Low‐accommodation and backwater effects on sequence stratigraphic surfaces and depositional architecture of fluvio‐deltaic settings (Cretaceous Mesa Rica Sandstone, Dakota Group, USA)

The adequate documentation and interpretation of regional‐scale stratigraphic surfaces is paramount to establish correlations between continental and shallow marine strata. However, this is often challenged by the amalgamated nature of low‐accommodation settings and control of backwater hydraulics on fluvio‐deltaic stratigraphy. Exhumed examples of full‐transect depositional profiles across river‐to‐delta systems are key to improve our understanding about interacting controlling factors and resultant stratigraphy. This study utilizes the ~400 km transect of the Cenomanian Mesa Rica Sandstone (Dakota Group, USA), which allows mapping of down‐dip changes in facies, thickness distribution, fluvial architecture and spatial extent of stratigraphic surfaces. The two sandstone units of the Mesa Rica Sandstone represent contemporaneous fluvio‐deltaic deposition in the Tucumcari sub‐basin (Western Interior Basin) during two regressive phases. Multivalley deposits pass down‐dip into single‐story channel sandstones and eventually into contemporaneous distributary channels and delta‐front strata. Down‐dip changes reflect accommodation decrease towards the paleoshoreline at the Tucumcari basin rim, and subsequent expansion into the basin. Additionally, multi‐storey channel deposits bound by erosional composite scours incise into underlying deltaic deposits. These represent incised‐valley fill deposits, based on their regional occurrence, estimated channel tops below the surrounding topographic surface and coeval downstepping delta‐front geometries. This opposes criteria offered to differentiate incised valleys from flood‐induced backwater scours. As the incised valleys evidence relative sea‐level fall and flood‐induced backwater scours do not, the interpretation of incised valleys impacts sequence stratigraphic interpretations. The erosional composite surface below fluvial strata in the continental realm represents a sequence boundary/regional composite scour (RCS). The RCS’ diachronous nature demonstrates that its down‐dip equivalent disperses into several surfaces in the marine part of the depositional system, which challenges the idea of a single, correlatable surface. Formation of a regional composite scour in the fluvial realm throughout a relative sea‐level cycle highlights that erosion and deposition occur virtually contemporaneously at any point along the depositional profile. This contradicts stratigraphic models that interpret low‐accommodation settings to dominantly promote bypass, especially during forced regressions. Source‐to‐sink analyses should account for this in order to adequately resolve timing and volume of sediment storage in the system throughout a complete relative sea‐level cycle.

The Mesa Rica Sandstone (Dakota Group, USA) encompasses an exhumed low-accommodation fluvio-deltaic system along its ~400 km depositional-dip oriented profile. Several parts of the transect have been studied in key localities (Holbrook, 1996(Holbrook, , 2001Holbrook et al., 2006;Van Yperen, Poyatos-Moré, Holbrook, & Midtkandal, 2020) but a regional-scale synthesis has not been presented before. Complemented with newly collected data, the depositional-dip oriented profile serves as a useful testing ground to F I G U R E 1 (a) Schematic drawing of a delta plain which illustrates frequently used terms in the paper. Lower order distributary channels include distributary channels with 1-3 successive bifurcations. (b) Projection of (c) onto graded stream profile and enveloping accommodation. The upper and lower buffer profiles track the highest surface of aggradation and the lowest depth of incision, respectively. These profiles converge towards sea level (modified from Holbrook et al., 2006). (c) Scheme for predicted scour patterns in the backwater zone under variable flow conditions. Averaged river bed elevation prior to scouring (solid lines) or after backwater induced scouring (dashed line). L b = backwater length, h c = bankfull flow depth, h scour = maximum scour depth (modifed from Trower et al., 2018). Pink solid line is inferred from Blum et al. (2013, Figure 4 VAN YPEREN Et Al. incorporate recent insights on hydrodynamic behaviours of the fluvial realm with the establishment of a sequence stratigraphic framework for time-equivalent fluvio-deltaic strata deposited in a low-accommodation setting. Specific research objectives of this study are: (a) to describe and discuss downdip changes in facies and depositional architecture and discuss their relationship with backwater effects and changes in base-level, (b) to establish a regional-scale (~400 km) sequence stratigraphic framework and discuss the challenges of correlating continental to shallow marine strata in a lowaccommodation setting, and (c) to discuss wider implications of the diachronous character of interpreted sequence boundaries.

| GEOLOGICAL SETTING AND PREVIOUS STRATIGRAPHIC FRAMEWORK
The Dakota Group is one of the eastward-prograding sedimentary systems of the US Western Interior basin that were sourced from the Sevier fold-and-thrust belt (e.g. MacKenzie & Poole, 1962;Pecha et al., 2018). The fold-and-thrust belt formed during the Cordilleran orogeny, with subduction of the Farallon plate beneath west North America, causing back-arc compression in the Late Jurassic (DeCelles, 2004). The Dakota Group also received minor sediment volumes F I G U R E 2 (a) Regional paleogeography showing the approximate location of the Western Interior Seaway (light blue, from Blakey, 2014) and main basins formed during Laramide and Colorado orogenies (modified after Van Yperen, .  Scott et al. (2004) and measured sections and 'locations where facies were identified and described but not measured' in Holbrook and Wright Dunbar (1992), Holbrook (1996Holbrook ( , 2001 and Holbrook et al. (2006). Main structural elements are indicated (from Broadhead, 2004;Suleiman & Keller, 1985). Schematic representation of the river pathway is based on previous work (e.g. Holbrook, 1996Holbrook, , 2001 and reflects the extent of the depositional system (lower Mesa Rica) during regressive phase. The indicated zones (proximal, transitional, distal) are based on the study profile and explained in the text (a) (c) VAN YPEREN Et Al. from the Bravo Dome and Siera Grande Uplift (Holbrook & Wright Dunbar, 1992;Kisucky, 1987). The Tucumcari Basin forms the depocentre for marine strata of the fluvio-deltaic Mesa Rica Sandstone (hereafter referred to as 'Mesa Rica'; Figure 2a), the oldest formation within the Dakota Group in Colorado and New Mexico (e.g. Holbrook & Wright Dunbar, 1992). The Tucumcari Basin formed during the late Carboniferous and early Permian as a tectonic element of the Ancestral Rocky Mountains (Broadhead, 2004). At times of Dakota Group deposition, the study area was located at ~35°N latitude, with a prevailing warm and humid climate (Chumakov et al., 1995). An overall NNW-to SSE-oriented depositional profile characterizes the Cenomanian Dakota Group (Scott, Oboh-Ikuenobe, Benson, Holbrook, & Alnahwi, 2018) in southeast Colorado and northeast New Mexico. The group is underlain by the Albian marine Glencairn Formation in Colorado and equivalent Tucumcari Shale in northeast New Mexico, and overlain by the Cenomanian Graneros Shale (e.g. Holbrook et al., 2006). The Dakota Group is further subdivided into the Mesa Rica, Pajarito (Dry Creek Canyon member in south-central Colorado and northeastern New Mexico) and Romeroville formations (Figure 2b). These represent phases of predominantly fluvial and paralic deposition. Regional sequence boundary SB3.1 (Figure 2b) forms the base of the Mesa Rica and is linked to a late Albian-early Cenomanian forced-regression, which caused widespread erosion in southeast Colorado and northeast New Mexico (Holbrook, 1996(Holbrook, , 2001Holbrook & Wright Dunbar, 1992;Oboh-Ikuenobe et al., 2008;Scott et al., 2004). In east-central New Mexico, the Mesa Rica is subdivided into lower, middle and upper units (Scott et al., 2004;Van Yperen, Line, Holbrook, Poyatos-Moré, & Midtkandal, 2019). The lower Mesa Rica shows a down-dip transition from fluvial to deltaic deposits at the northwestern rim of the Tucumcari Basin, recording the most proximal shallow-marine deposits within the system (Holbrook & Wright Dunbar, 1992;Van Yperen, Line, et al., 2019;Van Yperen et al., 2020). Regional sequence boundary SB3.2 forms the base of the upper Mesa Rica and is linked to another forced regression after a transgressive event that caused deposition of the paralic middle Mesa Rica (Oboh-Ikuenobe et al., 2008). These two transgressive-regressive cycles are interpreted to record higher frequency relative sealevel fluctuations than the whole Mesa Rica composite cycle (e.g. Holbrook, 1996;Holbrook & Wright Dunbar, 1992;Oboh-Ikuenobe et al., 2008;Scott et al., 2004). The SB3.2 is indistinguishable in southern Colorado where the lower and upper Mesa Rica merge into a single sandstone unit (Figure 2b;Holbrook, 2001). The down-dip extent of the SB3.1 and SB3.2 has received minimal attention to date, with the SB3.1 expression not directly mapped but interpreted as a correlative conformity at the base of the deltaic Mesa Rica (Holbrook & Wright Dunbar, 1992).

