Structural controls on the pathways and sedimentary architecture of submarine channels: New constraints from the Niger Delta

In submarine settings, the growth of structurally influenced topography can play a decisive role in controlling the routing of sediments from shelf‐edge to deep water, and can determine depositional architectures and sediment characteristics. Here we use well‐constrained examples from the deep water Niger Delta, where gravity‐driven deformation has resulted in the development of a large fold and thrust belt, to illustrate how spatial and temporal variations in the rate of deformation have controlled the nature and locus of contrasting depositional styles. Published work in the study area using 3D seismic data has quantified the growth history of the thrust‐related folds at multiple locations using line‐length‐balancing, enabling cumulative strain for individual structures over time and along‐strike to be obtained. We integrate this information with seismic interpretation and facies analysis, focusing on the interval of maximum deformation (15 to 3.7 Ma), where maximum strain rates reached 7%/Ma. Within this interval, we observe a vertical change in depositional architecture where: (1) leveed‐confined and linear channels pass upward in to (2) ponded lobes with erosionally confined channels and finally (3) channelised sheets. Our analysis demonstrate that this change is tectonically induced and diachronous across the fault array, and we characterise the extent to which structural growth controls both the distribution and the architecture of the turbidite deposits in such settings. In particular, we show that leveed‐confined channels exist when they can exploit strain minima between growing faults or at their lateral tips. Conversely, as a result of fault linkage and increased strain rates submarine channels become erosional and may be forced to cross folds at their strain maxima (crests), where their pathways are influenced by across‐strike variations in shortening for individual structures. Our results enable us to propose new conceptual models of submarine channel deposition in structurally complex margins, and provide new insights into the magnitude of fault interaction needed to alter depositional style from leveed to erosionally confined channels, or to deflect seabed systems around growing structures.

driven deformation has resulted in the development of a large fold and thrust belt, to illustrate how spatial and temporal variations in the rate of deformation have controlled the nature and locus of contrasting depositional styles. Published work in the study area using 3D seismic data has quantified the growth history of the thrust-related folds at multiple locations using line-length-balancing, enabling cumulative strain for individual structures over time and along-strike to be obtained.
We integrate this information with seismic interpretation and facies analysis, focusing on the interval of maximum deformation (15 to 3.7 Ma), where maximum strain rates reached 7%/Ma. Within this interval, we observe a vertical change in depositional architecture where: (1) leveed-confined and linear channels pass upward in to (2) ponded lobes with erosionally confined channels and finally (3) channelised sheets. Our analysis demonstrate that this change is tectonically induced and diachronous across the fault array, and we characterise the extent to which structural growth controls both the distribution and the architecture of the turbidite deposits in such settings. In particular, we show that leveed-confined channels exist when they can exploit strain minima between growing faults or at their lateral tips. Conversely, as a result of fault linkage and increased strain rates submarine channels become erosional and may be forced to cross folds at their strain maxima (crests), where their pathways are influenced by across-strike variations in shortening for individual structures. Our results enable us to propose new conceptual models of submarine channel deposition in structurally complex margins, and provide new insights into the magnitude of fault interaction needed to alter depositional style from leveed to erosionally confined channels, or to deflect seabed systems around growing structures.

| INTRODUCTION
The structural development of a shelf margin or basin impacts sea-floor topography in a range of ways which can be defined by two end-members: static and dynamic. In the static case pre-existing but stable topography drives the distribution of facies and sedimentary architecture for instance in late rift-topography and fold belts (e.g., Gawthorpe & Leeder, 2000;Mitchell et al., 2021bMitchell et al., , 2022. In contrast, in the dynamic case structural development generates topography on the sea-floor which changes in time and space due to, for example, salt diapirism, and the growth of active faults (Clark & Cartwright, 2009, 2012a, 2012bDeptuck et al., 2012;Don et al., 2020;Doughty-Jones et al., 2019;Howlett et al., 2021;Jolly et al., 2016;Jones et al., 2012;Kane et al., 2010;Mayall et al., 2010;Mayall & Stewart, 2000;Pinter et al., 2018;Pirmez et al., 2012;Pizzi et al., 2020). These end-members are clearly influenced by sediment supply, where the generation of dynamic topography can be dampened by high sedimentation rates (Christie et al., 2021;Sylvester et al., 2015).
More recently, a number of studies have started to quantify the rate of growth of structures such as thrusts and salt diapirs (e.g., Pizzi et al., 2020;Totake et al., 2018) and relate this directly to the behaviour of deep-water channels, depositional lobes and mass transport complexes (e.g., Doughty-Jones et al., 2019;Jolly et al., 2016Jolly et al., , 2017Jones et al., 2012;Pizzi et al., 2021). This is important because independent reconstructions of deformation rates in time and space are a pre-requisite to derive unambiguous links between structural growth, channel behaviour and facies architecture. For instance, Jolly et al. (2016Jolly et al. ( , 2017 documented the evolution of a selection of gravity-driven fold-thrust structures on the southern lobe of the Niger Delta and showed how changes in channel morphology and long-profile were linked to the structural template. They were able to demonstrate that submarine channels cross structures at points of minimum strain. Don et al. (2020) use three-dimensional kinematic models of fold growth to predict how channel geometry and architecture will change as the structures K E Y W O R D S sedimentary architecture, structural deformation, submarine channels, thrusts, turbidites Highlights • Submarine channel pathways and facies are influenced by across-strike variations in deformation rate. • Results quantify magnitude of fault interaction needed to alter depositional style of channels. • New conceptual models proposed of submarine channel deposition in structurally complex margins.
grow. While they did not explicitly quantify the structure growth rate, their work provides predictive models directly linked to whether the thrust-fold evolves with a fixed length or propagates laterally. Mitchell et al. (2021a) also established how changes in channel morphology and dimensions across several contrasting structural zones drove increases in predicted bed shear stresses and incisional capacity for channels crossing growing structures. Individual fold-thrust structures have been shown to modify the kinematics and time-integrated architecture of submarine channels, increasing the ratio of migration to aggradation resulting in the deposition of high aspect ratio channel complexes (Mitchell et al., 2021b). Recently, Pizzi et al. (2020) extended the work in the Niger Delta by quantifying comprehensively the growth of fault propagation folds in the region over a wider area of ca. 2500 km 2 from 15 Ma to present day. From this Pizzi et al. (2021) generated an extensive database of nearly 200 crossing points of channels across 11 fold-thrust structures. Their statistical analysis clearly proved that in response to increasing deformation the channels are driven to zones of lower strain rates, resulting also in a reduction and focusing of channel crossing locations. However, this work did not address the nature, characteristics and facies of submarine channels and related elements. These studies give some much-needed clarity regarding the relationships between structural growth and submarine channel behaviour. Nevertheless, from a structure-facies perspective, studies with independent constraints on structural deformation have mostly focused on the morphology or location of channels, rather than their sedimentary characteristics; or where sedimentary facies are considered only a limited number of structures or relatively short time periods are presented (Mitchell et al., 2021b(Mitchell et al., , 2022. Yet a key finding of studies in submarine fold and thrust belts is that structural evolution in time and space is complex and dynamic (e.g., Totake et al., 2018) with strain rates influenced by interactions both along-strike and between faults (Pizzi et al., 2020). To date, the complex interaction of the development of topography on the sea-floor and sedimentation patterns driven by such an evolving structural template has never been comprehensively quantified over durations of up to ca. 10 My. Here we address this outstanding research challenge. We utilise the quantitative structural growth history work of Pizzi et al. (2020) as a unique template on to which we can characterise and interpret depositional facies within a dynamically deforming thrust belt in the deep-water of the Niger Delta since the Miocene. In particular, we (i) assess the impact of structural development on stratigraphic patterns, depositional facies and channel pathways by integrating structural analyses with seismic facies characterisation on a regional scale; (ii) document and interpret how temporal variations in strain rate control channel orientation, characteristics including erosional versus aggradational geometries and the development and distribution of depositional lobes; (iii) develop a 3D model that explains how across and along strike structural variations, including detachment linkage, dictate the observed evolution of the stratigraphic architecture of the area. Our results give novel insights into the way in which deformation at or near the sea-bed influences deep water sedimentation in time and space and allows us to improve predictive models of sedimentary architecture and characteristics in structured margins and deep water basins.

