Contrôles sur la répartition des argiles organiques dans les bassins profonds

 Influence of chemical weathering and marine authigenesis on provenance proxies 

The REE contents, Eu anomaly, elemental ratios (Cr/Th and Th/Sc) and Sr-Nd isotopic compositions of sedimentary rocks prove to be useful tools for determining the provenance of sediments (McLennan et al., 1993). However, before drawing conclusions on provenance, the effects of chemical weathering and marine authigenesis should be carefully assessed as these processes may control the REE contents and radiogenic isotope composition of marine sediments. Despite the fact that individual samples have variable and high CIA values (Table 2.2., Figure 2.2.), the absence of any correlation with REE content, Eu/Eu*, Cr/Th, Th/Sc, Sr and Nd isotopes (see supplementary material Annexe 4) suggests that chemical weathering has not modified these provenance proxies. Any influence from carbonates, Fe-Mn oxides or organic matter formed under marine conditions may also be ruled out since these constituents are removed by the sequential leaching steps. Moreover, the REE patterns of authigenic smectites formed in deep-water environments are normally characterized by LREE depletion, HREE enrichment and strong negative Ce anomalies (ΩCe < 0.50, De Baar et al., 1985) typical of seawater (Piper, 1974). Here, all the analysed DSDP samples have Ce anomalies higher than or equal to 0.66. This suggests that, even if authigenic smectites were incorporated during early diagenetic processes, its proportion was much less than the detrital fraction. REE abundance patterns support this conclusion, with the exception of one sample from DSDP Site 137 situated close to the mid-oceanic ridge (Figure 2.1B.), since all the other analysed samples (Figure 2.6.) plot outside the seawater field, suggesting a negligible influence of marine authigenesis. The Early Cenomanian sample from DSDP Site 137 plots in the seawater field, and is therefore considered as strongly affected by marine authigenesis. This sample is not taken into account for the provenance interpretations. values are calculated following Martin et al. (2010). The MREE bulge and “bell-shaped” REE profiles correspond to REE patterns observed in fish teeth, Fe-Mn oxides, organic matter and pore waters, while “HREE-enriched” profiles correspond to modern sea water (Huck et al., 2016; Moiroud et al., 2016). These end-members reflect the REE contents of marine sediments influenced by seawater or authigenic phases, while “flat” REE patterns are characteristic of continental clays (see Huck et al., 2016; Moiroud et al., 2016 for a review). Samples from each DSDP Site are represented by specific symbols used in Figures 2.4. and 2.5. High Eu anomalies and Th/Sc ratios are characteristic of felsic and more highly differentiated source rocks, whereas high Cr/Th and low Eu anomalies suggest more mafic and less differentiated source rocks (e.g., McLennan et al., 1993; Cullers, 2000). Here, the samples from the DSDP and CM1 sites have higher Cr/Th, but lower Th/Sc values and Eu anomalies compared to PAAS (Table 2.2.). This suggests that the detrital supply came from less differentiated sources than those of PAAS, the latter being considered as representative of the Upper Continental Crust (Taylor and McLennan, 1985). If we exclude the 104 diagenetically altered Early Cenomanian sample from DSDP Site 137, no temporal or spatial variations are observed in CIA values, Eu and Ce anomalies or in the elemental ratios used for provenance studies (Cr/Th and Th/Sc; Figures 2.4., 2.5. and Table 2.2.). Only the İ Nd(0) and 87Sr/86Sr values are found to vary through time, with a trend toward lower values during the Late Cretaceous (Figure 2.2.). Hence, this decrease reflects a progressive shift in the average age of source rocks rather than changes in the nature of the source (felsic as against basic). When plotted in a 87Sr/86Sr vs. İ Nd(0) diagram and compared with relevant source fields (Precambrian, Hercynian/Paleozoic and CAMP sources) and with present-day Central Atlantic Ocean sediments (Figure 2.7.), the Sr-Nd isotopic compositions of analyzed Cretaceous sedimentary rocks do not show any covariation with 87Sr/86Sr ratios. In detail, the two CM1 samples with the highest İ Nd(0) values do not yield the least radiogenic Sr isotopic composition, and the samples with the lowest İ Nd(0) values do not show the highest Sr radiogenic composition (Figure 2.7.). As Nd is not fractionated by sedimentary processes, this indicates some loss or gain of radiogenic 87Sr during the erosion, weathering, transport and deposition of the sediments. Studies of the geochemistry of present-day Suspended Particulate Matters (SPM) transported by South American Equatorial rivers have pointed out the much higher variability of SPM Sr isotopic composition when compared with corresponding Nd isotopic composition over a one-year hydrological cycle (Viers et al., 2008; Rousseau et al., submitted). These studies also highlight that the SPM Nd isotopic composition is a much more robust provenance tracer than the Sr isotopic composition. Hence, the provenance interpretation given here is mainly based on the Nd isotopic composition of the analysed samples.