| METHODS
In this work we integrate previously published log data (n = 112), correlation panels, and interpreted photopanoramas (Holbrook, 1996(Holbrook, , 2001Holbrook et al., 2006;Holbrook & Wright Dunbar, 1992;Oboh-Ikuenobe et al., 2008;Scott et al., 2004;Van Yperen, Line, et al., 2019;Van Yperen et al., 2020) with 11 newly measured stratigraphic sections. We summarize different sedimentary facies types, their associations, the occurrence of architectural elements and the extension of key stratigraphic surfaces. Together, these form the basis of a large, regional-scale (~400 km) and depositional-dip parallel correlation panel, which covers the Mesa Rica transect along its NNW-SSE oriented profile from southeast Colorado to central-east New Mexico (Figure 2c). The panel is used as the main tool to describe and discuss down-dip changes in facies distribution, depositional architecture and the sequence stratigraphic interpretation. We selected representative trunk channel (i.e. not tributary, nor distributary) elements for grain-size sampling (see 'Backwater length and its components') based on newly collected UAV (unmanned aerial vehicle, shot with a Phantom 4 Pro ® ) imagery at four locations. The UAV imagery allowed assessment of channel dimensions and hierarchy.

| Backwater length and its components
The backwater length (L b ) scales approximately with L b = h bf /S, where h bf is bankfull flow depth and S the river bed slope (Paola & Mohrig, 1996). S and h bf are evaluated upstream in a reach of normal flow (e.g. Trower et al., 2018). L b is the approximate part of the river comprised between the river mouth and the point where mean sea level intersects the riverbed profile. To calculate the slope, we use an empirical equation (Holbrook & Wanas, 2014;Trampush, Huzurbazar, & McElroy, 2014): where S is slope, P is submerged dimensionless density of sandgravel sediment, and h m is the average bankfull channel depth of the trunk river. D 50 is average grainsize for the lowermost portion of a channel, which represents the coarsest material transported as bedload. Note that the average bankfull channel depth is one-half of the maximum bankfull thalweg depth (Bridge & Tye, 2000;Holbrook & Wanas, 2014;Leclair & Bridge, 2001) and not the average of multiple maximal bankfull measurements (cf Lin & Bhattacharya, 2017). The Shields number for dimensionless shear stress (τ* bf50 ) is 1.86 (Holbrook & Wanas, 2014 and references herein). Sediment density is assumed to be 2.65 g/cm 3 , given that the sediment is quartzose VAN YPEREN Et Al. (e.g. MacKenzie & Poole, 1962;Van Yperen, Line, et al., 2019). This gives a submerged density (P) of 1.65 g/cm 3 that is entered into Equation (1) as dimensionless number of 1.65. D 50 grain-size values are derived for four samples (Figure 2c), taken from approximately 10-15 cm above the basal scour surface of selected trunk channel-fill sandstone bodies in the lower Mesa Rica.
Bankfull channel depth was measured directly at completely preserved trunk channel deposits from outcrop and from ortho-rectified drone imagery. Where stories recorded incomplete preservation due to episodes of cut and fill, cross-set thicknesses were measured. We used these to calculate mean dune height (Leclair & Bridge, 2001) and subsequent bankfull paleoflow depths (Allen, 1982;Best & Fielding, 2019;Bradley & Venditti, 2017). By using these bankfull paleoflow depths with respect to valley scour depths, allogenic or autogenic backwater effects as the forcing mechanism for large erosional surfaces can be discussed (Fernandes et al., 2016;Ganti et al., 2019;Lamb et al., 2012;Trower et al., 2018).

| FLUVIAL CHANNEL ST YLE
Previous publications provided extensive descriptions about fluvial architectural style at different locations within the study area (Holbrook, 2001;Van Yperen et al., 2020). Based on sandstone-body dimensions and vertical and lateral spatial arrangements, we distinguish six different types of channel deposits ( Figure 5): multivalley sheet (channel type I), single-story sheet of trunk channels (channel type II), isolated fluvial distributary channels and channel belts (channel type III), incised valley (channel type IV), fluvial distributary-channel sheet (channel type V), and marine-influenced distributary channels and channel belts (channel type VI). Figure 5 provides a summary of their main characteristics. Incised-valley deposits are distinguished from channel deposits based on their multistory and multi-lateral infill (Fielding, 2008;Holbrook, 2001) and their estimated channel tops below the surrounding topographic surface (Martin, Cantelli, Paola, Blum, & Wolinsky, 2011;Strong & Paola, 2008;.