| Location and structural setting
We focus on the Niger Delta system (Figure 1, upper panel), a 12-km thick sequence which started prograding in the early Eocene and comprises timetransgressive deep-marine, shelf, deltaic and fluvial facies (Damuth, 1994;Doust & Omatsola, 1990;Evamy et al., 1978). The area of focus for this study is the outer fold and thrust belt of the southern lobe of the delta (sensu Corredor et al., 2005) where deformation is interpreted to have commenced at ca. 15 Ma (Jolly et al., 2016;Pizzi et al., 2020) with peak deformation being reached by ca. 9 Ma (Jolly et al., 2016;Krueger & Grant, 2011;Sun & Liu, 2018). The interval of interest is the offshore deep-marine Agbada Formation which includes typical facies of channels, lobes, mass transport complexes and hemipelagic shales (e.g., Pizzi et al., 2021). As a result of gravity-driven tectonics, extensional deformation in the upper part of the sedimentary wedge is matched by folding and thrusting in the lower slope with the underlying and overpressured Akata shales acting as the decollement layer (Figure 1, lower panel; Figure 2a; Corredor et al., 2005;Damuth, 1994;Rouby et al., 2011). Using 3D seismic reflection data, Pizzi et al. (2020), building on earlier work by Jolly et al. (2016), presented a quantitative analysis of the growth history of 11 fault-propagation folds within the study area over an area of 2500 km 2 through a time period of ca. 15 Ma (Figure 2a). The analysis included measuring the total shortening and strain, including both fault displacement and fault-related folding. In particular, this allowed the interval strain rates and their variation through time to be measured in the folded syn-kinematic strata. The strain was measured using line-length balancing (Dahlstrom, 1969) on 17 dip seismic sections, approximately 3 km apart and 30-50 km long. Comprehensive details of the approach are presented in Pizzi et al. (2020).
The result of this analysis shows that all the thrust and associated folds had nucleated as short segments by about 15 Ma and that between 15 and 12.8 Ma the segments were beginning to grow along strike and link laterally. This mode of growth continued between 12.8 and 9.5 Ma with thrusts increasing in length along strike. Peak deformation with highest strain rates was reached by 9.5-7.4 Ma in the inboard parts of the study area. Within this time interval most of the inboard faults became fully linked laterally and the mode of fault linkage started to be dominated by detachment linkage (sensu Bergen & Shaw, 2010) where the thrusts link across-strike via the common underlying detachment. The highest deformation rates were reached on the outboard thrusts between 7.4 and 3.7 Ma with the thrusts continuing to grow both laterally and by detachment linkage. By 3.7 Ma all thrusts had linked via the detachment with deformation occurring on a fully linked fold/thrust array. From 3.7 Ma to the present day the strain rate decreased across the fold belt with deformation only continuing on short segments of four fold/thrusts. These evolutionary stages of the outer deep-water fold and thrust belt were documented in a series of maps illustrating how the interval and cumulative strains have changed through time (Pizzi et al., 2020(Pizzi et al., , 2021). These constraints and maps therefore form the starting framework for this study, enabling us to evaluate how this structural template controlled the deposition and characteristics of the deep-water facies over a ca. 10 million year timescale.

| Data and structural template
The 3-D seismic data volume was processed to near zero-phase and is displayed using Society of Exploration Geophysicists (SEG)-normal polarity. The data were migrated using Kirchhoff pre-stack migration and bending ray post-stack migration to generate a 12.5 m by 12.5 m grid with a 4 ms twtt (two-way travel time) sampling interval. The data were provided every four in-lines and cross-lines, giving a bin size of 50 m, which corresponds to the maximum horizontal resolution. Based on a frequency content of 35 Hz and average seismic velocities   ranging from 1800 m/s (Plio-Pleistocene) to 4000 m/s (late Oligocene-early Miocene), the vertical seismic resolution (limit of separability) is estimated to range from 13 to 30 m respectively. The seismic volume was depth converted as described by Pizzi et al. (2020). Seven key horizons subdividing 6 units (representing intervals of time from 15 to 3.7 Ma) were identified and interpreted throughout the dataset, the latter labelled unit 1 to unit 6 ( Figure 2a,c). These units form the basis for this study and are the same units presented in the statistical analysis of Pizzi et al. (2021). A depth-structure map of the 9.5 Ma horizon, which is within the syn-kinematic section identified by Pizzi et al., 2020 (yellow horizon in Figure 2a), illustrates the significant structural control exerted by gravity-driven shortening on sedimentation in the region, with depocentre development partitioned by fault propagation folds, with a structural relief of >2 km for the 9.5 Ma horizon.
For each unit, one or more RMS (root-mean-squared) amplitude extractions were made in isoproportional windows. Together with seismic sections, these were used to make an interpretation of the seismic facies in the study area (e.g., channels, lobes, etc.), following the classification in Figure 3, which is outlined in the section below. To assess and quantify the relationship between growing structures and coeval sedimentation we derived isopach and sedimentation accumulation rate maps from the depth-migrated 3D seismic volume, for each of the 6 units, unit 1 to unit 6, from top to base. This allowed the changes in the thrust belt depocentres to be documented through time as well as identifying local areas where sediment thickened between thrusts and thinned onto thrust hanging-walls as a function of evolving thrust segment activity. An example of an isopach for the oldest unit studied (unit 6; 15-12.8 Ma) is shown in Figure 4a. Erosional truncation of horizons (e.g., Figure 2, F17) occurs in areas of maximum cumulative strain (and uplift) after 3.7 Ma which is the end of the study interval. This is reflected by blank areas on the isopach maps where stratigraphy has been eroded.
Our results are presented as palaeogeographical maps, which are overlain on the relevant interval strain rate map for the region, where the maps for units 6, 5 and 4 are directly from Pizzi et al. (2020) (e.g., Figures 4 and 5). In the work presented here, however, the number of interval and cumulative strain maps have been increased for the inboard parts of the study area deformed by faults 22-19. The thicker stratigraphy allowed an extra two horizons (6.5 and 5.5 Ma respectively) to be mapped and used for strain analysis and additional strain maps for the intervals between 7.4-6.5, 6.5-5.5 and 5.5-3.7 Ma have been calculated, comprising units 3, 2 and 2 respectively. Age constraints for these additional two horizons are based on the assumption of constant sedimentation (see Pizzi, 2019).
For the outboard parts of the array (faults 18-11), the finer subdivision was not possible, so the interval and cumulative strains are averaged for the longer time period of 7.4-3.7 Ma. Therefore, the three youngest intervals have composite strain maps, where the strain for the outboard thrusts is averaged over the three youngest intervals. This means the strain maps for the younger intervals have new more detailed strain data for the inboard thrusts, 22-19, but the outboard thrusts are shown with rates averaged for units 3-1 as calculated in Pizzi et al., 2020). Our approach allows a direct comparison between fault activity and the evolution of facies elements such as slope channels and lobes both along, and across, the strike of the thrusts within each time interval. By comparing sequential maps through time the architectural style of the slope facies can be documented for the evolving fold and thrust belt, from the early formation of isolated thrust segments through to the fully linked thrusts in youngest times.

