3.1. Study area and large-scale architecture

The Lower Sandstones have been investigated from the Deli-Borototou mountain (basal contact) to the Wadi Torbo area (upper contact), 40 km to the south and 35 km to the WNW of the town of Fada, respectively (Figure 1 and Table 1). The corresponding submeridian outcrop band dips slightly to the NE and consequently exposes progressively younger strata from south to north. Sandstones emerge from the sand cover in the form of either poorly accessible escarpments, mesa and pinnacles or, alternatively, as extensive horizontal exhumation surfaces (Figure 3). In the Fada area, the total thickness of the Lower Sandstones is c. 200 m. Here we successively describe four allostratigraphic units (Units 1–4) bounded by erosion surfaces from the base to the top of the Lower Sandstones (Figure 4). In addition, an overlying Unit 5 is described, which essentially corresponds to the shaly Bedo Fm. of De Lestang [1965]. Indeed, the Bedo Fm. is still potentially uppermost Ordovician in age, provided the correlation with the fine-grained interval studied by Le Hérissé et al. [2013] and Page et al. [2013] is valid.

Note that the deep, glaciation-related, erosion surface marking the base of Unit 4 may locally cut down into Unit 2, in which case, Unit 3 is absent. Also, Unit 2 is absent (or at least non-individualized) in places, and in these cases, Unit 3 amalgamates with Unit 1 along a poorly identifiable contact. Unit 4 and Unit 5 are locally very thin or absent, where Silurian sandstones of the Gouring Fm. rest almost directly onto, or locally amalgamate with Unit 3 (Figure 4). This architecture probably explains the various stratigraphic assignments proposed in the literature (Figure 2). Additional work is needed to specify at the regional scale the maximum erosion depths for each of the erosional boundaries.

3.2. The basement-cover unconformity

The Lower Sandstones rest in sharp contact with the underlying Pan-African basement comprising granites, gneiss or quartzite ridges [Saharan metacraton of Liégeois et al. 2013]. The geometry of the contact in the Deli area is essentially horizontal at the km-scale (Figure 3A), but in places displays an undulating surface comprising up to 5 m-deep depressions. Here, the sandstone cover shows gentle onlapping relationships. Consisting of granites, the uppermost basement rocks show a well-defined tripartite succession, from base to top; massive, weathered granites, outcropping as m-scale balls; deeply weathered granites, corresponding to a soft, up to 5 m-thick, whitish, clayey horizon; a “stratified” unit, 3–8 m in thickness made up of weathered and more or less rubified granites. At the basement-cover contact, a pebbly lag shows dispersed quartzite cobbles up to 15 cm in diameter.

Interpretation. A thick weathering profile had developed before the onset of sand accumulation, and has been largely preserved. Part of the rubefaction is however suspected to correspond to a later diagenetic imprint. Nevertheless, the nature of the pebbles underlining the basement-cover contact—only quartzite, no granite or gneiss—also substantiates a former, intense, weathering phase that left behind only clasts representative of the more resistant lithologies. Depressions might indicate an intervening erosion phase developing a paleo-drainage network after weathering but prior to deposition of the quartz-rich sediment cover. The weathering profile is reminiscent of the lateritic profile described by Germann et al. [1993] in NW Sudan.

4.1. Unit 1 (lower part of the Lower Sandstones, Figures 3, 5 and 6)

The cliff-forming Unit 1 is nearly 80 m thick. It shows a sandstone-dominated, mainly coarse-grained and cross-stratified succession. At the very base of Unit 1, a specific 3–6 m-thick subunit is individualized. The latter is characterized by alternating gravel conglomerates—which are relatively well sorted (no 1 in Figure 4)—, and coarse-grained sandstones either intensively bioturbated (Skolithos piperock ichnofabric) or made up of 5–20 cm-high, distinctively fining-up, cross-stratified sets showing low-angle laminae. Some of these horizons are radioactive [high U and Th contents; Schneider and Wolff 1992]. Above, coarse to very coarse sandstones with prevailing cross-stratification are dominant. Active burrowing is still noted, with some recurrences of piperock ichnofabric. Cruziana tracks, reported from the base of the succession by Wacrenier [1958] have not been found.

In the middle part of Unit 1, larger-scale and coarser-grained trough cross-strata show less burrowing, with dispersed, rare, or absent(?) Diplocraterion. The troughs are consistently oriented toward the NNE (rose diagrams, Figure 4; satellite imagery, Figure 6).

