Sediment routing systems to the Atlantic rifted margin of the Guiana Shield

Sediment routing systems shape the relief of continents by erosion and ensure the export of the sediments to the oceans where they are preserved in sedimentary basins at geological time scales (1–100 m.y.; e.g., Martinsen et al., 2010). Sediment routing systems have been studied mostly in orogenic domains because their relief, topography, and erosion rates are high and primarily driven by tectonic uplift (e.g., Métivier and Gaudemer, 1999; Clift and VanLaningham, 2010; Whittaker et al., 2011; Allen et al., 2013). Nonorogenic and cratonic domains have been less studied because their relief and erosion rates are low (Beauvais and Chardon, 2013; Grimaud et al., 2014, 2018). Nevertheless, their vast dimensions allowed them to contribute a significant proportion of the clastic sediments exported to the global ocean, which should not be underestimated in global studies (~30%–40%; Maffre et al., 2018; Milli man and Farnsworth, 2013).

Among the most studied anorogenic domains, northwestern Africa has undergone a homogeneous, slow and steady denudation at geological time scales (<7 m/m.y.; Beauvais and Chardon, 2013; Grimaud et al., 2018). The Equatorial Atlantic rifting, which achieved the separation of Africa and South America in the Early Cretaceous, impacted denudation of West Africa only locally and transiently. Indeed, the erosion of the rift-related relief took place essentially within a narrow coastal strip (100–300 km wide) of the rifted margin over a few million years after breakup that occurred at ca. 104 Ma (Wildman et al., 2019, 2022). Despite slow and steady denudation of the cratonic hinterland, accumulation in the basins of the Equatorial Atlantic rifted margin has varied significantly in time and in space (e.g., Grimaud et al., 2018). These variations must therefore primarily reflect modifications in the drainage network organization distributing sediment along the margin and/or longterm climate change. For example, the increase in accumulation rate in the Late Oligocene–Miocene was triggered by an increase in coastal catchment size due to the inland migration of the continental drainage divide (Chardon et al., 2016; Grimaud et al., 2018). This catchment reorganization was driven by the inland topographic growth of the Hoggar hotspot swell (Chardon et al., 2016; Grimaud et al., 2018). Conversely, the very low Paleocene–early Oligocene accumulation rates recorded along the Equatorial Atlantic rifted margins of Africa reflect greenhouse climate–induced chemical weathering of the continent (Grimaud et al., 2018). Indeed, that climate favored sediment production and storage as regolith in forest-covered weathering profiles while clastic sediment exports by river systems were reduced compared to solute exports produced by chemical weathering (e.g., Fairbridge and Finkl, 1980; Tardy and Roquin, 1998; Beauvais and Chardon, 2013; Grimaud et al., 2015).

To build on these results and gain further insights into the behavior of cratonic sediment routing systems at geological time scales, their transport capacities, and controlling factors, we investigated the Guiana Shield and its Atlantic rifted margin. The northern South American cratonic domain has undergone more tectonic and epeirogenic deformation than its conjugate West African counterpart. In particular, it has been continuously bounded to the west by the active margin of the Pacific subduction since the Neoproterozoic (Torsvik and Cocks, 2013), which evolved from a paired arc–back-arc system beginning in the early Paleozoic to the Andean cordilleran orogenic system from the latest Cretaceous onward (e.g., Ramos, 1999; Fig. 1). The Atlantic rifted margin of the Guiana Shield formed in two stages: first, in the Late Jurassic, as the southern tip of the Central Atlantic rift (e.g., Klitgord and Schouten, 1986; Pindell and Kennan, 2009; Schettino and Turco, 2009) and then, in the Early Cretaceous, as part of the Equatorial Atlantic oblique rift system (e.g., Moulin et al., 2010). This polyphase history resulted in a complex three-dimensional pattern of crustal thinning along the Atlantic margin with, in particular, alternating normal and transform margin segments showing contrasting necking styles (i.e., wide normal segments and very narrow transform segments; e.g., Loparev et al., 2021). The crustal necking style, in particular the margin width, controls the amplitude and wavelength of the rift-related topography, which in turns impacts the sediment routing systems (e.g., Rouby et al., 2013; Braun, 2018). Atlantic rifting produced rift-related shoulders and marginal upwarps that divided sediment fluxes between cratonic and rifted margin basins (Bajolet et al., 2022). This configuration lasted until the Late Miocene, when the transcontinental drainage of the modern Amazon River connected the retro-foreland basin of the Andes and the Atlantic rifted margin (ca. 10–6 Ma; e.g., Hoorn et al., 2010, 2017).