Provenance of Cretaceous sediments in the eastern 

Central Atlantic Ocean Based on İ Nd(0) variation, we can propose a three-step evolution for the paleogeographic evolution of the WAC (Figure 2.8.). Neodymium isotope values are presented on three distinct paleogeographic maps illustrating the reorganization of drainage patterns inferred for Northwest African Margin during the Albian-Middle Cenomanian period (Figure 2.8A.), in the Late Cenomanian−Turonian interval (Figure 2.8B.) and during the Campanian−Maastrichtian period (Figure 2.8C.).

 Albian−Middle Cenomanian 

Overall, the Albian−Middle Cenomanian DSDP samples yield the highest İNd(0) values and the largest range of variation (-5.5 to -15) when compared with other Cretaceous DSDP samples (Figure 2.8A.). This spatial heterogeneity implies distinct provenances and hence the co-existence of several small drainage basins restricted to peripheral domains of the WAC. The lowest İ Nd(0) values (-14.9 and -15) correspond to samples from DSDP Site 369 situated off the Mauritanian margin. These values are similar to those of modern detritus deposited in the Atlantic Ocean (Grousset et al., 1998; Meyer et al., 2011) and the Late Cretaceous-Early Eocene sedimentary rocks of the Tarfaya Basin in South-West Morocco (Ali et al., 2014). In comparison to present-day African catchment areas, such İ Nd(0) values are similar to the clays from the Congo River supplied predominantly by Precambrian terrains (Bayon et al., 2015). This suggests an unradiogenic Precambrian source as the main source of terrigenous sediments to the area of DSDP Site 369. The Proterozoic domains of the eastern Reguibat Shield seem the best candidates for this clastic supply since they have İ Nd(0) values that are closely similar to the DSDP site samples(Peucat et al. 2005). In the northern WAC domains, the İ Nd(0) values are more radiogenic and increase from -11.2 to -5.5 at DSDP Site 370 and from -9 to -7.2 at DSDP Site 415A, suggesting a regional change in the provenance (Figure 2.8A.). The Albian İ Nd(0) values of -11.2 (DSDP Site 370) and -12.5 (DSDP Site 416) and the Eu anomalies (Eu/Eu*~0.71 to 0.73) are similar to those measured in modern clays from the “Niger sub delta” (İ Nd(0) = -11.9 and Eu/Eu*~0.71; Bayon 107 et al., 2015). As the Niger River catchment drains both Paleozoic and Precambrian terrains (Milesi et al., 2010), we suggest that Albian sedimentation at the DSDP Site 370 was initially supplied by a mixed Paleozoic/Precambrian source. The lack of a significant change in Eu anomalies, or in Cr/Th and Th/Sc ratios, implies that the rise in İ Nd(0) is not caused by an increasing supply of more mafic detrital material from the same mixed Paleozoic/Precambrian source, but is rather linked to enhanced inputs from younger crustal sources. This could be related to a greater contribution from Hercynian and older Paleozoic rocks of the Meseta, Anti-Atlas and High Atlas Mountains, which show an average İ Nd(0) value of -5.6 (Figure 2.8A. and supporting information Annexe 2). The shallow marine samples of CM1 well yield İ Nd(0) values intermediate between modern Nile River clays (İ Nd(0) = -7.1; Bayon et al., 2015) and modern Niger River clays (İ Nd(0) = -11.9; Bayon et al., 2015), which are derived from the erosion of Paleozoic/Hercynian units and mixed Paleozoic/Precambrian domains, respectively (Milesi et al., 2010; Bayon et al., 2015). This suggests that the sediments in CM1 well were supplied by a drainage system eroding both Paleozoic and Precambrian rocks. On the one hand, the Precambrian sources may be located in the northwestern part of the Leo-Man domain (Figure 2.8A.). On the other hand, erosion of the Paleozoic sedimentary rocks of the Bowe Basin, as well as the Hercynian massifs of the Mauritanides could supply Paleozoic/Hercynian terrigenous detritus to the CM1 site, since these domains are considered to have been emerged from the Albian to the Cenomanian (Guiraud et al., 2005, Figure 2.8A.). The deep-water DSDP samples from Sites 137 and 367 have İ Nd(0) values higher than those from the shallow marine CM1 well, suggesting that sediments from the latter were not supplied by the same drainage basin. These less negative İ Nd(0) values suggest an increased input of younger detritus to these deep-water sediments compared with the shallow marine CM1 well. We exclude a South American provenance because Cretaceous sediments deposited on the South American margin have much lower İ Nd(0) values (-15.2 to -16.2, Martin et al., 2012). A North American provenance can also be excluded because, if this 108 hypothesis were correct, the samples from DSDP Sites 138 and 137 should all have the same isotopic signature. As the samples from deep-water DSDP Sites 137 and 367 have İ Nd(0) values close to the DSDP Site 415 samples, this rather suggests a similar provenance. If correct, this implies the influence of oceanic currents transporting detritus from the Moroccan Atlantic coast to the central part of the Equatorial Atlantic Ocean, while preventing the arrival of material from the Senegalese continental shelf (CM1). This scenario agrees very well with the southwestward oceanic dispersion of palygorskite clay minerals from Morocco (Pletsch et al., 1996).