In the upper part of Unit 1 (Figure 4), sandstones show a cyclic pattern corresponding to the superimposition of 3–6 m-thick, erosion-based, fining-up packages. Large (1–2 m high) trough cross-bedded sets of very coarse to granular sandstones, including quartz gravels and showing in place compound cross-stratification and overturned laminae, evolve upwards in progressively thinner and finer cross-bedded coarse-grained sandstones. The latter show frequent to intense, vertical to oblique burrowing. Skolithos may form dense networks and Diplocraterion is frequent. Cross-stratified sandstones are truncated by a gravel lag or a distinctive, cm- to dm-thick, fining-up granular sandstone bed. Above, but only if preserved beneath the overlying erosion surface marking the base of the next package, is a finer-grained interval. Such intervals, at the origin of discontinuous morphological re-entrants that extend at the 100 m scale, correspond either to a single, 5–30 cm-thick horizon of silty very fine-grained sandstone or to a m-thick succession of siltstones and fine- to coarse-grained sandstones (Figure 5A). The latter can be thin-bedded (<5 cm), cross-stratified or entirely rippled, with in places a distinctive lenticular bedding and undulating erosional contacts. Fine-grained, mm-thick laminae may underline some of the oblique laminae and reactivation surfaces are frequently observed. Such intervals show Harlania ( = Arthrophycus) everywhere, a favorable heterolitic sedimentation has developed.

Interpretation. Unit 1 is dominated by coastal deposits, at the transition from fluvial to nearshore environments. Basal conglomerates and related sandstones are more specifically understood as gravelly shoreface deposits [Dashtgard et al. 2009; Hart and Plint 1995] in an overall transgressive setting; the latter being also indicated by active burrowing and the concentration of radioactive minerals [Dabard et al. 2015].

In the middle part of Unit 1, the fluvial signature prevails but then decreases again upwards in parallel with occurrences of thin but apparent shallow-marine incursions showing a tidal signature (gravel lag indicating ravinement processes, drapes of fine-grained sediments, lenticular bedding, intense bioturbation). Rippled sandstones in these finer-grained intervals are tentatively interpreted as estuarine mouth-bar deposits.

From base to top of Unit 1, a continuously aggrading framework is suggested, nevertheless recording a long-term transgressive/regressive/transgressive evolution. Being described as erosion-based fining upward packages, high-frequency “transgressive” cycles are suggested as building blocks of the upper part of Unit 1. However, the cyclic pattern might alternatively represent the superimposition of “highstand” regressive (large cross-bedded sets) then transgressive (bioturbated cross-bedded sandstones and overlying tide-influenced intervals) deposits. In the latter case, the erosion surface at the base of any individual package would correspond to channel bases rather than to a regional-wide subaerial erosion surface. This may explain the lack of indication for weathering in between individual cycles. An overall marginal-marine depositional environment, comprising fluvial, bay-head delta and estuarine deposits, is inferred on the basis of both primary sedimentary structures and associated bioturbation [Buatois and Mángano 2011]. Whether the high-frequency alternation of fluvial-dominated and nearshore deposits represent autogenic or allocyclic processes—e.g. controlled by eustacy or a varying rate of sediment supply—remains to be elucidated [Zecchin et al. 2007].

4.2. Unit 2 (middle part of the Lower Sandstones, Figures 3B and 5B)

Unit 2 is a relatively thin, bipartite succession resting onto an erosion surface truncating Unit 1. The main part of Unit 2 is a <15-m thick, fining and thinning-upward sequence of cross-bedded sandstones. At the base, sandstones are organized in large-scale trough cross-stratified sets, up to 4 m in thickness (Figure 4/3). A rhythmic pattern can be identified when <5 cm-thick granular laminae repeatedly alternate with up to 40 cm-thick medium- to coarse-grained sandstones including dispersed gravels. Upward, cross-bedded sandstones typically show thinner, m-thick sets. Bioturbation has not been observed. Cross-bedded sandstones are sharply truncated by an essentially flat erosion surface underlined by a gravel lag including some quartz pebbles at the base of a 10–20 cm thick distinctive graded sandstone bed, including poorly defined burrows (Skolithos?). The uppermost part of Unit 2 consists of 1 m-thick, highly ferruginized siltstone, from which primary structures have been erased. Siltstones are deformed by decameter-scale load casts affecting the immediately overlying sandstones of Unit 3.

Interpretation. A fluvial origin could be proposed from the lower cross-bedded sandstones, even if the rhythmic organization might suggest a tidal cyclicity; the advance of coarse-grained, bar-top megaripple trains would, however, also produce such a rhythmic pattern without any tidal signal. The truncation and the related coarse-grained bed is interpreted as a marine ravinement surface overlain by a horizon of “basal transgressive sand”. The overlying siltstones may reflect either offshore deposition or fine-grained sedimentation in the middle, low-energy segment of an estuarine system. However, diagenetic ferruginization precludes a more advanced interpretation.

