1 Introduction

Les montagnes de l'Anti-Atlas, entre le massif du Kerdous à l'ouest et le massif El Graara à l'est, sont drainées vers le sud par les rivières Seyad, Tamanart, Aguemmamou, Akka, Tata, Myit, Tissint et Alougoum. Toutes ces rivières sont des affluents de la rive droite de l'oued Draa (Fig. 1). Là où ces rivières quittent leurs vallées encaissées dans les monts et plateaux du Précambrien métamorphique et dans les formations calcaires et schisteuses du fini-Protérozoı̈que et du Cambrien qui le recouvrent, elles traversent un paysage de plaines alluviales, de cônes, de glacis, de regs et de cuestas, développé dans l'Ordovicien–Dévonien plissé. Entre le front montagneux et le haut escarpement du Jebel Bani s'observe la première des vastes plaines alluviales, appelées « feijas ». Les rivières traversent le Jb. Bani dans des gorges ou « foums ».

Ces feijas contiennent d'importants dépôts, sur 30–380 km², de limons finement sableux de 5 à 30 m d'épaisseur, qui sont couramment disséqués par les rivières transportant des graviers en topographies de badland et de vastes terrasses à sommet plat. À la bordure des feijas, les limons débordent et enterrent les cônes et éventails alluviaux, ainsi que les glacis des pentes de colline adjacentes. La trace des limons peut être retrouvée dans les montagnes sous forme de terrasses, qui contrastent nettement par leur texture fine, des lits de cailloux des rivières actuelles, et des regs, éboulis ou roches nues de montagne. Les formations limoneuses se poursuivent au travers des foums du Jb. Bani et forment de vastes plaines sédimentaires, telles que celles au sud de Tata, Akka et Foum el Hassan, qui se développent à l'aval de l'oued Draa.

Comme le montre la Fig. 2, la plupart des coupes effectuées dans les formations limoneuses comportent trois membres : un membre basal limoneux, un membre de chenal graveleux inter-limon et un membre sommital limoneux. Les formations limoneuses reposent de manière abrupte sur d'épais conglomérats de cailloux bien arrondis, cimentés en une calcrète massive ; la cimentation gagne sur plusieurs mètres les limons susjacents.

Les surfaces des terrasses limoneuses sont fréquemment recouvertes par de fins niveaux de cailloux et galets arrondis, marquant le retour à des systèmes fluviatiles transportant des graviers, et par des clastes anguleux près des pentes de glacis, marquant une réactivation des processus de pente.

Les formations alluviales sablo-limoneuses et limoneuses sont le résultat de conditions climatiques et géomorphologiques très différentes de celles qui les précèdent ou qui les suivent durant l'Holocène. Elles se sont déposées pendant des périodes prolongées de vastes inondations et de flux de basse énergie, à forte concentration de sédiments en suspension. Les sédiments étaient issus d'épais manteaux d'altération, de colluvions de pente riches en matrice, dans les montagnes, et de chutes de poussières et de lœss, dont les sources sont encore à déterminer. Le membre graveleux inter-limon indique un retour à des débits fluviaux de haute énergie et à de fréquents événements d'érosion de pente.

Peu de travaux, excepté ceux de Dijon [4] et d'Andres [2], ont été publiés sur ces vastes formations limoneuses de l'Anti-Atlas du Sud marocain. Leur stratigraphie n'est pas très différente de celle des sédiments alluviaux soltaniens connus ailleurs au Maroc, qui ont été résumés et datés par Delibrias [3], Rognon [6] et Weisrock [8]. Cependant, Andres [2] fait état d'une seule détermination d'âge ¹⁴C de $12760\pm 270$ ans BP conv., obtenue à partir de nodules de carbonates de calcium prélevés non loin de la base de la formation limoneuse à Foum el Hassan, qui jette le doute sur un âge Soltanien.

2 Déterminations chronologiques

3 Discussion

4 Conclusions

The Anti Atlas mountains, between ‘Massif du Kerdous’ to the west and Massif El Graara 250 km to the east, are drained southwards by the rivers Seyad, Tamanart, Aguemmamou, Akka, Tata, Myit, Tissint and Alougoum. All, save the Seyad, are right bank tributaries to the Oued Draa (Fig. 1).