Late Cenomanian−Turonian 

A lowering and homogenization of İ Nd(0) values is recorded in deep-water environments of the eastern Central Atlantic Basin during the Late Cenomanian−Turonian interval (Figure 2.8B.). Compared with data for the previous interval, the İ Nd(0) values drop by -4.1 to -4.7 units for DSDP Site 367 and by -5.1 units for DSDP Site 415A. This contrasts with drops of -0.8 to -1.5 units recorded at DSDP Site 137 and -0.8 to -1.8 units on the Demerera Rise (Martin et al., 2012, Figure 2.8B.). A positive İ Nd(0) excursion of +8.3 units is observed in the Late Cenomanian−Turonian sediments from the CM1 well. Owing to the Late Cenomanian and Turonian hiatus at DSDP Sites 370, 416 and 369, this interval is only sparsely represented in Morocco (Figure 2.8B.). Nevertheless, one sample from DSDP Site 415A displays a low İ Nd(0) value of -12.3 which is in the same range as recorded at DSDP Sites 367 and 368. This suggests that a long-term decrease of İ Nd(0) values also occurred in the northern part of the WAC. When compared to the catchments of modern African rivers, the lack of significant differences in İ Nd(0) values between the sedimentary rocks from DSDP Sites 415A, 137 and 367 and modern Niger River sediments suggests the existence of mixed Paleozoic/Precambrian sources. When compared with older sedimentary rocks, the shift toward more negative İ Nd(0) values argues for increasing inputs of Precambrian detrital material to the eastern Central Atlantic Basin during the Late Cenomanian−Turonian. The reduced contribution of Paleozoic sources and increasing contribution of Precambrian sources may result from an extension of drainage areas toward the cratonic basement of the Reguibat and Leo-Man shield areas. This is because most Paleozoic source areas such as the Meseta (Morocco) or the Bowe Basin were flooded during this time interval (Guiraud et al., 2005, Figure 2.8B.). However, this long-term decrease in İ Nd(0) values is not recorded by samples from the shallow marine CM1 well, which show a positive excursion of +8.5 to +6 epsilon units. This suggests a greater local contribution from younger Paleozoic/Hercynian terrains probably corresponding to the Mauritanides (Figure 2.8B.). Moreover, the discrepancy 111 between the İ Nd(0) values from CM1 well and DSDP Site 367 indicate that sediments at this deep-water site were still not sourced by clastic supply from the Mauritanides.