Within such an evolving framework for the Meso-Cenozoic sediment routing of the Guiana Shield (Bajolet et al., 2022), the accumulation history of its outlet, the Atlantic rifted margin, remains to be established. The stratigraphic architectures of the rifted margin basins have been investigated only individually (e.g., Brandão and Feijó, 1994; Figueiredo et al., 2007; Zalán and Matsuda, 2007; Yang and Escalona, 2011; Basile et al., 2013; Sapin et al., 2016; Casson et al., 2021). The aim of the present study is to characterize, at the scale of the whole rifted margin, the stratigraphic architecture of the sedimentary basins and map their evolving paleogeography and main depocenters. This allows us to evaluate the capacity of the Guiana Shield to export sediments to the Atlantic Ocean through time and examine the consequences of the modifications of onshore sediment routing and denudation history in the offshore sedimentary record. We used an extensive set of subsurface data (two-dimensional seismic and well data provided by Total Energies) to quantify the volumetric sediment accumulation, evolving lithologies, and paleoenvironmental conditions of the offshore basins. We also estimated the denudation history of their continental source areas from published low-temperature thermochronology of the Guiana Shield (Derycke et al., 2021). We incorporated the onshore denudation and offshore accumulation histories into an integrated platescale kinematic and continental-crust thinning and faulting framework (after Loparev et al., 2021). Our results allow evaluation of the respective influence of the Central and Equatorial Atlantic rifting, the Andean orogenesis, and the response of the cratonic domain to longterm climate change on the organization of the sediment routing systems of the Guiana Shield toward its rifted margin since the Jurassic.

The rifted margin of the Guiana Shield is located between, to the northwest, the Barbados accretionary prism forming the Caribbean subduction front (Pindell and Kennan, 2009; Yang and Escalona, 2011) and, to the southeast, the Saint Paul fracture zone (Cordani et al., 2016; Figs. 1 and 2). The northwestern portion of the margin (i.e., between the Caribbean subduction front and the Demerara Plateau) is floored by Jurassic oceanic crust accreted following rifting of the Central Atlantic (Pindell, 1985; Klitgord and Schouten, 1986; Reuber et al., 2016; Fig. 2). That portion of the margin hosts the Guiana-Suriname Basin (GS Basin). The remaining southeastern portion of the margin (i.e., between the Demerara Plateau and the Saint Paul fracture zone) is floored by Cretaceous oceanic crust formed following rifting of the Equatorial Atlantic (Moulin et al., 2010; Heine et al., 2013; Figs. 1 and 2). That portion of the margin hosts the northern and eastern Demerara basins (DEM basins) and the Foz do Amazonas Basin (FOZ Basin; Figs. 1 and 2).

During the Triassic and Early Jurassic (ca. 230–190 Ma), thinning of the continental crust in the study area took place at the southern tip of the Central Atlantic rift (e.g., Yang and Escalona, 2011; Ye et al., 2017; Fig. 1). This formed the GS Basin that includes a wide NNE-trending rift system at the location of the presentday Demerara Plateau and a very narrow NNW-trending rift along the presentday Guiana transform margin segment (e.g., Yang and Escalona, 2011; Casson et al., 2021; Loparev et al., 2021; Museur et al., 2021; Trude et al., 2022; Figs. 1, 2, and 3A). During the Late Jurassic and Early Cretaceous (ca. 190–130 Ma), rifting evolved to lithospheric breakup and oceanic accretion to the northwest of the Demerara Plateau (Pindell, 1985; Klitgord and Schouten, 1986; Pindell and Kennan, 2009; Schettino and Turco, 2009; Nemčok et al., 2016; Figs. 1 and 2).

In the Early Cretaceous (ca. 130–103 Ma), the Equatorial Atlantic rift system developed (Brandão and Feijó, 1994; Figueiredo et al., 2007; Zalán and Matsuda, 2007). It formed an en échelon system of dextral pullapart basins separated by transfer faults (e.g., Mascle et al., 1988; Guiraud et al., 1992; Benkhelil et al., 1995; Basile et al., 2005; Ye et al., 2017). In the Albian (ca. 107–100 Ma), the rift system evolved to lithospheric breakup. Oceanic accretion occurred to the northeast of the Demerara Plateau (DEM basins; Mercier de Lépinay, 2016; Sapin et al., 2016; Loncke et al., 2020) and farther to the southeast (FOZ basin; Brandão and Feijó, 1994; Figueiredo et al., 2007; Figs. 1 and 2). This formed isolated “oceanic crust domains” within the former pullapart basins separated by transform or transfer faults (Basile et al., 2005; Gillard et al., 2017; Ye et al., 2017). In the Late Cretaceous (ca. 103–83 Ma), transforms accommodated further accretion and formed the equatorial part of the rifted margin.

As a consequence of the superimposition of the rifts and their obliquity, the rifted margin of the Guiana Shield shows alternating segments of variable obliquity (Basile et al., 2005; Loparev et al., 2021; Fig. 2). They can be categorized in three types depending on the angle between the segment direction and the direction of the rifting represented by directions of the transform faults: transform, oblique, and divergent segments (Ye et al., 2019; Loparev et al., 2021; Fig. 2).