| Seismic facies definition and structural domains
The seismic facies observed in the 3D volume were classified based on their cross-sectional external and internal geometry and architecture, plan view form, character and seismic amplitude of internal reflections, dimensions and relationship with underlying stratigraphy (erosional or concordant) and type of interaction they exhibit with the structures (c.f. Pizzi, 2019, Pizzi et al., 2021. Our interpretations are similar to those of other studies of the fold and thrust belt of the Niger Delta (Deptuck et al., 2003;Mitchell et al., 2021b;Zhang et al., 2018). Four main facies types are identified: channel forms, lobe forms, masstransport complexes and background slope sediments ( Figure 3). These facies are further divided into five types of submarine channels based on channel morphology (leveed channels, erosional channels, low relief channels, linear channels and channelised sheets) and lobate elements that may be unconfined or ponded (c.f. Spychala et al., 2017). The sand prone facies are set in a background of hemipelagic sedimentation. A summary of the key seismic facies is presented in Figure 3 with further details in Pizzi (2019). The deep-water facies have been mapped as far as possible on each palaeogeograhic map. However, in some areas identification is limited due to local degradation of the data quality. This is particularly an issue in the most highly deformed areas. In identifying these facies for this regional study we recognise that there are many detailed aspects of the facies geometries and associations that we have not included and are beyond the scope of this study. No detailed work has been undertaken on the section below 15 Ma (i.e., beneath unit 6) but seismic F I G U R E 3 Summary of the seismic facies used in this study. The scheme recognises a range of channel types, several sheet sands and lobes, mass transport complexes (MTCs) and hemipelagic deposits. Full details are in Pizzi (2019).

Seismic Facies
Description O ccurrence Continuous background deposition.
Channel axis within deep erosional surface. Cycles of downcutting and deposition; presence of terraces. Compex internal architecture with low to high amplitude reflectors that are continuous to chaotic.
Channel axis with gull-wing levees on sides. Channel axis has high amplitude internal reflections; levees low to moderate amplitudes.
Vertically stacked high amplitude, semi-continuous reflections dissected by numerous small, low-amplitude, u-shaped channels.
Elongated, kilometer-scale in section, high amplitude features where top and base amplitude responses increase and dim simultaneously.
Low amplitude (mud filled) channel axis with high amplitude levees or wings. Occasionally, levees are missing.
Laterally extensive and generally continuous high-amplitude reflections with tabular to wavy geometry. Internally organised into a hierarchy of lobe complexes, lobes and lobe elements.
Laterally extensive high-amplitude reflections with tabular geometry. Plan-view geometry controlled by shape of depocentre. Lobes onlap hangingwall of structure which limits their downdip extentent Highly chaotic, discontinous low to moderate amplitude reflections. The basal surface can be erosional; upper surface is generally smoother and can be mounded.
Low to moderate amplitude packages reflections that are almost transparent. They generally drape and smooth older topography. They have some variable lateral continuity but some packages can be correlated for tens of kilometres.
Occur from just before 7.4 until 6.5 Ma (Unit 3) but never dominant.
Occur from 15 to 12.8 Ma (Unit 6) most commonly but not dominant.
Most frequent from 5.5 -3.7 Ma (Unit 1). Occur occasionally between 9.5 and 7.4 Ma (Unit 4) and from 6.5-5.5 Ma (Unit 2). amplitude maps through this section indicate that the facies are dominated by unconfined lobes on an area of the lower slope with no significant topography (Pizzi, 2019).
To facilitate the results and discussion presented in the following sections, we sub-divide the outer fold and thrust belt into four areas (Figure 2b), 1 being the most proximal and 4 the most distal. Area 1 is the largest subregion and represents the most inboard part of the fold belt. It is bounded by the seismic data limit to the north, a tear fault to the east (Mitchell et al., 2021a(Mitchell et al., , 2021b and thrust 21 to the south and it incorporates thrust 22. Areas 2 and 3 represent the core of the fold belt and are bounded by thrust 21 to north and by the structural lineament of thrusts 13, 15 and 17 to the south. Internally, these areas incorporate thrusts 20, 19 and 18. Areas 2 and 3 have been separated due the presence of the central the north and thrust 11 to the south. Internally, it incorporates thrusts 12 and 14. Finally, we investigate how submarine channels reacted to across and along strike variations in fault strain in time and space. To do this, we used the published cumulative strain variations along strike for the faults in the study area (see Pizzi et al., 2020, their Figure 10). For this analysis, we focus on selected faults in areas 1 and 4 which are documented to have different strain rate histories. We plot locations of submarine channel crossing points, and their seismic character/morphology ( Figure 3) on to the cumulative strain profiles over time. We use these data to establish how the spatial distribution of the channels, and their nature (e.g., erosional or aggradational) is controlled by the temporal and spatial evolution of faults in the array as a whole.

Isopach and accumulation rate maps
The interval has an average unit thickness of 240 m with a general trend of decreasing thickness from 450 m in the NE to 180 m in the SW (Figure 4a). Localised NE-SW thickening and thinning trends (Figure 4a) are caused by the presence of submarine channels either within this unit (thicker) or eroding into this unit from the unit above or creating topography from the unit below (thinner). The thickness distribution within this interval does not show the impact of tectonic activity as clearly as some of the later units and the isopach is largely interpreted to represent the general seaward thinning of a distal delta wedge. The sediment accumulation rates ( Figure 4b) average ca. 0.1 mm/yr (triangle in Figure 4b), ranging from 0.06 mm/ yr in the S and SE to 0.22 mm/yr to the N-NW.

Structural context
The strain analysis shows that the deformation started near 15 Ma, initially as individual segments although a few fault segments were partially linked (e.g., 20, 17, 11; Figures 4c and 5b). By 12.8 Ma ( Figure 4d) the fault segments across all the fault arrays had laterally propagated and started to link into longer segments. Strain rates for this interval were generally low across all the faults (up to 1%/Ma; Figure 5b) with the central portions of the fault arrays shortening at higher rates of 2%/Ma and exceptionally up to 3%/Ma along faults 19, 17 and 15 (Figure 5b). Cumulative strain across the region was still typically in the order of 1%-5%; however, more segments accumulated 6%-10% of shortening and two faults (17, 15) had segments shortened by 11%-15% (Figure 4c,d).

Seismic facies and structural impact
Attribute maps extracted from this unit show that seismic facies primarily consist of linear channels in areas 1, 3 and 4 with unconfined lobe complexes present in area 1, inboard of fault 21 (Figure 5a,b). In the older large lobe complex (LC1) to the SE of area 1, at least four individual lobes can be identified. Each lobe is composed of four to five smaller lobe elements (Figure 5a,c). A second younger lobe complex (LC2) occurs inboard of fault 21 and 22 to the NW of area 1, which appears to be composed of several lobes and lobe elements of comparable size to LC1 (Figure 5a,b). The influence of structures on the location of LC1 and LC2 is inferred from fault 21 limiting their southern extent ( Figure 5). The lateral extent of LC2 is focussed between two active segments of fault 22. The strain rate map suggests that rates of deformation in the order of 1%-2%/Ma were enough to control the distribution of the lobe complexes. In addition to the lobe complexes, widespread linear channels are present throughout the interval. On a regional scale, these channels flow downslope without showing major deflections (Figure 5b,d). The channels, in places, were deflected either to the side or between fault segments for strain rates as low as 1%/Ma (Figure 5b).

