2.2 Characterization
In this paper, we show 4 sections taken from the 3D seismic data located in the map in Fig. 3e as well as a 3D base sediment horizon (Fig. 3d). Continentward (i.e. on the western side), the base sediment horizon displays prominent, positive, linear topographic features proximal to the continent trending NNW-SSE. The seaward side of these features are characterized by scarps that show clear striations (or corrugations) trending in the direction of extension (WSW-ENE) and terminate into a rough, chaotic surface. This “floor” is marked by escarpments and its depth compared to the rest of the study area. Seaward (westward) of these features is a high, dome-like structure also covered by striations trending in the same direction (WSW-ENE). The continentward face of this dome is smooth, gently sloping, and corrugated parallel to the striations. On the seaward face of the dome, the slope is gentle but rough and lacks obvious striations. Note that a positive gravity anomaly in Fig. 3b is consistent with the location of the corrugated “dome.” The most distal part of the top- basement surface is deep and hummocky, slightly shallower than the basement flooring the opposite side of the dome, and riven by escarpments trending NNE-SSW. Importantly, this last set of escarpments are not perpendicular to the spreading direction.
The 2D seismic sections crosscutting the surface further reveal the basin and crustal structure across the study area and show the transition between the hyperextended continental lithosphere and the oceanic lithosphere (Fig. 3e). The interpretation of the Ivory Coast seismic sections was done by distinguishing eight seismic units and 2 main set of faults as described below.
Unit 1 (transparent/white): This unit is characterized by thinly spaced, high amplitude, sigmoidal to parallel reflectors that are ubiquitous across the upper sections of each profile. Strata imaged in this seismic unit show clear onlap relationships above the units located below. This pattern of reflectors as mostly flat lying along with its position in the section, informs the interpretation that this is a post-tectonic sedimentary depositional sequence made of continental slope and oceanic sediments.
Unit 2 (green): Unit 2 is characterized by thinly spaced, slightly diverging reflectors with varying dip angle that decrease in seismic amplitude with depth. Seismic reflectors display fan-shaped growth strata in association with seaward-dipping normal faults (as indicated by offset reflectors). Fan-shaped growth strata associated to normal fault in a domain where continental crust can be identified (see below) suggest this unit to be a syn-rift depositional sequence.
Unit 3 (purple): Characterized by a mix of chaotic reflectors and low amplitude, widely spaced complexly layered reflectors. Both sets of reflectors become fewer with increased depth. Due to the parallel lying stratigraphic reflections, this unit is interpreted to be pre-kinematic sedimentary layers in respect of the rift-related normal faults.
Unit 4 (light brown): This is characterized by a homogenous seismic facies with only some diffractions and chaotic reflectors in association with the lateral continuation of normal faults (shown by offsets in units or by bright reflectors). The fairly uniform features and position in relation to other units possibly indicates a crustal basement lithology.
Unit 5 (black): A unit with complex layering of high-amplitude reflectors and some regions with few to no reflectors. The upper boundary of this unit corresponding to the base of Unit 4 is a high-amplitude, thick reflector indicating a high impedance contrast referred to as the “M” reflector in this paper (mapped in 3D as the base sediment horizon). We propose the “M” reflector to be the continental Moho capping the subcontinental mantle corresponding to Unit 5. This interpretation is geometrically and isostatically consistent with a regional seaward rise of this reflector going along the deepening of the base sediment.
Unit 6 (light red): Characterized by wedge-shaped bodies of discontinuous, diverging reflectors mixed with chaotic reflectors, all of which vary in amplitude. We interpret them as syn-kinematic lava flows as reflections are shifted by normal faults (i.e. pillow basalts, hyaloclastites and/or basaltic breccias). They are observed on top of Unit 2 and Unit 7 where they correspond to the classical Layer 2a of Penrose type-oceanic crust (e.g., Cann, 1970; Nicolas 1989; Gómez-Romeu et al., 2022).
Unit 7 (medium red): A unit characterized by a seismic facies of low-amplitude reflectors. Because of its location below volcanic flows of Unit 6, we interpret this unit as sheeted mafic dikes equivalent to Layer 2b of Penrose type-oceanic crust (e.g., Cann, 1970; Nicolas 1989).
Unit 8 (dark red): This unit exhibits a similar characterization as Unit 5, but with the addition of sigmoidal reflectors interpreted to be sheared gabbros and other gabbroic bodies. The gabbroic bodies are in turn interpreted as intruding the mantle in a manner consistent with Gillard et al. (2019) and are part of a region equivalent to Layer 3 of Penrose-type oceanic crust (e.g., Cann, 1970; Nicolas 1989).
