Figs. 6F, 9, 12E; e.g., Hubbard et al. 2014;Hofstra et al. 2015). The debrites capping unconformities could be halokinetically-derived or related to channel-margin collapse induced by diapir movement (Rodriguez et al. in press). The wide range of clast rock types in these debrites suggests that they are dissimilar to other FIG. 9.-Variability of facies and architectures within large-scale depositional elements at Bakio West Bay, located in Figure 3. A, B) Stratigraphic logs, which are located in Part D and blown up for clarity. Thickness is in meters. Key is in Figure 8. BW# refers to stratigraphic unit discussed in the text. Dashed lines between logs highlight correlation. C) Pie chart showing debrite data divided by predominant clast type and debrite type (Table 2), and relative proportions of all debrites at this section, and average thickness of each type is shown. D) Uninterpreted and E) Interpreted large-scale architectures and facies details at Bakio West Bay. Unconformities are from Poprawski et al. (2014), and are highlighted in red where they divide packages of stratigraphy. U4 of Poprawski et al. (2014) is not laterally extensive and appears to represent an isolated, erosionally based depositional element (Fig. 9E). Based on these observations and the presence of channel-axis facies associations we suggest that U4 represents the base of a channel cut and not a halokinetic angular unconformity (sensu Giles and Rowan 2012). Black box in photograph locates the channel-axis facies association (Fig. 5F). F) Clasts of Gaztelugatxe Limestone which form out-runner blocks forming sea-floor topography. The deepest point of each lenticular geometry in BW3 appears to step eastward towards the Bakio diapir (Figs. 6F, 7A); this could indicate lateral-accretion deposits from a meandering submarine channel (e.g., Peakall et al. 2007;Kane et al. 2010;Janocko et al. 2013). The concaveupward geometry of the turbidites and the undulose contact with the debrite below (Figs. 6F, 7B, 9) could represent channel fills or scour fills influenced by previous debrite topography (e.g., Cronin et al. 1998;Jackson and Johnson 2009;Kneller et al. 2016). The thick beds, concaveupward geometry, erosional bases, and coarse grain size suggests that these deposits are channel fills rather than scour fills (Hubbard et al. 2008;Romans et al. 2011;McArthur et al. 2020).

SALT-INFLUENCED DEEP-WATER SUCCESSIONS
The repeated facies change between pebbly chaotic debrites and channelized turbidites is interpreted to represent periods of rapid diapir  Figure 3. Missing sections are indicated; thickness is in meters. Key for all logs is provided in Figure 8. CM# indicate stratigraphic units discussed in the text. Roughly 500 m of missing section separates South and North Cabo Matxitxako. The pie chart shows debrite data divided by predominant clast type and debrite type (Table 2), and relative proportions of all debrites at this section, and average thickness of each type is shown. Z.A. CUMBERPATCH ET AL. 50 J S R growth, evidenced by debrites overlying halokinetic unconformities . This is suggested to have been followed by periods of relative diapir quiescence, which permitted submarine channels to infill debrite topography and migrate around the diapir due to reduced seafloor topography ).

Cabo Matxitxako
Cabo Matxitxako provides an extensive section (c. 600 m) through the BFG. In this locality, we subdivide the group into eight lithostratigraphic units (CM1-8; Fig. 10). There is c. 500 m of missing section between Cabo FIG. 11.-Evidence for topography and paleoflow direction. Black arrow shows orientation and peach indicates scale, and lens cap is 52 mm. A) Rose diagram for 284 paleocurrent indicators (ripples, sole marks, cross-stratification) from the Black Flysch Group. Readings have been corrected for tectonic tilt yellow arrow indicates dominant paleoflow direction some radial spread is due to ripple reflection. Gray arrows indicate regional (to the south) and local (to the west) paleoflow directions, discussed in the text. B) Evidence for opposing-direction ripples suggesting ripple reflection. C) Triassic-age Keuper Group outcrop of clays, carbonate, and gypsum at Bakio Beach, thought to be part of the Bakio diapir. D) Composite halokinetic sequence associated with the western flank of the Bakio diapir; HS, halokinetic sequence. E) Pinch-out and onlap of lowermost Black Flysch Group thin-bedded turbidites onto a Gaztelugatxe Limestone clast on the eastern flank of the diapir. F) High-density turbidite abruptly terminating against a block of Gaztelugatxe Limestone in HS3 (BW3). Section is two times vertically exaggerated. Sollube and Jata Basins are indicated. Note change in orientation at the Bakio diapir.

