Towards the end of the African Humid Period there were some dry peaks due to fluctuations in rainfall and increasing dryness, with serious impacts on human settlements. Demographic modelling suggests a relative human population maximum in the central and western Sahara near the end of the AHP (between 4700–4300 BC), followed by a population collapse coinciding with the end of the AHP. This crisis is coincident with the emergence of the Ashakar-Skhirat ware along the Moroccan north-Atlantic coast, and also with the abrupt decrease of humidity and rainfall leading to sudden changes in vegetation, particularly the Gineo-Congolian taxa, in the western Sahara ca. 4500 BC (Martínez Sánchez et al. 2018).
The appearance of Ashakar-Skhirat ware parallels the expansion of cattle pastoralism in the Sahara Desert, which has been interpreted as an adaptive mechanism against the arid conditions after the end of the AHP. Pastoralism required frequent relocations in search of fresh wild grasses for livestock, resulting in the spread of this ceramic technology throughout central and northern Sahara during the late 6th and 5th millennium BC. Similar types of distribution have been linked to southward and westward dispersal of pastoral groups from the Sahara (Martínez Sánchez et al. 2018).
Megalithic stone monuments with different architectural features and various ceremonial purposes, which possess characteristics of social relevance and temporality as to be considered ‘places to be remembered’, also appear and diffuse at the end of the AHP (6th-5th millennium BC) and spread afterwards (di Lernia 2013). After ca. 4500 BC, herding spread south and east, and pastoralists built megalithic structures at Wadi Khashab in the Red Sea Hills, and established large cemeteries at Kadreo, Kadruka, and R12 in the Nile Valley. Elaborate mortuary traditions continued for more than a thousand years at sites like Jebel Moya, as agricultural lifeways and early states developed.
Before 5500 BC, population curves for the eastern Sahara, the Atlas & Hoggar, and central Sahara follow broadly synchronous variation. After the population decline period ca. 5500–4500 BC, there are divergent responses: the eastern Sahara, extremely arid today, underwent a rapid population decline, with occupation shifted towards the Nile Valley, probably giving rise to the Hamito-Semitic isolation. The finding of hg. R1b1b2a2a-V1589/Y7771 and subclade R1b1b2a2a1-V69 in modern populations of the Arabian Peninsula may indicate its wide distribution in eastern Africa, just before the Semitic and Hamitic expansions.
To the north and west, the Atlas & Hoggar mountain region declined equally rapidly, possibly isolating the pre-Berber-speaking population (Brierley, Manning, and Maslin 2018). An important population replacement happened in north-west Africa between the Early Neolithic and Late Neolithic samples from Kelif el Boroud (ca. 3700 BC), with ancestry shifting (up to 50%) from Iberomaurusian to European Early Neolithic. Reported mtDNA haplogroups include K1, T2, and X2, proper of Anatolian and European Neolithic populations; and Y-DNA T-M184, observed in European Neolithic individuals. This is probably related to contacts between both sides of the Gibraltar strait at this time, before the Bell Beaker expansion (Fregel et al. 2018).
Central Sahara, on the other hand, had a much more gradual decline in population, attesting flexible and adaptive strategies that co-evolved with the drying environment, living in balance with the available pasture (Brierley, Manning, and Maslin 2018). The persistence and expansion of haplogroup R1b1b-V88 in this region dominated by modern Chadic speakers is thus most likely the result of less population pressure and continued expansion of Pre-Chadic peoples, contrasting with the harsh environment, faster life history, and haplogroup replacement experienced to the north and east.
Two Neolithic individuals from Takarkori in the central Green Sahara (radiocarbon-dated to the early 5th and mid–5th millennium BC, respectively) show a basal mtDNA haplogroup N that branched off immediately after the Palaeolithic sample Oase 1, and before all present-day N-derived mtDNAs. This finding could be explained as from a local subclade that branched off just after the differentiation from L3 within Africa, or as a back-migration after its expansion out of Africa (Vai et al. 2019).