Distal margin detachment fault system (green): Faults defined by offsets in Units 2, 3, 4, and 5 that sole into the high-amplitude reflector atop the Unit 5 dome. This system’s single decollement dips towards the ocean and recorded a top-to-the ocean senses of shear as shown by syn-rift wedging bounded by synthetic normal faults (generally dipping seaward as well) that are the shallow expressions of this system (exclusively found in the continentward part of the rifted margin).
Out-of-sequence detachment fault system (red): This system of faults offsets Units, 2, 3, 4, 5, 6, 7, and 8. It also crosscuts the distal margin detachment system and is therefore considered out-of- sequence with respect to this latter. These out-of-sequence faults are geometrically distinct, with high-angle normal faults locally soling into an extensional duplex that dips towards the continent. The duplex contains sigmoidal boudins of Units 5 and 8 bounded by anastomosing high-amplitude reflectors interpreted as shear zones.
2.3 Interpretation
Based on the seismic reflection images provided across the DICB, the Ivory Coast rifted margin records the transition from magma-poor continental rifting to seafloor spreading. Near the continent there is a seaward dipping, distal margin detachment faults system which bounds blocks of continental crust (extensional allochthon blocks), accommodates syn-rift basins above tapering continental crust and exhumes mantle further outboard. Towards the ocean basin, this set of faults are crosscut by out-of-sequence, continentward-dipping top-basement faults; anastomosing shear zones in the mantle; magmatism and volcanism; further exhumation of the mantle; and layered oceanic crust.
The earliest evidence of tectonic activity is the formation of the distal margin detachment system that accommodates the final extension of the continental  and exhumes its underlying sub-continental mantle (Units 3, 4, and 5). The fanning reflectors in Unit 2 is indicative of growth structures and suggests that it’s a syn-kinematic depositional system coeval with the earlier extensional phase in the area covered by the seismic survey (late rifting at the scale of the entire margin). The high-angle normal faults sole into a low-angle detachment fault with sub-continental mantle in its footwall. This is interpreted as relating to the ultimate tapering of the continental crust leading to mantle exhumation.
The second major phase of deformation is the development of an out-of-phase sequence of detachment faults with opposite vergence, hereby referred to as OCT detachment systems. The volcanic wedges of Unit 6 appear to be coeval with the shallow, high-angle faults associated with this set of faults. The seaward increase in Unit 6 thickness correlates to increased evidence of plutonic bodies in the mantle (Unit 8). Unlike distal margin detachment faulting, the OCT detachment system soles into multiple levels of decollement to form extensional duplexes. Blocks within these duplexes contain sub-continental mantle while the footwall of the whole system contains mantle material with sigmoidal reflectors interpreted to be sheared gabbroic bodies (similar to anastomosing shear zones at Lanzo, Nurali, and elsewhere). These gabbroic bodies increase in frequency and thickness in the seaward direction until they form Unit 8.
Unit 8 is the lowermost unit of the oceanic crust pseudo-stratigraphy (Nicolas 1989) seen in the most distal parts of the seismic section. Atop the sheared gabbros (Unit 8) are the sheeted dikes identified by the lack of reflectors and the presence of diffractions (Unit 7) and the pillow basalts and volcanics indicated by the mix of chaotic and divergent reflectors (Unit 6). These three units are Layers 3, 2b, and 2a of oceanic crust (Nicolas 1989). Layer 1, oceanic sediments, is represented by Unit 1, the most recent geologic unit in the study area.
3 Numerical modeling methods
3.1 Methods justification
While geological observations and seismic experiments provide observational constraints on rifted margin evolution, they are limited because they only provide snapshots of rifting stages. Numerical modeling is necessary to link these localities in a spatio-temporal context to construct a framework for the evolution of a continental rift. The program GeoFLAC (FLAC stands for Fast Lagrangian Analysis of Continua) has been a useful tool for researchers for the past three decades by providing a means to explore the rheological, petrological, thermal, and kinematic evolution of rifting (Davis & Lavier, 2017; Detournay & Hart, 1999; Geoffroy et al., 2015; Poliakov et al., 1993). Recent work using geodynamic modelling purports to show that doming and exhumation of mantle peridotite via out-of- sequence detachments is the result of a “strength competition” between weak, frictional-plastic shear zones and thermal weakening beneath the necking domain of a continental rift (Theunissen & Huismans, 2022). Theunissen & Huismans (2022) also suggest that the mode of deformation is a consequence of varying extension rate and fault strength. Ruh et al. (2021) invoke dynamic grain recrystallization as a key process of lithospheric rupture in their own numerical modeling work. In this case, grain size reduction in olivine weakens the mantle lithosphere and allows for the formation of continentward-dipping shear zones. Our work can test these hypotheses by characterizing the roles that dynamic grain recrystallization, melt production and migration, and initial thermal conditions play in rifting.