Depositional environments of Unit 2 do not actually depart from the ones identified in the upper part of Unit 1 (and overlying Unit 3). A deeper basal erosion surface, larger cross-bedded sets and a (residual) 1 m-thick siltstone interval each appear as exaggerated versions of all the features characterizing cyclical packages described in the upper part of Unit 1. A larger-scale pulsation (higher amplitude? higher rates of variation?) of the factors controlling coastal sedimentary dynamics in the Ennedi is thus inferred at this level.

4.3. Unit 3 (upper part of the Lower Sandstones, Figures 3B, 5 and 6)

Like Unit 1, Unit 3 is a cliff-forming, up to 100 m-thick succession of essentially cross-bedded sandstones. Where Unit 2 ends with siltstones, the basal erosional contact of Unit 3 is underlined by a distinctive blue-grey, coarse-grained gravelly sandstone showing large-scale load structures (Figure 4/4). Unit 3 is characterized by coarse- to very coarse-grained cross-stratified sandstone beds with dispersed gravels. Mixed large trough and planar cross-stratification are identified (Figure 5C; Figure 4/5). Neither bioturbation nor fine-grained intervals have been observed in the lower 30 m of Unit 3 but fine-grained intervals including Arthrophycus reappear upwards (only observed in scree deposits).

The upper part of Unit 3 is only exposed in a few preserved interfluve areas beneath the deep erosion surface characterizing the base of the overlying, erosion-based Unit 4 (Figure 4). Medium- to coarse-grained cross-strata are observed, corresponding to 10 m wide troughs; the larger being distinguishable from satellite images and evidencing a NW-oriented dispersal pattern (Figure 6). As in Unit 1, cross-bedded sets are organized in fining- and thinning-up packages. In contrast of the ones of Unit 1, some of the intervening bounding erosion surfaces are deeply (>2 m) incised with pebbly lags and steep margins suggesting large-scale gutters. In addition, cross-bedded sandstones show less bioturbation than those if Unit 1. A dm-thick, distinctively well-sorted, coarse-grained sandstone bed underline the boundary in places between cross-bedded sandstones and the overlying fine-grained interval. In the latter, bioturbation includes small Cruzianids, generally Rusophycus forms, rarer Cruziana forms, which are associated, or not, with Arthrophycus showing well-defined annulations (Figure 5D).

Interpretation. A large-scale transgressive trend is recorded from the base to the top of Unit 3, as evidenced by the greater development of fluvial deposits in the lower part of the unit (including, frequent planar cross-stratification), the overall fining-up trend and ichnofacies changes (no bioturbation, then Arthrophycus, then Arthrophycus and Cruziana). Though it remains to be confirmed, a lesser development of the Skolithos ichnofacies seems to be noted in the whole of Unit 3 when compared to Unit 1. Well-sorted sandstone beds separating fluvial sandstones from marine, bioturbated intervals are interpreted as the signature of marine ravinement processes, possibly by tidal currents.

5.1. Unit 4 (uppermost part of the Lower Sandstones, Wadi Djoua Fm., Figures 3D and 7)

Unit 4 constitutes the infill of several tens of meters deep incisions truncating Unit 3, which are themselves truncated by the erosion surface marking the base of Unit 5 (Figure 4). Unit 4 consists of a variety of massively bedded to cross-stratified sandstones, with various abundance of dispersed pebbles. Sandstones locally grade into conglomerates with dispersed quartzite cobbles up to 20 cm in diameter. Unit 4 is also characterized by discontinuous chaotic outcrops that include cuesta-like slopes at the km-scale, large-scale (>100 m) sigmoidal and channelized sandstones bodies, sand injections in the form of 5–15 cm-thick ferrugineous dikes or sills (Figure 4/7), and three-dimensional folding structures affecting 1–3 m thick sandstone horizons.

No paleovalley network has been properly identified so far, yet WNW-EWE-oriented depressions have been inferred, in agreement with paleocurrent trends (Figure 4). In “interfluves” area, Unit 4 is essentially missing and a thin or absent Unit 5 rests directly on Unit 3. Unit 4 is equivalent to the Wadi Djoua Fm. of De Lestang [1965] (Figure 2). However, the author reported dolerite intrusions in this formation, which we did not observe.

