ABSTRACT
Between the two 100 mm isohyets presently at c. 17°N and c. 31°N, the Sahara desert is hyperarid. Vast areas to the East (Nubia) and the West (Taoudenni basin) receive less than 25 mm annual rainfall, even less than 5 mm in places (Dubief 1963) (Fig. 1) with a potential evaporation of 5–6 m. Except for the mountainous areas, the presence of surface water as well as the development of a vegetation and of a significant animal life are impossible. Due to the absence of wells and fodder for the camels, human settlements are absent and nomadic life is rare.
Throughout this area covering c. 9,500,000 km2, the Holocene humid phase is documented by 1870 radiocarbon ages from proxy markers for a very different, more humid environments: lacustrine or paludal sediments, shells of fresh water molluscs and ostracodes, organic matter, wood or charcoal, pollen, bones of large herbivorous animals and also neolithic sites testifying to the possibility of a sedentary or a semi-sedentary life. The landscape, at least up to the Tropic of Cancer at 23°N was, on the whole, a grassland covered with widely scattered water-bodies and a diversified fauna: fish, crocodiles, large turtles, hippopotamus, elephants, rhinoceros, buffaloes, phacochoerus and even a wooded grassland with giraffes up to the 20th parallel (Gautier 1989, 1991, Hooghiemstraa 1988, Kröpelin 1993, Neumann 1989, Pachur et al. 1990, Petit-Maire et al. 1995, Petit-Maire et al. 1993, Petit-Maire & Riser 1981, Petit-Maire 1986, 1989, Schulz 1991, Van Neer & Uerpmann 1989).
Five hundred sixty dates from fresh water lakes and swamps, 140 dates from vegetal remains, organic matter and bones of herbivorous mammals and 400 dates from archaeological sites cover the period between 3 and 5 ka within the general Holocene pattern (Petit-Maire & Guo 1996).
The number of observations and their distribution throughout the area (Petit-Maire et al. 1993) allow to study the changes in their frequency in time. Hence, given the importance in archaeological research of the c. 4 ka period, as related to the collapse of many socio-economic cultures (Kukla 1994, Sinclair & Bryson 1994), we shall consider the evolution between 5 and 3 ka of the direct climatic parameters (as presence of fresh water reflecting the precipitation/evaporation ratio) compared with the frequency of human settlements in the same zone. Time intervals of 200 years, which cover the chronological errors, were chosen for determining graphically the climatic evolution within this span of time (Figs 2a, b).
