Introduction. It is believed that an inlandsis covered the northern half of the African Gondwana at the end of the Ordovician. After a review of the stratigraphic framework and the methodology used, an attempt is made to reconstruct the successive stages of the advance of the sea at the end of the Ordovician and in the early Silurian in a region believed to have been close to the pole. Only the Algerian Sahara is taken in consideration (fig. 1). Some suggestions are made on the role of the glacio-eustatism in the « Silurian » transgression of the Algerian Sahara.
Lithostratigraphy. Because of an inadequate biostratigraphy and facies variations in the uppermost Ordovician of Sahara, several lithostratigraphic successions have been proposed. To emphasize the more important features of the glaciation, a generalized lithological column has been prepared [Legrand, 1999] (fig. 2).
The biostratigraphical framework. Graptolites are the classical fossils used to construct the biostratigraphy of the uppermost Ordovician and the Lower Silurian. In the Algerian Sahara, the use of the standard graptolite zones has been handicaped by the fact that graptolite faunas are endemic and the typical, diagnostic species occurring in this interval of geologic time are absent. A new regional biostratigraphy had to be constructed based on the diplograptid graptolites present [Legrand, 1999]. The new zones are essentially distribution zones. Brachiopods and chitinozoans have also been used as additional correlating tools. Based on the new graptolite zonation, and the recorded sedimentary events of the region, such as transgressions, regressions and so on, new regional stages and substages have been defined (fig. 3) allowing us to reconstruct the paleogeographic evolution of the Algerian Sahara.
Problems with the uppermost Ordovician. The uppermost Ordovician, as herein defined, comprises all the strata belonging to the n4 regional subsystem of the Saharan stratigraphy. This subsystem includes all the “glacial” formations disconformably laid down on the Saharan platform before the beginning of the Silurian. Many problems have been encountered with the uppermost Ordovician of the Algerian Sahara and adjacent regions, but only are considered in this paper : (a) epeirogenic movements and erosion ; (b) the nature of the glacial, periglacial, deltaic and fluviatile sediments ; (c) one glaciation, multiple glaciations or only a polyphase glaciation ; (d) the varying importance of unconformities ; (e) the age of the uppermost Ordovician formations, i.e. whether they are uppermost Ashgillian or upper Caradocian-upper Ashgillian and the precise age of the last Hirnantia fauna. In this paper only the hypothesis of one late Ashgillian glaciation is taken into consideration.
Paleogeography of the latest Ordovician (regional subsystem n4). Stage n4a (Cautleyan p.p. and Rawtheyan p.p.). The exact time of the beginning of glaciation is uncertain. Stage n4b (upper Rawtheyan-earliest Hirnantian) (fig. 5). At Djado, shales with the graptolite “Glyptograptus” ojsuensis overlie the glacial strata. Trinucleidae trilobites attest to the withdrawal of the ice followed by a marine transgression in this region. Elsewhere, because of the lack of biostratigraphic data, nothing is definitely known as to what went on in the Algerian Sahara. One may suppose that the ice sheet went on growing in some regions and continental deposits may help in dating this period. Substage n4c1 (early and mid-Hirnantian) (fig. 5). One may suppose that the ice sheet advanced and regressed several times before beginning to melt and the continent began to rebound. However, there are some remarkable regional differences. In the Djado area, the silty-argillaceous sedimentation went on, locally interrupted, perhaps, by the return of glacial sedimentation. Farther north, at Oued In Djerane, the last dropstone shales are replaced by normal, marine graptolitebearing shales, while elsewhere the upper sandstones of the Felar-Felar formation, consisting of more or less continental periglacial facies with “cordons” are deposited. The return to marine, littoral conditions is represented by the Hirnantia– or Plectothyrella-bearing sandstones extending from the Ougarta Mountains to the central Tassili N’ Ajjer. Locally, fluviatile beds mark the end of Ordovician sedimentation. Substage n4c2 (Late Hirnantian) (fig. 6). A new transgression begins with this sub-stage, marking the true beginning of the « Silurian » transgression. It should be noted that it is quite possible that the Hirnantia- and Plectothyrella-bearing sandstone dated as marking the end of substage n4c1 could also indicate the beginning of substage n4c2.