A detailed facies succession in Unit 4 remains to be established and only preliminary observations are provided below. The basal bounding surface and immediately overlying sandstones were generally affected by heavy ferruginization, which hinders the identification of probable onlaps relationships. Conglomeratic facies associations appear preferentially in the lower part of the unit, while the initial deposits that directly seal the basal erosion surface of Unit 4 are, in places, made up of relatively well-sorted coarse-grained sandstones. Cut-and-fill structures 0.5–5 m-deep and 5–50 m in width show alternation of well- to faintly-laminated coarse-grained sandstones and clast- or matrix-supported pebbly conglomerates with dispersed cobbles. Undulating laminae, backsets and pebbly lags are also noted (Figure 7E). Clasts are subangular to subrounded and clast lithologies are both extrabasinal (vein quartz, fine-grained, granular or gravelly quartzites, rare cherts) and intrabasinal (sandstones and siltstones). The latter may constitute intraformational conglomerates, with siltstones intraclasts up to 60 cm-long and a poor representation of extra-basinal clasts. Such conglomeratic deposits progressively grade upward in coarse- to medium-grained, trough cross-stratified sandstones. Together they constitute tens of m-thick, fining- and thinning-upward sandstone sequences. Such successions include in places m-thick intervals of thinly bedded, fine- to medium-grained sandstones with rippled tops and small-scale load casts, and are locally truncated by sandstone beds characterized by abundant rip-up clasts at their top. Though the former have a limited extent (10s to 100s of meters), the latter constitute extensive, horizontally stratified marker beds at the km- to 10  km-scale, which sharply truncate the underlying chaotic and locally deformed sandstones forming the main infill of the incisions.

Interpretation. The deep basal erosion surface, the overall chaotic organization, the conglomerates with numerous extra-basinal clasts, and the stratigraphic position immediately beneath a well-identified glacial surface (see Unit 5) permit Unit 4 to be interpreted as a glaciation-related depositional unit. Similar facies associations including subglacial sedimentary injections and large-scale channels have been documented in the Felar-Felar Fm. of the Djado Basin [north Niger; Denis et al. 2010] and in the Mamuniyat Fm. of the Kufrah Basin [southeastern Libya; Le Heron et al. 2010, 2015]. Large intrabasinal clasts and bedding/lamination patterns that indicate supercritical flow processes noticeably depart from the other fluvial deposits described in units 1–3 and also support the glacigenic interpretation [Girard et al. 2015; Lang et al. 2021]. No striated pebbles have been, however, identified so far. This is usually the case in bedload dominated deposits, where tractive processes rapidly erase potential glacial striae. Cryptic intraformational glacial erosion surfaces might be suspected owing to the occurrence of large-scale folds and massive sandstones, which could represent deformation tills especially where they are intruded by networks of sandstone dikes [Ravier and Buoncristiani 2018]. Channelized and sigmoidal sandstone bodies at the 100 m scale in one hand, and intervening finer-grained intervals including flaser bedding and load casts are reminiscent of end-Ordovician meander belts [Rubino et al. 2003]. Fluvio-glacial dynamics in an ice-marginal context is therefore predominant in Unit 4, though intervening subglacial deformation events are also suspected. Sandstone marker beds are tentatively interpreted as subordinate transgressive horizons, a better identification of which should help to establish a more detailed stratigraphic succession in Unit 4.

5.2. Unit 5 (Bedo Formation, Figures 5 and 7)

Unit 5 was observed close to Wadi Torbo, where it is discontinuously but frequently exposed over a c. 100 km² area. Elsewhere, Unit 5 might not be individualized (for instance, south of Fada; Figure 3A), in which case the identification of the contact between the amalgamated Lower and Upper Sandstones is challenging (Figure 4). Unit 5 is mainly comprised of a fine-grained succession including an intervening sandstone interval. This succession onlaps at the outcrop scale a corrugated surface shaping the ferruginized top of an underlying sandstone (white arrows in Figure 7D).

The corrugated surface is shaped in parallel, north-south (N 04 ± 1°) oriented ridges and troughs (Figure 7B). Wavelengths and amplitudes are in the 250–500 m and 5–30 m ranges, respectively. Some of the corrugations can be followed individually on distances >3 km (Figure 7A). Rectilinear, smaller-scale (<10 m wide), second-order and parallel sandstone ridges show higher elongation ratios (Figure 7D). Beneath the corrugated surface, the fine-grained succession is almost everywhere underlined by a continuous, <1 m-thick, sandstone/conglomerate bed, which therefore appears to be also conformable to an underlying, undulating erosion surface (Figure 4/11). Small-scale extensional step fractures are ubiquitous in the coarse-grained bed. Only one isolated (displaced) stone bearing striae was discovered. In a single outcrop, N–S-oriented poorly defined intraformational grooves have been observed in the underlying Unit 3, one meter below the Unit 5/Unit 3 contact (Figure 4/8). The sandstone/conglomerate bed locally shows a pebbly lag at its upper surface and is systematically capped by trains of current ripples atop a cm-thick veneer. The lag and the ripple trains mold m- to several m-scale gullies or depressions that add local complexity to the otherwise relatively smooth corrugated surface. Satellite images reveal that the corrugation field is disrupted by 25–60 m wide furrows with lateral and frontal berms. Their orientation ranges from N350 to N30, the frontal berm being positioned to the north (Figure 7C).