Isopach and accumulation rate maps
This unit is approximately 280-300 m thick in area 1 and thinner (150-200 m) in areas 3, 4 ( Figure 6a). On a regional scale this unit appears to broadly compensate the thickness distribution of the underlying unit 6. Localised, short-wavelength variations in thickness are caused by the presence of submarine channels either within this unit (thickening) or eroding into this unit from the unit above or creating topography from the unit below (thinning) (Figure 6a). The sediment accumulation rate (Figure 6b) averages 0.06 mm/yr during this interval, which is half of that during unit 6, resulting in a substantially thinner unit.

Structural analysis
During this interval all the faults in the area were active at low strain rates (up to 1%/Ma), while a few segments along the strike of faults 17 and 20 deformed at 2%/Ma (Figure 6c). Many of the faults had continued to propagate laterally and by 9.5 Ma, the faults in the core of the fold belt (20, 17, 18, 15) had almost fully linked along their entire strike length. The cumulative strain had now reached 6%-10% for many fault segments and locally up to 16%-20% (Figure 6d). Away from the central area, some faults still had unlinked segments.

Seismic facies and structural impact
This interval is dominated by channels with occasional small lobate forms (Figure 6e,f). Significantly, sedimentary systems continued to be widespread across the region for along-strike distances of more than 40 km. The channels entered the study area from the north-east of area 1 and crossed it along two main pathways. One set of channels (12-5, 7, 8) with low sinuosity, and identified as linear channels, flowed south-west (towards areas 3 and 4), while another set of more sinuous leveedchannels (12-1, 2, 3, 4) flowed in a more westerly direction (area 2; Figure 6e,f). The structural influence is evident on both styles of channels. The linear channels (12-5 and 12-8) were diverted around the active segments of fault 22, shortening at 1%/Ma (Figure 6f), predominantly exploiting the inactive zones along the strike of this fault and passing through strain minima between unlinked segments ( Figure 6f). The greater strain rates along faults 20 and 17 also diverted the linear channels to the south and south-east, where they exploited areas of lower strain rate (1%/Ma) along the strike of the faults. The linear channels also exploited relay zones between faults, where faults overstep, as between faults 17 and 18 (Figure 6f). The leveed channels were diverted to a more westerly route by fault 22, terminating with the deposition of lobes up-dip of segments of fault 20 and 17 that were deforming at greater strain rates (2%/Ma, Figure 6e,f).

Isopach and accumulation rate maps
Unit 4 records a period of major change in the basin configuration as the thickness of sediment varies markedly across the study area ( Figure 7a). The eastern portion of area 1 shows a north-east to south-west trending zone of >700 m thinning to 400-500 m. Thickness minima of 100-150 m are present along strike of most of the faults, and on the back limbs of faults in area 2.
Sediment accumulation rates in this interval (Figure 7b) were up to three times greater than the underlying Unit 5, reaching 0.3 mm/yr, decreasing to 0.06 mm/yr in the core of the fold belt.

Structural context
This period records a marked increase in regional strain rates (Figure 7c). In the inner domain of the fold belt (areas 1, 2 and 3) all the faults reached background deformation rates up to 3%/Ma and with many segments shortened by 4%-5%/Ma. However, area 4, the outer domain of the study area, was still deforming at lower strain rates (up to 1% and 2%/Ma) as the deformation front had not propagated fully into this area (Figure 7c). By 7.4 Ma (Figure 7d), all the faults in the inner domain had fully linked, although their cumulative strain was generally not higher than 10%-15% with only a few fault segments reaching a total shortening of 21%-25%. By contrast, faults 11, and 14 in the outer domain (area 4), had not fully linked and their cumulative strain did not exceed 5%-10%.

Seismic facies and structural impact
This interval remained dominated by leveed-channel complexes (Figure 8), particularly in the north-west, with minor lobate forms and low-relief channel levees. Notably, the areal distribution of the sedimentary systems become more focused compared to the underlying unit 5. In the lower part of the interval the sedimentary systems (time 4a) flowed in a similar NE-SW direction to unit 5 (Figure 8a), but subsequently (time 4b; Figure 8b,c) entered the area from the N and flowed to the S or SW. In the lowest part of this interval (time 4a Figure 8a) leveed channels were deflected to the west avoiding the areas of greater strain rates in the centre of faults 20, 21 and 22 (up to 4%-5%/Ma). Channel 9-1 terminated with a lobe on the back-limb of fault 20 which was deforming at a rate between 2% and 3%/ Ma. Subsequently (time 4b, Figure 8b) a large channel levee complex (9-8) is clearly seen responding to the developing structure as it diverted around the eastern tip points of faults and the max strain rate zone of fault 21. Further downslope it is deflected towards the southwest, exploiting a relay zone between faults 18 and 17. In all cases the channel can be seen to be avoiding the fault segments with the highest strain rates. At the western end of fault 22 two leveed channels (9-4 and 9-5) avoid the highest strain rates of 4%-5%/Ma. Lobes were deposited up dip of fault 22, which was deforming at strain rates of 4%/Ma (Figure 8c) and at the termination of channel 9-6 on the back of fault 21.

Isopach and accumulation rate maps
The interval shows significant changes in the distribution of sediment thicknesses across the slope (Figure 9a

Structural context
Unit 3 records the peak deformation for the growth history of the fold belt, particularly in the inner domain (areas 1, 2 and 3; Figure 9c). Here, faults 21, 20 and 19 achieved deformation rates higher than 7%/Ma along much of their strike length. The outer domain (area 4) also underwent higher deformation rates, although lower than in the inner domain. Faults 13 and 11 deformed at rates up to 6%/Ma, indicating that the deformation front was propagating basinward. By 6.5 Ma, all the faults in the area had completely linked (Figure 9d). On average all the faults had accumulated a maximum shortening of 15%-20%, except for faults 17 and 13, in areas 3 and 4 of the fold belt, that reached 26%-35% shortening. The maps for the structures in the outboard area, faults 11-18, have interval and cumulative strain rates for the entire unit 3-to-1 interval as discussed in the methodology section.

Seismic facies and structural impact
This interval is initially dominated in the inner domain by lobes ( Figure 10) and subsequently by a series of mainly erosional channel complexes. This interval also records the maximum restriction on the areal distribution of the sedimentary systems. Initially (time 3a; Figure 10a,b), lobes were deposited inboard of fault 22 and onlap the back limb where interval strain rates reached a maximum of 5%/Ma (Figure 10a,b). Erosional channels (7-1, 7-7) were deflected around the lateral tips of fault 22. Channel 7-7 terminated with the deposition of a ponded lobe within the syncline formed between faults 22 and 21 (Figure 10a). Subsequently (time 3b; Figure 10c) erosional or mixed erosional-aggradational channels (7-3, 7-4, 7-5) and low-relief channel levees (7-0, 7-2) entered the area from the north-east. Fault 22 directed their courses towards different parts of the basin. The channels passed through locations at either the lateral tips of the fold (channels 7-2, 7-5) or the points of older segment linkage along its strike, which can be seen in comparison with strain rate maps for units 5 and 6 (channels 7-3, 7-4;) avoiding the areas deforming at the greatest strain rates in the centre of fault 22 (Figure 10c). Channels 7-2 and 7-0 deposited lobes in the syncline that formed between the footwall of fault 22 and the hangingwall of fault 21 where strain rates were up to 7%/ Ma. Although the channels all exploited the locations of relatively lower strain rate along the strike of each fault and early zones of linkage between segments of faults, they were, in some places, forced to exploit relatively high strain rate locations due to the diversions further up dip. The uppermost interval (time 3c; Figure 10d,e) shows a single mixed erosional-aggradational channel, 7-6, entering area 1 from the north-east and being diverted to the west following the strike of fault 22. The channel is then deflected south-west around the lateral termination of fold 22 before exploiting a passage through fault 21 at a location of minimum strain rate along its strike (Figure 10c).