The stratigraphic architecture of the GS Basin first receives the Central Atlantic syn-rift deposits that form a wide volcaniclastic depocenter below the presentday Demerara Plateau divergent margin segment and much thinner deposits along the transform-oblique segment (Figs. 2 and 3A; Casson et al., 2021; Loparev et al., 2021; Trude et al., 2022). The syn-rift infill is truncated by the Central Atlantic breakup unconformity, above which an Upper Jurassic carbonate platform developed along the eastern boundary of the syn-rift basin (Fig. 3A; e.g., Yang and Escalona, 2011; Mercier de Lépinay, 2016; Casson et al., 2021; Loparev et al., 2021). Above that unconformity, Cretaceous post-rift clastic systems, which are widely spread from the proximal to the oceanic domains, show a longterm aggrading trend and depocenters progressively migrating into the distal margin domain and onto the oceanic crust (Fig. 3A). Thin Paleogene and Miocene deposits indicate a reduced clastic input to the GS Basin between ca. 66 and 6 Ma. Sedimentation resumed in the Pliocene and deposits are significantly thicker (Fig. 3A; Yang and Escalona, 2011; Loparev et al., 2021). Indeed, they are fed not only by Berbice River but mostly by the Orinoco River whose outlet followed Caribbean trench eastward retreat and routed sediments from the Andean retroforeland to the GS Basin (Fig. 1B; e.g., Yang and Escalona, 2011; Loparev et al., 2021).

The stratigraphic architecture of the DEM and FOZ basins comprises Early Cretaceous thin and narrow clastic syn-rift deposits under continental and shallow marine depositional environments (Fig. 3B; e.g., Mercier de Lépinay, 2016; Sapin et al., 2016; Loparev et al., 2021). These deposits are truncated by the Equatorial Atlantic breakup unconformity above which post-rift deposits are widely spread from the proximal to the oceanic domain (Fig. 3B; e.g., Mercier de Lépinay, 2016; Sapin et al., 2016; Loparev et al., 2021). Upper Cretaceous deposits first filled up existing relief and then evenly spread over the entire margin width. From the Paleogene onward, deposits became thicker in the oceanic domain than in the proximal domain (Fig. 3B; e.g., Loparev et al., 2021). During the Paleogene and Miocene, the margin was starved of clastic influx, allowing for the development of a wide carbonate platform (e.g., Carozzi, 1981; Figueiredo et al., 2007; Sapin et al., 2016; Loparev et al., 2021). The clastic input was renewed in the Pliocene, with a platform delta fed by the Maroni River on the Demerara Plateau (Fig. 3B) and, mostly, the giant Amazon Delta burying the carbonate platform under very thick clastic series (Figs. 1 and 3B; e.g., Hoorn et al., 1995, 2010; Watts et al., 2009; Shephard et al., 2010; van Soelen et al., 2017). The delta formed after the establishment of the modern course of the Amazon River in the Late Miocene, routing sediments from the Andean retroforeland across the Amazonian plains to the FOZ Basin (Hoorn et al., 1995, 2010; Shephard et al., 2010; van Soelen et al., 2017).

To determine the sedimentary history of the rifted margin basins, we interpreted a dense grid of industrial seismic reflection profiles (between the Caribbean subduction front and the Saint Paul fracture zone; extent of data set on Fig. 2 and data set map on Fig. S1 in the Supplemental Material¹). We calibrated our seismic interpretations and sedimentary facies using 33 exploration wells (located in the shelf and bathyal domains; Fig. S1). To estimate the denudation history of the Guiana Shield at the regional scale, we used the published thermal history of Derycke et al. (2021) established from inversions of low-temperature thermochronology (LTT) data coupling apatite fission tracks and (U-Th)/He dating (location of samples in Fig. 2).

We mapped nine stratigraphic horizons across the studied area (Fig. S3) using seismic and sequence stratigraphy methods (see details in Section S2.1 [see ^{footnote 1}]; Figs. S5 and S6). We illustrate the margin geometry by two interpreted cross sections (Fig. 3). Additional cross sections can be found in the Supplemental Material (Figs. S3 and S4). From the nine horizon maps, we constructed eight isochore maps in twoway travel time that we depth-converted into thickness maps. Details on the depth-conversion can be found Section S2.2. Note that for the Central Atlantic syn-rift and early post-rift intervals (200–165 Ma), evaluated thicknesses are minimal values because the volcanic content of these deposits is unknown. We integrated the isopach maps within a plate kinematic and crustal thinning framework provided by the reconstructions of Ye et al. (2017), Loparev et al. (2021), and Bajolet et al. (2022) (Fig. 4). That framework includes the position of the landmasses at the end of each considered interval and the main active faults (Fig. 4).

For each isopach map, we calculated the raw incremental volumes of sediments accumulated in the GS and FOZ basins. We followed the method of Rouby et al. (2009) to correct the raw volumes from remaining porosity and in situ production (carbonates or volcanics) and compute solid siliciclastic volumes (Figs. 5A–5C). Details of the approach can be found in Sections S2.3 and S2.4 (Fig. S6, see ^{footnote 1}). We calibrated the horizons in absolute ages to calculate incremental volumetric accumulation rates (Figs. 5D–5F; Section S2.5). We also estimated the variability of the results due to the uncertainties in the calibration of horizons in absolute ages, the depth conversion, and the corrections for porosity and in situ carbonate production using the method of Guillocheau et al. (2012; Fig. 5). Details of uncertainty estimation can be found in Section S2.6.