The fine-grained succession above the corrugation field starts with a thin (5–20 cm) muddy sandstone or siltstone horizon including in places isolated quartz granules. Above, and constituting the main part of Unit 5, are laminated, whitish, clayed siltstones including various proportions of fine-grained, thin-bedded sandstone (mm-scale “laminae” to cm-scale beds). Rhythmical patterns are frequently observed at the 10–20 cm scale. A sandstone boulder has been found isolated in the lower part of Unit 5 (Figure 4/13). A more indurated horizon is in addition usually observed within the soft, fine-grained succession. Its nature and position are variable from place to place, yet usually observed in the lower part of Unit 5 (Figure 4/14). Additional work is needed to confirm it represents a correlatable horizon. In troughs, this horizon may not fundamentally differ from the laminated siltstones (less clayed?) or corresponds to a few m-thick, heretolitic interval characterized by a flaser to lenticular bedding at the mm- to cm-scale. In this case, Kinneyia-like wrinkle marks repeatedly characterize bed tops, showing a consistent N20–N200 orientation (Figure 5E. Figure 4/14). Laterally, where Unit 5 thins out over “paleohighs” (the highest ridges of the corrugation field, i.e., less eroded interfluve areas), a single coarse-grained sandstone bed, 20–40 cm-thick, relatively but distinctively well-sorted, shows cross-stratification and, in some places, megaripples with preserved primary relief (Figure 4/15). An intraformational conglomerate of rip-up siltstone clasts occasionally underlines the lower bed contact. Above, dm-thick, fine-to medium-grained sandstone beds show in places hummocky cross-stratification (HCS) laminations with parting lineations and/or Arthrophycus.

The uppermost 2 m of Unit 5 consists of shales (Figures 5F and 4/16). This is the finest-grained interval of Unit 5, enhancing an overall fining-upward pattern. Unit 5 is sharply truncated by a dm-thick gravelly sandstone bed underlining the trough cross-bedded Silurian–Devonian Upper Sandstones. The latter consist of medium- to coarse-grained sandstones including recurrences of siltstone horizons and bioturbations, mainly tracks along horizontal truncation surfaces. Palaeocurrent trends noticeably deviate to the west or SW (Figure 4). Upper Sandstones grade upwards into coarse- to very coarse-grained deposits, with resumption of palaeocurrent trends toward the north.

Note that due to onlaps onto the corrugation field (Figures 4 and 7), Unit 5 is most often incomplete; Unit 5 being reduced to a few m-thick interval above some of the highest ridges. The lowermost Upper Sandstones have a wider distribution but might also be absent (or undifferentiated?). It has not been established whether their absence is related to non-deposition or to truncation beneath one of the erosion surfaces locally underlined by intrabasinal conglomerates within the lower part of the Upper Sandstones (question marks in Figure 4).

Interpretation. The corrugation field is interpreted as an exhumed a glacial surface. The ridge-and-trough organization and superimposed second-order ridges correspond to streamlined glacial lineations developed over a soft sediment substrate [Deschamps et al. 2013; Le Heron et al. 2022; Moreau et al. 2005]. In this context, the sandstone/conglomerate bed draping the undulating basal truncation likely corresponds to a subglacial fluidized horizon [Denis et al. 2010]. In association with late step fractures and underlying intraformational shear structures such as grooves, this bed highlights a soft-sediment subglacial shear zones [Denis et al. 2010; Girard et al. 2015; Le Heron et al. 2005]. In map view, the Chadian corrugation field resembles assemblages of Quaternary megascale glacial lineations (MSGL), but we note however that: (i) the width of the highly-elongated second-order ridges is noticeably smaller than that of typical MSGL; and (ii) conversely, the height of the first-order corrugations, in excess of 20 m, is largely greater than that of typical MSGL [Spagnolo et al. 2014]. This suggests that the corrugations fit a specific category of glacial lineations, characterized by vertical amplitude greater than 20 m, which can represent forms that are transitional with elongated drumlins [Spagnolo et al. 2014]. It is in addition noted that the core of the corrugations is made up of non- to slightly deformed Unit 3 sandstones. Glacial lineations observed in Wadi Torbo are therefore interpreted as having been formed according to a drumlinization model accounting for the erodent layer hypothesis [Eyles et al. 2016]. Initial drumlins essentially shaped by carving in pre-existing bed material were progressively reworked into ridges under an accelerating ice stream. Delicate features preserved at the former ice-sediment contact (striae, grooves…) were not preserved due to active and generalized post-glacial winnowing evidenced by ripple trains (tidal pumping close to a retreating grounding line?). Only the largest structures were preserved, such as gullies (formed by gravity-driven instabilities?), furrows and associated berms (ploughmarks of northward-drifting icebergs) or depressions (marks of keels of stranded icebergs?). A potential analogue of this palaeo-landsystem is the Barents Shelf [Newton and Huuse 2017], considering the intracontinental setting, moderate water depths, MSGL size and cross-cutting relationship with iceberg ploughmarks.