J S R 53
Matxitxako North and South Beach (Fig. 3). Cabo Matxitxako is located in the Sollube basin, c. 4.5 km northeast and c. 5 km northwest of the Bakio and Guernica salt structures, respectively.
Description.-CM1 is a 110-meter-thick package of mudstones, with minor thin-to medium-bedded siliciclastic turbidites and debrites. CM2 is 60 m thick and is dominated by meter-scale debrites with subordinate thin-to medium-bedded turbidites, hybrid beds, and mudstones (Fig.  10). Slump axes, where present, indicate a range of paleoflow directions (forming two clusters: 80-1608 and 280-3208; Fig. 3). CM3 is 50 m thick and contains meter-thick packages of stacked medium to thick bedded, dewatered turbidites and slumps, meter to tens of meters thick containing rafts of thin-bedded turbidites. Two beds of granular sandstone (27 and 19 cm thick) with lenticular geometries are observed at the top of CM3. CM4 is 32 m thick, is distinguished from CM3 as it is debrite-poor and mostly comprises thick-bedded, amalgamated highdensity turbidites that stack into 3 to 6-meter packages. CM5 is 124 m thick and consists of roughly equal proportions of 1 to 3-meters-thick amalgamated turbidites, which are normally graded on a bed scale, and meter-scale slumps and debrites, which occur every 2-10 m and contain rafts of thin-bedded turbidites. CM6 has a minimum thickness of 18 m, is observed at the northern part of South Beach, and is composed of 1 to 3 m debrites and 1 to 80 cm turbidites and mudstones (Fig. 3, 10). CM7 has a minimum thickness of 38 m, similar to CM6; however, the units are separated by c. 500 m of missing stratigraphy, much of which is assumed to have been removed due to Pyrenean deformation and recent landslides (Vicente Bravo and Robles 1995), so have been separated. CM8 is 40 m thick and consists of predominantly meter-scale, normally graded thickbedded turbidites with erosional bases, cross-stratification, amalgamation, mud-clasts, and dewatering structures common throughout (Fig.  10).
Debrites and slumps with clasts of thin-and medium-bedded stratigraphy are present throughout CM2, 3, 5, 6, and 7 suggesting that the seafloor was periodically unstable, possibly due to relatively high rates of diapir rise and related seafloor deformation. CM1 is dominated by background suspension fallout and dilute low-density turbidites in a lobecomplex distal fringe setting (Fig. 12D). CM2 represents higher-energy, more proximal lobe conditions compared to CM1, based on facies, hybrid beds, geometry, stacking patterns, and thickness and is interpreted as proximal lobe fringe deposition ). The depositional sub-environment of CM3 is interpreted as an off-axis lobe complex, based on facies and bed thicknesses, with small distributive channel fills, evidenced by lenticular granular sandstones (Normark et al. 1979;Deptuck and Sylvester 2017). CM4 is dominated by stacked, amalgamated, highdensity turbidites (Fig. 10) and is interpreted to represent deposition in the axis of a lobe complex ( Fig. 12F; Prélat et al. 2009;Spychala et al. 2017). CM5 contains distributions of facies similar to those of CM3, so is interpreted to represent lobe-complex off-axis deposition with some distributive channel fills (Normark et al. 1979;Deptuck and Sylvester 2017). Debrite dominance in CM6 and CM7 suggests remobilization due to diapir growth throughout deposition (Fig. 12F). These deposits are interpreted to represent proximal-fringe to lobe-off-axis deposition , which is highly modulated by halokinetic debrites. The coarse grain size, cross-stratification, thick beds, and lack of debrites and mudstones (Figs. 6E, 10), suggests that CM8 was deposited in either a lobe-axis or channel-lobe transition zone. This unit has previously been interpreted as a channel-lobe transition zone (Vincente Bravo and Robles 1995) and shares characteristics similar to channel-lobe transition zones reported elsewhere (e.g., Brooks et al. 2018). This unit shows little evidence for active topography, suggesting that the sedimentation rate had increased with respect to the diapir rise rate, possibly associated with uplift of the Landes Massif (Rat 1988;Martín-Chivelet et al. 2002), or welding of the salt source layer. Any remaining seafloor topography was filled by CM8 (Fig. 12G).