From the seismic and sequence stratigraphy interpretation of the subsurface data set, we constructed depositional environment maps for each time interval (Fig. 6). We mapped fluvial, coastal, and deltaic plain facies as continental environments, continental platform facies as transitional environments, bathyal facies (continental slope) as shallow marine environments (<200 m), and abyssal facies as deep marine environments (>200 m). On the continent, sedimentary paleoenvironments are represented by the area of preserved deposits (dark colored in Fig. 6) and the minimal areal extent beyond that preserved area at the time of deposition (light colored in Fig. 6) (after Ye et al., 2017; Bajolet et al., 2022). Note that for the 200–165 Ma interval, we mapped occurrences of volcanics in the syn-rift depocenter although the volcanic and clastic content is unknown (Fig. 6A). We used Ye et al. (2017) for the African rifted margin paleoenvironment reconstructions. Additional details of the mapping criteria and method can be found in Section S3 (see ^{footnote 1}).

We also constructed maps of main lithologies (Fig. 7). We defined six types of dominant lithology: sand (sand content >90%, usually the fluvial, coastal, and deltaic plain), sandclay (sand content >60%), claysand (sand content <40%), shale (sand content <10%, usually the abyssal domain), carbonates (usually the platform domain), and marl (usually deposits in bathyal environments at the toe of carbonate platforms). We calibrated lithologies on well data (Fig. S3). For the 200–165 Ma interval, we mapped the syn-rift depocenter of the Central Atlantic rift although its volcanic/clastic ratio remains unknown (Fig. 6A). Additional details of the mapping criteria and method can be found in Section S3 (^{footnote 1}).

We selected the published preferred thermal history of model of Derycke et al. (2021) from the inversion of LTT dating for six series of samples (see location on Fig. 2). Following the method of Wildman et al. (2019), we assumed that onshore sample cooling or heating was primarily driven by unroofing or burial and chose a constant paleo–thermal gradient of 25 °C/km (geothermal gradient within the first 3–5 km of Earth’s surface in a normal continental crust; DiPietro, 2013).

We derived the denudation/burial from thermal history using:

where D is denudation (km), d is denudation rate (km/m.y.), T₁ and T₂ are the expected temperatures (°C) at time t₁ and time t₂ (Ma), and G is the geothermal gradient (°C/km). We also used the 95% confidence interval of the thermal histories of Derycke et al. (2021) to estimate the uncertainties in denudation and burial estimations (Fig. 8). Denudation and burial values and associated uncertainties can be found in Table S6 (see ^{footnote 1}).

We organized our results for visualizing the spatial and temporal evolution of depocenters (Fig. 4), accumulation history (Fig. 5), depositional environments (Fig. 6), main lithologies (Fig. 7), and denudation history of the cratonic domain (Fig. 8). The paleomaps of Figures 4, 6, and 7 are arranged by time intervals from panel A to panel H. We also provide in PowerPoint File S1 (see ^{footnote 1}) an alternative presentation of the maps, with depocenters, depositional environments, and lithologies shown together on a same figure for a given time interval.

Between 200 and 165 Ma, crustal deformation was driven by the Central Atlantic rifting and early post-rift phases (Fig. 4A). The main depocenter is very thick (at least 20 km) and occupies a wide rift zone that formed below the presentday Demerara Plateau (Fig. 4A). The associated volume is at least 9 × 105 km3, including an unknown proportion of volcanics (200–165 Ma interval; Fig. 5A). Besides this volcaniclastic syn-rift infill, depositional environments are continental to transitional (Fig. 6A). Deposits are clastic (claysand) along the South American rifted margin (Fig. 7A). In contrast, a carbonate platform developed along the African part of the rifted margin (Fig. 7A). This is consistent with little clastic input to that part of the margin at the time, although no data are available for its clastic accumulation history. To the northwest of the study area, the Central Atlantic breakup and oceanic accretion took place. The associated distal basin developed under shallow and deep marine depositional environments (shale dominated; Figs. 6A and 7A). Inland, the SE-trending rift branch separating the Guiana Shield and the West African craton, as well as the Marajó Basin, received sandclay deposits under fluviodeltaic environments (Figs. 6A and 7A).

After Central Atlantic rifting (165–130 Ma), the main depocenter remained below the presentday Demerara Plateau but spread northwestward to the distal part of the rifted margin (Figs. 3A and 4B). It was also much thinner (<7 km) than during the previous time interval (Figs. 3A, 4A, and 4B). A continuous strip of transitional depositional environments continuously fringed the rifted margin in Africa and northern South America (Fig. 6B). That strip mostly coincides with a carbonate platform (Fig. 7B). This is consistent with the very low clastic input to the margin at the time (165–130 Ma interval; ~0.2 × 102 km3 and 60 km3/m.y.; Figs. 5A and 5D). In the distal domain of the rifted margin, depositional environments have remained deep marine (shale dominated) until the present day (Figs. 6B–6H and 7B–7H).