3.2 Boundary conditions
All models presented here share initial geometries. The starting box is a two-dimensional slice of the lithosphere, 300 km wide and 150 km deep. The crustal thickness is initially 30 km for all models. To localize deformation in the center of the model space we impose thinning of the sub-continental mantle. Extensional velocity boundary conditions (1 or 2 cm./yr-1) allow the box to extend passively until remeshing is needed and the box’s width reset to 300 km. Remeshing occurs when the minimum angle in the triangular elements in the mesh drops below 15o. The Winkler foundation and the addition of material during remeshing at the lower boundary allows for asthenosphere to be upwelled from below. The foundation simulates isostatic balance for each column of elements in the model box. The compensation depth is assumed to correspond to the bottom of the model box and the elastic, viscoelastic and plastic properties of the model simulate the regional isostatic response. The top boundary is free in both temperature and stress and the side boundaries experience no heat flow. The bottom boundary temperature is varying across the box as it is calculated using the adiabat geotherm for the asthenosphere (0.003ºC/km) (Davis & Lavier, 2017).
The boundary between the lithosphere and the asthenosphere is initially defined as the 1300ºC isotherm, though elements which subsequently cross the isotherm remain labeled as asthenosphere or lithosphere even as their rheology changes in response to temperature variations (this is necessary to visually demonstrate the provenance of mantle material during the exhumation and transition phases of rifting). Melt is produced by the method described by (Katz et al., 2003) based on anhydrous decompression melting of fertile mantle and adapted by Davis & Lavier (2017) for GeoFLAC. In the numerical experiments presented here we assume that for pre-rift conditions the mantle is anhydrous and is equivalent to normal mantle lherzolite, though we use the slightly damp solidus parameterized by Hirschmann et al. (2009). Once produced, melt moves with the solid mantle and is not extracted but recrystallizes when its temperature is below the solidus (Schmeling, 2010). When the melt crystalizes, the material in the element undergoes a phase change that assigns the physical properties of plagioclase and olivine in proportions corresponding to the actual melt fraction in the particle.
The ten models were chosen to showcase a range of mantle potential temperatures from 1300ºC to 1400ºC, extension rates from 1 cm/yr to 2 cm/yr, and 45 mW/m2 to 75 mW/m2. From all those possible combinations of parameters, we chose the ones that exhibited homologous features to rifted margins and excluded models that failed to complete the rift-to-drift transition resulted in margins that were not reaching seafloor spreading or too melt-dominated to be considered magma-poor margins (> ~ 40% melt).
3.3 Rheological assumptions
In the crust, mantle lithosphere and asthenosphere we mainly use the same rheological assumptions as Lavier et al. (2019). To simulate the formation of ductile shear zones by dynamic recrystallization of olivine in the mantle lithosphere we use temperature and dissipation thresholds to initiate grain size reduction and viscous flow by diffusion creep (Bickert et al., 2020). This localization mechanism simulates the formation of high temperature anastomosing shear zones that are observed at Lanzo Massif, Nurali Massif, and elsewhere (e.g., Kaczmarek & Müntener, 2005; Kaczmarek & Müntener, 2010; Spadea et al., 2003). We use dislocation creep laws to calculate the deformation of mantle, crust, serpentinized mantle, weakened crust, basaltic crust, and sedimentary phases (Table 1) (Lavier et al., 2019). Localization of deformation in the brittle lithosphere is modeled by cohesion loss as a function of accumulated plastic strain in an elastoplastic material with a Mohr-Coulomb yield criterion (Lavier et al., 2000). The formation of ductile shear zones in the lower crust is achieved by simulating reaction or compositional weakening using a yield criterion with a work threshold dependent on both brittle (plastic) and ductile (dislocation creep) work (Lavier et al., 2019). This assumes that there always is a sufficient amount of meteoric or metamorphic fluids for a reaction to occur up to 15 km depth in the continental crust and 10 km depth for mantle serpentinization (Fricke et al., 1992). The work and temperature thresholds are given in Table 1.