Evolution of a mixed siliciclastic‐carbonate deep‐marine system on an unstable margin: The Cretaceous of the Eastern Greater Caucasus, Azerbaijan

Mixed siliciclastic‐carbonate deep‐marine systems (mixed systems) are less documented in the geological record than pure siliciclastic systems. The similarities and differences between these systems are, therefore, poorly understood. A well‐exposed Late Cretaceous mixed system on the northern side of the Eastern Greater Caucasus, Azerbaijan, provides an opportunity to study the interaction between contemporaneous siliciclastic and carbonate deep‐marine deposition. Facies analysis reveals a Cenomanian–early Turonian siliciclastic submarine channel complex that abruptly transitions into a Mid Turonian–Maastrichtian mixed lobe‐dominated succession. The channels are entrenched in lows on the palaeo‐seafloor but are absent 10 km towards the west where an Early Cretaceous submarine landslide complex acted as a topographic barrier to deposition. By the Campanian, this topography was largely healed allowing extensive deposition of the mixed lobe‐dominated succession. Evidence for irregular bathymetry is recorded by opposing palaeoflow indicators and frequent submarine landslides. The overall sequence is interpreted to represent the abrupt transition from Cenomanian–early Turonian siliciclastic progradation to c. Mid Turonian retrogradation, followed by a gradual return to progradation in the Santonian–Maastrichtian. The siliciclastic systems periodically punctuate a more widely extensive calcareous system from the Mid Turonian onwards, resulting in a mixed deep‐marine system. Mixed lobes differ from their siliciclastic counterparts in that they contain both siliciclastic and calcareous depositional elements making determining distal and proximal environments challenging using conventional terminology and complicate palaeogeographic interpretations. Modulation and remobilisation also occur between the two contemporaneous systems making stacking patterns difficult to decipher. The results provide insight into the behaviour of multiple contemporaneous deep‐marine fans, an aspect that is challenging to decipher in non‐mixed systems. The study area is comparable in terms of facies, architectures and the presence of widespread instability to offshore The Gambia, NW Africa, and could form a suitable analogue for mixed deep‐marine systems observed elsewhere.

Mixed systems developed in deep-marine (below storm wave base) settings are usually formed of material shed from a shallower carbonate-producing shelf that periodically also received terrigenous input (Figure 1) (Crevello & Schlager, 1980;Dunbar & Dickens, 2003;Mount, 1984). This material is deposited in the deep-marine by a spectrum of sediment gravity flows types, from turbidity currents to submarine landslides (Dorsey & Kidwell, 1999;Miller & Heller, 1994;Moscardelli et al., 2019;Tassy et al., 2015). Generic concepts have been proposed in the literature for mixed systems and their potential relationship with deep-marine systems (e.g., Francis et al., 2008;Francis, Dunbar, Dickens, Sutherland, & Droxler, 2007;McNeill, Cunningham, Guertin, & Anselmetti, 2004); while valuable, these studies rely on limited exposures and incomplete subsurface data sets leading to models that do not account for the vertical and lateral stratigraphic variabilities observed in recent subsurface studies of deep-marine mixed systems (Casson, Calvès, Redfern, Huuse, & Sayers, 2020;Moscardelli et al., 2019).

K E Y W O R D S
Azerbaijan, Caucasus, deep-marine, mixed system, siliciclastic-carbonate

Highlights
• We document the evolution of a mixed siliciclastic-carbonate deep-marine system in the Eastern Greater Caucasus of Azerbaijan. • A siliciclastic submarine channel complex abruptly transitions into an overlying mixed siliciclastic-carbonate lobe succession. • Deep-marine depositional systems were affected by submarine landslide topography during deposition. • Interaction between the two contemporaneous systems makes typical lobe stacking patterns difficult to decipher. • The evolution of a mixed siliciclastic-carbonate deep-marine system offshore The Gambia, NW Africa is suggested as an offshore analogue.

Caucasus
The EGC forms the easternmost extent of the NW-SE trending Greater Caucasus orogenic belt, which runs from the Black Sea in the west to the Caspian Sea in the east ( Figure 2) (Mosar et al., 2010;Phillip, Cisternas, Gvishiani, & Gorshkov, 1989). The EGC sits on the southern-edge of the Scythian Platform, which represents the southern margin of the Eastern European continent ). The exposed EGC is mainly composed of Jurassic and Cretaceous deep-marine sediments that accumulated within the eastern strand of the Greater Caucasus Basin. Multiple phases of extension and compression are recorded in the basin fill and were driven by the sequential closure of the Tethys Ocean whose active margin was situated farther to the south (Adamia et al., 2011;Golonka, 2004;Nikishin et al., 1998Nikishin et al., , 2001Vincent, Braham, Lavrishchev, Maynard, & Harland, 2016;Vincent, Morton, Carter, Gibbs, & Barabadze, 2007). Late Triassic-Early Jurassic compression was followed by the main rifting phase in the evolution of the Greater Caucasus Basin, in the Early-to Mid-Jurassic (Mosar et al., 2010;Nikishin et al., 2001;Vincent et al., 2016). This tectonic event is recorded by major thickness variations across the Middle Jurassic interval (Bochud, 2011).
The Lower Cretaceous succession in the EGC was deposited within an unstable marine environment, as recorded by frequent mass-wasting events (Bochud, 2011;Egan, Mosar, Brunet, & Kangarli, 2009). This was triggered by renewed rifting (e.g., Vincent et al., 2016;Vincent et al., 2018). The rifting was potentially associated with opening phases in the Black Sea Basin farther to the west, although the exact timing of Black Sea Basin opening is still debated (e.g., Maynard & Erratt, 2020;Nikishin, Okay, et al., 2015;Nikishin, Pokay, et al., 2015;Sosson et al., 2016). This resulted in deep-marine deposition of extensive mudstones interspersed by submarine landslide deposits and terrigenous sediments.
The remainder of the Mid and Upper Cretaceous sequence was deposited during a period of thermal subsidence on a southward-dipping slope (Bochud, 2011). The Mid-Upper Cretaceous stratigraphy is dominated by calcareous and siliciclastic turbidites, and mass failure deposits, interbedded with hemipelagic marls and mudstones. A number of intra-Cretaceous unconformities occur within the Greater Caucasus Basin and are related to periods of compression (Egan et al., 2009) or sea-level fluctuations. The Cretaceous sequence is capped by a base-Cenozoic unconformity that may have formed during Paleogene compression (Bochud, 2011).
Cenozoic collision of the Arabian and Eurasian plates inverted the Mesozoic and early Cenozoic basinal succession (Mosar et al., 2010;Vincent et al., 2007Vincent et al., , 2016. In the EGC a series of exhumed synclines are bound by major southerly verging thrusts and associated back thrusts (Mosar et al., 2010). These faults separate distinct structural zones, including the Qonaqkend Zone in the northern EGC ( Figure 2) (Bochud, 2011).

| The Buduq Syncline
This study focuses on Late Cretaceous strata exposed in the Buduq Syncline, or Buduq "Trough" (Bochud, 2011), which is located between the NW and SE striking thrust faults that bound the Qonaqkend Zone (Figure 3), and encompasses the villages of Buduq, Cek and Qonaqkend ( Figure 2). Late Cretaceous strata have been interpreted to be deposited in a "palaeo-valley" incised into Early Cretaceous deep-marine sediments and Late Jurassic limestones (Bochud, 2011) following a period of Cretaceous compression (Figure 2b) (Egan et al., 2009 conglomerates (the Kemishdag Formation) that crop out in the east (Kapaevitch, Beniamovskii, & Bragina, 2015;. These sediments conformably overlie Aptian-Albian sediments at Mt. Kelevudag (Kopaevich et al., 2015) but rest unconformably on Barremian strata at Khirt (Figure 2). The overlying mid Turonian-Maastrichtian succession is represented by mixed turbidites and is conformable with the Cenomanian-early Turonian interval in the east. In the west, near Cek, the Cenomanian-Santonian interval is absent, with Campanian strata directly overlying Aptian-Albian thinly bedded depsoits. Campanian-Maastrichtian strata consist of submarine landslide deposits, comprising remobilised Upper Jurassic blocks, and mixed siliciclastic-carbonate turbidites (Bochud, 2011;Kopaevich et al., 2015). Oceanic red beds (CORBs) occur throughout the Upper Cretaceous sequence, particularly in Coniacian-Campanian turbidites and marls, indicating periodically oxic deep-marine conditions (e.g., Hu et al., 2005). The overlying Paleogene succession has largely been removed by Cenozoic compression and erosion (Bochud, 2011), which folded the remaining Cretaceous into a shallow syncline (Figures 2 and 3). Recent erosion of the study area has carved a number of narrow valleys through this syncline exposing the Cretaceous succession ( Figure 2).