Between 130 and 103 Ma, crustal deformation was due to the Equatorial Atlantic rifting (syn-rift and early post-rift) in the DEM and FOZ basins (Fig. 4C). Accordingly, in the GS Basin, only normal faults restricted to the sedimentary cover were active, mostly accommodating collapse of the platform, westward along the divergent margin segment and northward along the oblique-transform margin segment (Loparev et al., 2021; Figs. 2, 3A, and 4C). The main depocenter shifted westward into the eastern GS Basin (Figs. 3A and 4C). In the Equatorial Atlantic rifting domain (DEM and FOZ basins), depocenters were controlled by en échelon dextral pull-apart basins formed in between the Guiana Shield and West African craton (Fig. 4C). Farther south, the Marajó rift received sediments as well (Fig. 4C). The clastic supply initiated in the DEM and FOZ basins was renewed in the GS Basin (130–103 Ma interval; Figs. 5A and 5B). Deposits in fluvial-deltaic and transitional environments (sandclay) covered most of the previous carbonate platform except in a few local patches (Figs. 6C and 7C). The African conjugated margin showed more pronounced continental environments (sandclay deposits) fringed oceanward by a very narrow carbonate platform (Fig. 7C). The Equatorial Atlantic rift domain was under shallow environments with claysand deposits except in the newly formed isolated patches of oceanic crust that reach abyssal environments with shale-dominated deposits (Figs. 6C and 7C).

Between 103 and 66 Ma, crustal thinning ceased and separation of the Guiana Shield and the West African craton became accommodated by accretion and transform faults (Figs. 4D and 4E). Only very local and limited gravity-driven deformation occurred in the main depocenters of the GS and FOZ Basins between 103 and 83 Ma and ceased afterwards (Loparev et al., 2021; Figs. 4D and 4E). Volumes of clastic sediment accumulated in the GS Basin decreased over the Late Cretaceous (from ~2.7 to ~1.6 × 10⁵ km³) while they increased in the FOZ Basin (from ~2.4 to ~6.5 × 10⁵ km³; 103–66 Ma interval; Figs. 5A and 5B). The Guiana Shield rifted margin was dominated by transitional and bathyal environments (Figs. 6D and 6E). Except for a few isolated patches of carbonates, deposits were clastic with higher sand contents in the proximal domain of the FOZ Basin (sandclay) than in the GS Basin (claysand; Figs. 7D and 7E). During that period, turbiditic systems developed, in particular at the junction of the oblique and normal margin segments in the GS Basin and in the FOZ Basin (Figs. 2, 7D, and 7E). These systems exported sanddominated deposits across the shelf break from the platform to the shale-dominated abyssal plain (Figs. 7D and 7E).

Thus, after Equatorial Atlantic breakup, the Guiana rifted margin was dominated by clastic sedimentation with depocenters controlled by post-rift subsidence as well as the main river outlets (Figs. 4D–5E, 6D–6E, and 7D–7E). In the GS Basin, the main depocenter was fed by sediment routed through the intersection between the oblique and divergent segments, potentially from the Essequibo, Berbice, and/or Maroni Rivers or their ancestors (Figs. 2B, 4D, and 4E). In the DEM and FOZ basins, the main depocenter was fed by sediments routed through the former Marajó rift, probably from a paleo–Tocantins and paleo–Oyapok Rivers (Figs. 4D and 4E).

Between 66 and 6 Ma, clastic accumulation was very limited (66–6 Ma interval; Fig. 5). The proximal domains of the GS Basin and the Demerara Plateau underwent marl- or shale-dominated deposition while a carbonate platform was installed in the FOZ Basin (Figs. 6F, 6G, 7F, and 7G). The number of turbiditic systems decreased drastically with respect to the Late Cretaceous (Figs. 7F–7H). Only small calciturbiditic lobes were emplaced in the abyssal plain east and west of the Demerara Plateau (Figs. 7F and 7G) as well as a few sand lobes at the transition between the oblique and divergent margin segments of the GS Basin (Figs. 7F and 7G).

Between 6 and 0 Ma, clastic supply to the margin drastically increased through the Amazon, Maroni, Essequibo-Berbice, and Orinoco deltas (66–6 Ma interval; Figs. 4H, 5, and 6H) following establishment of the transcontinental drainage linking the Andean retro-foreland to the Atlantic rifted margin by the modern Amazon River and Orinoco River systems (between 10 and 6 Ma; Hoorn et al., 1995, 2010; Shephard et al., 2010; van Soelen et al., 2017). Accumulations were mostly clastic in fluvial and transitional depositional environments (sand-clay and clay-sand facies). These deposits covered the former carbonate platform except at the southeastern tip of the study area (Figs. 6H and 7H). Turbiditic export resumed in the FOZ Basin (Fig. 7H).

The six batches of LTT samples are located within 300 km of the Guiana Shield rifted margin (Fig. 2). Overall, the samples show heating between 200 and 145 Ma (burial rate as high as −35 m/m.y.), rapid cooling between 145 and 100 Ma (denudation rate of 30 to 60 m/m.y.), and very slow and steady cooling afterwards (denudation rate of <20 m/m.y. or burial; Fig. 8).