Granules dispersed in siltstones and muddy sandstones draping the glacial lineations most likely record ice-rafting processes. Interestingly, no outwash facies have been identified. Overlying laminated siltstones, lacking ice-rafted clasts—with, however, the notable exception of the sandstone boulder—, lacking storm-generated facies and bioturbation, but showing rhythmical stratification pattern are tentatively interpreted as the settling of hypopycnal meltwater plumes beneath a perennial sea-ice cover [Rüther et al. 2012]. In this case, and accounting for high ice-marginal accumulation rates, the time span represented by the fine-grained Unit 5 might be relatively short [Lucchi et al. 2015]. The sea-ice cover hypothesis is favored relative to an ice-shelf model because of the shallow water depths, the lack of ploughmarks from large tabular icebergs, and the lack of facies marking the retreating calving-line [Smith et al. 2019]. The more conservative interpretation that would simply consider a restricted-marine depositional environment for the Bedo Fm. cannot be ruled out.

In this context, the significance of the intervening coarser-grained interval observed in the lower part of the Bedo Fm. remains unclear: the break-up of the ice cover and restoration of storm processes; temporary re-advance of a groundling line; or transgressive deposits involving a healing phase [Proust et al. 2018]? Over the adjacent highs, HCS beds and megaripples, which are the signature of combined flow deposits in fine- to medium-grained, and coarse-grained sands, respectively [Leckie 1988], may be the proximal counterpart of such a healing phase. In any case, wave processes suggest that the Ennedi was affected at that time by relatively open-marine conditions.

The presence of Kinneyia-like structures, most likely linked to microbial mats [Aubineau et al. 2018] suggests a tidal influence with cyclical, microbially-mediated, hydrodynamic instabilities [Thomas et al. 2013]. Crest orientation of the wrinkle marks conforms to submeridian depositional slopes, i.e., dipping almost parallel to the axis of former glacially cut overdeepenings. The development of microbial mats is, however, inconclusive regarding the presence, or the termination, of a sea-ice cover. The overall Unit 5 might be the signature of an ice-front retreat during a glacio-isostatically induced forced regression rather than that of a postglacial transgression [Dietrich et al. 2019]. Only the 2 m-thick shale horizon at the very top of the Bedo Fm. suggests an ultimate, post-glacial sea-level rise.

The truncation of the Bedo Fm. by cross-bedded sandstones showing paleocurrents departing from regional trend (Figure 4) is interpreted as an extensive marine ravinement surface overlain by tidal sandstones. An unconformity and a major hiatus are thus inferred at the top of the late- to postglacial shale-dominated sedimentation. Similar relationships are also been identified in the analogous succession of the southeastern Kufrah Basin [Turner 1991], which interestingly differs from more continuous regressive trends interpreted from more basinal areas [Lüning et al. 2010; Gindre et al. 2012].

6.1. The preglacial succession

The study area is located at c. 600 km to the SSE and SSW from the closest outcrop belts fringing the Kufrah Basin (Jebel Eghei and Arkenu area, respectively), and at c. 900 km to the ESE from the Djado (Niger). Considering NNE- to NW-ward oriented paleocurrents, the study area is representative of one of the innermost segments of the Gondwana shelf (Figures 1 and 4). In the absence of biostratigraphic dating, only attempts at regional correlation can be proposed for the Lower Sandstones of the Ennedi. The latter are comparable in terms of depositional environments to the Lower Sandstones around the southeastern Tuareg Shield in northern Niger [also known as the Timesgar Sandstone; Joulia 1959] and SE Algeria [sometimes referred to as the Tin Taharadjeli Member; Beuf et al. 1971]. The stratigraphic assignment of these strata has long been controversial. While a Cambrian to lowermost Ordovician age is often proposed [e.g., Eschard et al. 2010; Greigert and Pougnet 1967; Perron et al. 2018], the same succession is regarded in the Tassili area as Ordovician by Legrand [1974, 2002], Riché and de Charpal [1998] and Fabre and Kazi-Tani [2005]. The Lower Sandstones of the Ennedi are also similar to successions exposed in the Kufra Basin (fluvial sandstones with overturned cross-stratification, tidal bars, superimposed m-scale cycles, among others; see detailed description by Turner [1980]). Originally attributed to a Cambrian Hasawnah Fm. [Bellini and Massa 1980], the upper portion of the Lower Sandstones of the Kufrah Basin is today mostly correlated with the shallow-marine, (Lower to?) Middle Ordovician Hawaz Fm. of the Murzuq Basin [Anfray and Rubino 2003; Gil-Ortiz et al. 2022]. This revision results from the generalized observation in SE Libya of bioturbated sandstones including Skolithos, Arthrophycus and Cruziana Ghienne et al. 2013; Le Heron and Howard 2012; Lüning et al. 1999; Seilacher et al. 2002]. Of particular interest is a change in sedimentation identified at c. 200 m beneath the glacially-related strata in the Jebel As Zalmah area, involving a basal conglomerate, a significant transgressive episode and an overall change from prevailing trough cross-strata below, to planar cross-strata above [Le Heron and Howard 2012]; all of these features being reminiscent of Unit 2 and its relationships with Unit 1 and Unit 3 of the Ennedi (Figure 4). Unit 2 might also correlate with the middle part of the Lower Sandstones of the Jebel Arkenu [Turner 1980] and the lower Araske Fm. of the Tibesti [De Lestang 1965]. It is worth noticing that in the Ennedi, the Lower Sandstones include bioturbated coastal deposits since their very base, lacking the thick fluvial deposits characterizing the lower/lowermost part of the Cambrian–Ordovician succession in the more distal Murzuq Basin [Elicki and Altumi 2019; Ghienne et al. 2013].