| DATA AND METHODS
The data set comprises 23 sedimentary logs, totalling 500 m, collected across the Buduq Syncline (see Supporting Information). Logs were generally collected at 1:25 scale. Bedding, structural data ( Figure 3) and palaeocurrent data ( Figure 4) were collected to ground-truth the geological F I G U R E 3 Equal area stereographic projection showing bedding orientations for Cretaceous strata across the study area. Bedding planes shown as lines and poles to bedding shown as dots. Coloured by stratigraphy and location; LC, Lower Cretaceous (Aptian-Albian and Cenomanian-early Turonian stratigraphy); UC, Upper Cretaceous (mid Turonian and younger). Structural data reveal shallow-moderate structural dips to the north-northeast and south-southwest, in agreement with the east-southeast-west-northwest trending structural zones of the Eastern Greater Caucasus 618 | EAGE CUMBERPATCH ET Al. map and cross-sections of Khain and Shardanov (1960) and Bochud (2011). Palaeocurrent readings were quite rare and were taken only where sedimentary structures were clear enough to permit unambiguous data collection. Sparse biostratigraphic data hinders precise correlation across the study area. Chrono-stratigraphic subdivisions of the Buduq Syncline stratigraphy are still being refined (cf. Bochud, 2011;Bragina & Bragin, 2015;Khain & Shardanov, 1960, Kopaevich et al., 2015, due to the litho-stratigraphic similarities between the units and the complex palaeobathymetry in which they were deposited (Egan et al., 2009). Therefore, we use mapped stratigraphic units (J 1 , J 2 , K 1 , K 2 etc.), lithostratigraphy and cross-cutting relationships to suggest associated ages. Combining the work discussed above, we group stratigraphy into early (Aptian-Albian), mid (c. Cenomanian-early Turonian) and late (c. mid Turonian and younger) Cretaceous units. Sedimentary logs were used to develop a lithofacies scheme ( Figure 5, Table 1) and facies associations ( Figure 6).
Over 10,000 sedimentological measurements (e.g., bed thickness, grain-size, facies) were collected and quantitatively analysed (see Supporting Information). Stratigraphic logs were assigned one of seven facies associations ( Figure 6) in order to quantitatively compare bed statistics across deep-marine sub-environments (Figures 7-10).

| Lithofacies
Carbonate and siliciclastic lithofacies presented in Table 1 and Figure 5 represent beds deposited by individual events (event beds) and are classified based on outcrop observations. For simplicity and ease of comparison to their siliciclastic gravity flow counterparts, here we define calcareous mudstones/siltstone/sandstones as resedimented mudstones-wackestones comprising detrital carbonate grains in accordance with the Dunham classification of limestones (Dunham, 1962). Grain-sizes range from silt-to very fine sand-sized in a micritic matrix. Lithofacies show evidence for bed-scale stratal mixing (sensu Chiarella et al., 2017) (Table 1).

| Facies associations
Facies associations have been interpreted based on the dominant lithofacies ( Figure 5, Table 1) and architecture of a given succession and are subdivided into siliciclastic and mixed associations ( Figure 6). Letters in brackets refer to lithofacies described in Table 1. Facies associations (FA) 1 is Aptian-Albian (early Cretaceous) in age, 2-3 are Cenomanian-early Turonian (mid Cretaceous) in age and FA 4-7 are mid Turonian-Maastrichtian (late Cretaceous) in age. Facies association nomenclature commonly used for lobes (Prélat et al., 2009;Spychala, Hodgson, Prélat, et al., 2017) and channels (Hubbard, Covault, Fildani, & Romans, 2014;Kane & Hodgson, 2011) best fit our observations. Observations. FA 1 is dominated by metre-scale packages of thin-bedded siliciclastic siltstones to fine-grained sandstones (H) with subordinate mudstones (J) and mediumbedded siliciclastic sandstones (F) (Table 1, Figure 6a). Beds are laterally extensive for 100's of metres and are commonly flat based and flat topped, often being normally graded from fine-grained sandstone to siltstone. Planar and convolute laminations are observed in the upper part of many beds ( Figure 5e). Poorly sorted clast-rich deposits, bi/tri-partite beds, conglomerates and thick-bedded sandstones are absent. Deposition from debris flows having cohesive as well as frictional strength (Fisher, 1971;Nemeck & Steel, 1984). The grading of conglomerates into thick-bedded sandstones reflects the transition of hyper-concentrated submarine debris flows into highly concentrated turbulent flows (Mulder & Alexander, 2001;Sohn, Choe, & Jo, 2002), due to the deposition of the coarsest size fraction and the entrainment of ambient water (Postma et al., 1988).