EVIDENCE FOR SEAFLOOR TOPOGRAPHY
There is widespread sedimentological evidence for seafloor topography during deposition of the BFG, which as we discuss below reflects the interplay between active salt growth and depositional processes controlling the available accommodation.
Ripple cross-lamination, cross-stratification, and sole marks indicate a regional southwestward paleoflow direction (Fig. 11). This direction is consistent with a northward source for the BFG, supporting the Landes Massif as a potential regional source area (Rat 1988;Robles et al. 1988;Ferrer et al. 2008;Puelles et al. 2014). At Cabo Matxiatxako, a secondary westward paleoflow orientation (Fig. 11A) is comparable to findings by Robles et al. (1988), who suggest that this reflects the passage of gravity flows that spilled across the Guernica structure into the Sollube basin. Therefore, paleocurrent data (Fig. 11A) suggest modulation of a regional (primarily south-trending) paleoflow by salt-induced topography (westtrending flows off east-facing slopes) (Fig. 12). Analogously, a west-southwestward paleoflow observed at Bakio West Bay (Fig. 3) may reflect the passage of gravity flows that spilled from the Sollube basin, across the Bakio structure into the Jata basin. This west-south-westward paleoflow could alternatively represent the westward deflection of regional paleoflow around the Bakio diapir.
Turbidites that pinch out up depositional slope (Fig. 7C) reflect the thinning towards topography (e.g., Ericson et al. 1952;Gorsline and Emery 1959) as the low-density part of the turbidity current ran up the topography farther than the high-density component (e.g., Al-Ja'aidi 2000; Bakke et al. 2013). Hence, thicker sandstones are more likely to be confined to localized salt-withdrawal minibasins (Figs. 12, 13, 14), whereas thinner sandstones may drape halokinetically influenced slopes (Figs. 12, 13, 14;Straub et al. 2008;Soutter et al. 2019). Based on a bedscale thinning rate of 10 cm/m at Cabo Matxitxako (Figs. 3, 10), we calculate the slope angle to be 2-38 (Fig. 7C). Due to the distance (c. 5 km) from the present-day Bakio diapir, it is unlikely that this slope is solely related to diapir growth, but rather caused by a combination of salt withdrawal or welding from the salt source layer at depth, and saltstructure growth, as observed in similar-sized subsurface examples (e.g., Doughty-Jones et al. 2017).
Multiple paleoflow directions, hybrid beds, and abrupt pinch-out of beds can suggest the presence of (static) topography. However, the number of debrites intercalated with lobes, and the tapered composite halokinetic sequence observed at Bakio West Bay (Fig. 11D) suggest that this topography and salt growth was active at the time of deposition.

Debrites
Description.-Debrites, slumps, and slides (the vast majority being debrites) account for 23% of stratigraphy across all measured sections, with an average thickness of 111 cm. Assuming that all the debrites (Table  1 Soutter et al. (2019). One salt body has a carbonate roof and one a siliciclastic roof purely for clarity. Part D shows compensational stacking occurring during system progradation. Part E shows how progradation may be accelerated by parallel topography, based on flume tank experiments by Soutter et al. (2021) and Bakio West Bay. Part F shows how progradation is further accelerated as gravity-flow deposits are funneled through dual-confinement. Both siliciclastic and carbonate failures are shown in Parts E and F to indicate diapiric influence on axial deposition, as the depositional system shifts away from the salt structure to avoid the debrite topography. LDT, low-density turbidite; HB, hybrid bed; MDT, medium-density turbidite; HDT, high-density turbidite.