Isopach and accumulation rate maps
This interval is similar to the underlying unit, except in area 1 where a thickness maximum of 400-450 m is F I G U R E 9 Unit 3, 7.4-6.5 Ma, structure maps. (a) Isopach map. (b) Sedimentation rate map. (c) Strain rate map during unit 3. (d) Cumulative strain map at 6.5 Ma. Note that faults 18-11 have not had strain rate and cumulative strain calculated separately for this interval but show the average for units 3 to 1 (7.4-3.7 Ma). present as a north-east to south-west trending depocenter (Figure 11a).
In the rest of the inboard domain (areas 1, 2 and 3) north-west to south-east trending thrusts reflect where thickness minima (50 m or less) lie in the hangingwall of the fold-thrust structures (Figure 11a). Sediment accumulation rates during this interval decrease to an average 0.4-0.5 mm/yr in area 1, with minima of F I G U R E 1 0 Unit 3, 7.4-6.6 Ma, facies maps, all maps superimposed on interval strain rate map; scale for interval strain rate maps is as per Figure 9c.

Structural analysis
This unit records a period of decreasing strain rates, particularly in the inner domain (Figure 11c,d) where only the central part of fault 22 was still deforming, at low F I G U R E 1 1 Unit 2, 6.5-5.5 Ma, structure maps. (a) Isopach map. (b) Sedimentation rate map. (c) Strain rate map during unit 2. (d) Cumulative strain map at 5.5 Ma. Note that faults 18-11 have not had strain rate and cumulative strain calculated separately for this interval but show the average for units 3 to 1 (7.4-3.7 Ma). rates of 1%/Ma. Along fault 21, rates decreased to a maximum of 4%-5%/Ma on its north-western side, and to 2% to 4%/Ma to the south-east. Likewise, fault 20 had high strain rates (6%-7%/Ma) in the NW decreasing to 1%-2%/ Ma on the SE portion of the fault. Along fault 19, strain rates were consistently between 1% and 3%/Ma. By 5.5 Ma (Figure 11d) along the strike of faults 21, 20 and 19 segments had accumulated up to 25% of shortening with a maximum of nearly 30% in the centre of fault 21. Faults 18-11 have not had strain rate and cumulative strain calculated separately for this interval, but show the sum for units 3 to 1.

Seismic facies and structural impact
The results of the amplitude analysis within this interval show great variability in the type and architecture of seismic facies. These range from high amplitude sheets and ponded lobes to erosional and mixed erosional-aggradational channels, low relief channel levees, highly aggradational channel levees and mud-dominated sequences ( Figure 12). The lower two intervals (Figure 12a,b,d), show several mixed erosional-aggradational (6-3) and low relief channel levees (6-1, 6-2 and 6-4) converging towards the centre of fault 21, coinciding with the location of earlier fault linkage (compare to maps of unit 5 and 6), and avoiding the higher strain rates developed to the north-west and south-east. These channels were also able to cross fault 22 which at this time was deforming very slowly (strain rate of 1%/Ma). The remainder of the lower two intervals are characterised by two deeply erosional channels (6-5 and 6-10). Channel 6-5 bypasses the tip of fault 21 and develops a sinuous route exploiting locations of low strain rates along faults 20 and 19 and relay zones between faults 18, 17, 15 and 13 (Figure 12dii). Lobes were deposited up dip of zones of greater strain rate developed along fault 21 (Figure 12a,b).
The uppermost interval ( Figure 12c) contains a meandering low-relief channel levee belt (5-1) developed within the north-east to south-west trending thick depocenter in the centre of area 1 (Figure 11a). The channel is deflected towards the northern tip of fault 22 (Figure 12c) and is deflected into a zone of local minimum strain rate along the strike of fault 21 (Figure 12c). This unit ends with the deposition of a large mass-transport complex (MTC) in the centre of area 1 which is compensationally stacked to the south of channel 5-1 (Figure 12c). The grooves formed by its basal erosional surface indicate that the MTC is deflected into the area of lowest strain rate in the centre of fault 21 and 20 (Figure 12c). The deposition of the MTC resulted in infilling of residual relief created by fault 22. This interval again demonstrates that the channels cross structures at zones of relative lower strain rate with respect to the adjacent fault segments, but also clearly dependent upon the diversions and avulsions they took further up dip. Erosional channels are able to cut through structures with strain rates of 2%-4%/Ma, and lobes are located up dip of zones above 4%/Ma.

Isopach and accumulation rate maps
A large depocenter, up to 450 m, occurs extensively across the entire area inboard of fault 21 with the crest of fault 21 remaining an area of minimum thickness (50-100 m) (Figure 13a). Depocenters are also developed in areas 3 and 4 (both up to 450 m) with thickness minima along the crests of faults 11, 14, 19 and 20 and over most of area 2. Sediment accumulation rates during this interval continued to decrease to values of 0.2-0.25 mm/yr within the depocenters and to less than 0.1 mm/yr along the crests of the thrust-folds (Figure 13b).

Structural analysis
During the deposition of this unit strain rates decreased in the inner domain, while higher strain rates (summed for units 3-1) were still present in the outer domain (Figure 13c,d). In particular, fault 22 completely stopped growing and fault 21 slowed down to values of 1%/Ma in its NW part and locally 2%/Ma in its eastern half. Only a few segments along faults 20 and 19 were still active with strain rates of 1%/Ma. By the end of unit 1, faults 11-18 record a total amount of cumulative strain of up to 35%, similar to structures in the inner domain. As per units 3 and 2, Faults 18-11 have not had strain rate and cumulative strain calculated separately for this interval, but show the sum for units 3 to 1.

Seismic facies and structural impact
This interval is dominated, in the inboard domain, by channelised sheets, low-relief channel levees and MTCs, while erosional channels are present in the outboard domain (area 4; Figure 13e). The distribution of the numerous channelised sheets and low-relief channel levees is controlled by both structural interaction with fault 21 and compensational stacking with the MTCs, which continues from unit 2c (Figure 13e,f). The largest MTC is deposited above the unit 2 MTCs and although a few channels cross this area most are focussed to the northwest and southeast of these deposits. The channelised facies flow in a northeast to south-west direction and are widespread due to the low strain rates throughout areas 1 and 3. These facies can be grouped into three main sets, each consisting of three to six individual channels. They converge towards and cross fault 21, which was deforming at a strain rate of 1%-2%, at a number of locations, each marked by small troughs on the isopach map along the strike of the fault (Figure 13a). They are then are largely diverted to the north-west and south-east of faults 19, 17 and 15, because strain rates on the faults reached over 2%/Ma, and through the inactive segments of fault 20. In the outer domain, a series of erosional channels (Figure 13e) are present in the centre of area 4 following the location of the depocenter. The erosional channels are observed to exploit the relay between faults 14 and 15.