| Siliciclastic facies associations
Poorly sorted clastrich deposit (B) 0.1-1 + m thick poorly sorted deformed, matrix-supported units. Matrix can range from mudstone to coarse-sandstone sized clasts and is often poorly sorted and sheared. Clasts include cm-m scale carbonate and siliciclastic blocks, folded thin-bedded sandstones, sporadic pebbles and granules and frequent mud clasts.
These deposits are commonly non-graded, but can show weak normal-grading.
Thick-bedded sandstones (C) 0.5-1 + m thick brown siliciclastic fine-granular sandstones. Normally graded or non-graded and typically lacking primary depositional structures. Bases are often sharp and erosive. Parallel laminations are sometimes present at bed tops and mud-clasts can be observed throughout. Weak cross-lamination is infrequently observed. The general massive nature of these deposits suggests that they represent rapid aggradation beneath a highly concentrated but dominantly turbulent flow, and are thus interpreted as high-density turbidites (Kneller & Branney, 1995;Lowe, 1982;Mutti, 1992). In some instances, these may have been deposited by flows that evolved from laminar to turbulent, following the deposition of the coarsest grain fragment (Postma et al., 1988).
Mixed siliciclastic and calcareous sandstones (D) 0.1-1 m beds of medium-bedded calcareous sandstones with punctuated interbeds of cm-scale thin-bedded siliciclastic sandstone, either as continuous beds or lenses. The medium-bedded calcareous sandstones are massive, and the siliciclastic beds are often erosively based and show tractional structures (ripple and planar lamination). Siliciclastic beds can be amalgamated with each other or isolated between calcareous siltstones or sandstones.
Medium-bedded calcareous sandstones are interpreted to represent deposition from a slowly aggrading dilute turbidity current. Periodic, thin-bedded siliciclastic sandstones represent deposition from a relatively quickly aggrading dilute turbidity current, which interacted with a much slower aggrading calcareous turbidity current. These facies represent lithofacies-scale mixing consistent with decimetre-scale alternations between siliciclastic and carbonate layers (Chiarella et al., 2017).
Medium-bedded calcareous sandstones (E) 0.1-1 m thick beige beds of calcareous siltstone -fine sandstone. Carbonate grains are normally graded or non-graded. Planar lamination may be present, but other tractional structures are rare. Beds can be amalgamated. Based on their tractional structures and normal-grading, beds are interpreted as having been deposited from dilute, slowly aggrading medium-density turbidity currents. These beds were deposited by thicker or more sustained flows than (G) (Remacha & Fernández, 2003).
Medium-bedded siliciclastic sandstones (F) 0.1-0.5 m thick brown beds of very fine-granular grained, commonly normally graded, sandstones. Inverse-grading is infrequently observed. Basal parts of the bed are often structureless containing infrequent cm-scale mud-clasts while tops are rich in tractional structures including parallel, ripple and hummock-like laminations. Bed bases are often erosive and can be amalgamated. 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 medium-density turbidites due to their bed thickness and common lack of structures in the lower part of the bed (e.g., Soutter et al., 2019).
Thin-bedded calcareous sandstones (G) 0.01-0.1 m thick beige beds of calcareous siltstone-fine sandstones. Carbonate grains can be normally graded, often increasing in micrite percentage upwards, or non-graded. Planar laminations are infrequently observed but other tractional structures are typically absent. Individual beds are often amalgamated. Thin-beds, fine grain-size and weak planar laminations represent deposition from a lowconcentration turbidity current (Jobe et al., 2012;Mutti, 1992;Talling et al., 2012), indicating these beds are low-density turbidites. Fine grain-size, thicker beds compared to thin-bedded siliciclastic sandstone (H) and absence of ripple laminations suggest deposition by slowly aggrading, dilute remnants of a turbulent flow, (Bell, Stevenson, et al., 2018;Remacha & Fernández, 2003), which did not reach sufficient velocity to generate ripple laminations (Baas et al., 2016). Alternatively, ripples may not have been preserved, or may be extremely difficult to decipher due to lack of variety in grain-size or colour and early lithification (Imbrie & Buchanan, 1960).
Commonly normally graded, occasionally inverse-graded. Tractional structures (planar, ripple, hummock-like and convolute laminations) and sporadic mudclasts are observed. Bases can be flat or weakly erosive and sometimes contain granules. Bed tops are often flat. Ripples can show opposing palaeoflow directions.
Bi or tri-partite beds (I) 0.05-0.5 m thick beds that contain multiple parts. Typically consisting of a lower fine-to coarse-grained sandstone (division 1) overlain by a poorly sorted muddy siltstone to medium-grained sandstone (division 2). Division 3 is sometimes present consisting of a siltstone-fine-sandstone loaded into division 2. Divisions 1 and 3 sometimes contain planar laminations and sporadic cm-scale mud-clasts. Division 2 is often highly deformed and rich in mud-clasts and very-coarsegrained sandstone to pebble-grade clasts. Sandstones can be calcareous or siliciclastic.
Tractional structures in division 1 and 3 indicate formation under turbulent flows.
Poor sorting and mud content suggest division 2 was deposited under a transitionallaminar flow regime. These bi/tri-partite beds are hybrid beds (Haughton et al., 2009), generated by flow transformation from turbulent to laminar. Such transformation occurs through flow deceleration (Barker, Haughton, McCaffrey, Archer, & Hakes, 2008;Patacci, Haughton, & McCaffrey, 2014) and by an increase in concentration of fines during flow run-out .
Mudstone (J) 0.005-8 m thick pale grey or red mudstone-fine-siltstone beds, which are friable and often inferred in areas of missing section. Planar laminations, discontinuous drapes and lenses of siltstone may be present. Commonly calcareous in composition. Red mudstones are common at the base of the Campanian stratigraphy. "Mud" is used here as a general term, for mixtures of clay, silt and organic fragments.
Low-energy conditions, representative of background sedimentation via suspension fallout. Laminations may be present below the scale visible in outcrop, representing deposition from a dilute turbidity current (Boulesteix et al., 2019(Boulesteix et al., , 2020. Pale colour indicates low total organic carbon (TOC). Red mudstones are similar to Cretaceous Oceanic Red Beds (CORBS) described across Europe Wagreich & Krenmayr, 2005;Wang, Hu, Sarti, Scott, & Li, 2005) and represent deposition below the carbonate compensation depth (CCD) in a deep oceanic basin.

FA 2: Channel axis
Observations. FA 2 is composed of metre-scale conglomerates (A) and thick-bedded medium-grained to pebbly sandstones (C) with lesser medium-bedded sandstones (F) and rare thin-bedded sandstones (H), mudstones (J), poorly sorted clast-rich deposits (B) and bi/tri partite beds (I) ( Figure 6c). FA 2 has the highest density of conglomerates (44%), thick-bedded sandstones (10%), and bi/tri partite beds (10%) of all Cenomanian-early Turonian facies associations ( Figure 8). Conglomerates often grade normally into thickbedded sandstones, commonly associated with a grain-size break, with the granular-grade sandstone often missing. Where conglomerates do not grade into thick-bedded sandstones they are amalgamated or separated by thin beds of mudstone. Conglomerates are poorly sorted, clast-supported and contain sub-angular-sub-rounded clasts of wackestone, micrite, sandstone and mudstone that often crudely grade from cobbles to pebbles upwards ( Figure 10). Conglomerates also often contain small amounts of disarticulated shelly fragments (<5% clasts). Sandstone and conglomerate bases are almost always erosional. Thick-bedded sandstones are commonly normally graded but can occasionally be massive or inversely graded. Decimetrescale mud-clasts are common throughout thick-bedded sandstones and low-angle cross-stratification is infrequently observed ( Figure 5h). Thin-to medium-bedded sandstones often have erosional bases and contain convolute, hummock-like and planar laminations, and are normally graded, with rare examples of inverse-or non-grading. These sandstones are either amalgamated or separated by decimetre thick mudstone layers, and often contain mud-clasts throughout the bed with granules concentrated at the bed base ( Figure 5e). Sporadic poorly sorted clast-rich deposits are also seen within FA 2; these have a deformed mudstone matrix and contain clasts of limestone and sandstone. Bi/tri-partite beds are amalgamated into 30-50 cm packages, with individual beds commonly consisting of a 2-4 cm thick fine-to medium-grained sandstone overlain by a clast and shelly fragment rich 8-12 cm thick muddy, very fine sandstone poorly sorted deposit.
Channel "off-axis" sequences, are infrequently observed in outcrop, so are not divided into a separate facies association. Where interpreted to be present, they have fewer thick-bedded sandstones (C) and conglomerates (A) and more thin-to medium-bedded sandstones (H, E) than channel axis successions ( Figure 8).
Interpretations. The thick-bedded nature, coarse grain-size, amalgamation, erosion and entrainment of clasts within the sandstones suggest that the parent flows were highly energetic and capable of eroding and bypassing sediment (Mutti, 1992;Stevenson, 2015). Therefore, these beds are interpreted as high-density turbidites (Lowe, 1982). The poorly sorted nature of the conglomerates suggests that they were initially deposited by laminar flows (Sohn, 2000), however, apparent grading of conglomerates into thick-bedded sandstones could reflect the transition of hyper-concentrated submarine debris flows into highly concentrated turbulent flows (Mulder & Alexander, 2001) due to entrainment of ambient water (Kane, McCaffrey, & Martinsen, 2009;Postma, Nemec, & Kleinspehn, 1988). Limestone clasts are interpreted to have been remobilised from Jurassic platforms (Bochud, 2011).
The transition from conglomerates, to medium-to very coarse-grained sandstone is associated with a grain-size break, often missing the granule fraction, which could suggest bypass of flow (Stevenson et al., 2015). The coarse grain-size and F I G U R E 5 Facies photographs. Due to the interbedding and mixing of these facies, it is not possible to document all facies in individual photographs and, therefore, multiple lithofacies appear in each photograph. Abbreviations on the figures refer to interpretations of lithofacies (Table 1): C, Calcareous; Db, debrite (poorly sorted clast-rich deposit); LDT, low-density turbidite, MDT, medium-density turbidite; Tb, Turbidite, S, Siliciclastic. Scale is either lens cap (52 mm), person (1.74 m) or indicated. Letters in square brackets below refer to lithofacies definitions (Table 1)  basal location of the conglomerates with respect to thick-bedded sandstones suggest these beds could have been deposited as channel-base lags (Hubbard et al., 2014). Erosionally based lenticular bodies grading from cobble-rich conglomerates to fine-grained sandstones are interpreted to represent submarine channel fills (Bell, Stevenson, et al., 2018;Jobe et al., 2017). This facies association is consistent with gravelly conglomeratic deposits reported elsewhere to represent submarine channel axis deposition (Dickie & Hein, 1995 Kneller et al., 2020;Li, Kneller, Thompson, Bozetti, & dos Santos, 2018;McArthur, Kane, Bozetti, Hansen, & Kneller, 2019;Nemec & Steel, 1984;Postma, 1984aPostma, , 1984bSurlyk, 1984).
While typically related to storm deposits (e.g., Hunter & Clifton, 1982), hummock-like cross-lamination has been interpreted in deep-marine environments elsewhere as anti-dune stratification (Mulder et al., 2009), bottom current deposits (Basilici, 2012;Furhmann et al., 2020) and reworking of an initial deposit by a subsequent flow (Mutti, 1992; Tinterri, Laporta, F I G U R E 7 Quantitative facies analysis for the mid Turonian-Maastrichtian stratigraphy. The three columns represent three different logged sections from north to south that are representative of northern margin, axis and southern margin of the Buduq Syncline respectively. Charts compare bed number (with 1 being at the base of the log and 200 at the top) to bed thickness (linearly in the top column and logarithmically in the middle column) and logged grain-size (in the basal column). Where grain-size varies within the bed average grain-size is used. In the top column thick beds are highlighted with a dashed line. Scales for bed number vary across the rows. Colours are for visual separation of data with greyscale used to for bed thickness (i.e., thicker beds are darker) and blue to orange used for grain-size (i.e., orange is coarser) & Ogata, 2017). The channel axis interpretation of FA 2 speculatively suggests anti-dunes formed by supercritical flows may be the most probable interpretation of these hummock-like structures (Alexander, 2008;Araya & Masuda, 2001).