| STRATIGRAPHIC ARCHITECTURE
The study interval is represented by a tabular and laterally extensive package of strata across the ~400 km depositionaldip profile (Figures 6a,b and 7a-e). The studied transect is divided broadly into three geographical zones, proximal, transitional, and distal, based on the dominant facies associations and depositional style that distinguish them (Figure 6a,b). The characteristics of the defined zones are described below and interpreted in terms of changes in depositional mechanisms and/ or available accommodation. As this study only focuses on the Mesa Rica deposits, the stratigraphic relationships with underlying and overlying strata are only locally incorporated to provide stratigraphic context, and not described in detail.  Table 1). In the updip reaches of this zone, 11-22 m-thick multivalley-sheet deposits (channel type I; Figure 5a) are present. In the downdip reaches of the proximal zone, single-story sheet of trunk channels (channel type II; Figure 5b) of the lower Mesa Rica form a >80 km wide, laterally continuous 10-15-m-thick sheet that thins to 6-10-m-thick towards the transitional zone (Figure 6a,b;Holbrook, 1996;Van Yperen et al., 2020). The continuous sandstone sheet is one story thick and channel-fill elements locally aggrade into the overlying fine-grained facies (Holbrook, 1996). Trunk-channel fill deposits have an average aspect ratio (width-to-thickness) of 16.7. Fine-grained paralic strata (FA6) separate the lower from the upper Mesa Rica (Figure 6b). The latter forms one-story-thick localized channel belts (channel type II, Figure 5a) with a total thickness 4-7 m. In the lower Mesa Rica, interfluve facies such as overbank fines, splay deposits and/or abandoned channel-fill facies are rare. Cross-bedding orientations (FA4) indicate unidirectional palaeocurrents with a mean SSE-orientation ( Figure 6b).

| Interpretation
The multivalley-sheet deposits (channel type I) represent buffer valleys (sensu Holbrook et al., 2006) and amalgamation of the lower and upper Mesa Rica into one sandstone unit (Holbrook, 2001). Temporal fluctuations in upstream sediment and water discharge control incision and aggradation and hence the internal architecture of the buffer valleys (Holbrook, 2001). They form outside the influence of downstream controls ( Figure 5). The laterally continuous sheet of single-story trunk channel deposits (channel type II) reflects significant avulsion. We interpret this as evidence for deposition in the updip reaches of the backwater zone, because entering of the backwater zone increases avulsion and limits channel incision and/or aggradation (e.g. Chatanantavet et al., 2012;Jerolmack & Swenson, 2007). The localized channel belts (channel type III) of the upper Mesa Rica represent reoccupation of preferred channel paths and sedimentation patterns indicating higher A/S ratios than in the lower Mesa Rica.

| Description
The transitional zone encompasses the area over which riverdominated delta-front deposits (FA2) replace fluvial deposits (FA4) of the lower Mesa Rica (Figure 6b; Table 1). These delta-front facies form a sandstone-prone, sharp-based 6-10-m-thick deltaic package (Figure 7a Figure 6b). Channel belt deposits (channel type III) have average axial thickness of 4 m and true cross-stream widths of 100 m, which gives an average aspect ratio of 25. Tideinfluenced channel-fill deposits (channel type VI) have an average aspect ratio of 25 as well, with average axial thickness of 2 m and true cross-stream widths of 50 m, respectively.

| Interpretation
The transitional zone represents the fluvial-marine transition zone of the Mesa Rica depositional system. The delta-front deposits represent deposition close to the river outlet, based on the dominance and near-absence of upper flow regime bedforms and fine-grained facies, respectively (Van Yperen et al., 2020). The resemblance of prodelta deposits pinch out and the location of the Tucumcari Basin rim indicates a close relationship between basin configuration and openmarine sediment deposition (e.g. Holbrook & White, 1998;Holbrook & Wright Dunbar, 1992;Kisucky, 1987). The underlying estuarine deposits represent transgressive infill of topographic lows (Holbrook, Wright, & Kietzke, 1987;Van Yperen, Line, et al., 2019). In the incised valleys (channel type IV), erosion and deposition occurred at depths below the topographic surface of the valleys (see 'incised valleys; paleoflow depth and knickpoint migration' for further details). The dispersed trunk channel deposits (channel type II) represent continued progradation and feed a more distal part of the delta. The upper Mesa Rica represents an upper to lower delta plain depositional environment.

| Description
The distal zone is where the lower Mesa Rica represents its fully deltaic development (Van Yperen,   (Figure 6b). Here, prodelta mudstones (FA1; Figure 3f; Table 1) are up to 21 m thick, with a discontinuous pebble lag at their base. These dark grey to black fissile mudstone deposits grade vertically into river-dominated, wave-reworked delta-front deposits (FA3) of the lower Mesa F I G U R E 3 Photographs of prodelta (FA1), river-dominated delta front (FA2) and river-dominated, wave-reworked delta front (FA3) deposits in the transitional (a-e) and distal (f-h) zones. (a) Muddy bioturbated (BI 4-5) siltstone within prodelta deposits (FA1). (b) Tabular and sharpbedded fine-grained sandstones in river-dominated delta front deposits (FA2). Bioturbation is non-uniform, but basal bedding planes are thoroughly bioturbated . (c) Detail of bioturbated basal bedding planes in (b). (d) Plane-parallel laminated sandstone with sparse (BI 1) opportunistic Ophiomorpha in river-dominated delta front deposits (FA2). (e) Detail of traces in (d). (f) Black fissile mudstone prodelta deposits (FA1). (g) Symmetrical (wave) ripples overlain by single and double mud-draped asymetric (current) ripples, in river-dominated delta front deposits (FA2). (h) Coarsening-upward delta front deposits consisting of prodelta mudstones (FA1) gradually transitioning to river-dominated wave-reworked delta front sandstones (FA3)   Rica ( Figure 3h). The delta front deposits form 6-14-mthick sheet-like sandstone unit throughout the distal zone (Figures 6b and 7c-e). The overlying sand-filled distributarychannel deposits (FA4) are laterally amalgamated, rework the upper delta-front deposits and form a continuous sheet in places (Figure 5f, channel type V). Their individual channelfill elements have average aspect ratios of 17.5. In the distal reaches of the distal zone, downstepping delta-front strata are 2-8 m thick and overlain by lagoonal deposits (FA9; Figures 4d and 6b). Erosional composite surfaces bound the multi-storey infill of incised valleys (FA4; channel type IV), incise deeply into underlying deltaic strata, and have thicknesses between 8 and 12 m and total widths between 90 and 250 m (i.e. aspect ratios of 7.5-31; Figures 5e and 6b). Their sediment infill is sandstone-prone and predominantly fluvial, although sparse sandstone beds with Skolithos trace fossils (BI 1-2) occur. Drone survey imagery reveals the rare occurrence of incised-valley fill deposits (FA4; channel type IV) fining upwards to mud-or silt-dominated facies.
The upper Mesa Rica consists of a laterally varying spectrum of interdistributary bay deposits (FA6; Figure 4e), beach deposits (FA8) and laterally disconnected fully fluvial (FA4, channel type III; Figure 5c) or marine-influenced distributary channel deposits (FA5, channel type VI; Figure 6b). Isolated channel belt deposits (channel type III) have average axial thickness of 2.5 m and cross-stream width of 50 m (aspect ratio of 20; Figure 5a). Marine-influenced distributary channel deposits (channel type IV) have average axial thickness of 2 m and cross-stream widths of 30 m (aspect ratio of 15; Figure 5b). Palaeocurrent measurements (FA4) indicate an average SSW orientation (Figure 6b).

| Interpretation
The increased thickness of prodelta mudstones towards the SE is consistent with the deepening of the basin. The sheet-like delta-front sandstone geometries overlain by sandfilled amalgamated distributary channel deposits (channel type V) result from enhanced mouth-bar depositional cycles and highly avulsive distributary channels. The low-accommodation setting favoured these depositional mechanisms (Olariu & Bhattacharya, 2006;.
The upper Mesa Rica represents a dynamic lowerdelta-plain with setting in which short-lived marine incursions locally caused weak tidal influence. The A/S ratio was higher than in the lower Mesa Rica, as the upper Mesa Rica does not form continuous sheet of amalgamated sandstone body deposits. Deflection of the main paleocurrent trend mimics the basin orientation (Van Yperen, .