| Location versus architecture of deep-water facies interacting with growing structures
Our data set provides compelling evidence for how proximity to structures, and their relative growth history determines the channel pathways and architectures documented here. These relationships are exemplified by the along and cross strike differences in channel F I G U R E 1 2 Unit 2, 6.5-5.5 Ma, facies maps, all maps superimposed on interval strain rate map; scale for interval strain rate maps is as per Figure 11c

| Area 1
The stratigraphic evolution in area 1, the most inboard part of the fold belt, is represented by four broad facies associations: (1) Unconfined lobes between 15 and 12.8 Ma; (2) Linear and leveed channels, low-relief channel levees and sheets between 12.8 and 7.4 Ma; (3) Ponded lobes, erosional and low-relief channels between 7.4 and 5.5 Ma; (4) Channelised sheets, low-relief channel levees and MTCs between 5.5 and 3.7 Ma. Within this overall pattern, the behaviour of slope channels relates to the strain profile of the faults studied, which is clearly exemplified in fault 22, located in the north of area 1 (Figure 14aadapted from Figure 10 of Pizzi et al., 2020). The early slope channels (12.8-7.4 Ma; green and light blue dots) are distributed along the entire strike-length of the fault. They have aggradational, leveed forms, and interacted with a number of unlinked fault segments that likely created only subtle topography (c.f. Pizzi et al., 2020). Therefore, the channels were generally able to exploit relative topographic lows which represented zones of minimal strain between the active segments (c.f. Pizzi et al., 2021). The occurrence of erosional channel forms (red and orange dots; Figure 14a) directly relates to the time of complete fault linkage along strike at 7.4 Ma, when (a) the fault strain profile approached its present-day shape and (b) there was high cumulative strain and strain rate in the centre of the fault that progressively decreased towards the lateral tips. The occurrence of ponded lobes (after 7.4 Ma) is restricted to the south-eastern half of fault 22 (Figures 10a and 14a), which is a relative area of low strain rate along fault 22, but up dip of the high strain rates recorded for fault 21 after 7.4 Ma.
Early aggradational leveed slope channels (Figure 14b, green and light blue dots; 12.8-7.4 Ma) also interacted with the segmented fault 21, located directly to the southwest, and were widely distributed. This fault ultimately accumulated greater strain than fault 22. In the same way as for fault 22, the subsequent channels (7.4-6.5 Ma; red and orange dots) became erosional as the fault segments interacted, grew and accumulated strain. However, these channels were generally concentrated within the zone, at ca. 7.4 Ma, that was a relative strain and strain rate low, albeit against a background of high average strain rate at this time. The distribution of ponded lobes is generally up dip of areas of greater structural growth (Figure 14b).
It is important to note that the area of maximum strain in the centre of fault 22 balances the zone of strain deficit at the same strike distance on fault 21 (c.f. Pizzi et al., 2020), particularly between 9.5 and 7.4 Ma. Until 9.5 Ma the two faults were mostly segmented, with leveed channels widely distributed along both faults, but after 9.5 and 7.4 Ma the greater development of fault 21 at the south-east and north-west portions, with large cumulative strain, deflected the incoming flows towards its lessdeformed centre (Figure 14a). However, in order to achieve this, some slope channels were forced to cut through the crest of fault 22, situated up-dip of this strain minimum (orange dots, Figure 14a) but with a demonstrably lower growth rate at this time. These insights suggest that the location of submarine channels in an evolving fault array cannot always be understood at the scale of a single structure and consequently, the behaviour of the submarine channels, and their facies architectures, may reflect the across and along strike interplay of multiple faults in the array. We return to this issue in the discussion.

| Area 4
The evolution of the stratigraphic architecture of area 4 (outer domain) is also characterised by a distinct succession of facies associations: (1) Background hemipelagic sediments interbedded with linear channels and minor sheets between 15 and 9.5 Ma; (2) Minor MTCs followed by mostly leveed channels between 9.5 and 6.5 Ma; (3) Erosional channels and MTCs between 6 and 3.7 Ma. However, the vertical changes in architectural style in area 4 occurred later with respect to the inner domain. This reflects the later accumulation of high strain rates, owing to the basinward migration of the deformation front towards area 4 from area 1. Consequently, leveed channels comprised the dominant style for a longer time interval until 6.5 Ma, resulting in the simultaneous occurrence of contrasting depositional styles across the slope as a function of the growth history of the fault array. Across-strike and along-strike interactions between the faults also influenced the pathways of the channels (Figure 15). For instance, almost all channels are located in the relay zone between fault 14 and fault 12 throughout the life time of the array (Figure 15a). However, this results in channels being directed towards the centre of Fault 11 (shown as Fault 16 in the original data of Pizzi et al., 2020, their Figure 10), even once this structure starts to develop significant strain after 6.5 Ma (Figure 15b). Consequently, this phase coincides with the development of erosional channel forms. Aggregating the cumulative strains of fault 14, 12 and 11 (Figure 15c), channels remained located in a zone of lower relative strain, which lay down-dip of the maximum strain accumulated in the core of the fold and thrust belt (areas 2 and 3).

| DISCUSSION
The spatio-temporal analyses in the previous section demonstrate that the occurrence and changes between different seismic facies coincided with the development of thrusting in the four are studied. An example of this is the broadly diachronous evolution of the seismic facies where it followed the basinward propagation of the deformation front, as seen when the leveed channels persist for longer through time in area 4 versus areas 1, 2 and 3 where erosional channels were dominant in the same time interval.
While strain rates, shortening rates, or cumulative strains are not necessarily a direct measure of the topography that may have developed on the sea-floor during the evolution of the fold-thrust array (c.f. Christie et al., 2021;Sylvester et al., 2015), structural growth clearly influenced bathymetry during the evolution of these submarine channel systems. The published seismic data for the region (Jolly et al., 2016(Jolly et al., , 2017Pizzi, 2019;Pizzi et al., 2020) and our results demonstrate: (i) thinning and draping of stratigraphy across structures (seen in all intervals after 9.5 Ma) (Figure 2a, Figure 10); (ii) onlap of strata on to the back of growing structures (e.g., unit 3 in area 1) ( Figure 2a) and (iii) pronounced topographic expression on the present day sea-floor across some structures, for which its control on channel geometry and architecture has already been documented (Jolly et al., 2017;Mitchell et al., 2021a). Because of the way strain has been measured, by line length balancing in sequential units, high strain rates imply that clear growth sequences are present on both the limbs of a given fold-thrust structure (c.f. Burbank & Verges, 1994;Clark & Cartwright, 2012a, 2012bDon et al., 2020;Shaw et al., 2004;Suppe et al., 1992). Consequently, given the clear relationships between the F I G U R E 1 4 Cumulative strain (e) in percent over time against distance along strike for (a) fault 22 and (b) fault 21. Data are adapted from Figure 10 of Pizzi et al. (2020). The approximate location of submarine channel-fault crossing points are located along strike, taken from the unit maps presented in Figures 6-12, at a cumulative strain equal to half the difference between the successive time slices. However, erosional channels from 6.5 Ma are plotted in (a) on the present-day cumulative strain profile as this fault had accumulated the majority of its total strain by this time. Approximate shortening is plotted on the right-hand y axis. architecture and routing of the seabed channels across the fault array as a whole-where channels predominantly cross faults in regions of low strain rate-we deduce that the evolution of fold-thrust structures has controlled channel behaviour and facies in space and time by influencing sea bed topography, likely over multi-kilometric scales. The following sections consider two major points: First, we document how across-and along-strike F I G U R E 1 5 Cumulative strain (e) in percent over time against distance along strike for (a) faults 14 and 12; (b) fault 11 and (c) the aggregate of faults 11, 12 and 14, presented in terms of normalised strain. Data are adapted from Figure 10 of Pizzi et al. (2020), noting that fault 11 is numbered as fault 16 in this publication. The approximate location of submarine channel-fault crossing points are located along strike, taken from the unit maps presented in Figures 6-12, at a cumulative strain partitioned equally between the successive time slices bracketed by the channels. interactions between neighbouring thrusts in the evolving array influence the stratigraphy; and second, we discuss what conceptual models can be built regarding the temporal evolution of strain rate and its impact on the distribution of depositional facies and architecture in structurally complex settings such as these.