FA 3: Channel margin
Observations. FA 3 comprises thin-to medium-bedded, fine-grained to granular sandstones (H, F) in 30-80 cm packages interbedded with 10-90 cm thick dark mudstones (J) (Figure 6c). Within the siliciclastic Cenomanian-early Turonian succession, FA 3 is dominated by thin-and to a lesser extent medium-bedded sandstones (72%) (Figure 8). Conglomerates (A) and thick-bedded sandstones (C) are rare in FA 3 ( Figure 8). Thin-bedded sandstones and the upper part of medium-bedded sandstones can be argillaceous and micaceous, and are often planar-, ripple-and convolutelaminated, with rarer hummock-like laminations. Sandstones are often normally graded but inverse-grading is also observed. Beds of medium thickness are rich in mud-clasts and commonly amalgamated along mud-clast laden surfaces. Bed bases can be highly erosive and scour-like, removing a significant proportion of the underlying bed. Thin-bedded sandstones can be flat or erosively based, and are commonly scoured; where bases are erosional the lowermost part of the bed is commonly rich in granule-grade material (Figure 6c). Granules and coarser fragments are composed of limestone and sandstone. Infrequent bi/tri-partite beds (I) are composed of medium-coarse grained siliciclastic sandstone, overlain by a muddy, occasionally marly fine sandstone poorly sorted deposit.
Interpretations. The thin-bedded nature and presence of tractional structures indicate that this facies association was deposited by a low-density turbidity current (Lowe, 1982). A deep-marine origin is interpreted based on the presence of thick, dark mudstones and frequent sediment gravity flow deposits (Mutti, 1992). Anti-dune formation (Mulder et al., 2009) and tractional reworking of an aggrading deposit (Bell, Stevenson, et al., 2018;Mutti, 1992;Tinterri et al., 2017) have both been interpreted to form similar hummock-like lamination in deep-marine environments, similar to those indicating storm wave influenced deposition (Harms, Southard, Spearing, & Walker, 1975). Clean sandstones that grade into argillaceous, micaceous sandstones could indicate transitional flow deposits (Baas, Best, Peakall, & Wang, 2009;Kane & Pontén, 2012;Sylvester & Lowe, 2004). The thinbedded, coarse grain-size and erosive nature of these deposits, along with the presence of supercritical bedforms, are similar to the overbank deposits seen adjacent to bypass-dominated channels (Kane & Hodgson, 2011;Hubbard et al., 2014;Jobe et al., 2017;Li et al., 2018;McArthur et al., 2019). These similarities, coupled with the along strike location of FA 3 adjacent to FA 2 (channel axis), have led to the interpretation of FA 3 as a channel margin (Figure 8). The lateral transition of FA 2 and 3 is indicative of the continuum between axis and margin channel facies, and is similar to the "on-axis" to "offaxis" shifting of channel-belts (Kane, Dykstra, et al., 2009).
interbedded with thin-to thick-bedded, very fine-to very coarse-grained sandstones (H, F, C). Within the mid Turonian-Maastrichtian succession, the thickest conglomerates are found within FA 4 (Figure 7). The conglomerates are laterally discontinuous over decametres, erosionally based, and are either flat topped when onlapping, or convex-up when downlapping the slope (Figures 6 and 11). Conglomerates increase in frequency, clast size (up to cobble-grade) and thickness up stratigraphy (Figure 7) and contain sub-angular to rounded clasts of carbonate (wackestone and micrite) and siliciclastic (sandstone and mudstone) material ( Figure 10). Within the mid Turonian-Maastrichtian stratigraphy, the greatest number of amalgamated beds is in FA 4 ( Figure 9) and the largest grain-size range (majority of beds between very fine to medium-grained sandstone) is observed (Figure 7). Within FA 4, a coarser grain-size class (of coarse-grained sandstone or above) is observed which is almost absent in other mid Turonian-Maastrichtian facies associations (FA 5, FA 6, FA 7) (Figure 7).

| Mixed facies associations
Alternations between siliciclastic and carbonate beds in the mixed facies association indicate most mixing was on the stratal-scale (bed-to architecture-scale). Smaller-scale (lamination-scale) fluctuations between carbonate and siliciclastic are observed in the mixed siliciclastic and calcareous sandstones facies (D), indicating subordinate compositional mixing processes (Chiarella et al., 2017).

FA 5: Lobe off-axis
Observations. FA 5 is represented by erosively based, thin-to medium-bedded, fine-to coarse-grained siliciclastic sandstones (H, F), thin-to medium-bedded fine-grained calcareous siltstones (G, E), (A) and mudstones (J) (Figures 6  and 7). Sandstones with siliciclastic bases that appear to transition into calcareous topped are present throughout. They are often amalgamated with siliciclastic and calcareous sandstones, forming packages separated by mudstones and silty mudstones. Calcareous beds are typically flat based when overlying mudstones, whilst siliciclastic beds are commonly irregularly based. Calcareous siltstones and sandstones are structureless (Figures 5b,d and 6d), whilst siliciclastic sandstones may contain planar, convolute and ripple laminations, but can also be structureless ( Figures  5d,e and 11b). Poorly sorted clast-rich deposits (B) are interspersed, often comprising remobilised thin-bedded calcareous siltstones and sandstones. Bi/tri-partite beds (I) are rare (Figure 8).
Interpretations. The presence of both calcareous and siliciclastic sandstones suggests deposition in a mixed system (Figures 1 and 9) (Al-Mashaikie & Mohammed, 2017; Chiarella et al., 2017;Walker, Jobe, Wood, & Sarg, 2019). Structureless medium-bedded calcareous siltstones and sandstones with normal grading or tractional bedforms at the bed top are interpreted to record deposition from waning turbidity currents transitioning relatively continuously from higher-to lower-densities over a single point on the seabed (e.g., Talling et al., 2012), and are, therefore, termed "medium-density" turbidites to differentiate them from high-to low-density turbidites (Soutter et al., 2019). These sandstones could also represent amalgamation of deposits formed by multiple low-density currents, with amalgamation surfaces difficult to decipher due to the lack of grain-size, colour and mineralogical variation within the sandstones (Imbrie & Buchanan, 1960). This depositional process is complicated within the calcareous medium-bedded deposits, which appear to have aggraded much more slowly than their siliciclastic counterparts, as evidenced by thin-bedded and medium-grained siliciclastic beds (D) being deposited within medium-bedded and fine-grained calcareous beds. These beds may, therefore, be derived from low-density carbonate-dominated flows that were depositional over longtime periods, resulting in thick, but fine-grained, beds. This may be driven by the presence of, and depositional distance from, two contemporaneously active sediment source areas (Chiarella et al., 2017;Moscardelli et al., 2019). The presence of medium-density turbidites, their relatively coarse grain-size and common amalgamation suggest lobe off-axis deposition (Prélat et al., 2009;Spychala, Hodgson, Prélat, et al., 2017).  Observations. Primarily composed of normally graded, thin-to medium-bedded, calcareous very fine-to fine-grained sandstones and siltstones (G, E), with subordinate thin-bedded siliciclastic fine-to medium-grained sandstones (H) and mixed siliciclastic and calcareous sandstones (D) (Figures 6-8). Calcareous siltstones and sandstones are flat based when overlying mudstones (J), but are often irregularly based at amalgamation surfaces (Figure 5d). Siliciclastic sandstones, either isolated or within mixed beds, are frequently <3 cm thick, with flat to weakly irregular bases (Figure 5d). Poorly sorted clast-rich deposits (B) are interspersed within FA 6 and often rework the thin-bedded calcareous siltstones and sandstones. Planar laminations are common within the thin-bedded siliciclastic and calcareous sandstones (G) (Figure 5d,e). Less common ripple laminated sandstones show multiple and/or opposing palaeocurrent orientations (Figures 4 and 11b). Bi/tri-partite beds (I) are rare (Figure 8).
The preservation of both structured and structureless sandstones suggests an off-axis location of deposition because; similar preservation of both deposit types has been interpreted in proximal lobe fringes elsewhere (Prélat et al., 2009;Soutter et al., 2019;Spychala, Hodgson, Prélat, et al., 2017). FA 6 is differentiated from FA 5 based on its thinner beds and less abundant erosional events and is, therefore, interpreted as being more distal and deposited within the proximal fringe. Hybrid beds are rare throughout the studied system and a distinction between frontal fringe and lateral fringe is not possible (e.g., Spychala, Hodgson, Prélat, et al., 2017).

