Abstract Several types of stratigraphic successions are found at the top of the glacial to periglacial “Complexe terminal” forming the uppermost Ordovician of the North-African border of Gondwana. Locally, there may be a progressive transition from microconglomeratic clays (diamictites) with dropstones to more normal marine clays accompanied by the almost immediate reappearance of graptolites. Study of the graptolites has revealed a distinctive composition at the specific level, differing from contemporaneous faunal associations in adjacent regions, which consist of such classical species as “Glyptograptus” persculptus, Akidograptus ascensus, Parakidograptus acuminatus, etc. The faunal composition has long posed a chronostratigraphic problem, which has now been largely resolved, the fauna being interpreted as Hirnantian to early Rhuddanian in age. On the other hand, the clear evidence of faunal specificity poses a number of problems in the post-glacial context, namely: the disappearance of graptolites in the mid-Caradoc from these regions poses a problem as to the origin of the new specific fauna. Three hypotheses are considered, none of which is satisfactory. Most available data point to an eastern communication, although it is possible that the rejuvenation of the Mauritanides to the west “erased” all prior evidence, giving a wrong idea as to the communications which existed in the area under consideration; what mechanisms caused and maintained this isolation? The hypothesis of a wide, east-west trending depression produced by the overloading of the frontal ice sheet and its progressive disappearance concomitant with the glacial rebound is being considered; what were the effects of isolation on the morphology of graptolites and their population? The virgella and the virgula in several species are remarkably long and this could be attributed to a reduction in water density. The size of monospecific populations also attests to an adaptation to a restricted regime with sandy deposition; how can one explain why some species like N. pseudovenustus, N. inazzaouae, N. normalis brenansi and Ps. kiliani, occur also in other parts of the world? If these species occur where the classical species are absent, the opposite is even more difficult to explain, leading one to postulate the presence of a selective “filter”; how did this faunal specificity disappear progressively? The extent of the sea with Nd. africanus and Nd. fezzanensis put an end to isolation, although it respected the east-west trend. However, there was an opening to the adjacent regions corresponding to present-day Libya. To conclude, if pelagic faunas are considered to be poor paleogeographic tools, faunal specificity rather than endemism should be regarded as the starting point for further reflection. However, all faunal specificity must be fully documented and the results integrated in a framework that includes all aspects of sedimentology, tectonics and climatology. Seven new species are briefly described: Normalograptus nseirati sp. nov., Normalograptus gelidus sp. nov., Normalograptus arrikini sp. nov., Normalograptus pretilokensis sp. nov., Neodiplograptus inezzani sp. nov., Neodiplograptus incommodus sp. nov. and “Glyptograptus” saharensis sp. nov.