SALT-INFLUENCED DEEP-WATER SUCCESSIONS
J S R derived from far field, a 1:4 ratio of halokinetically to axially derived deposition exists. While this is clearly an oversimplification, and we provide evidence for local, diapirically derived debrites and regional, slope-derived debrites, it does suggest that locally, extra stratigraphy may be present in salt-confined basins compared to unconfined settings. Therefore interpreting stacking patterns and correlating between basins may be challenging.
Debrites on the flank of the Jata basin have an average thickness of 140 cm, compared to 119 cm and 73 cm in the axis and flank of the Sollube basin, respectively. 31% of measured stratigraphy on the flank of the Jata basin comprises debrites, compared to 22% and 18% in the flank and axis of the Sollube basin, respectively. Debrite composition shows siliciclastic dominance in the axis of the Sollube basin; debrites with carbonate clasts or matrix become more common on the flank of the Sollube basin and are dominant on the flank of the Jata basin (Figs. 8,9,10), in agreement with models of halokinetic sequences . Limestone clasts of similar composition in debrites on both the Jata and Sollube flanks support the presence of a carbonate platform growing on top of the Bakio diapir (García-Mondéjar and Robador 1987; Poprawski et al. 2014Poprawski et al. , 2016, indicating that these debrites are related to local diapir failures. The proportion of carbonate clasts decreases from the Sollube basin flank to axis (39% in Fig. 8 and , 1% in Fig. 10) due to distance from the carbonate platform growing on top of the Bakio diapir. Isolated limestone megaclasts (Figs. 9F, 11E, F) derived from the Gaztelugatxe Limestone could be out-runner blocks (e.g., De Blasio et al. 2006;Soutter et al. 2018) or fractured blocks of platform limestone that have toppled off during diapiric growth (e.g., Alves et al. 2002Alves et al. , 2003Martín-Martín et al. 2016). Younger deposits progressively onlap these limestone clasts (Fig. 11E, F), showing that subsequent turbidity currents later interacted with this additional seafloor topography (e.g., Kilhams et al. 2012Kilhams et al. , 2015Olafiranye et al. 2013).
Interpretation.-Debrites can be sourced from either collapse on top of the diapir or its flanks, or from failures of the shelf-edge and/or slope (Doughty- Jones et al. 2019;Rodriguez et al. in press;Wu et al. 2020). However, the source is difficult to decipher without a dataset covering the coeval shelf and slope succession. The presence of more debrites in the basal active part of the Sollube basin fill (Fig. 8, 9, 10, 13) compared to the upper passive part of the basin fill (Fig. 10) suggests that mass failures are more common during initial development of salt-confined depocenters (Wu et al. 2020). The Sollube basin is more confined than the Jata basin, and therefore can be expected to receive a larger proportion of material derived from diapir growth. In fact, the Jata basin received more and thicker debrites. This could suggest asymmetric growth of the Bakio diapir, or the carbonate platform above it, creating a steeper slope on the boundary of the Jata basin. However, the apparent difference in the number and thickness of debrites could also be due to limitations of the outcrop, in that the Sollube basin flank location (Gaztelugatxe Island) is more axial than the Jata basin flank outcrop (Fig. 3). The absence of carbonate clasts in the axis of the Sollube basin, compared to the Sollube and Jata basin flanks, could suggest that diapirically derived mass failures did not extend into the axis of the basin. However, the exposed minibasin stratigraphy is younger than the flanking stratigraphy, and therefore carbonate clasts are expected to be rarer due to a decrease in exposure and availability to be remobilized through time.
The presence of complicated variations in thickness, clast type, and debrite style across the study area suggests that mass flows are likely to be both locally derived (from the salt structures) and regionally derived (from the shelf), and therefore debrites were likely both allogenic and halokinetic in origin (Moscardelli and Wood 2008;Doughty-Jones et al. 2019;Wu et al. 2020). The proximity to salt structures, reduction in mass-failure deposits away from salt structures, and the remobilized intrabasinal clasts within them suggests that the majority of debrites in the study area are locally derived from the Bakio diapirs. The wide variety of clasts types in Jata basin debrites (Table 3: Fig. 9) and the association of these deposits with channelized turbidites above could suggest that some of the Jata basin debrites were regionally derived, and deposited as part of channel-axis depositional elements. The absence of these possible regionally derived debrites in the Sollube basin could be due to confinement by the offshore part of the Bakio diapir, or different up-dip sediment routing. The difference in thicknesses and numbers of debrites between the Jata and Sollube basins suggests that they developed, at least partially, as separate depocenters influenced by different controls (Figs. 8, 9, 10).

DISCUSSION
The discussion initially focuses on the Sollube basin, then compares the Sollube and Jata basins, before comparing our observations to similar depocenters developed in other salt-influenced basins.

Architecture of the Sollube Basin
The Sollube basin is of a similar size and geometry to previously reported subsurface minibasins ( Fig. 13; e.g., Pratson and Ryan 1994;Booth et al. 2003;Madof et al. 2009;Doughty-Jones et al. 2017). Therefore, this rare, exhumed example of a halokinetically influenced deep-water succession provides an excellent exposure of fine-scale minibasin depositional architecture, providing an analogue for subsurface minibasins.