SURFACES
In the Mesa Rica depositional system, several stratal discontinuities can be distinguished based on underlying and overlying facies, and stacking patterns of adjacent strata. These key sequence stratigraphic surfaces were described and interpreted at separate key localities (Holbrook, 1996(Holbrook, , 2001Oboh-Ikuenobe et al., 2008;Scott et al., 2004;. In this study, their proposed correlation and expansion provides improved understanding of their regional extent (Figures 6c and 7).   VAN YPEREN Et Al. in the distal zone, and commonly represent a sharp sandstone-mudstone contact (Figure 7a-d; e.g. Holbrook, 1996Holbrook, , 2001. These surfaces are locally rooted (Figure 4b), show evidence of oxidization and/or display moderate to high bioturbation (BI 2-6).
In the distal zone, deposits overlying this surface consist of ~50-cm-thick finer-grained sandstone interbedded with mudstone, overlain by ~50 cm of dark grey mudstone (Figure 7f). Lagoon deposits (FA9; up to 4-m-thick) overlie this surface in the most distal outcrops (Figures 4d and  6b).

| Interpretation
Top surfaces bounding fluvial and deltaic strata are overlain by more distal facies. These surfaces correspond to the end of a regressive phase and are therefore interpreted as maximum regressive surfaces (sensu Catuneanu, 2006; MRS1, MRS2; Figure 7a-f). Roots and oxidization suggest subaerial exposure. MRS1 marks the top of the lower Mesa Rica, and is traceable for ~300 km throughout the study area, but cannibalized by overlying fluvial sandstone in the upper reaches of the proximal zone (Figures 5a and 6c). MRS2 marks the top of the upper Mesa Rica, and is traceable throughout (>400 km). These stratigraphic surfaces are essentially equivalent to previously published transgressive surfaces TS3.1 and TS3.2 (e.g. Oboh-Ikuenobe et al., 2008;Scott et al., 2004), and are used as correlation data ( Figure 6). Locally, some channel fills grade vertically into the overlying finer-grained facies, which complicates an interpretation of whether their top surface was formed during lowstand normal regression or subsequent transgression. Consequently, the maximum regressive surface is potentially diachronous in some places. In the most downdip exposures, MRS1 underlies the lagoonal deposits (Figure 6b), as these are interpreted to represent transgression with respect to their underlying distributary-channel deposits (Figure 6c). Where transgressive deposits are not preserved, MRS and MFS coincide. Regional traceability of the MRSs suggests allogenic forcing (Beerbower, 1964;Holbrook & Miall, 2020;Paola, Ganti, Mohrig, Runkel, & Straub, 2018). However, the lagoon deposits at sub-regional scale can be also ascribed to localized transgressive conditions due to lateral switching of active delta progradation locations in the distal zone (e.g. Bhattacharya, 2010;.

| Regional composite scours and sequence boundaries
Earlier work on the Mesa Rica system recognized and labelled two sequence boundaries (SB3.1 and SB3.2) in the proximal zones of the study area (e.g. Scott et al., 2004). In this paper, we will use the term Regional Composite Scour (RCS; sensu Holbrook & Bhattacharya, 2012), because of the increasing evidence for the diachronous/composite nature of sequence boundaries (e.g. Bhattacharya, 2011;Holbrook & Bhattacharya, 2012;Strong & Paola, 2008). Thus, we change the previously used SB3.1 and SB3.2 into RCS3.1 and RCS3.2, to acknowledge the time-transgressive character of these surfaces. By definition, the RCS excludes the interfluve component of sequence boundaries (Holbrook & Bhattacharya, 2012).

| Description
An erosional composite scour forms the basal surface of the multivalley sheet (channel type I) and the single-story sheet of trunk channel strata (channel type II) in the proximal zone (Figures 5a,b and 6b,c; Holbrook, 1996Holbrook, , 2001. Additionally, erosional composite surfaces bound the multi-storey infill of incised valleys (channel type IV) in the transition and distal zone (Figures 5d,e and 6b,c), where they separate fully fluvial deposits (FA4) from underlying deltaic facies associations F I G U R E 6 Regional-scale (~400 km), depositional dip-parallel correlation panel of the Mesa Rica fluvio-deltaic system throughout southeast Colorado to central-east New Mexico. The colour code for the logs indicates the data source, similar as in Figure 2c. Key stratigraphic surfaces and distribution of facies associations and architectural elements are based on all available log data, drone surveys and descriptions of architectural elements from both this study and previous work (Holbrook, 1996(Holbrook, , 2001Holbrook et al., 2006;Holbrook & Wright Dunbar, 1992 (FA2, FA3, Table 1

| Interpretation
The composite basal surface in the proximal zone is the expression of the regional sequence boundary RCS3.1 (SB3.1 in Scott et al., 2004;Oboh-Ikuenobe et al., 2008) and relates to late Albian -early Cenomanian forced regression (Holbrook, 1996(Holbrook, , 2001Holbrook & Wright Dunbar, 1992;Oboh-Ikuenobe et al., 2008;Scott et al., 2004). The basal surface of dispersed single-story trunk channel deposits (channel type II) in the transitional zone, and the erosional composite surfaces bounding the incised-valley fills (channel type IV) in the transitional and distal zones, are all interpreted as different expressions of the RCS3.1 regional sequence-bounding scour (Figure 6c) & Bhattacharya, 2012). This insight forms the conceptual base for the Regional Composite Scour, which forms by progradation and scouring of fluvial systems above marine strata, and expands laterally and seaward throughout the transgressive/regressive cycle (Holbrook & Bhattacharya, 2012). The incised-valley walls were shaped continuously and there was continuous deposition during relative sea-level fall. This contradicts non-deposition during valley formation as often suggested (e.g. Van Wagoner et al., 1988). The RCS3.2 (SB3.2 in e.g. Scott et al., 2004;Oboh-Ikuenobe et al., 2008) represents a regional surface as well ( Figure 6c) and relates to a second regressive phase of the Mesa Rica system.