Derycke et al. (2021) related the heating trend between 200 and 140 Ma (“burial” on Fig. 8) to a Central Atlantic magmatic province (CAMP)–related magmatism heating rather than to burial. However, heating overlaps with the Central Atlantic rifting (ca. 230–165 Ma). Basile et al. (2020) suggested that the magmatic activity of this rifting might have been related to a hot spot (ca. 170–180 Ma). This could provide an alternative origin for this heating episode. The following increase in denudation rates between 145 and 100 Ma (to as much as ~40 ± 10 m/m.y.; Fig. 8; Table S6 [see ^{footnote 1}]) is coeval with the Equatorial Atlantic rifting. It is interpreted by Derycke et al. (2021) as due to the erosion of the rift-related relief (Figs. 4C and 8).

After ca. 90–100 Ma, samples were at <2 km depth, implying a less constrained thermal history because they recorded temperature too low to be interpreted as due to denudation or burial (Derycke et al., 2021). Paleogene (>30 Ma) lateritic weathering profiles are currently preserved over most of the Guiana Shield (Théveniaut and Freyssinet, 2002). Their formation since probably the Late Cretaceous and their current preservation are consistent with limited denudation (<10 m/m.y.) or burial of the shield since ca. 70 Ma Fig. 8; Table S6). This is also consistent with estimations of denudation made for the same period for the conjugated West African craton from both thermochronological and geomorphological data (Beauvais et al., 2008; Grimaud et al., 2018; Wildman et al., 2019, 2022).

The accumulation history of the Guiana Shield margin is controlled primarily by the rifting of the Central Atlantic in the Early–Middle Jurassic (200–165 Ma; Fig. 4A) and of the Equatorial Atlantic in the Early Cretaceous (130–103 Ma; Fig. 4C). The contrasted crustal and stratigraphic architectures of the associated basins along the Guiana Shield rifted margin show that the Central Atlantic and Equatorial Atlantic rifts resulted from very different types of crustal thinning processes (Fig. 3).

For instance, the Central Atlantic syn-rift depocenter is much larger than the Equatorial Atlantic syn-rift depocenter (>9 × 10⁵ km³ and 2.5 × 10⁵ km³ respectively; Figs. 3, 4A, 4C, 5A, and 5B). Also, the Central Atlantic syn-rift deposits include an unknown but probably large proportion of volcanics (i.e., seaward-dipping reflectors; Fig. 7A) while the Equatorial Atlantic syn-rift deposits do not (Fig. 7C). Additionally, the thermal history of the Guiana Shield suggests that Central Atlantic rifting was coeval with a heating episode, tentatively related to a magmatic event (CAMP or hot spot), while Equatorial Atlantic rifting was coeval with a cooling episode related to the erosion of the rift-related relief. Finally, the Central Atlantic rifting has involved a larger proportion of ductile deformation during thinning than the Equatorial Atlantic rifting (e.g., Loparev et al., 2021; Museur et al., 2021).

In terms of depositional environments (outside the volcanic depocenter), syn-rift clastic deposits of the Central Atlantic are finer grained (clay-sand; Fig. 7A) than the Equatorial Atlantic syn-rift deposits (sand dominated; Fig. 7C). This suggests that the Equatorial Atlantic syn-rift sediments resulted from the erosion of steeper rift-related reliefs than the Central Atlantic syn-rift sediments. Furthermore, in the Central Atlantic domain (165–130 Ma; Fig. 4B), a wide carbonate platform developed immediately after the breakup (Fig. 7B), while accumulated clastic volumes were extremely low (Figs. 4B, 5A, 5D). This suggests that the rift-related reliefs were mostly eroded away immediately after breakup and/or too low to provide a significant clastic supply to the Central Atlantic rifted margin (Figs. 4B, 5A, 5D, and 7B). By contrast, along the Equatorial Atlantic rifted margin, large volumes of sand-dominated deposits associated with turbiditic systems accumulated throughout the Late Cretaceous (i.e., during rifting, breakup, and early post-rift; 103–66 Ma; Figs. 5B, 5E, and 7C–7E). This suggests that Equatorial Atlantic rifting formed a relief (upwarp) high and persistent enough to sustain an entrenched drainage delivering coarse-grained clastics to the Equatorial Atlantic rifted margin for 40 m.y. after breakup. This drainage also allowed renewal of the clastic supply to the GS Basin (former Central Atlantic rifting) throughout the Late Cretaceous (i.e., 100 m.y. after the Central Atlantic rifting), burying the former carbonate platform under sand-dominated and clay-dominated deposits (Figs. 7C–7E). To summarize, the Equatorial Atlantic rift-related relief (and drainage system) sustained a larger, coarser-grained, and longer-lasting clastic supply (>40 m.y. after rifting) than that feeding the Central Atlantic rift, which was eroded away immediately after rifting (Figs. 4A, 5A, and 7B).