Distribution of Facies and Architecture
The Sollube basin is 8 km wide and confined to the east and west by the Guernica and Bakio structures, respectively ( Fig. 4; Robles et al. 1998;Poprawski and Basile 2018). Repeated stratigraphy and facies distribution on either side of Cabo Matxitxako, the change in bedding angle between the LBF and UBF, and sedimentological evidence for syndepositional topography support the presence of a broadly symmetrical basin confined by the Bakio and Guernica structures.
Early stratigraphy in the siliciclastic fill of the Sollube basin is dominated by thin-bedded sandstones, with localized debrites on the flanks (Figs. 8,10,13). As the basin developed, thicker-bedded sandstones representing channel fills and lobes were deposited in topographic lows (basin axis), consistent with subsurface analogues (e.g., Booth et al. 2003;Madof et al. 2009;Mayall et al. 2010;Doughty-Jones et al. 2017) and numerical models (e.g., Sylvester et al. 2015;Wang et al. 2017). Towards the flanks, the lower-density parts of the flows responsible for the thickbedded sandstones may run up topography, depositing thin-bedded sandstones. Therefore, allocyclically controlled, axially derived, and often the thickest stratigraphy is observed in the axis of the minibasin. Halokinetically controlled (e.g., debrites) or -influenced (e.g., thickness variations) stratigraphy occurs towards the basin margins. These interpretations are consistent with subsurface studies (e.g., Doughty-Jones et al. 2017;Rodriguez et al. in press;Wu et al. 2020). Oluboyo et al. (2014) suggest that a fundamental control on the type of confinement developed is the incidence angle between the strike of the salt structure and the paleoflow direction. ''Fill-and-spill'' architecture is observed in deep-water environments where topographic highs strike perpendicular to the gravity-flow direction (i.e., at a high incidence angle) (e.g., Piper and Normark 1983;Hay 2012;Prather et al. 2012;Soutter et al. 2019). This study documents a rare example of an exhumed halokinetically influenced deep-water succession where paleoflow is at a low incidence angle to structural strike (i.e., oblique-parallel). In such scenarios, spill between basins is rare, and sedimentary systems are deflected to run broadly parallel to salt walls in minibasins for several kilometers (Figs. 13, 14, 15;e.g., Oluboyo et al. 2014 (Sinclair and Tomasso 2002) is not appropriate where paleoflow is oblique or parallel to salt structures, and there is a down-dip exit (i.e., not confined in all directions). Our study indicates that the dominant style of interaction between gravity flows and topography is lateral confinement, with channels and lobes in the Sollube basin being thickest in the axis and elongated parallel to the margins of the saltcontrolled basin. The presence of MTDs is somewhat overlooked in both the confinement model proposed by Oluboyo et al. (2014), and the earlier ''fill-and-spill'' model (Winker 1996;Prather et al. 1998;Sinclair and Tomasso 2002). MTDs sourced from either up-dip (i.e., extrabasinal; detached) or from local mass failures (i.e., intrabasinal; attached) related to growing salt structures, can generate additional syndepositional relief and flow confinement (Fig. 14;Moscardelli and Wood 2008;Rodriguez et al. in press).
The hierarchical scheme for classifying deep-water systems developed in the Karoo basin (Prélat et al. 2009) is widely accepted, but must be used with caution, or be adapted for confined systems (Prélat et al. 2010;Etienne et al. 2012;Marini et al. 2015). Prélat et al. (2010) recognized that width-tothickness ratios and areal extent-to-thickness ratios are different for confined and unconfined systems, with width-to-thickness ratios of 100:1 measured in selected subsurface confined settings, compared to 1000:1 in unconfined settings, and areal extent-to-maximum-thickness ratios 30 times greater in unconfined systems compared to confined systems (Prélat et al. 2010). All examples used in Prélat et al. (2010) are from settings where paleoflow is perpendicular (high angle) to topographic strike.
In terms of areal extent, stratigraphy in the Sollube basin (c. 8 km wide) would be best classified as a lobe element (c. 5 km wide) using the Prélat et al. (2009)  This suggests similar basic geometries for confined lobes regardless of incidence angle between paleoflow and topographic strike. The dimensions observed in the salt-influenced minibasins in the study area are comparable to those of intraslope lobes complexes documented in the Karoo (6-10 km wide, 10-15 m thick; Spychala et al. 2015), which are an order of magnitude smaller than their basin-floor counterparts (Prélat et al. 2009).
Confined-lobe complexes are predicted to have smaller areal extents because radial spreading is minimal due to topographic confinement (e.g., Marini et al. 2015;Soutter et al. 2019). The ratio of axis to fringe deposition is likely to be increased in confined settings where flows stay turbulent for longer, flow deceleration is limited, and development of a wide fringe is hindered due to presence of topography (Etienne et al. 2012;Soutter et al. 2019). The presence of thicker axial deposits due to confinement by topography, and less space for lateral lobe switching to occur, may make axis definition easier in a confined setting; however, this may change through time if topography is healed and depositional systems become less confined (Marini et al. 2015).
This study supports recent work (Oluboyo et al. 2014;Rodriguez et al. in press) which suggests that elongate systems are common adjacent to topography, on all scales. This is in contrast with the roughly equant geometries predicted in unconfined systems (Prélat et al. 2009). Multiscale analysis suggests that all confined hierarchical elements would have greater lengths than widths, and lesser areal extents and greater thicknesses (lower aspect ratios) than unconfined settings when deposited adjacent to oblique-parallel topography (Booth et al. 2003;Oluboyo et al. 2014;Rodriguez et al. in press).