| Description
Erosional composite scours bound sheets of amalgamated distributary-channel deposits (channel type V) in the distal zone (Figures 5f, 6c, 7g,h). They mark sharp facies boundaries that represent the culmination of the typical shallowing-upward character of the deltaic succession. However, newly visited localities (Figure 2)  F I G U R E 7 Overview of stratigraphic architecture and key stratigraphic surfaces in the transitional and distal zones. For the proximal zone, see Figure 5a,b. (a) Photograph showing the Cretaceous stratigraphy in the transitional zone. (b) Interpretation of (a). Note that the RCS excludes interfluve. The contact between the estuarine (FA7) Campana and deltaic (FA2) lower Mesa Rica represents a turnaround from transgressive to regressive conditions. Note the limited thickness of the delta front deposits (FA2) compared to the deltaic succession of the lower Mesa Rica in the distal zone (Figure 7c,g,h).

| Interpretation
Basal composite scours bound distributaries that are younger than the deltaic deposits they incise, and which fed a more distal part of the delta system. These scours form a surface named basal distributary composite scour (BDCS; Figure 6c; . However, the deposits they bound are localized to discrete deltaic localities and consequently they are not part of the regional scour surface, which is formed by larger channel cut-andfill-cycles and forms the regional sequence-bounding scour . The basal distributary composite scour is interpreted to have rather formed by the autogenic process of distributary-channel avulsion and deposition. Such autogenic surfaces commonly have limited lateral extent (Morshedian, MacEachern, Dashtgard, Bann, & Pemberton, 2019), and their recognition is quite uncommon (Pattison, 2018).

PALEOSLOPE CALCULATIONS
In order to investigate the potential backwater effects on surface generation and down-dip changes in depositional architecture, it is key to establish the landward limit of this marine influence. To do so, we distinguish two datasets for the backwater length calculations in this study: samplebased estimates and outcrop-based estimates. Sample-based estimates provide backwater lengths resulting from empirical Equation (1) using the grain-size samples representative for the coarsest material transported as bedload within trunk rivers (Holbrook & Wanas, 2014;Trampush et al., 2014).
Outcrop-based estimates are inferred from changes in fluvial architectural style observed in the studied outcrop profile, and hence a direct measurement within the basin. Median grain-size values (D 50 ) for four trunk channel-fills (channel type II) of the lower Mesa Rica were derived from

F I G U R E 9
Stepwise evolutionary model for the Mesa Rica depositional system. The profile represents a simplified version of the correlation through the study area ( Figure 6) VAN YPEREN Et Al. cumulative grain-size distribution curves. They have a D 50 grain-size value of 0.17-0.28 mm, which fall within the fine sand category (0.125-0.25 mm). We consider a bankfull depth of 11 m as representative for the trunk channel deposits (channel type II) in the lower Mesa Rica (Holbrook, 1996). This gives one-half bankfull depth of 5.5 m. Using empirical Equation (1), resultant paleoslopes are 0.9 × 10 −4 -1.6 × 10 −4 ( Table 2). Sample-based estimates of backwater lengths are consequently 71-117 km, which places the maximum backwater length ~30 km south of the New Mexico-Colorado border at onset of deltaic deposition in the Tucumcari basin (Table 2; Figure 8).
Outcrop-based estimates indicate a backwater length of ~180 km, which is the distance between the rim of the Tucumcari basin and the most updip evidence of backwater conditions (Carizzo Canyon, Figure 8). The latter is inferred from the updip limit of single-story trunk channel deposits (channel type II) forming a laterally continuous and extensive sheet. This occurrence is taken as evidence for deposition within the updip reaches of the backwater zone (e.g. Chatanantavet et al., 2012;Jerolmack & Swenson, 2007). Farther updip, the presence of multivalley deposits formed by smaller (likely tributary) channel-fill elements indicate incision and aggradation independent of relative sea-level changes and suggest deposition updip of backwater influences (Figures 5 and 8; e.g. Blum et al., 2013).
The outcrop-based estimate of the backwater length (~180 km) is significantly longer than the sample-based backwater length range of 71-117 km. This mismatch between the two different datasets can be explained by one or a combination of the following reasons: (a) the channels in the most updip evidence of backwater conditions (Carrizzo Canyon) fed a shoreline farther upstream that predates regression to the rim of the Tucumcari Basin. (b) Errors in slope estimates up to a factor 2 are intrinsic to the used calculation method (Holbrook & Wanas, 2014); therefore, outcrop-inferred estimates would be within the error range of the sample-based calculations. (c) Increased avulsion started up dip of the calculated backwater length. We cannot further eliminate uncertainties based on the limited grain-size samples, the studied outcrop profile or the state-of-the-art for backwater calculations.
Backwater length calculations can also be used to estimate the position of the maximum regressive shoreline. This is done by taking the most downdip occurrence of sheet-forming single-story trunk channel deposits (channel type II) and assume that this position approximates the updip reach of the backwater length at times of maximum regression. The sheet-forming single-story trunk channel deposits disperse around the basin rim, which implies that the upstream limit of the coeval backwater zone was close to this location. Based on this, shoreline progradation made it as far as ~117 km (sample-based) or ~180 km (outcrop-based) south of the basin rim, a position beyond the outcrop window (Figure 8).