FA 7: Distal fringe
Observations. Dominated by laterally extensive, metrescale packages of thin-bedded amalgamated calcareous sandstones (H) that are normally graded from very fine-or fine-grained sandstone to siltstone and are interbedded with metre-scale mudstones and silty mudstones (J) (Figures 6-8). Beds are flat based, flat topped and frequently contain both parallel and convolute laminations. Medium-bedded calcareous siltstones to fine-grained sandstones (E) are present and may reflect amalgamated thinner beds that are difficult to decipher. Poorly sorted, clast-rich deposits (B), siliciclastic thin-bedded sandstones (H) and bi/tri-partite beds (I) are rare (Figure 8). The smallest grain-size range (between siltstone and very fine-grained sandstone) is observed in FA 6 and FA 7 (Figure 7) and amalgamation is infrequent (Figure 9). More thin beds are seen in FA 7 than elsewhere in the stratigraphy (Figures 6-8).
Interpretations. Thin-bedded laminated sandstones with tractional structures are interpreted to be deposited from low-concentration turbidity currents (Jobe et al., 2012;Mutti et al., 1992;Talling et al., 2012). The presence of thin-to medium-bedded calcareous siltstones and fine-grained sandstones and a lack of ripple laminations suggest slow aggradation rates from low velocity flows (Bell, Stevenson, et al., 2018;Remacha & Fernández, 2003). Alternatively, ripples may not have been preserved, or may be difficult to recognise due to a lack of grain-size or colour contrast (Imbrie & Buchanan, 1960). The infrequency of siliciclastic beds suggests deposition in a carbonate-dominated environment. The thin-bedded nature, lateral-extent, fine grain-size, rare hybrid beds and lack of ripple stratification suggest a distal lobe fringe depositional setting (Marini et al., 2015;Mutti, 1977;Prélat et al., 2009;Spychala, Hodgson, Prélat, et al., 2017).

| STRATIGRAPHIC EVOLUTION
This section briefly describes the spatial and temporal distribution of facies associations ( Figure 6) and palaeocurrent changes throughout the Cretaceous (Figure 4). These observations, along with evidence for palaeobathymetry (Figures 11-14) are later used to interpret the Cretaceous evolution of the Buduq Syncline ( Figure 15).
The Aptian-Albian succession comprises sheet-like stacked units of FA 1 which are extensive across the scale of the outcrop (100's of metres laterally) (Figure 6a). Palaeocurrent data are not available due to limited accessibility.
An abrupt transition from distal fine-grained Aptian-Albian deposition to conglomeratic slope channels indicates a shallowing-upwards (regression) or change in sediment supply configuration into the Cenomanian. The Cenomanian-early Turonian succession is dominated by FA 2 and FA 3. FA 2 erosionally overlies FA 1, has a concave-up geometry and can be extensive for 100's metres, in both length and width (Figure 6b). Metre-scale thick packages of FA 2 often erode into older channel axis facies below. The sheet-like architecture of FA 3 is laterally extensive for at least decametres, and is often laterally adjacent to FA 2. FA 2 can erosionally incise into FA 3 and appears to transition laterally into FA 3, in agreement with its interpretation as marginal channel deposition (e.g., Li et al., 2018;McArthur et al., 2019). Cenomanian-early Turonian palaeocurrents are variable, consistent with sinuosity in submarine channels (e.g., Kane, McCaffrey, & Peakall, 2008;Peakall, McCaffrey, & Kneller, 2000;Peakall et al., 2012). There is a predominant SSW-trend in the Cenomanian-early Turonian (Figure 4), consistent with the presence of a terrigenous sediment source to the NNE (Gómez-Pérez, Morton, Kelly, & Stewart, 2005). A significant amount of northerly dominated flow indicators could indicate palaeoflow reflections in agreement with flow deflection from a bathymetric high to the south (Figure 11b) (e.g., Kneller et al., 1991). From the mid Turonian onwards mixed facies associations are common and the percentage of carbonate intraclasts in conglomerates increases ( Figure 10) illustrating a transition from Mid Cretaceous terrigenous to Late Cretaceous chalk-dominated shallow-water carbonate Scythian and southern Russian platforms (Baraboshkin, Alekseev, & Kopaevich, 2003). This, coupled with a change from slope channels to submarine lobes, indicates sea-level rise or a reduction in sediment supply. FA 4-7 are laterally extensive and stack together. The facies associations often transition vertically into each other in a non-predictable manner (Figure 17). FA 6 and 7 are extensive at Cek, in the NW of the Buduq Syncline. FA 4 and 5 are more common in the centre, in agreement with Cek being at a more distal location with respect to the proposed Scythian Platform source. FA 6 and 7 are sheet-like and laterally extensive across 100s-1,000s of metres. Where FA 6 and 7 are in contact with trough margins or Jurassic clasts, units are steepened and thin towards the contacts (Figures 11a and 12a,b), suggesting slopes of up to 6° in places.
Conglomeratic bodies within FA 4 and FA 5 (Figure 6d,e) are discontinuous over metres-decametres and can be convex-up or down in geometry and amalgamated or erosive with the beds below. These small conglomeratic channel fills have similar composition to the underlying, Cenomanian-early Turonian, more extensive slope channels (100s metres wide) (FA 2). Conglomerate body frequency increases throughout the mid Turonian to Maastrichtian stratigraphy, suggesting progradation (Figures 6d and 7).
Limited palaeocurrent observations from the mid Turonian-Santonian stratigraphy at Qonaqkend (Figure 4) indicate E-W flow. This may reflect the interruption of siliciclastic input immediately to the north and the continued presence of a bathymetric high to the south resulting in the axial flow of mixed systems within a sub-basin. Given the limited number of palaeocurrent measurements, however, this hypothesis must be considered speculative. Campanian-Maastrichtian palaeocurrent data support palaeoflow in a broadly SW to WNW direction (Figure 4). This is consistent with palaeoflow measurements from older strata and is in broad agreement with regional palaeogeographic maps (e.g., Barrier et al., 2008;Nikishin et al., 1998). It is considered to represent flows derived directly from a northerly Scythian Platform source, along with those from input points farther east that were deflected to flow parallel to the structural grain of the Greater Caucasus Basin. Southerly verging folds in debrites (Figure 11e) are further evidence for the presence of a northerly slope.
Changes in the geometry, exact extent and thus source area of the platform are probably linked to the compositional transition from a terrigenous to a carbonate-dominated platform (Nikishin et al., 1998(Nikishin et al., , 2001 rather than a geographical  (Bochud, 2011), which moved up to 20 km without significantly effecting the internal stratigraphy (Gavrilov, 2018). (e) Block with both margins exposed and onlapped by mixed stratigraphy, black box locates C EAGE CUMBERPATCH ET Al.
shift in source area. The majority of palaeocurrent measurements were from siliciclastic units, and therefore the change in palaeocurrent could reflect a switching of dominance of various siliciclastic point-feeding conduits on the platform through time, rather than a change in overall platform geometry or position (e.g., Casson et al., 2020;Tcherepanov et al., 2008).