Campanian−Maastrichtian 

At the end of the Campanian−Maastrichtian, the sedimentary rocks of the Northwest African Margin from DSDP Sites 367, 368 and 369 yield a narrow range of İ Nd(0) values (- 14.3 to -15, Figure 2.8C.). Compared with the İ Nd(0) values for the previous time interval, this implies a negative shift of -1.9 to -4 units. A decrease of -6 units is also recorded in the CM1 sample on the platform, as well as in the Campanian−Maastrichtian sedimentary rocks of the Demerara Rise showing changes by -0.6 and -2.6 units towards more radiogenic neodymium isotopic composition (Martin et al., 2012). The İ Nd(0) values of Campanian−Maastrichtian DSDP sedimentary rocks are close to the Precambrian end-member (Figure 2.7.) and similar to values observed in present-day marine sediments off the Northwest African Margin (Grousset et al. 1998; Meyer et al., 2011) and modern sediments deposited by the Congo River (Bayon et al., 2015). As present-day sediments are supplied by Precambrian sources, this suggests that Precambrian terrains represent the predominant source of clastic supply along the Western African Margin during the Campanian−Maastrichtian. The Campanian−Maastrichtian sedimentary rocks of DSDP Site 137 show an opposite trend with a positive shift of +2.6 units (İ Nd(0) ~ -7.7, Figure 2.8C.). Compared with other values, this suggests that the source of the sedimentary supply may not be the West African Craton. This could reflect an increasing contribution of volcanic sediments/rocks from the mid-ocean ridge. However, the REE pattern of this sample is flat with respect to PAAS and we find no difference in Eu anomaly compared with the other analysed samples (Table 2.2.). Hence, we can rule out a significant contribution from volcanic sources. The potential source area for this sedimentary rock could lie farther to the North, since the İ Nd(0) value of -7.7 is within the range of NW Tethyan sediments (i.e., -12 to -6 İ Nd units; Dera et al., 2015) or North American Appalachian rocks (-10 to -5 İ Nd units, Patchett et al., 1999). 