Where these rivers leave their valleys incised into the mountains and plateaus of the Precambrian metamorphic massifs and their Late Proterozoic and Cambrian limestone and schist cover rocks, they cross a terrain of alluvial plains, fans, glacis, regs and cuestas in the folded Ordovician to Devonian formations to the south. Between the mountain front and the high escarpment of Jebel Bani is the first of the extensive alluvial plains locally called ‘feijas’. The rivers pass through the Jb. Bani in gorges or ‘foums’.

These feijas contain extensive deposits, 30–380 km², of fine sandy silts 5–30 m thick that are currently being dissected by the gravel transporting rivers into ‘badland’ topographies and extensive flat-topped terraces. At the feija margins, the silts onlap and bury the fans, cones and glacis of the adjacent hill slopes. The silts can be traced back into the mountains as terraces that contrast sharply in their fine textures from the cobble beds of the present rivers and the regs, screes, and bare rocks in the mountains. The silt formations continue through the Jb. Bani foums and form vast flat sediment plains such as those south of Tata, Akka and Foum el Hassan that extend down to the O. Draa.

The silt formations vary in thickness. In the Bani feijas, the thickness increases from ca 5 m near the mountain front to 30 m at the Bani foums. Most sections (Fig. 2), show three members – a basal silt member, an inter-silt channel gravel member and a top silt member. The ‘silt’ formations are abruptly underlain by thick conglomerates of well-rounded cobbles cemented into a massive calcrete and cementation extends several metres up into the overlying silts. The silt terrace surfaces are frequently covered by thin (<0.3 m) sheets of well-rounded pebbles and cobbles, marking the return to gravel transporting fluvial systems, and by angular clasts near the glacis slopes, marking a reactivation of slope processes.

The bottom silt member, 3–5 m thick, comprises thin, flat lying and planar bedded clayey silts, silty clays, with rare sandy silt layers. In the Tissint feija, the lower silts contain sodium chloride evaporite beds and marly clays. Colours are varied and range from red, beige to white.

The inter-silt gravel member is generally 1–3 m thick and contains flat bedded to cross bedded silty sands with gravel and pebble lenses, multiple trough cross bedded gravels characteristic of braided channels and planar bedded sandy matrix cobble gravels. Where these channel facies occur near the feija margin hillslopes, the fans, cones and glacis' merge into the ‘inter-silt’ fluvial channel strata. Thus hill slope processes were ‘active’ during the period when gravel bedded rivers flowed through the feijas.

The top silt member comprises flat bedded, silty fine sands and sandy silts which become slightly more sandy towards the top of the succession. Down-feija this upper silt member increases in thickness from several metres to 20 m, bedding becomes more pronounced and silty clay beds appear. Calcium carbonate enriched zones occur in the upper silts and generate indurated benches in the walls of the trough shaped wadis.

These silty and silty sand alluvial formations are the result of climatic and geomorphological conditions greatly differing from those that preceded them and those that followed them during the Holocene. They were deposited during prolonged, wide-flooding, low energy flows with high concentrations of suspended sediments. The sediments were sourced from the erosion of deeply weathered mantles and matrix-rich slope colluvia in the mountains and from dust or loess fallouts whose source areas have yet to be identified. The inter-silt gravel member indicates a return to high-energy river flows and frequent slope erosion events.

Apart from Dijon [4] and Andres [2], little has been published on these extensive formations of the southern Anti Atlas. Their stratigraphy is not unlike that of the alluvial Soltanian sediments elsewhere in Morocco summarised and dated by Delibrias et al. [3], Rognon [6] and Weisrock [8]. However, Andres [2, p. 46] reports a single ¹⁴C determination of $12760\pm 270$ conv. yr. BP from calcium carbonate nodules near the base of the silt formation in Foum el Hassan that casts doubt on a Soltanian age.

In the course of the present investigation, shells of the land snail Rumina decollata have been found at shallow depths below the surface in the Seyad feija east of Tarjijcht. Two samples from one site at N29°03.99, W9°19.17, taken from −1.0 and −1.5 m below the surface, in fine sandy silts, were dated by standard radiometric counting and by AMS respectively (Table 1).