Development of Axial Systems
The overall upward thickening of beds and coarsening of grain size from south to north at Cabo Matxitxako is associated with a transition in depositional environment from lobe distal fringe to channel-lobe transition zone (Figs. 10,12). Coarsening and thickening upwards elsewhere is widely interpreted to represent progradation (e.g., Mutti 1974;Macdonald et al. 2011;Kane and Pontén 2012); however, this could also represent (a component of) lateral stacking of lobes (e.g., Prélat and Hodgson 2013).
Throughout the Pyrenean, the BFG is interpreted to have been controlled by allocyclic progradation (Robles et al. 1988;Agirrezabala and Bodego 2005), driven by increases in sediment supply following the uplift of the Landes Massif (García-Mondéjar et al. 1996;Martín-Chivelet et al. 2002;Puelles et al. 2014) and/or increased flow efficiency from confinement ). This regional progradation, along with our field observations, provides compelling evidence that, on a lobe to lobe-complex scale, the stratigraphic architecture of the study area is controlled primarily by system progradation. Lateral switching may be reduced due to confining topography decreasing the amount of space available for switching to take place (e.g., Mayall et al. 2010;Oluboyo et al. 2014).
By definition, only two lobe elements would be able to fit laterally within the Sollube basin during LBF deposition (Prélat et al. 2009), suggesting that due to confinement lateral stacking in our study area is feasible only at the bed scale (Marini et al. 2015). The apparent retrogradation observed between CM5-7 could be a result of bed to lobe-scale lateral shifting and compensational stacking modulating an otherwise progradational lobe complex (e.g., Gervais et al. 2006;Pickering and Bayliss 2009;Etienne et al. 2012;Morris et al. 2014).
The lack of space for lateral stacking to occur suggests that lateral topography reduces flow loss to overspill and deceleration, and therefore the system remains turbulent for longer. This causes a basinward shift in deposition effectively acting to magnify the effects of progradation (e.g., Kneller and McCaffrey 1999;Talling et al. 2012;Patacci et al. 2014). This concept accounts for numerous, thick, high-density turbidites (UBF) along the axis of the Sollube basin formed by gravity flows that were funneled between the two structures (Figs. 13, 14C, F, 15; e.g., Scott et al. 2010;Oluboyo et al. 2014;Counts and Amos 2016;Howlett et al., in press).
Using observations from Bakio, a range of stacking patterns may form during progradation of a deep-water system in an unconfined, partially confined, and dual-confined setting (Fig. 15). Unconfined fans have a higher aspect ratio, surface area, and avulsion angle than confined systems as the ability of the flows to spread radially was not restricted by topography (Prélat et al. 2009;Spychala et al. 2017). Where only one lateral confinement is present, deposits may be asymmetrical, as flows are confined by diapir topography in one direction but are able to spread radially away from it, as is seen in the deposits of the Jata basin ( Fig. 9; Soutter et al. 2021).
In settings with lateral confinement, deep-water systems are elongated axially, subparallel with bounding relief (Figs. 13, 14, 15;Oluboyo et al. 2014;Soutter et al. 2021). Funneling of axial gravity flows, and therefore bed amalgamation, is interpreted to be more enhanced where two salt structures are present, resulting in thicker deposits but areally smaller depositional architectures than unconfined settings (Kneller and McCaffrey 1999;Patacci et al. 2014;Soutter et al. 2021).

Active or Passive Topography
The geometry and number of the debrites, and thin-bedded sandstones that pinch-out, is controlled by the presence of actively growing topography during LBF deposition. The absence of debrites in the UBF suggest that diapir growth ceased following uplift of the Landes Massif (García-Mondéjar et al. 1996;Puelles et al. 2014).
We infer that, following the cessation of diapir growth, an underfilled synclinal basin remained due to remnant topography of the buried Bakio and Guernica structures, which appears to have constrained UBF deposition until it was filled (Figs. 3, 12G, 13). The LBF represents early-stage ''active'' deposition, perhaps comparable to synkinematic megasequences observed in the subsurface, whereas the UBF represents late-stage ''passive'' deposition, infilling antecedent topography, comparable to postkinematic megasequences observed in the subsurface (e.g., Pratson and Ryan 1994;Warren 1999Warren , 2006Jackson and Hudec 2017). UBF deposits are less confined due to the lesser influence of saltinfluenced topography during deposition, and therefore through time may evolve to represent semiconfined or unconfined deep-water systems, more capable of lateral stacking (Marini et al. 2015).