 Possible mechanisms for drainage reorganization 

Our provenance results indicate that the Albian−Middle Cenomanian interval is characterized by at least three main paleodrainage basins with restricted extensions toward the hinterland: two of them draining Paleozoic/Hercynian units in the northern and central parts of the WAC and another draining Precambrian units in the Reguibat Shield region (Figure 2.8A.). This drainage partitioning implies that catchment basins could have been separated by areas of relief on the periphery of the WAC. The existence of topographic highs acting as natural barriers dividing and limiting the extension of drainage areas suggests that significant uplift may have affected the WAC during the Albian−Middle Cenomanian. The mechanisms causing these uplifts may be related to the onset of opening of the South and Equatorial Atlantic Ocean during the Early Cretaceous and Late Albian (Förster, 1978; Flicoteaux et al., 1988; Moulin et al., 2010) or increasing compressive stress in northern, central, eastern and southern region of Africa induced by the anticlockwise rotation of Africa during this period (Guiraud and Maurin, 1991; Guiraud and Bosworth, 1997; Moulin et al., 2010) or a combination of all these factors. Whatever the mechanisms, the processes of uplift are likely to have created topographic barriers that may have led to the reorganization and separation of drainage basins supplying DSDP Sites 369, 415A, 370 and 367. During the Late Cenomanian−Turonian, with the exception of the Senegalese continental shelf (CM1 well, Figure 2.8B.), there was an increased supply of Precambrian detritus to the deep-water basin (DSDP Sites 367, 137 and 415A), as shown by lower İ Nd(0) values compared with values for the Albian−Cenomanian (Figure 2.8B.). This change may be due to a reduced contribution of proximal Hercynian/Paleozoic units probably caused by paleogeographic and physiographic changes in the Mauritania-Senegal basin. These changes could be triggered by the decrease and subsequent cessation of the flexure of the western edge of the basin, resulting in the burial of its substratum and progressive flattening of the relief (Flicoteaux et al., 1988; Leprêtre et al., 2015). These events are coeval with crustal thinning occurring in the Mauritania-Senegal basin and in Guinea that could be related to the opening of the Equatorial Atlantic Ocean (Förster, 1978; Flicoteaux et al., 1988; Latil-Brun 113 and Lucazeau, 1988). In addition, the Cenomanian−Turonian boundary records the Cretaceous maximum flooding which led to the expansion of shelf seas (Schlanger and Jenkyns, 1976) towards the Mauritanides and the Reguibat Shield. Associated with a more flattened topography, this marine transgression may have flooded some of the Paleozoic and Hercynian proximal source areas (e.g., the Bowe Basin, Meseta, Guiraud et al., 2005) (Figure 2.8B.). This flooding may be responsible for paleogeographic changes in the MauritaniaSenegal basin and may have induced a reorganization of the drainage pathways. Indeed, Barnett-Moore et al., (2017) point out that the combined action of dynamic changes in topography and global sea-level fluctuations led to major incursions of the peripheral regions of Northwest Africa during this period. Thus, the maximum flooding of the Cenomanian−Turonian boundary, associated with changes in the paleo-topography, could also partly explain the decrease in Paleozoic inputs when compared with the Albian−Middle Cenomanian. The Campanian−Maastrichtian is characterized by increasing inputs of Precambrian detritus along the Northwest African Margin (except at DSDP Site 137 and CM1, Figure 2.8C.). This implies an expansion of drainage areas toward the east and the inner units of the WAC. As recorded by low-temperature thermochronology data (Leprêtre et al., 2015), the Late Cretaceous to Early Paleogene uplift of the Saharan region of South Morocco may have caused increasing erosion of Precambrian rocks of the Reguibat Shield, which could explain the lowering of the İ Nd(0) values along the northern margin of the WAC. This uplift has been related to the onset of Africa/Europe convergence (Leprêtre et al., 2015) and could be correlated with the “Santonian Compressional Event” (84 to 80 Ma) (Binks and Fairhead, 1992; Guiraud and Bosworth, 1997). This compressional event is related to the convergence between Europe and Africa (Olivet et al., 1984; Rosenbaum et al., 2002) as well as the change in poles of rotation for the opening of the Atlantic Ocean (Klitgord and Schouten, 1986; Binks and Fairhead, 1992; Guiraud et al., 1992; Guiraud and Bosworth, 1997). Indeed, this compressive event led to the inversion of several sedimentary basins in Africa and the reactivation of some faults related to the Panafrican and Hercynian sutures in the WAC. Such 114 an event may also have impacted the drainage pathways leading to the severe erosion of Precambrian units in the hinterland of the WAC (e.g., the Reguibat Shield, the Leo-Man Shield, and the Taoudeni Basin, Figure 2.8C.). Finally, our results on provenance also have some implications regarding oceanic circulation in the Central Atlantic basin during the Cretaceous. The difference in İ Nd(0) values between the shallow-marine CM1 samples and deep-water DSDP samples suggest that the Central Atlantic Ocean never received clastic inputs from the Senegal-Mauritanides (i.e., the drainage basin that supplied the CM1 samples) during the Cretaceous. This implies that, as early as the Middle Aptian, ocean currents may have existed along the eastern margins of the Central Atlantic Ocean which were capable of preventing the arrival of Senegal-Mauritanides clastic inputs to the Central Atlantic Ocean deep-water domain.