| Impact of fault growth and linkage on channel architecture
Our results illustrate how the distribution and architecture of slope channels in the deep-water Niger Delta system are controlled by the temporal evolution of the fault array. But the data presented in Figures 14 and 15 demonstrate that a complete understanding of the spatial distribution and architecture of sedimentary systems is dependent upon simultaneously analysing the effect of multiple structures across strike. In other words, to understand fully how sedimentary systems respond to growing seabed deformation, structures cannot simply be analysed in isolation.
In thrust fault arrays, as Pizzi et al. (2020) demonstrate explicitly, across-strike interaction through detachment linkage is particularly important to consider. This process can efficiently transfer strain across strike in a complex 3D interaction that can create highly asymmetric straindistance profiles (Pizzi et al., 2020) and which can be revealed by the aggregate strain profiles of multiple faults (e.g., Figure 15). These fault interactions likely modify seabed topography over a length-scale greater than individual structures considered in isolation. Consequently, the effect of multiple interacting faults is an under-recognised but critical control on the routing and architecture of deepwater channel systems on structurally complex slopes.
When faults are segmented, slope channels can be widely distributed across the basin as they exploit multiple linkage zones along their strike (e.g., Pizzi et al., 2021). The channels may follow a tortuous route through a population of incipient folds and their overall distribution can be partially dictated by compensation of the topography created by the previous systems. In fact, because they largely avoid the crests of the incipient folds, they aggrade with time and develop levees. At the early stages of deformation, none of the faults has a dominant effect on the creation of topography as they are not yet linked through a common detachment, as we discuss in the following section. However, when faults completely link, both along strike and across-strike through detachment linkage, our results allow us to develop two end-member cases that can simultaneously exist across slope systems affected by contractional deformation (Figure 16). In the first case, a smaller, and potentially more buried thrust-fold (A) is located up dip, resulting in a linked strain deficit in the centre of a larger down dip structure (B) in the footwall of A. As the small fold A dies out laterally, the large structure B shows two corresponding strain maxima (B1 and B2). Because the two strain maxima on B represent considerably greater shortening and relief than the crest of A, which may affect bathymetric gradients updip from the fault crests over a spatial scale of several kilometres, channels will respond primarily to fault B-created topography and will be deflected towards the centre of this structure, where a strain deficit (and a relative topographic low) is present and where the down-system 'exit' point from the local fault network is located. As a result, channels cross smaller fold A near its crest, eroding into it, which explains why channels may be located at the structural culmination of individual fold-thrust structures even where these were not fault segment boundaries earlier in the evolution of fault array. The channels that erode into the fold crest typically are u-shaped and may offset laterally. In contrast the channels that are focussed into the area of strain deficit become vertically stacked with an overall erosional architecture because of the topographic forcing. This type of fold interaction has been dominant in the inner domain of the fold belt and is shown by fault 22 and 21 in area 1 (Figure 14).
The second type of fold interaction ( Figure 16) shows an opposite scenario where a large fold (C) is present up dip of two or more smaller, unlinked, structures (e.g., D, E) in its footwall. The large amount of shortening accommodated by C leaves a relatively undeformed area down dip of C where small structures D and E tip out laterally. In this end member, slope channels are diverted around C, which is expected to have marked topographic relief, and are subsequently deflected towards its relatively undeformed footwall, which is a locus of sediment deposition. However, their course is also confined by the topography of folds D and E that focus the channels around their lateral tip points. This causes the channels to develop a highly aggradational architecture as subsequent flows are persistently focused on the same location at the lateral tips of D and E, as long as Fold C dominates the accommodation of strain in the local area ( Figure 16). This mode of interaction is documented in area 4 (outer domain; Figure 15). During the time of peak deformation (7.4 and 6.5 Ma) both the two modes outlined above coexisted in the study area and resulted in the simultaneous occurrence of different architectural styles across the slope.
The driving mechanism for focussing channels into the areas of lower strain rates has been shown to be a function of a subtle interaction of the timing of the development of the sea-floor topography and the origin of the channels (e.g., Clark & Cartwright, 2009Mayall et al., 2010). Channels will deflect around pre-existing topography into areas of lower strain rates and/or will migrate laterally into these zones as the topography is developing. Mitchell et al. (2021b) developed a series of models showing this effect for channels in the shallow subsurface of the same area. We suspect that the channels in the 15-3.7 Ma interval will also show this variation in architecture depending F I G U R E 1 6 Schematic models of submarine channel development influenced by across-and along-strike interaction of growing faults: (1) channel development where a thrust-related fold up-slope with smaller topographic expression (A) is followed by a larger structure down-slope (B1, B2). Fold A is detachment-linked to B, but had achieved maximum growth rates earlier and is now partially buried. (2) channel development where a larger thrust-related fold up-slope (C) is followed by smaller structures (D, E) down-slope, which have not yet linked. With time faults D and E link. When they accrue enough cumulative strain the channels will become erosional as in scenario 1.
Channels erode into the crest of smaller structures (A) due to topographic effect of larger folds nearby (B1 & B2).Channels are laterally offset with flat or U-shaped base.
Channels are directed towards the central depression of a linked fold with larger amounts of shortening and bathymetric relief.
Channels are vertically stacked with V-shaped bases because of steep lateral gradient changes in the zone of strain deficit between folds B1 and B2.
High lateral gradients between unlinked faults D and E focus the flows towards the central underformed footwall of fault C.
on the detailed timings of the sea-floor topography and channel development locally.

| Distribution and architecture of deep-water facies during the evolution of a thrust array
Our data present a compelling reconstruction of how sediment routing across the fold-thrust belt of the Niger Delta changed through time in response to the evolving, structurally-induced slope configuration. A key finding is that the distribution of facies strongly correlates with the variations in strain rate and shortening along the strike of the faults. This outcome can be seen as the corollary of a statistically based study in the same area, which illustrated that the routing of submarine channels was through regions of lower-than-average strain rate throughout much of growth history of the fold and thrust belt (Pizzi et al., 2021). The results also allow us to build a new conceptual model to understand the behaviour of slope systems in these settings. Figure 17 presents our general stratigraphic and channel behaviour model for the area which links strain rate (as an indirect measure of topography) to the evolution of fault linkage (as a measure of lateral extent of the topography and the maturity of the fold belt).
The initial two stratigraphic intervals (unit 6-15 to 12.8 Ma; unit 5-12.8 to 9.5 Ma) are dominated by linear channels and leveed-channels and, to a minor extent, unconfined or ponded lobate forms (Figure 17, point A). Importantly, the channels are evenly distributed across the lower slope as they interact with subtle topography created by numerous incipient, highly-segmented thrusts. Leveed channels continue to be the dominant architectural style through to unit 4 (9.5-7.4 Ma). However, by 7.4 Ma channels interact with fault segments that have propagated laterally and linked (in the inner domain), forming continuous thrusts that deform at faster strain rates along most of their strike length (up to 5%/Ma). This results in the channels becoming more spatially restricted/ focused, locally larger in dimension and with some major deflections away from the faster growing fold segments ( Figure 17, point B). The channels are forced to exploit the same pathways for long periods of time, resulting in the stacking and aggradation of multiple flows, forming highly aggradational channel levees. Linear and leveed channels are, therefore, characteristic of the initial stages F I G U R E 1 7 Synthetic diagram summarising the key changes in channel pathways and architectures documented in this study as a function of strain rate and fault linkage through time.