Comparison of Sollube and Jata Basins
Different facies distributions on either side of the Bakio diapir varied during BFG times according to the degree of confinement and distance from diapir crest (Figs. 12,13). In the BBF, clast-and matrix-supported breccias are common in the Sollube and Jata basins, respectively (Poprawski et al. 2016), suggesting that minibasin individualization is long-lived. The lack of confining topography to the west of the Jata basin may have caused flows to dilute, resulting in muddier, more-matrix-rich breccias (e.g., Hampton 1972;Sohn et al. 2002;Baas et al. 2009).
LBF stratigraphy in the Jata basin thins towards the Bakio diapir (Fig.  11D), showing more evidence for topography than comparable strata in the Sollube basin, supporting the idea that halokinetic deformation is greatest closest to salt structures (Giles and Lawton 2002;Giles and Rowan 2012). Richness in limestone clasts in LBF MTDs in the Jata basin (Fig. 9C, Table  3) could indicate asymmetric buildup and failure of the carbonate platform above the Bakio diapir, preferentially filling the Jata basin (Rosales and Pérez-García 2010;Li et al. 2016). Asymmetric growth and failure of either the carbonate platform or the diapir itself is further supported by thicker, more prevalent debrites observed in the Jata basin compared to the Sollube basin (discussed above: Figs. 8,9,10). The diverse range of clast types in the Jata basin (Table 3; Fig. 9) could suggest different depositional routing relative to the Sollube basin, possibly due to the presence of salt topography causing different source areas to be tapped (e.g., Mayall et al. 2010;Oluboyo et al. 2014).
Another key difference is the architecture of thick-bedded sandstones. In the Jata basin, individual depositional elements are often erosionally based, concave-upwards, and thinner, and show more tractional structures (e.g., ripple lamination and planar lamination) than those in the Sollube basin. Where lateral confinement occurs along one margin, sandstones could represent sinuous low-relief channel fills that ran subparallel to topography (e.g., Mayall et al. 2010;Oluboyo et al. 2014). These channels were able to spread laterally and migrate because they were only partially confined (e.g., Mayall et al. 2010;Oluboyo et al. 2014). Such systems are less modulated by halokinetic controls than those that develop under duallateral confinement (e.g., Oluboyo et al. 2014;Rodriguez et al., in press). Subsurface observations indicate that channels migrate away from growing structures (Mayall et al. 2010;); however, those at Bakio West Bay appear to step towards the diapir in 2D (Figs. 6F, 7B). This could suggest a periodic reduction in diapir growth, due to the episodic nature of halokinesis , which may be unresolvable in subsurface data. Alternatively, the debrites underlying the thick-bedded sandstones could form pathways that controlled sandstone deposition, and therefore the apparent movement towards the diapir was in fact controlled by the deep-water systems infilling debrite-related paleotopography (Armitage et al. 2009). However, deciphering detailed interpretations of 3D sinuous channels from 2D exposure remains challenging (Li et al. 2016).
The Jata and Sollube basins had unique tectonostratigraphic histories throughout the deposition of the BBF and LBF due to the interplay of halokinetic, allocyclic, and autocyclic controls. Inaccessible UBF stratigraphy to the west of Bakio (Fig. 11D), from UAV photographs, appears to exhibit facies and geometries similar to UBF stratigraphy at northern Cabo Matxitxako (Fig. 10: Vincente Bravo and Robles 1991aRobles , 1991bRobles , 1995 suggesting that influence of halokinesis decreased through time. This supports the idea that sediment accumulation rate ultimately outpaced diapir growth rate, possibly due to the increase in sediment supply associated with the uplift and erosion of the Landes Massif (Martín-Chivelet et al. 2002;Puelles et al. 2014). Partial or complete welding of salt bodies could also be, at least partially, responsible for the reduction of influence of halokinesis through time (Jackson and Hudec 2017).

Comparison to Other Depositional Settings
Here, for ease of comparison with subsurface examples, we use masstransport deposit (MTD), instead of debrite, to describe deposits from varied subaqueous mass flows, including a mixture of slides, slumps, and debris flows in agreement with previous subsurface studies (e.g., Nardin et al. 1979;Posamentier and Kola 2003;Pickering and Hiscott 2015;Doughty-Jones et al. 2019;Wu et al. 2020).
Deposition of thick-bedded sandstones along the axis of the Sollube basin, and thinner beds and mudstones on the flanks of the Sollube and Jata basins, is comparable to fluvial facies distribution (Banham and Mountney 2013a, 2013b, 2014Ribes et al. 2015) where channel-fill sandstones dominate axial settings and floodplain mudstones are observed closer to the diapir.
Individual beds in the BBF are comparable in size (tens-to hundreds-ofmeters packages) and composition to stacked MTDs reported overlying bounding unconformities in halokinetic sequences in the La Popa Basin (10-120 m in thickness) associated with remobilization of diapir roof or cap rock (Giles and Lawton 2002;Poprawski et al. 2014Poprawski et al. , 2016. Smaller carbonate breccias with wedge-shaped geometries (meter-scale packages; Fig. 7E) are similar in geometry and composition to ''lentils'' (1 meter to 100s of meters thick) described by McBride et al. (1974), but differ in thickness and areal extent. Lentils, MTDs, and breccias represent talus-like failure from diapir roof stratigraphy (Giles and Lawton 2002;Poprawski et al. 2014Poprawski et al. , 2016.
Presence of evaporite clasts in fluvial successions (Banham and Mountney 2013b; Ribes et al. 2015), suggest that the nearby diapir was exposed during deposition. Such clasts are not observed in our study area, suggesting that the Bakio and Guernica structures may have only episodically been exposed at the seabed, if at all. This fits the interpretation of carbonate-platform growth above the structures, preventing salt SALT-INFLUENCED DEEP-WATER SUCCESSIONS exposure (García-Mondéjar 1990; Rosales and Pérez-García 2010;Poprawski et al. 2014Poprawski et al. , 2016.
The consistency of our observations and previously described halokinetically influenced settings suggests that the criteria for recognizing halokinetically influenced systems is similar regardless of depositional environment, suggesting that halokinetic controls are dominant over allocyclic ones. Multiple directions of ripple lamination, presence of hybrid beds, range of MTD types, and abrupt juxtaposition of deep-water depositional facies can be used to identify halokinetically influenced deepwater systems in core and outcrop.