Samples for optical luminescence dating (Table 2) have been taken from a number of sites throughout the feijas and this paper reports on the trial analysis of three samples from the Tissint alluvia at long. N29°54.5, lat. W7°21.7 (Fig. 2). Sample 1 came from the top silts at 0.4 m below the surface, sample 2 from a sand lens within the inter-silt gravel member, and sample 3 from clayey silts in the lower silt member.

For optical dating, quartz grains in the 90–150 μm size fraction were extracted from the samples and ca 3.4 mg aliquots of refined quartz were mounted onto 10 mm diameter aluminium discs. Assessment of long-lived radioisotope concentrations, and corresponding environmental dose rates were measured from the samples via inductively coupled plasma mass spectroscopy (ICP–MS). Moisture within the sediment matrix attenuates a typically small portion of the radiation dose, which would otherwise been absorbed by the sediment. However, as the saturation moisture contents (W) and the average field moisture conditions during the burial periods (F) of the samples are not well known, we have chosen to develop two age models based on generally moist and on generally dry sediment burial conditions. Conversion from radioisotope concentrations to dose rate followed Adamiec and Aitken [1].

All optically stimulated luminescence measurements were conducted using an automated RISO TL/OSL DA-15 reader fitted with a blue (λ=470 nm) diode array ($\mathrm{P}=18\mathrm{mW}{\mathrm{cm}}^{-2}$), and a calibrated ⁹⁰Sr/⁹⁰Y beta radioactive source. The luminescence signal emitted from the samples was measured with a photomultiplier (type 9235QA, series 101482) filtered with two Corning U-340 glass filters. Equivalent dose (D_e) estimates were calculated using the recently developed single aliquot regeneration technique (SAR) [5]. The SAR procedure uniquely allows for the determination of a D_e using a single aliquot, which avoids any need for inter-aliquot normalisation, improves the precision of D_e estimates by incorporating interpolative (versus extrapolative) estimation techniques and includes a series of internal checks, which monitor and correct any changes in the behaviour of the aliquot during the procedure. An average of nine aliquots was used to estimate the mean sample D_e and its associated standard error. The results are summarised in Table 2.

The ¹⁴C dates are stratigraphically coherent. The ages reported may be maximums, given the potential for inclusion of older non-organic carbonates in the shells, although the isotopic ratios indicate that this contamination may be limited. Such considerations, in fact, improve their ‘correlation’ with OSL sample 1.

The optical ages depend partly on the moisture model assumed for the samples. Assuming that sample 1 has been dry since deposition the dry scenario date of 11.3–9.9 ka accords well with the 11.34–11.96 ka cal. ¹⁴C BP date stratigraphically ca 5 m lower. Sample 2, from a free-draining stratum, can be assumed to have been dominantly dry since deposition and it is suggested that the true age lies nearer to the 27.6–31.6 ka date than to the wet model age. Sample 3, from a silty clay bed in the lower silt member may have been dominantly moist until the rivers incised to their present level some time in the Late Holocene. It would seem reasonable, therefore, to accept an age for the bottom silts nearer to the wet model of 41.3–48.7 ka than to the dry model age.

These dates accord very closely with the maximum and minimum published dates for Soltanian I and II sediments elsewhere in Morocco of $>40000$ ¹⁴C yr BP [3,7] and $12680\pm 100$ ¹⁴C yr BP [9]. Furthermore, Delibrias et al. [3] suggest that the upper silts of Soltanian II range are bracketed by $27770\pm 1100$ and $13900\pm 250$ ¹⁴C yr BP (Gif 3621 similarly on Rumina dec.), whilst Weisrock et al. [10] identify, in the O. Tamdroust succession, a hiatus between their lower silts FR1 and middle silts FR2 at ca 30 000 ¹⁴C yr BP. These dates accord with the dates reported here: 29.6±2.0 ka, for the inter-silt gravels and 11.34–11.94 cal. yr BP and 10.6±0.7 ka for the top of the upper silt member.

It is tempting to correlate the lower silt member and the inter-silt channel facies with oxygen isotope stage 3 and the upper silts with the Glacial maximum of oxygen isotope stage 2. However, it is premature to suggest a climato-genetic interpretation on these formations while sedimentological analyses are in progress and before the further batch of optical age determinations is completed.