| INCISED VALLEYS: PALEOFLOW DEPTHS AND KNICKPOINT MIGRATION
Incised valleys form where regression exposes a slope steeper than the contemporary river equilibrium profile, and have been interpreted to evidence relative sea-level fall (e.g. Blum et al., 2013;Catuneanu, 2006;Van Wagoner et al., 1988). Consequently, their adequate recognition influences the understanding of a depositional system. In the Mesa Rica system, incised-valley fills (channel type IV) are on average 16 m thick in the transitional zone. Bankfull paleoflow depth (Allen, 1982;Best & Fielding, 2019;Bradley & Venditti, 2017) of average channel fills within these valleys was ~7.4 m, based on cross strata thicknesses and mean dune height calculations (Leclair & Bridge, 2001). This indicates that valleys were cut by channels that had undergone approximately two bifurcations (Yalin, 1992), or they were initially smaller because they carried less discharge than the largest trunk channels. Their water surface was 8.6 meter below the topographic surface (16 m valley depth minus 7.4 m depth of active channel). In the distal zone, complete incised-valley fills are on average 11 m thick. Applying the same method, their average channel story thickness within these valley scours is ~5.9 m. Consequently, their water surface was ~5.1 m below the concurrent topographic surfaces.
The updip extent of valley incision relates to updip knickpoint migration over time Wescott, 1993). The dataset allows estimates for both minimum and F I G U R E 1 0 (a, b) Simplified depositional profile illustrating different possible sequence stratigraphic correlations between the fluvial and marine realm. Numbers indicate relative time relationships. Both models focus on the sequence boundary (SB)/Regional Composite Scour (RCS) and its marine extent. Model I extends the SB below the first downstepping deltaic deposits as a correlative conformity (Posamentier et al., 1992). Model II extends the SB beneath the lowstand deposits of the last downstep (Hunt & Tucker, 1992. Both wheeler diagrams show that there is limited temporal or genetic relationship between the fluvial and deltaic deposits. Labels 'SB1' and 'SB2' are only meant to illustrate chronological order and do not relate to the nomenclature of the identified sequence boundaries of the Mesa Rica depositional system in New Mexico and Colorado (Figure 2b). (c) Simplified depositional profile and Wheeler diagram showing the dispersive nature of the Regional Composite Scour (RCS) in the marine realm. Discrete parts of the composite, highly diachronous and amalgamated erosional composite surface below the fluvial deposits in the proximal zone, are time-equivalent to individual regressive marine surfaces. Each segment of the RCS is contemporaneous to the clinoform surface underlying the genetically-related clinothem. Similarly, segments of the composite scour bounding an incised valley are formed contemporaneously with deposition in the valley, trunk channel deposition in the proximal zone, and clinothem deposition in the distal zone. The regional composite scour is generated in the fluvial realm throughout the T-R cycle. Therefore there is no single correlatable surface in the marine realm, but rather multiple, dispersed segments. Faded deltaic wedges t3 (in a, b) or t7 (in c) are not documented in this study. See text for further discussion 536 | EAGE VAN YPEREN Et Al. maximum updip occurrence of knickpoints. The maximum updip occurrence is inferred from extensive mapping and architectural-element analysis just south of the Colorado-New Mexico border (i.e. Dry Cimarron Valley in Holbrook, 1996). Here, incised-valley deposits are absent which confirms the lack of knickpoint migration to this distance up dip (Figure 8). The minimal updip occurrence of knickpoint incision is the southernmost location without any valleys observed and hence no evidence for knickpoint migration (Figure 8). However, this is based on local sampling of discontinuous outcrops with drone surveys and not the systematic examination of continuous outcrops executed further north. The localized nature of this dataset (Figure 8) leaves room for incised-valley deposits missed by drone coverage. The resultant range between the minimum and maximum updip occurrence of valley knickpoints is approximately 115 km.
The maximum updip occurrence of valley knickpoints is situated in between the sample-based and outcrop-based backwater lengths at onset of deltaic deposition in the Tucumcari Basin ( Figure 8). During maximum regression, the maximum updip occurrence of valley knickpoints scales to ~2× the backwater length from the maximum regressive shoreline (Figure 8). This scaling relationship is used to discuss the forcing mechanism for these large erosional surfaces (see Section 10; Fernandes et al., 2016;Ganti et al., 2019;Lamb et al., 2012;Trower et al., 2018).

| Relative sea-level control on depositional architecture
Evidence for relative sea-level fall during deposition of the Mesa Rica system is threefold: (a) downstepping delta-front geometries in the distal zone ( Figure 6b); (b) key stratigraphic surfaces (MRS1, MRS2) extend over regional distances (>300 km, Figure 6c), which cannot be explained solely by autogenic behaviour ; (c) multi-storey sandstone bodies (channel type IV, Figure 5d,e) represent incised valleys, based on their regional occurrence, their multi-storey and multi-lateral infill, and their estimated channel incisions at least two channel depths below the concurrent topography (Fielding, 2008;Holbrook, 2001;Martin et al., 2011;Strong & Paola, 2008;. Flume modelling results show improbable autogenic formation of multi-storey sandstone bodies with more than two channel depths (Strong & Paola, 2008). Despite all this, a potential other scenario for autogenic multi-storey sandstone body generation is the scouring by trunk channels and later reoccupation and deposition by distributaries, creating a multi-storey infill. However, the coeval downstepping delta front geometries in the Mesa Rica evidence an externally-forced drop in sea level (Van Yperen, . This, and concurrence with the incised-valley scours, is conclusive for a fall in relative sea level. The sea-level drop needed for the formation of the documented valleys in the lower Mesa Rica is ~9 m. This is based on average bankfull channel depths of 7.4 m within the 16-m-thick valleys, which implies that their water surfaces had dropped ~9 m. The subsequent transgression covered a distance of roughly 250 km, based on the occurrence of paralic middle Mesa Rica deposits in the distal reaches of the proximal zone ( Figure 6b) and the reconstruction of weak brackish influence in southern Colorado (Oboh-Ikuenobe et al., 2008). The estimated minimum and maximum slopes values for the single story trunk channels of 0.9 × 10 −4 and 1.6 × 10 −4 would have required a relative sea-level rise between 23 and 40 m, to cause this flooding, respectively. In addition to the ~9 m sea-level drop this means a total of 32 to 49 m rise in relative sea-level is likely for flooding of the lower Mesa Rica system.

| A stepwise model for the Mesa Rica depositional system
In the lower Mesa Rica, multivalley deposits (channel type I) appear ~240 km upstream from the Tucumcari basin rim, which equals ~2× the sample-based maximum backwater length (i.e. 117 km), and ~1.5× the outcrop-based backwater length (i.e. 180 km). The multivalley deposits thin downstream to a single-story-thick channel sheet (channel type II; Figures 5 and 6a,b) which also thins towards the rim of the marine basin. This is consistent with the anchoring of the graded stream profiles, causing convergence of the upper and lower buffer profiles (Figure 1b;Holbrook et al., 2006) accompanied with vertical limits on aggradation and incision (e.g. Holbrook et al., 2006;Mackin, 1948;Quirk, 1996). Channel thinning in the transitional and distal zones results from repetitive bifurcation (Edmonds & Slingerland, 2007;Yalin, 1992). Onset of deltaic deposition occurred close to the rim of the basin (Figure 9a). However, low-accommodation conditions limited the preservation of deltaic sediments, as younger prograding fluvial channels were forced to use the same accommodation ( Figure 9b). Consequently, these channels almost completely eroded the deposits that recorded the facies change from shallow-marine to fluvial settings, which is now preserved as a rather abrupt transition. Thickness values of the delta-front deposits suggest water depth abruptly increased basinwards in the transitional and distal zones (Figure 9b). Here, single-story trunk channels and incised valleys (channel type II and IV, Figure 5) incise locally into underlying delta front strata or distributarychannel deposits (channel type V; Figure 9b). Basal surfaces of these sheet-forming distributary-channel deposits (basal distributary composite scour) eroded most upper delta-front | 537 EAGE VAN YPEREN Et Al. sediment and indicate that accommodation was still limited. The single-story trunk channel elements (channel type II) were deposited during continued normal progradation and feed a more distal part of the delta. Later forced regression and progressively less accommodation resulted in downstepping delta-front geometries (Figure 9c). Subsequently, this fall in relative sea level caused valley incision (channel type IV) as the equilibrium profile adjusted to steeper gradients (Figure 9c; e.g. Talling, 1998). After a period with steepened depositional gradients, the equilibrium profile shallowed during subsequent relative sea-level rise. Incised valleys filled and facies belts shifted ~250 km landwards, based on the occurrence of paralic middle Mesa Rica deposits in the distal reaches of the proximal zone (Figure 9d), although fully-marine conditions were not established over this entire length (Oboh-Ikuenobe et al., 2008).
Onset of upper Mesa Rica deposition by renewed normal progradation caused fluvial and interdistributary bay deposition ( Figure 9e). The main differences with the lower Mesa Rica are two: first, the upper Mesa Rica is characterized by a higher A/S ratio ( Figure 6). This can be a consequence of insufficient time to form a sheet of laterally amalgamated channel-fill elements (channel type II), as characteristic for the lower Mesa Rica. Another explanation is a higher profile gradient for the lower Mesa Rica than for the upper Mesa Rica, as the first prograded into the Tucumcari Basin, whilst the latter prograded over a shallower flooding surface (Figure 9e). Such low gradient conditions are accompanied with the relative increase in preservation of delta plain fines. Low profile gradients promoted this preferred upstream deposition of the sand-fraction (Holbrook & Bhattacharya, 2012). Secondly, incised valleys of the upper Mesa Rica formed during a subsequent relative sea-level fall, but their knickpoints did not migrate into the transitional zone ( Figure 9f). The genetically related delta front deposits to this down step accumulated beyond the outcrop window and thus away from the study profile.
In general, this model suggests that cut-and-fill cycles of all channel types occurred continuously throughout a relative sea-level cycle, and during deposition of both the lower and upper Mesa Rica. Changes in relative sea level triggered the equilibrium profile to adjust, which in turn determined the vertical limits of erosion and deposition along the lower reaches of the depositional profile.