Paleogeography of the early Silurian (regional subsystem g1) Lower Llandovery. Substages g1a1, g1a2–3 (lower Rhuddanian) (fig. 7). The substage g1a1 is characterized by the local P. (?) kiliani Zone, which approximately corresponds to the A. ascensus Zone of the classical zonation of the Silurian. The sub-stage g1a2–3 is characterized by the presence of endemic graptolite species of the genus Neodiplograptus believed to correlate the Pk. acuminatus zone of the British zonation. Strata attributed to these substages are common in the western Tassili N’Ajjer, the Tassili of Tafassas-set and the Tassili Ouan Ahaggar. At the Oued In Djerane, the first substage begins with a return to argillaceous sedimentation in an anoxic environment. The second substage is marked by a local regression and the beginning of a northward transgression. Remnants of this substage are present in the eastern Tassili Ouan Ahaggar. Substage g1a4 (upper Rhuddanian) (figs. 8 & 9). This substage holds the sedimentary record of an important phase of the Silurian marine transgression. In the eastern Tassili N’Ajjer, strata assigned to this substage gradually overlie the Ordovician and extend far northwards becoming thinner by progressive transgression – and not because of erosion – to disappear finally south of Al Awaynat (Serdeles). Strata referred to this substage occur also in the eastern Tassili Ouan Ahaggar as far as Ted-jert and possibly In Guezzam. Remnants of this substage occur also, but as outlayers, near the mole of Amguid, at Ers Oum El Lil, Tassili of Tarit and Adrar Tikkadouine. In the area of Bled el Mass the last meters of the Aïn ech Cheikr sandstones are attributed to this substage. Beds of this age are not known on the northern border of the Eglab with the possible exception of the region of El Rhers to the west of Bou Bernous.
Middle Llandovery [according to Toghill, 1968 - Legrand, 1996]. Stage g1b (in part Aeronien) (figs. 10 & 11). The middle Llandovery marks the return to marine sedimentation along the Algerian-Libyan border after a short regression at the end of the early Llandovery. The sea covered many parts of the Tassili Ouan Ahaggar. In the east, a regression clearly took place. On the other hand, the extension of the sea to the west, beyond In Guezzam, appears to overlap the preceding substage. In the central Tassili N’Ajjer, a transgression probably took place after an emergence at the end of the Ordovician. This transgression seems to be part of the process already observed farther east in the preceding substages during which the sea appears to abandon a domain after having invaded another one. Thus this transgression appears to correspond to the regression on the Algerian-Libyan frontier and in the eastern Tassili Ouan Ahaggar. All this leads us to think that at this time the sea covered much of the Ahaggar. The middle Llandovery reappears in the Bled el Mass (Aïn ech Cheikr) and core data indicate that it is also present in the eastern limit of the Tanezrouft. The transgression on the northern border of the Eglab probably began at this time, with the possible exception of the region of El Rhers to the west of Bou Bernous, where it was of a very short duration.
Upper Llandovery [according to Toghill, 1968 - Legrand, 1996]. Stages g1c, g2a, g2b (in part Aeronian, Telychian). The movements of the sea and the evolution of the sedimentation initiated at the lower and middle Llandovery continue during the upper Llandovery.
Conclusion. Many questions can be asked on the development of an inlandsis centered on Africa at the end of the Ordovician. On the question of the melting of the inlandsis, things are somewhat clearer. Everything, or almost everything, took place before the end of the Ordovician. The « microconglomeratic clays » began probably to settle during the stability phase of the inlandsis. The melting of the inlandsis was accompanied by the accumulation of « microconglomeratic » clays, followed by silty clays. The resulting sea level rise caused a transgression from North to South. This rise was compensated by the accumulation of sediments and the glacial rebound causing the filling of the basins. The movements of the sea did not stop because the filling-up of the available space predicted by the principle of accomodation is constantly called into question by subsidence and epeirogenic movements. The mode and the time of formation of the Hoggarian basin are very much in the realm of hypothesis. Traces attributed to it are many in the east, as are those indicating a communication to the north. The sea invaded first the depressed zones of the southeastern Sahara to finally overflow them much like a wave, following the principle of sedimentary accumulation and the rebounding of the hinterland (the hypothesis of forced transgression of Legrand [1999]). All this demands some epirogenic adjustements. The marine domain expanded during the early Silurian in a northwesterly direction, but it appears to have contracted to the east, which was invaded mostly by marine sands. Thus the Silurian transgression, which is less the result of glacioeustatism than is generally admitted, progressed from the southeast to the northwest or locally from the south to the north (from the Hoggar basin), and not from the north to the south as one may have logically supposed (fig. 12). Naturally the irregular topography left from the older relief may have perturbed its progression to the north and northwest.