| Nature of the Late Cretaceous bathymetry
In the western part of the study area, Late Cretaceous deep-marine sandstones are observed to thin towards, and onlap, Late Jurassic platform limestones (Figures 11-13). Stratigraphy is observed to thin from metres to centimetres across the scale of the outcrop (10's-100's metres) towards Late Jurassic limestones around Cek (Figures 11a, 12b and  13). Late Jurassic limestones must, therefore, have formed 100s of metres of relief on the Cretaceous seafloor. The most likely mechanism for the generation of seafloor topography is through allochthonous block emplacement. These blocks, or "megaclasts" (e.g., Blair & McPherson, 1999), were likely derived from Late Jurassic carbonate platform limestones (Figures 12-14). The presence of decametre-scale allochthonous blocks and submarine landslide deposits throughout the Cretaceous stratigraphy indicates a highly unstable margin ( Figure 12). Interpretations of basin-scale submarine landslide deposits, which partially form the Qizilqaya and Shahdag mountains farther west, further validates this interpretation (Bochud, 2011;Gavrilov, 2018) (Figure 14), with the mega-clasts in the west of the study area possibly forming part of this deposit (Figures 13-15). Similar onlap relationships to those formed as the Cretaceous stratigraphy infilled the irregular surface created by earlier submarine landslide deposits, have been observed elsewhere at outcrop (e.g., Armitage, Romans, Covault, & Graham, 2009;Burbank, Vergés, Munoz, & Bentham, 1992;Kneller et al., 2020) and in the subsurface (Figure 16) (e.g., Casson et al., 2020;Soutter, Kane, & Huuse, 2018). These Late Jurassic blocks (Figures 12 and 14) within the Cretaceous stratigraphy can be interpreted as either: (a) Late Cretaceous failures from an exposed Jurassic shelf; (b) out-running blocks from Early Cretaceous failures (e.g., De Blasio, Engvik, & Elverhøi, 2006) that were subsequently onlapped during the Late Cretaceous, or; (c) blocks that were periodically remobilised throughout the Late Cretaceous from high-relief Early Cretaceous slope submarine landslides identified in the west (Figures 13 and 15). Differential compaction around these rigid blocks will have resulted in the steepening of strata adjacent to the block, which may contribute to the gradual rotation and steepening of stratigraphy identified (Figures 11 and 13). This has been reported elsewhere around allochthonous blocks (e.g., Burbank et al., 1992).

Buduq Syncline
Late Cretaceous deep-marine deposition within the Buduq Syncline began following a period of compression and folding in the mid-Cretaceous ( Figure 15) (Bochud, 2011;Egan et al., 2009). Evidence of this compression is preserved in the east of the study area. The early fill is represented by Cenomanian-early Turonian conglomeratic slope channels that either eroded into Barremian deep-marine mudstones or are conformable with thin-bedded Aptian-Albian siliciclastic turbidites. These basal Cenomanian stratigraphic relationships are suggested to be caused by channels preferentially infilling lows present on the seafloor, forming entrenched channel axes that pinch-out laterally against Barremian mudstones ( Figure 15). These lows may have formed during mid-Cretaceous compression and folding (Bochud, 2011;Egan et al., 2009) or through submarine slope failure and seafloor erosion.
It is possible that poorly preserved thin-bedded Aptian-Albian turbidites (FA 1) represent the distal extents of the slope channel systems evident in the Cenomanian that were either eroded by the channels during progradation or deposited within isolated lows on the Barremian slope. These lows may have formed in response to similar processes to those that entrenched the Cenomanian channels (FA 2 and 3). The abrupt nature of the transition from distal fine-grained F I G U R E 1 5 Evolutionary model for the Cretaceous of the study area. Studied stratigraphic sections highlighted. Topography, thought to be formed by a mega-clast, is present throughout the Cretaceous and influences deposition, discussed in text. Extract from the geological time scale, sea-level fluctuations (Haq, 2014) and local tectonic events highlighted on the left. Tectonic events are compiled by Bochud (2011), after Zonenshain and Le Pichon (1986), Philip et al. (1989), Nikishin et al. (2001), Brunet, Korotaev, Ershov, and Nikishin (2003),  and Barrier et al. (2008). The Pre-Albian was dominated by limestone blocks on a muddy slope. Thin-bedded siliciclastic turbidites of a distal lobe were deposited during the Aptian-Albian. Siliciclastic channels are prominent throughout the Cenomanian-early Turonian. In the mid Turonian-Maastrichtian mixed calcareous and siliciclastic lobes, of different sub-environments interact, and are likely sourced from the same northern margin, discussed in text. Locations in red boxes are stratigraphically, and not spatially, representative | 635 EAGE CUMBERPATCH ET Al. turbidite deposition to conglomeratic slope channels may correspond to either tectonic rejuvenation during the Mid-Cretaceous compressional event ( Figure 15) (Egan et al., 2009) and/or an abrupt relative sea-level fall. Such a sea-level fall has been identified in the mid-Cenomanian (Baraboshkin et al., 2003;Miller et al., 2003), resulting in a mid Cenomanian-early Turonian hiatus or condensed section on the Russian Platform to the north.
Evidence for bathymetry is present during deposition of the Cenomanian-early Turonian interval, with the sequence almost entirely absent 10 km to the west at Cek, indicating the presence of a relative high in this location (Figures 2, 12 and  15). Submarine landslide thicknesses also increase towards this high in the Barremian, suggesting the high influenced deposition from the Early Cretaceous until the Turonian. Previous work has shown the presence of a large c. Early Cretaceous submarine landslide, composed of remobilised Late Jurassic blocks, towards the west ( Figure 15) (Bochud, 2011;Gavrilov, 2018). The exact timing of this failure is uncertain, with Gavrilov (2018) suggesting this event may have occurred in the Late Cretaceous. The stratigraphic observations made by this study indicate that this failure occurred prior to the Cenomanian, with this submarine landslide complex, or a related basin-scale mass failure, forming the westerly high and the complex stratigraphic relationships described previously (Figures 12-14). It is also likely that this submarine landslide, and other more minor ones in the area, were emplaced during an earlier period of tectonism and instability related to Early Cretaceous compression ( Figure 15). Evidence for topography ( Figure 11) in the Late Cretaceous also exists on a smaller-scale through palaeocurrent reversals in low-density turbidites indicating a northward-dipping slope confining southward-directed flows (Figures 4 and 11), and through the deposition of Late Jurassic blocks within the Turonian succession, indicating slope instability during this period (Figures 12 and 14).
Following the Cenomanian regression, the study area began to deepen again during the mid Turonian, as represented by the deposition of laterally extensive, thin-to medium-bedded, mixed turbidites overlying the slope channels ( Figure 15). The mixed lithology of the turbidites contrasts with the dominantly siliciclastic Aptian-Albian turbidites underlying the slope channels, indicating the development of a carbonate factory along the northern margin of the Greater Caucasus Basin in the Late Cretaceous ( Figure 15). The presence of thinning and facies changes towards the present-day syncline margins, frequent debrites and out-runner blocks, and divergent palaeocurrent distributions indicate that basinal topography had an impact on mid Turonian and later deposition (Figures 4, 11, 12 and  14). This topography may have been formed by differential compaction over the rigid limestone megaclast, or external compression (Figures 13-15). Erosional contacts are seen within the succession at the base of metre-scale channel fills, which occur with increasing frequency through time. These small channel fills (FA 4) are filled by conglomerates and high-density turbidites with similar compositions to the underlying and much more extensive slope channels (FA 2). These metre-to decametre-scale channels are, therefore, interpreted as small distributary channels in the axes of lobes (FA 4 and 5) that formed at the distal ends of the underlying 100s metre-scale slope channels (e.g., Normark, Piper, & Hess, 1979). The increasing frequency and thickness of these conglomerates through the mid Turonian to Maastrichtian succession (Figure 7) may, therefore, represent gradual progradation of the slope channels following their abrupt backstep in the mid Turonian. Clasts within these younger conglomerates are also more carbonate dominated, which fits with the transition to a more carbonate-dominated system through the Late Cretaceous (Figures 10, 15 and 17). Mixed-deep-marine deposition continues in the Buduq Syncline throughout the remainder of the Cretaceous and into the Paleocene (Bochud, 2011).