These differences in the types of crustal deformation and rift-related reliefs may be related to the contrasted kinematic developments of the two rifts. Indeed, the Equatorial Atlantic rifting was faster and more oblique than the Central Atlantic rifting (Brune et al., 2016, 2018). Faster and more oblique rifts produce a narrower rifted margin with higher rift-related relief than slower and normal rifts (e.g., Svartman-Dias et al., 2015; Theunissen and Huismans, 2019; Wolf et al., 2022). This is also consistent with the Central Atlantic rifting involving a larger proportion of ductile deformation during thinning than the Equatorial Atlantic rifting (e.g., Loparev et al., 2021). Indeed, a weaker and/or warmer lithosphere generates lower rift relief than a colder lithosphere (e.g., Beucher and Huismans, 2020; Wolf et al., 2022).

In terms of source to sink, the sand content and hence the siliciclastic volume of sediments produced by rifting of the Central Atlantic is unknown. However, the Equatorial Atlantic rift preserved a solid volume of ~5 ± 1.9 × 10⁵ km³ between 130 and 103 Ma (Fig. 5; Table S5 [see ^{footnote 1}]). This volume is at first order consistent with the ~1.5 km of unroofing recorded by LTT data of Derycke et al. (2021) between 130 and 100 Ma if generalized to the area of the Guiana Shield (~500 × 500 km; Table S6).

Overall, the Equatorial Atlantic rift-related relief sustained a larger, coarser-grained, and longer-lasting clastic supply than the relief feeding the Central Atlantic rift, whose relief was eroded away rapidly after rifting. One would therefore expect larger, coarser-grained, and longer-lasting clastic supply in narrow and fast oblique rifts than in wider and magmatic rifts. Nevertheless, in the study area, the Central Atlantic and Equatorial Atlantic rifts were super-imposed, and the relative contribution of Central Atlantic rifting inheritance on relief dynamics of Equatorial Atlantic rifting remains to be evaluated.

After Equatorial Atlantic breakup, the Guiana rifted margin was dominated by clastic sedimentation, with depocenters controlled by post-rift subsidence as well as the main river outlets (Figs. 4D–4E, 6D–6E, and 7D–7E). Unroofing of the Guiana Shield remained low, steady, and spatially evenly distributed (Fig. 8). Meanwhile, the accumulated sediment volumes and accumulation rates increased during the Late Cretaceous (Figs. 5C and 5F). But the accumulation trends differ in the GS and FOZ Basins: Accumulated volumes and accumulation rates increased in the FOZ Basin, while accumulated volumes decreased and accumulation rates remained steady in the GS Basin (Figs. 5A, 5B, 5D, and 5E). Given the bulk steady unroofing of the Guiana Shield, this contrast in accumulation history cannot be related to an increase in rock uplift in the source area, whether driven by tectonics or mantle dynamics. Nor can it be related to a change to a climate more propitious to physical erosion given that the two basins show opposite accumulation trends. It therefore most probably results from change(s) in the organization of the drainage feeding the rifted margin. Few geologic data are available to further constrain such modifications. But several authors have suggested that the drainage supplying the GS Basin during the Early Cretaceous included the Takutu graben (McConnell, 1969; Crawford et al., 1985; Roddaz et al., 2021; Bajolet et al., 2022; Trude et al., 2022; Fig. 9). This implies that a drainage longer than at present day would have contributed to renewing siliciclastic accumulations in the GS Basin after ca. 145 Ma (Figs. 5A and 7C–7E). To account for the following decrease in accumulation (ca. 145–66 Ma; Fig. 5A), we suggest that the drainage area decreased in the Late Cretaceous, i.e., the watershed, initially located south of the Takutu graben, migrated oceanward (Fig. 9). Nevertheless, the mechanism driving this migration is yet to be determined.

The Marajó Basin contributed to the drainage that fed the FOZ Basin during Equatorial Atlantic rifting (paleo–Tocantins River; Figs. 1, 6C, and 7C). At the time (130–103 Ma), the Marajó Basin was separated from the Parnaíba Basin located to the southeast by an arch preventing sediment transfer between the basins (e.g., Petri, 1987; Bajolet et al., 2022). During the Albian and onward (<110 Ma), this arch became subdued, allowing for the connection of the two basins (Bajolet et al., 2022). To account for the increase in accumulation in the FOZ Basin in the Late Cretaceous, we suggest that the watershed of the drainage to the FOZ Basin migrated inland to incorporate the Parnaíba Basin area (Fig. 9).

In the Late Cretaceous (100–66 Ma), the drainage of the Guiana Shield was partitioning the sediment supply toward either inland cratonic basins or the rifted margin via northeastward-flowing coastal rivers (Bajolet et al., 2022; Figs. 6C–6E, 10, and 11D). We suggest that the watershed of this coastal drainage system migrated oceanward during that period, while the watershed of the coastal drainage to the FOZ Basin migrated inlandward (Fig. 9).