As a working hypothesis, and considering the occurrence of Cruziana petraea in the upper part of the Lower Sandstones of the Kufrah Basin [Lüning et al. 1999], an ichnospecies generally characterizing Middle to Upper Ordovician deposits reported from Chad [Seilacher 2007] as well, we suggest the following attributions for the pre-glacial Lower Sandstones units on the basis of the recognition of two large transgressive cycles, which we acknowledge as two large Ordovician retrogradations over the platform.

• Unit 1: mostly Lower Ordovician sandstones, which should correspond to the proximal counterpart of sandstones recurrently showing Cruziana tracks of the Rugosa group more to the west and north. The lowermost shallow-marine segment might be the only representative, if any, of an (upper?) Cambrian sedimentation, though an earliest Ordovician age if favored.
• Unit 3: upper Lower to Middle Ordovician sandstones, which we correlate with the Floian to Darriwilian retrogradation recorded to the west (In Tahouite Fm. of Algeria; Hawaz Fm. of Libya). The overall “transgressive” trend is highlighted by conditions favorable to a Cruziana ichnofacies development at the top of Unit 3.

An alternative correlation might have considered that the two transgressive wedges relate to Early-Middle (Unit 1) and Late (Unit 3) Ordovician cycles; the latter being well known for a major retrogradation over north and west Africa [Gatinskiy et al. 1966; Kichou-Braîk et al. 2006] and the Middle East [MFS O40 of Sharland et al. 2001]. Upper Ordovician strata are, however, most often truncated by the end-Ordovician glacial erosion surfaces, except in the more distal segments of the platform [Ghienne et al. 2010] or in interfluve areas beneath the end-Ordovician glacial erosion surfaces [Bataller et al. 2019; Gil-Ortiz et al. 2022]. Though these alternative attributions cannot be excluded, the change in paleocurrent trends noted from Unit 1 to Unit 3 (Figure 4) is reminiscent of the change in drainage patterns that occurred in an Early to Middle Ordovician time interval [Beuf et al. 1971], favoring the Early to Middle Ordovician age attribution considering the Lower Sandstones of the Ennedi.

The status of the intervening Unit 2 is puzzling. It might be understood as an ultimate short-term “parasequence” concluding the long-term transgressive stacking pattern initiated with Unit 1 and truncated by Unit 3, which represents an overlying transgressive wedge similar to Unit 1. This interpretation would suggest the Unit 3/Unit 2 contact corresponds to a relatively long stratigraphic hiatus, which should correspond to the development of highstand to lowstand conditions in more distal areas. It however does not conform to large-scale loading structures underlining the contact, which indicate still non-lithified shales when the overlying sandstones were deposited. Another possibility would be to suggest a flooding linked to a short-term, but significant regressive-transgressive interval, for instance to a glacio-eustatic event during the Early Ordovician. Going further, we note that the sequence of events linked to Unit 2 echoes the end-Ordovician development involving a glacio-isostatic flexure that allowed seawater to invade such a proximal area (see below). The Lower Ordovician inferred glaciers would have reached or at least been close to the Ennedi area. Although it required confirmation, this interpretation would explain well why the horizon of offshore shales characterizing the Unit 2 has been taken for a Lower Silurian interval by some previous geologists.