| Backwater effects in the Mesa Rica depositional system
The regional scale of the Mesa Rica outcrop profile provides a unique opportunity to study changes in architectural style and their relation to backwater effects. The observation of flood-induced scours up to 3× bankfull depth (Fernandes et al., 2016;Ganti et al., 2019;Lamb et al., 2012;Trower et al., 2018) poses potential challenges to differentiate large scours induced by drawdown effects in the backwater zone (e.g. Lamb et al., 2012), from allogenically-formed incisedvalley fills (e.g. Blum et al., 2013). Trower et al. (2018) showed that maximum scour depths of the Cretaceous Castlegate Sandstone range between 1 and 3× bankfull channel depth, and questioned the role of base-level fall in creating these erosional surfaces. The maximum scour depth of flood-induced erosion is proportional to flow variability in normal-flow depths (Chatanantavet & Lamb, 2014), which is typically 0.5 to 3× bankfull flow depth upstream of their backwater zone (Ganti et al., 2014). Therefore, allogenic scour depths must theoretically exceed bankfull flow depth (>3×) and occur over a greater distance than the backwater length in order to unambiguously distinguish allogenic signals from backwater-induced scours (Ganti et al., 2014(Ganti et al., , 2019Trower et al., 2018). In our study, incised valleys are on average 11-16 m thick and their infill indicates deposition in 5.1-8.6 m thick channels. Consequently, scouring happened at less than 3× below bankfull depth. Nevertheless, the observations that support the existence of a drop in relative sea level listed in the previous section (i.e. the downstepping delta front geometries) suggest these valleys formed as a response to an allogenically-induced steepening of the graded stream profile and not as a consequence to flood-induced scours within the backwater zone. The limited distance over which the knickpoints migrated (~1-2 L b ) and hence the incised valleys occur relates to the minor drop in sea level (~9 m) and a short-lived nature of this relative sea-level drop. The latter is inferred from the narrow incised valleys (indicating limited time for lateral migration or erosion of valley sidewalls), good preservation of delta plain deposits (which would otherwise be cannibalized in this low-accommodation setting), and the knickpoint of these valleys being close the upstream limit of the backwater zone. In summary, one of the main criteria offered by other authors (Ganti et al., 2014(Ganti et al., , 2019Trower et al., 2018) to unambiguously assign an allogenic origin to the incised valleys (i.e. occurrence of incised valleys over distances longer than the backwater length and scouring >3× bankfull flow depth) is not consistent with the results of this study, which evidence allogenic forcing of valley scours <3× bankfull flow depth occurring over one to two times the backwater length (~1-2 L b ). This emphasizes that decoupling autogenic and allogenic controls on erosional surface generation might be especially problematic, particularly in low-gradient river systems.
Other down-dip changes often linked to backwater effects are downstream fining channel belt deposits, decrease in sinuosity, and channel belt deepening and narrowing (e.g. Fernandes et al., 2016;Lamb et al., 2012;Martin et al., 2018;Nittrouer, 2013). Of these, this study has only documented channel-belt narrowing, but the lack of other downdip changes 538 | EAGE VAN YPEREN Et Al. linked to backwater effects in the Mesa Rica system can have several causes. Firstly, backwater analyses of the sedimentary record imply that backwater hydrodynamics must persist long enough for its signal to be recorded (Chatanantavet & Lamb, 2014;Ganti, Chadwick, Hassenruck-Gudipate, & Lamb, 2016). Low-accommodation settings lower preservation potential in general, which might lower the chances of such signals being recorded in addition. Secondly, the generally low preservation potential in low-accommodation systems might lower the chance to record this signal. This might be particularly challenging in low-accommodation systems, where preservation potential is generally low. Secondly, backwater concepts originated and are predominantly tested on the Mississippi river, and supported by numerical models assuming simplified input parameters (e.g. Chatanantavet et al., 2012;Fernandes et al., 2016;Lamb et al., 2012;Nittrouer, 2013;Nittrouer et al., 2012). Consequently, their full applicability in other settings is part of future research. Several other studies have documented results that contrast the 'expected' backwater effects, such as channel widening and shallowing in tide-dominated river deltas (Gugliotta & Saito, 2019), or absence of erosion in the distal part of the backwater zone during river floods (Zheng, Edmonds, Wu, & Han, 2019). The interplay of sediment type, depositional gradient, climate and (above all) time and preservation potential make trends in backwater effects difficult to predict.

| Sequence stratigraphic correlations in low-accommodation settings
Conceptually, there are several possible scenarios for correlation between fluvial and genetically-related deltaic deposits. One scenario places the sequence boundary below fluvial deposits and extends it below the first downstepping deltaic deposits as a correlative conformity (Figure 10a; e.g. Posamentier, Allen, James, & Tesson, 1992) or to the correlative conformity beneath the lowstand deposit is of the last downstep (Figure 10b; Hunt & Tucker, 1992. In another scenario, the sequence boundary is correlated with the flooding surface on top of the deltaic strata (Embry, 1995). In both scenarios, the normal-regressive deltaic deposits are included in the highstand systems tract and only late lowstand shallow marine deposits are time equivalent to the fluvial strata. Theoretically, temporal relationships between fluvial and marine strata would be distinctive, as the fluvial facies would gradually transition into deltaic facies in highstand and early falling stage strata. However, in this study, there is an abrupt change from fully-fluvial to deltaic deposits (Figure 6b), and so no true zone with gradational facies transitions is identifiable. But by principle, such facies transition must have been present at least at the onset of deltaic deposition. We argue that this facies transition was eroded at later time, when the fluvial system advanced over highstand strata and completely eroded the delta deposits to one channel depth up to approximately the northern margin of the Tucumcari Basin (Figure 9a,b). Consequently, the area that theoretically holds the physical evidence for a temporal relationship between fluvial and shallow marine highstand strata is nowadays eroded. This process by which prograding fluvial facies incise and remove the record of underlying highstand deposits is commonly referred to as compensation (Hajek & Straub, 2017;Holbrook & Miall, 2020;Straub & Esposito, 2013). Examples of complete compensation, as happened with lower Mesa Rica deltaic strata, are atypical (Holbrook & Miall, 2020).
Following Posamentier et al. (1992), Tucker (1992, 1995) or Embry (1995), an additional fall in relative sea level would be needed to explain the sequence boundary (SB2 in Figure 10a,b) that bounds the incised valleys (channel type IV) and incises into the single-story sheet of trunk channels (channel type III) that in turn is underlain by a sequence boundary (SB1 in Figure 10a,b). As studies on modern fluvio-deltaic also suggest (Blum et al., 2013), our model infers temporal relationships between fluvial and deltaic deposits (Figure 9a-d) and no additional fall in relative sea level is needed (see next section for further discussion).