From 66 to 6 Ma, sediment supply to the margin was very reduced (Figs. 4F, 4G, and 5) and allowed for the growth of a major carbonate platform in the FOZ Basin (Figs. 7F and 7G). Subdued Paleogene–Miocene clastic fluxes to the margin were coeval with intense lateritic weathering over cratonic northern South America (Vasconcelos et al., 1994) and particularly on the Guiana Shield, which preserves extensive lateritic or bauxitic weathering mantles formed between 70 and 5 Ma (e.g., Théveniaut and Freyssinet, 2002).

Very low clastic fluxes along the Guiana Shield margin during the Paleogene and Miocene are in agreement with similarly subdued clastic flux estimates along the conjugated African Equatorial margin (Sierra Leone to Benin: Grimaud et al., 2018; Wildman et al., 2022). Clastic fluxes were subdued as well in basins of the Central Atlantic (Senegal: Lodhia et al., 2019) and of the South Atlantic rifted margins at that time (Niger: Grimaud et al., 2018; southern Africa: Baby et al., 2020; Pelotas Basin, offshore Brazil and Uruguay: Rohais et al., 2021). These peri-Atlantic low clastic fluxes are attributed to the Paleogene greenhouse climate and the mid-Miocene thermal optimum that favored sediment production and storage as regolith in forest-covered weathering profiles. During these periods, clastic sediment exports (i.e., regolith) to river systems were therefore reduced compared to solute exports produced by chemical weathering (e.g., Fairbridge and Finkl, 1980; Tardy and Roquin, 1998; Beauvais and Chardon, 2013; Grimaud et al., 2015). These periods were accordingly propitious to marine chemical sedimentation (e.g., carbonates, phosphates, attapulgite clay accumulation; Millot, 1970).

After 6 Ma, the major increase in siliciclastic supply to the rifted margin basins records the establishment the modern Amazon River and Orinoco River courses linking the Andean retroforeland basins to the Atlantic rifted margins (Hoorn et al., 2010; Figs. 3, 4H, and 5C). Clastic-dominated sedimentary systems were reinstalled along the whole margin (Fig. 7H). This includes the short coastal drainages such as the Maroni and Oyapok Rivers (Fig. 6H). Nonetheless, the main drainage systems feeding the rifted margin were the Essequibo-Berbice and Orinoco deltas in the GS Basin, the Maroni delta on the Demerara Plateau, and the Amazon delta in the FOZ Basin.

Overall, the accumulation history of the Guiana Shield rifted margin shows phases of very low siliciclastic supply (Figs. 11B and 11E) alternating with phases of higher siliciclastic supply (Figs. 10, 11C, and 11F). Periods of low siliciclastic input are associated with the development of carbonate platforms (Figs. 11B and 11E). They correspond to either (1) very limited relief variations in the drainage areas such as during the Late Jurassic for the GS Basin (Figs. 10 and 11B) or (2) periods of intense lateritic weathering of the cratonic source area such as in Paleogene–Miocene time for the FOZ Basin (Figs. 10 and 11E).

In contrast, periods of higher siliciclastic input to the Guiana Shield rifted margin basins correspond to either (1) relief-topographic growth in the drainage basin supplying the margin such as during the Equatorial Atlantic rifting in the Early Cretaceous (Figs. 10 and 11C), (2) drainage reorganization over a steadily eroding cratonic domain such as in the Late Cretaceous (Figs. 10 and 11D), or (3) tapping of sediments produced in the Andean orogenic domain such as during the Plio-Pleistocene via the presentday Orinoco and Amazon Rivers (Figs. 10 and 11F).

We determined the sedimentary budget of the Guiana Shield rifted margin from the stratigraphic architecture of the Guiana-Suriname, northern and eastern Demerara, and Foz do Amazonas sedimentary basins. We investigated the implications of their histories of accumulation, depositional environments, and grainsize distribution in terms of sediment routing systems.

The accumulation history of the Guiana Shield rifted margin shows two phases of very low siliciclastic input associated with development of carbonate platforms. In the Late Cretaceous, reduced fluxes corresponded to a period of low continental relief variation following the rapid removal of the Central Atlantic rift-related relief. In the Neogene, low fluxes were due enhanced continental weathering that prevented solid exports from the Guiana Shield.

These periods alternated with phases of higher sediment supply to the rifted margin. In the Early Cretaceous, high clastic fluxes were due to the erosion of the Equatorial Atlantic rift-related relief that was steep enough to sustain a coarse siliciclastic supply to the margin for more than 40 m.y. In the Late Cretaceous, high clastic fluxes were due to changes in the organization of the drainage over the slowly eroding Guiana Shield: The drainage area of the Guiana-Suriname Basin decreased while the drainage area of the Foz do Amazonas Basin increased. In the Mio-Pliocene, the continental-scale reorganization of the drainage, establishing the presentday course of Orinoco and Amazon Rivers, massively increased the accumulation in the basins of the Guiana Shield rifted margin.

We thank the Bureau de Recherches Géologiques et Minières and Total Energies for funding this work within the framework of the S2S project and the R&D Total Energies group for providing subsurface data. Subsurface data are under restrictive proprietary license and thus not open for distribution. We thank Cecile Robin for contributing to the accumulation calculation and Yann Montico for contributing to the isochore depth conversion. We also thank Nicholas Hayman for his review.