Comparison between Quickly and Slowly Deforming Basins
The gravity flows responsible for the Eocene to Oligocene Annot Sandstone, SE France, were confined during deposition by Alpine foldand-thrust belt topography (Apps 1987;Sinclair 1994). When compared with the rate of topographic deformation associated with diapir growth, the rate of orogenic deformation is more minor.
Like the BFG, the stratigraphy of the Grès d'Annot is broadly progradational, and rapid facies changes occur over tens of meters towards pinch-outs (Soutter et al. 2019). Unlike Bakio, where paleoflow was consistently at a low angle to structural trend, sub-basins in the Annot area were eventually filled and bypassed sediment into down-dip basins (Sinclair and Tomasso 2002;Salles et al. 2014), indicating that paleoflow was perpendicular to at least one of the complex structural trends (Oluboyo et al. 2014).
Debrites in the Grès d'Annot were slope-derived and infrequent compared with our study area, which were sourced laterally from failures of stratigraphy above salt structures that are intercalated with axially derived deep-water deposits (Figs. 12,13,14,15). This is a reflection of a more active slope in diapiric settings.
The stratigraphy in this study is characterized by an axial deep-water depositional system and a series of lateral systems dominated by debrites fed from the growing salt structures. This interplay of two distinct depositional systems is common in deep-water environments influenced by active rift topography, such as the Gulf of Corinth, Greece (Leeder and Gawthorpe 1987;Pechlivanidou et al. 2018;Cullen et al. 2019) and the Gulf of Suez, Egypt Jackson et al. 2002Jackson et al. , 2005Leppard and Gawthorpe 2006). Here the continually evolving footwall scarps feed lateral MTD-rich systems coevally with axial, allocyclically controlled depositional systems. Deposits in synrift settings are often narrow and elongated parallel to the strike of normal-fault segments (Carr et al. 2003;Jackson et al. 2005;Cullen et al. 2019), indicating the control on stratigraphic architecture by footwall physiography. Analogous variability of depositional facies occurs due to salt-structure evolution in halokinetically influenced settings.

CONCLUSIONS
This study documents deep-water facies distributions, with variable amounts of topographic confinement, adjacent to growing salt structures from a rare exposed example. We compare observations from two minibasins, one confined (Sollube) and one partially confined (Jata), which are comparable in size and facies heterogeneity to subsurface minibasins in salt provinces globally.
Stratigraphic variability and juxtaposition of architectural elements in the Jata and Sollube basins is high and controlled by the interplay of halokinetic, autocyclic, and allocyclic processes. The low angle between the paleoflow and the strike direction of salt structures results in the depositional system being focused between two salt structures in the Sollube basin, and against one salt structure in the Jata basin, but with no evidence of downdip flow confinement. Confinement against topography increases the effects of allocyclic progradation. Failure of carbonate platforms developed above the crests of the active Bakio diapir and Guernica structure created lateral debrites in the flanking basins, and generated local topography to further constrain axial depositional systems. Debrites can also be sourced axially from up-dip failures on the shelf, and the compositional differences of debrites suggests that the Jata and Sollube basins were subject to different influences.
Indicators of active topography include hybrid beds, remobilized strata, lateral thickness changes over short distances, reversal in ripple crosslamination in beds and intercalation of debrites throughout the stratigraphy. These indicators individually are not diagnostic of salt-influenced topography, but collectively they provide a set of features that support interpretation of halokinetic modulation of a deep-water setting. Following the cessation of diapir growth, topography does not heal instantly, and the ''passive'' paleotopography continues to confine subsequent depositional systems despite diapir inactivity.
Closely related depositional systems can be highly variable depending on their complete or partial confinement. Stacked, amalgamated sandstones are observed between the confining barriers in the Sollube basin, whereas more variable architectures are observed in the Jata basin, where only partial confinement is present. These observations are due to the modulation of a broadly progradational system by halokinetically influenced lateral barriers and the coeval development of axial allocyclic and lateral debrite-rich depositional systems.
Utilizing outcrop analogues combined with a good regional understanding of source area and salt movement and extracting insights from depositional analogues are advised when exploring in the salt-sediment interface for carbon storage or geothermal or hydrocarbon reservoir targets.