Whatever the age within the Ordovician, the proposed correlation framework depicts an overall south- to southeastward-oriented, continent-scale, progressive encroachment of the Lower Palaeozoic strata. It assumes the non-deposition and progressive disappearance of Middle-Upper Cambrian strata to the south and east, though correlative strata are well expressed to the north (northern Libya, north-Saharan basins of Algeria). The relatively monotonous coastal deposits preserved in the proximal reaches of the platform (Niger, SE Libya, Chad) are interpreted essentially as correlatives of transgressive Ordovician fine-grained inner-shelf to condensed successions in the north or NNW. Similar coastal suites are identified around Gondwana. In Jordan, they characterize an older upper Cambrian to Lower Ordovician interval [Disi Fm. Ghienne et al. 2010; Meischner et al. 2020], a time range logically expected in this intermediate, i.e., less proximal segment of the platform. In South Africa, the Peninsula Sandstone [Turner et al. 2011] is assigned to a time interval comparable to that of the Lower Sandstones of the Ennedi. Interestingly, the Peninsula Sandstone characterizes a proximal setting where the succession wedge-out onto basement rocks at the scale of a few tens of km, as should also occurred to the south of the Ennedi. Note that the relatively limited thickness (<200 m) of the Ordovician in the study area might not be solely a signature of the proximality of the Ennedi, as along-strike thickness changes reflecting subsidence/uplift trends controlled by basement heterogeneities [Nkono et al. 2018] could also have played a significant role, as recognized elsewhere throughout the platform [Klitzsch 1970; Perron et al. 2021].

6.2. The end-Ordovician glacial record

The basal erosion surface and conglomerates of Unit 4, as well as the exhumed overlying glacial erosion surface and glaciomarine deposits at the base of Unit 5, allow both units to be interpreted as representatives of the end-Ordovician glacial record. Unit 4 (Wadi Djoua Fm.) appears as the diachronous(?) proximal counterpart of the Mamuniyat Fm. of the Kufrah Basin [Le Heron et al. 2010; Le Heron and Howard 2012]. A “channelised unconformity” and an “intra-Mamuniyat unconformity” in the Jebel Azbah area of the eastern Kufrah Basin might correlated with the basal bounding surfaces of Unit 4 and Unit 5, respectively.

The preservation of the glacial record is restricted to glacially-cut overdeepenings, possibly palaeovalleys, flanked by less eroded interfluve areas, where Silurian–Devonian sandstones directly superimposed the preglacial Ordovician sandstones. The contrast between Unit 4 and Unit 5 is striking, the former dominated by conglomerate-prone outwash deposits, the latter being mainly a fine-grained subaqueous succession. Fining-up outwash deposits of Unit 4 suggest limited interferences with glacio-isostatic patterns during this earlier deglaciation phase, which is surprising in such a proximal setting. This might have been the case, however, if Unit 4 was principally deposited in the forefield of an ice sheet of a lesser extent (a less extensive glacial maximum? or a major stabilization? see discussion in Girard et al. [2015]). Fine-grained sedimentation of Unit 5 is more easily understandable as a consequence of a rapid ice-front retreat and deposition in a relatively and temporarily deep setting, both accounting for a major glacio-isostatic deflection in tens to hundreds of meters that inevitably interfered at that time with glacio-eustacy [Dietrich et al. 2019]. The fine-grained sandstone interval within Unit 5 might relate in one way or another to the late glacial recurrence identified on the eastern side of the Tibesti, which occurred before the deposition of Rhuddanian “warm” shales [Le Heron et al. 2015; Meinhold et al. 2016], which in turn can correlated with the uppermost shale interval of Unit 5.

The temporal framework of the Bedo Fm. (Unit 5) has to be further discussed. The occurrence in the same stratigraphic position of an uppermost Ordovician to lowermost Silurian fauna at c. 500 km to the NNW [Le Hérissé et al. 2013; Page et al. 2013] and c. 650 km to the NNE [Thusu et al. 2013] suggests the Unit 5 might be as older, or even slightly older—i.e., strictly latest Ordovician. The corresponding more distal deposits in the Kufrah Basin show well-developed more proximal HCS sandstone beds and a regressive evolution [Gindre et al. 2012; Le Heron et al. 2015] lacking in the Ennedi record. A latest Ordovician post-glacial, glacio-isostatically driven forced regression in the Ennedi area, with a coeval condensation to the north (or NW) might explain the distinct records. Such a setting predated the overall, long-term progradational trend characterizing the Lower Silurian [Bellini and Massa 1980; Gindre et al. 2012]. Therefore, the Bedo Fm. and Libyan analogues would be older than the Lower Silurian Tanezzuft shales usually defined in the Murzuq Basin [Bellini and Massa 1980; Grignani et al. 1991; Le Hérissé et al. 2013; Lüning et al. 1999], as are similar records preserved in southeastern Algeria and northern Niger [Denis et al. 2010; Legrand 2002].