Buduq Syncline
A seismic-scale mixed system analogous to the Cretaceous succession in the Buduq Syncline has been identified and is used to support and increase 3D visualisation of the outcrop-based model. The continental margin offshore The Gambia, NW Africa, developed through the Late Cretaceous (summarised in Casson et al., 2020;Figure 16). Unconfined mixed systems developed on the deep-marine basin floor are interpreted to have been line-fed through s developed on an unconformity surface ( Figure 16c). Seismic geomorphology reveals the strata mixing (sensu Chiarella et al., 2017) of interdigitating siliciclasticdominated and carbonate-dominated systems (i.e., at X and Y in Figure 18), similar to that observed at outcrop-(facies and facies architecture) scale in the EGC (Figures 5 and 6).
Sediment gravity flows passing through the submarine canyons eroded into the underlying carbonate platform redepositing hundreds of metre-scale, seismically resolvable carbonate mega-clasts 20+ km from the escarpment, above the unconformity surface (Figure 16b,d). Our field work suggests that these blocks may be associated with a multitude of different types and sizes of submarine landslides and blocks that are below seismic scale (Figures 12 and 14). The presence of carbonate blocks, lobe-morphology and similarity in run-out length and volume to debris erosion to form mixed lithology flows (bed-scale strata mixing), and then through deposition of interdigitating systems (lithofacies-scale strata mixing; sensu Chiarella et al., 2017). Pervasively channelised siliciclastic-systems with single feeder channels show a distinct seismic geomorphological response to their carbonate counterparts (Figure 16d,e). The lateral location along the margin of siliciclastic-dominated systems is conceivably related to sediment input points (i.e., shelf-incising canyons) capturing an extra-basinal source of siliciclastic sediment from the shallow marine environment, away from shelfal carbonate factories. Basin floor topography is created by early deposits and influences subsequent lobe deposition (Figure 16), causing stacking and lateral migration of lobes, which cannot be resolved in the Buduq Syncline ( Figure 17) probably because the scale of the study area is smaller than the scale at which migration occurred. Ancient subsurface mixed systems have been described from seismic reflection data (e.g., Casson et al., 2020;Moscardelli et al., 2019). It may also be possible that transitions from calcareous-dominated to siliciclastic-dominated deep-marine systems, which are commonly associated with the rapid arrival (progradation) of the siliciclastic system (e.g., Kilhams, Hartley, Huuse, & Davis, 2012;Kilhams, Hartley, Huuse, & Davis, 2015;Scott et al., 2010;Soutter et al., 2019), may have been overlooked as "transition zones," and in fact represent short-lived mixed systems, which are often below the scale of seismic resolution. The role of mixed system interactions on a grain-scale and its implications in terms of reservoir quality remain unclear until such systems are studied in detail in the subsurface (Chiarella et al., 2017;Moscardelli et al., 2019), or at outcrop .

| Source area and mixing origin
Palaeoflow indicators are limited within the calcareous system due to a lack of ripple laminations developed in the finegrained and likely slowly accumulating deposits that build this system (Baas, Best, & Peakall, 2016), and because of a lack of contrasting mineralogies, which prevented preservation of defined structures in the fine-grained carbonates (Imbrie & Buchanan, 1960). It is, therefore, difficult to decipher whether these siliciclastic and calcareous systems were perpendicular, oblique or parallel to each other. The palaeoflow indicators that were collected, however, are consistent with a provenance to the north (Figures 4, 11 and 14). A northern provenance is also suggested from palaeographic maps for the Cretaceous, suggesting a Scythian-Russian Platform source area (e.g., Barrier et al., 2008;Nikishin et al., 1998). In Figure 15 it is assumed that the simplest explanation is the most likely, and that both systems are sourced from the Scythian Platform to the north. This is in agreement with the Mid-Cretaceous transition from terrigenous to chalk-rich deposition (Baraboshkin et al., 2003).
Disentangling specific types of mixing and interactions of mixing is possible in stratigraphy with excellent time constraints (e.g., Chiarella, Longhitano, & Tropeano, 2019) or well-established shelf-basin floor seismic data sets (e.g., Casson et al., 2020;Moscardelli et al., 2019;Tcherepanov et al., 2008). These remain uncertain in the study area which has a poorly exposed source-to-sink relationship and limited temporal constraints, however, it is possible to identify two different scales of stratal mixing (sensu Chiarella et al., 2017) of the two separate systems. Outcrop identification of mixed siliciclastic and calcareous sandstones (Lithofacies D, Table 1) indicate bed-scale stratal mixing, and observations of mixed facies associations (Figures 6 and 9), suggest facies-scale stratal mixing.
Below we speculate how these mixed lobes differ from their siliciclastic counterparts in terms of sub-environments, stacking patterns and controlling parameters.