-9.5 Ma
Segmented faults cause minor deflections of channels between active segments. Channels exploit topographic lows and aggrade.
As faults link, longer thrust segments cause major deflections around fault tips and channels become focused and highly aggradational When most faults have linked and strain rates are high, large ponded lobes are deposited, followed by channels that cut through structural culminations and become erosional. Channels still exploit earlier sites of linkage and sites of low cummulative strain After the peak deformation the topography that has been created causes major diversion around fault tips and channels decrease their depth of incision of deformation when the thrusts in the lower slope had accumulated no more than one third of the total shortening (10%-15%). Strain rates are low, and the resultant seabed topography is subtle. At this time, the segmentation of the thrusts allows the channels to avoid structural culminations and exploit undeformed areas between fault segments favouring their aggradational nature. However, with the subsequent increase in fault linkage and strain rate leveed channels are deflected from areas of higher topography, more spatially restricted and highly aggradational. Soon after 7.4 Ma, until 6.5 Ma (unit 3, Figure 17, point C), the inner domain undergoes the highest recorded deformation rates (≥7%/Ma) which causes a major change in the basin geometry as folds achieve their maximum topographic expression. Large depocenters are, at first, created becoming the locus of ponded lobes deposited up-dip of the fast-deforming controlling faults. The subsequent channels are continuously restricted by the location of strain deficits and relative lower stain rates along the strike of faults. However, they now develop an erosional morphology as their long profiles are perturbed by ongoing deformation (c.f. Mitchell et al., 2021a). Therefore, erosional channels are characteristic of strain rates in excess of 3%/ Ma and high degrees of both along strike and detachment fault linkage when the lower slope is divided into discrete piggy-back basins. The folds have higher topographic expression and channels cannot exploit lateral tips being directed towards structural culminations; some channels still exploit earlier sites of linkage. During this period, the complex across and along strike interaction of faults critically determines the pathways and architectures of slope channels, as depicted in Figure 16. Deposition of lobes can occur updip of strain rates as low as 1%-2%/Ma, but more typically they occur for strain rates in excess of 4%/Ma. Lobes, therefore occur in lower strain rate areas located up dip of the zones of greatest strain rate along the strike of faults, while channels generally avoid these areas. In the outer domain the peak of deformation was later and less intense than the inner domain. Aggradational channellevee complexes continued to be deposited and were deflected around the lateral tips of unlinked faults. In the interval between 6.5 and 3.7 Ma (units 2 and 1, Figure 17 points D and E) strain rates in the inner domain progressively decreased to values of <2%/Ma and eventually resulted in the development of large 'shallow-ponded' basins in areas 1 and 3 (unit 1 Figure 17e; Figure 11a) dominated by channelised sheets interbedded with MTCs. In the core of the fold belt, where strain rates are in excess of 2%-3%/ Ma, the channels are deflected and restricted to the side of the central core, and most probably are responding to the inherited seabed topography from the preceding phase of peak deformation. Nevertheless, for similar strain rates as those in the earlier 15-9.5 Ma (units 5 and 6) interval, the depositional style is different, dominated by channelised sheets versus leveed-channels. We interpret this to be the result of the different basin geometry, where channelised sheets are deposited in response to a pre-existing topography that is almost, but not completely healed.
The general model presented here for the facies development during the evolution of a slope fold-belt should be applicable to all fold belts that show a similar temporal and spatial pattern of development that is initiation of structures and lateral linkage, a period of maximum deformation followed by a reduction of strain rates. In the Niger slope example, the initial development of the inner fold belt, followed by increased deformation of the outer fold-belt has also resulted in down-dip changes in the dominant facies, particularly channel style, through time. Our analysis shows that a sophisticated prediction of sedimentary facies evolution in slope systems is now possible (c.f. Howlett et al., 2021). However, to do this it is important that shortening within the fault array is considered as a whole and we show that both along and across strike variations can control the pathways, evolution and architectures of submarine channels at a range of scales.
Finally, we note that while the summary figure focusses on the impact of structure and sea-floor topography, we recognise that there are additional potential controls up-dip on the evolution of the facies from: changes in sediment supply to the delta top (see Pizzi et al., 2020 for a discussion); local sedimentation rates; the impact of climate changes through this period and the evolution of sea-floor topography created in the up-dip extensional regime. Additionally, in presenting the results at a regional scale we recognise that there are detailed characteristics of the facies which are not discussed. This includes the relationship and process controls on the different styles of channels, the changes in stratigraphic architecture of channels around structures, the association of different channels as the foldbelt evolves, the initiation point of the lobes and the origin of the MTCs. The present study provides a robust framework for further detailed studies addressing these questions.

| CONCLUSIONS
Using 3D seismic data from the deep-water fold and thrust belt of the southern lobe of the Niger Delta and building on the previous work by Pizzi (2019) and Pizzi et al. (2020Pizzi et al. ( , 2021, we investigated the relationships and interactions between structural deformation and the distribution and architecture of deep-water sedimentary systems active coevally to deformation, over a period of ca. 11 Ma. Using this structural framework, a conceptual model has been derived for the distribution and architecture of slope channels and lobes through time. The main findings of this work are: 1. To understand the evolution of deep water facies, we demonstrate that it is vital to quantify and reconstruct fault-related deformation in four dimensions. The complex, but ultimately predictable along and across strike interactions of growing structures over time fundamentally determines the facies, characteristics and distribution of slope channels, lobes and MTCs. 2. The integration of the strain data with isopach maps and the mapping of the channel systems through time has demonstrated that leveed channels occur when they are able to exploit the (early) sites of linkage that exist at the early stages of structural evolution, when the fault array is immature and segmented. However, leveed-channels continue to occur as long as they can exploit the lateral tips of growing folds. Ponded lobes and erosional channels occur when the fault array has linked in 3D, strain rates increase and lead to high cumulative strains, which likely translate into more seabed topography. Channelised sheets dominate when strain rates across the fold belt decrease, structures start to be buried by sediments and the basin geometry is that of a shallow-ponded basin. 3. The change from leveed-channels into erosional forms occurs as a direct result of how structures are arranged in 3D and their relative size. The larger/largest structures in an area have dominant effects in the creation of seabed topography such that channels primarily respond to these large folds. Because structures transfer strain efficiently across strike once they link through the detachment, channels may be directed towards the strain deficits present along the large structures, however, being forced to erode into the corresponding strain maxima on the smaller up dip folds. Conversely, if the large folds are present up dip, they will leave a wide undeformed area in their footwall, such that channels will become aggradational as they enter this zone of low cumulative strains (and likely low topography). 4. The two types of channel-structure interactions can be simultaneously present across the slope, which leads to the coeval development of erosional channels up dip and leveed-channels further outboard. However, as the deformation front migrates basinward, the inboard structures become progressively abandoned and will develop a shallow-ponded configuration dominated by channelised sheets. As deformation concentrates on the outboard structures, these completely link and start to deform at high strain rates, which causes the development of erosional forms in the outboard domain; a process that leads to the diachronous development of the architectural styles. 5. When faults first link in 3D they inherit highly asymmetric strain-distance profiles, as strain is transferred between across-strike faults, and are under-displaced with respect to the length of the new linked structures. This asymmetry results in topographic changes where flows are directed towards the zones of strain deficits. These also correspond to the zones that subsequently undergo higher strain rates as faults grow by accumulation of strain at constant lengths. Ponded lobes are predicted to occur within the strain deficits present along the strike of faults as soon as they have linked in 3D. Additionally, they occur up dip of the zones of greater/ greatest strain rate and at the beginning of the interval.