Composite Scour in the marine realm
The extent of the traditional sequence boundary into correlative marine strata is often debated and causes practical problems in its application (e.g. Bhattacharya, 2011;Bhattacharya et al., 2019). The Regional Composite Scour (RCS) acknowledges the three-dimensional and diachronous nature of this surface in the continental realm. This includes along-strike variability, which is crucial to understand any depositional system (e.g. Amorosi et al., 2019;Madof et al., 2016;Martinsen & Helland-Hansen, 1995;Miall, 2015).
In the Mesa Rica system, the nature and predominantly fully fluvial infill of single-story trunk channel deposits (channel type II), incised valley fill (channel type IV) and amalgamated distributary channel deposits (channel type V) imply active filling of channels rather than passive backfilling, and suggests continuous reshaping and active deposition occurred at the delta plain and in incised valleys. Additionally, we mapped and physically traced several stratigraphic surfaces down dip of the Regional Composite Scour (RCS). These are (a) the Basal Distributary Composite Scour (BDCS) below amalgamated distributary channel deposits, (b) the basal surface below dispersed trunk channel deposits incising into deltaic deposits, (c) composite surfaces bounding incised valleys, and (d) a downstep in deltaic onlap (Figures 6c and 7). None of them is necessarily equivalent to the sequence boundary as defined originally in the fluvial | 539 EAGE VAN YPEREN Et Al. realm (e.g. Hunt & Tucker, 1992;Van Wagoner et al., 1988). However, recent understanding of the diachronous character of the sequence boundary/RCS (Holbrook & Bhattacharya, 2012;Martin et al., 2009;Strong & Paola, 2008) entails that individual segments of these stratigraphic surfaces each correlate with discrete parts of the RCS: the RCS was created in the proximal zone throughout the complete relative sea-level cycle, due to ongoing river erosion and virtually contemporaneously deposition within the channel (Figure 9a-d). The sediment that was not incorporated into the updip fluvial deposits bypassed this area and fed the coeval deltaic clinothem in the distal zone. Consequently, each part of the RCS is time-equivalent to the clinoform surface underlying each genetically-related clinothem ( Figure 10c).
Thus, the RCS results from multi-phase scouring throughout a relative sea-level cycle (Holbrook & Bhattacharya, 2012;Martin et al., 2009;Strong & Paola, 2008), contemporaneous to deposition in the shallow-marine realm. The composite nature of the RCS and the documented and physically traced stratigraphic surfaces evidence that the fluvial regional composite scour disperses into several surfaces in the shallow-marine part of the depositional system, rather than one single, correlatable surface. This also implies that no third-order sequence boundaries are necessary to correlate the incised valleys with the delta-front sandstones they incise into (Pattison, 2019). Dispersive key stratigraphic surfaces have been documented previously (Korus & Fielding, 2017). In their study, composite sequence boundaries split in downdip direction and are physically traceable as they pass into conformable surfaces. This differentiates from the dispersion of a single sequence boundary into several surfaces as highlighted in our study.
The application of this concept along a complete fluvio-deltaic system evidences the need to focus on dynamics and mechanisms creating key sequence stratigraphic surfaces, rather than debating their nomenclature or chronostratigraphic value. This debate seems an impossible quest for a single correlatable surface in the marine realm, given that regional composite scours may be generated in the fluvial realm throughout a relative sea-level cycle. Additionally, the active deposition and continuous reshaping of channels and incised valleys suggests that erosion and deposition occurred virtually contemporaneous at any point along the depositional profile, which implies that there is also no complete bypass at any given time or point in the system. This cautions against many stratigraphic models in which low-accommodation settings are interpreted to promote complete bypass, especially during forced regression, which results in extensive lowstand wedges (e.g. Emery & Myers, 2009;. Results of this study suggest that sediment is stored more continuously in the fluvial part of depositional systems than conventional models suggest. Basin reconstructions and source-to-sink analysis need to take this into account in order to adequately resolve the amount of sediment volume trapped temporally or permanently in the system throughout a complete relative sea-level cycle.

| CONCLUSIONS
• This work presents for the first time a regional-scale (~400 km) and depositional-dip parallel stratigraphic correlation of the low-accommodation Mesa Rica fluvio-deltaic system, and illustrates the complexities inherent to sequence stratigraphic interpretations of fluvial to marine systems. • The distribution, stacking patterns and dimensions of six distinguished channel types (i.e. multivalley-sheet, single story-sheet of trunk channels, isolated fluvial distributary channels and channel belts, incised valley, fluvial distributary-channel sheet, marine-influenced distributary channel) reflect their position along the equilibrium profile and a general trend of decreasing accommodation towards the paleoshoreline. • Evidence for relative sea-level fall during deposition of the Mesa Rica system is based on downstepping delta-front geometries in the distal zone, key stratigraphic surfaces extending over regional distances, and the regional occurrence of valley incised valley scours that correlate with the downstepping delta-front strata. • Incised valley scours <3× bankfull flow depth occurring over one to two times the backwater length (~1-2 L b ) resulted from allogenically-induced steepening of the graded stream profile and not as a consequence of flood-induced scouring in the backwater zone, as other authors have suggested. Even though decoupling autogenic and allogenic controls on erosional surface generation might be problematic, particularly in low-gradient river systems, it is better to differentiate flood-induced multi-storey channels from allogenically-formed incised-valley fills based on multiple observations rather than only scour depth and occurrence over a distance compared to backwater length. • The position of changing fluvial architecture from multivalley to single story channel fill deposits and the distance over which incised valley scour scale with ~1-2 backwater lengths (L b ). Within the backwater zone however, only limited changes in fluvial architecture observed in the Mesa Rica system (i.e. channel belt narrowing) fit the general model for backwater-mediated down-dip changes. This can be related to backwater hydrodynamics not persisting long enough for its signal to be recorded, and/or to their limited preservation potential in low-accommodation systems.