| Lobe sub-environments
In the study area, when independently observed, the siliciclastic part of the mixed system could be interpreted as stacked lobes with axis, off-axis and fringe sub-environments all being identified (e.g., Spychala, Hodgson, Prélat, et al., 2017). The calcareous part of the system in the study area, however, would be interpreted as being predominantly the product of lobe fringe deposition (Bell, Stevenson, et al., 2018;Remacha & Fernández, 2003). Since the systems are mixed, it is difficult to assign a single lobe sub-environment to a sequence of beds as they represent the inter-fingering of two interacting systems ( Figure 18). Hence, siliciclastic lobe elements (sensu Prélat et al., 2009) are likely to occur within calcareous lobe elements forming a succession of mixed event beds (Lithofacies D, Table 1) (Figures 1, 16 and 18 interpretations and correlations more difficult (Figure 17) (Braga, Martin, & Wood, 2001). Similar observations offshore of erosion and re-working of carbonate deposits have been made offshore The Gambia (Casson et al., 2020). Due to these complexities, it is perhaps necessary to refer to such systems with a more specific descriptor (e.g., mixed axis-fringe), or broadly refer to them as "mixed systems" in order to allude to their heterogeneity and contrast them from siliciclastic-dominated systems (Figure 18). Use of the siliciclastic lobe hierarchy of Prélat et al. (2009) is possible in mixed systems, but calcareous and siliciclastic descriptors are required (Figure 18). It is possible to decipher the different systems in our field and subsurface examples, due to their lithological differences being visually resolvable at outcrop (Figures 5 and 6), and showing different seismic characteristics in the subsurface (Figures 16 and 18). However, without detailed provenance and geochemical analysis it would be very difficult to decipher the mixing of two siliciclastic systems or two calcareous systems, due to similarity in depositional facies and thus seismic character. Unless an individual system can be followed from source to sink in outcrop or subsurface, the possibility of multiple systems interdigitating, interacting, modulating each other and complicating stacking patterns must always considered (Figure 18).
Our study shows that in mixed interdigitating systems it can be difficult to decipher stacking patterns within each individual system due to the interdigitation of each system by the other (Figures 17 and 18). Bed thickness trends within the calcareous turbidites are difficult to interpret, possibly due to their narrow grain-size range, lack of contrasting mineralogies and early pseudo-lithification preventing the identification of thinner beds, and amalgamation within thicker beds ( Figure 17). A lack of defined structures in fine-grained carbonates also hinders the identification of bed tops (Imbrie & Buchanan, 1960).
In "pure" siliciclastic systems, stacking patterns are frequently deciphered based on grain-size and bed thickness trends. Progradational cycles, for example are often manifested in deposits as coarsening-and thickening-upwards packages (e.g., Hodgson, Kane, Flint, Brunt, & Ortiz-Karpf, 2016;Prélat & Hodgson, 2013;Prélat et al., 2009). However, bed thickness and grain-size analysis for the mid Turonian-Maastrichtian do not show any thickness trends or stacking patterns within the calcareous or siliciclastic turbidites when treated separately, or the combined mixed deposits ( Figure 17). This suggests that in mixed systems it is difficult to disentangle the evolution of the systems individually, or together. It, therefore, may not be possible to describe the F I G U R E 1 8 Interpreted RMS maps showing potential interactions of calcareous (blues) and siliciclastic (yellows/browns) lobes in mixed systems. A and B are from Figure 16, and have been overlain by schematic lobe complex geometries, based on seismic facies analysis and understanding of regional source area (see Casson et al., 2020). X and Y are representative logs, based on field observations where the lobe complexes interact in A and B respectively. X crosses the lobe fringe of the calcareous system and the lobe axis of the siliciclastic system and Y crosses the lobe fringe of both systems resulting in a thinner and finer-grained succession when compared to X. This variability highlights difficulties arising from exporting sub-environment terminology developed in siliciclastic systems (e.g., Prélat et al., 2009)  progradation or retrogradation of an individual system, and only possible to describe the relative ratio between the two; the apparent dominance of the mixed system (e.g., if siliciclastic (s) > carbonate (c) this could be due to progradation of s or by the retrogradation of c, both of which are controlled by a number of external and internal forcings (e.g., Chiarella et al., 2017;Chiarella et al., 2019;Moscardelli et al., 2019).
It is also a possibility that the influence of topography promotes predominately aggradational stacking, explaining the lack of statistical significance between grain-size or bed thickness and bed number. However, aggradational stacking is often associated with a consistent bed thickness and grainsize throughout evolution (Grundvåg, Johannessen, Helland-Hansen, & Plink-Björklund, 2014;Hodgson, di Celma, Brunt, & Flint, 2011;Spychala, Hodgson, Prélat, et al., 2017) which is not observed in the Buduq Syncline ( Figure 17). Siliciclastic conglomerates become more frequent and thicker throughout the mid Turonian-Maastrichtian, perhaps reflecting a progradation of the siliciclastic system (Figures 7 and  15). These observations, rather than bed thickness or grainsize alone, are more helpful in disentangling these interdigitating systems and understanding subsurface spatial and temporal distribution of depositional elements.
On the scale of the outcrop (100s m), calcareous turbidites appear to be sheet-like, while the siliciclastic turbidites show more thickness variation, representing more typical channel and lobe geometries (e.g., Prélat et al., 2009). Conglomerates in FA 4 appear to be confined to isolated depocentres and pinch-out across meters or decametres, indicating the presence of subtle topography ( Figure 11). This suggests the deposition of the conglomerates may have been controlled by depositional topography, that is compensational stacking, and that the underlying calcareous turbidites do exhibit subtle, long-wavelength thickness changes over a greater scale than observed at outcrop, influencing subsequent sediment routing. Alternatively, the thinning of conglomerates is due to the basinal topography present at this time, preventing these highly concentrated flows running-out over great distances ( Figure 11).
In both cases the calcareous system is likely to represent the "background" sedimentation, dominated by suspension fall out and very low-density gravity flow deposits (sensu Boulesteix et al., 2019Boulesteix et al., , 2020 derived from a calcareous platform, most likely the Late Cretaceous chalk-dominated Scythian and Russian platforms (Baraboshkin et al., 2003;Barrier et al., 2008). The siliciclastic system is interpreted to have punctuated this slowly aggrading "background" deposition via canyonised conduits on the platform in agreement with the subsurface example (Casson et al., 2020). The extensive existence of the calcareous deposits throughout the study area, at the margins and adjacent to topographic highs, versus the siliciclastic deposits localised isolation in the central syncline, further supports this hypothesis.
The overall change in composition of the Scythian-Russian Platform, from terrigenous to calcareous (Baraboshkin et al., 2003) is controlled by a high-order sea-level rise, and associated climatic warming. Bed-scale mixing could suggest very high-frequency sea-level cycles, which are impossible to decipher without better age control (Moscardelli et al., 2019). Individual gravity flow deposits, are known to represent very short periods of deposition (hours, days) (e.g., Maier et al., 2019;Paull et al., 2018) highlighting the importance of keeping time scales in perspective when studying such high-frequency variations (Moscardelli et al., 2019).
Scalloped and serrated carbonate platform margins (e.g., Casson et al., 2020;Grant, Underhill, Hernández-Casado, Barker, & Jamieson, 2019;Saller, Barton, & Barton, 1989), like those that may characterise the northern margin of the Greater Caucasus Basin (Figures 12, 15 and 16), have been proposed as conduits for siliciclastic sediments without requiring a sea-level change (Al-Mashaikie & Mohammed, 2017; Braga et al., 2001;Francis et al., 2008;Puga-Bernabéu et al., 2014;Walker et al., 2019). The high-frequency variation in composition within the Buduq Syncline mixed stratigraphy is, therefore, likely to be best explained by this model rather than by sea-level fluctuations discussed above. This indicates that the calcareous deep-marine system exposed in the Buduq Syncline is part of a much more extensive and line-fed system (the Scythian and Russian platforms) derived from shedding of active carbonate factories perched on the shelf (Figure 16). The contemporaneous siliciclastic system may, therefore, have been derived from multiple point source conduits along this margin that either: (a) periodically punctuated this larger carbonate system or; (b) were longlived conduits, fed by the uplifting siliciclastic hinterland, | 641 EAGE CUMBERPATCH ET Al. permanently bound by, or incising into, carbonate highs on the platform (Figure 15) (Moscardelli et al., 2019;Mueller, Patacci, & di Giulio, 2017). Once sediment reached the upper slope, gravity flows would have been funnelled through sedimentary pathways, the location of which were controlled by the presence of underlying structures and topography (e.g., Moscardelli et al., 2019).

| CONCLUSION
This study of the Upper Cretaceous stratigraphy of the Buduq Syncline, Azerbaijan documents the characteristics of an unstable and mixed siliciclastic-carbonate sedimentary system. Deposition in the syncline is represented by a Cenomanianearly Turonian submarine channel complex, which transitions into a mid-Turonian to Maastrichtian mixed lobe succession. This sequence represents an abrupt Cenomanian regression, probably related to an abrupt mid-Cenomanian eustatic sea-level fall and/or a mid-Cretaceous compressional event; followed by a mid Turonian transgression. Siliciclastic systems fed to the deep-marine via point-sourced conduits along a carbonate platform and are interpreted to punctuate a more extensive calcareous system throughout the remainder of the Cretaceous, resulting in the deposition of a mixed deep-marine system.
A westerly bathymetric high, likely formed by an Early Cretaceous submarine landslide complex deposited during earlier compression, is interpreted to have prevented deposition of Cenomanian-Turonian sediments towards the west. This submarine landslide complex may also have provided a lateral source for landslides through secondary remobilisation perpendicular to the regional palaeoflow from the north. Bed pinch-out, thinning, ripple reflections and frequent debrite deposition provide further evidence for the presence of basinal topography during deposition.
The mid Turonian-Maastrichtian mixed siliciclasticcalcareous deep-marine system contains both siliciclastic and calcareous lobe elements, which represent different lobe sub-environments, requiring modification of terminology developed for siliciclastic lobes. Mixed systems are also shown to have unique facies, both at outcrop and in a subsurface analogue from offshore The Gambia, reflecting differing depositional processes between the systems operating contemporaneously. Interaction between the two deep-marine environments characterising the mixed systems has also made stacking patterns difficult to decipher, with each system attenuating the other.