In a recent talk, Peter B. deMenocal was showing results from this 2006 paper in Science. If you're interested in this stuff, I also recommend looking at Peter's work. [More paleo-climate/social papers here.]
Climate-Controlled Holocene Occupation in the Sahara: Motor of Africa’s Evolution
Rudolph Kuper and Stefan Kropelin
Abstract: Radiocarbon data from 150 archaeological excavations in the now hyper-arid Eastern Sahara of Egypt, Sudan, Libya, and Chad reveal close links between climatic variations and prehistoric occupation during the past 12,000 years. Synoptic multiple-indicator views for major time slices demonstrate the transition from initial settlement after the sudden onset of humid conditions at 8500 B.C.E. to the exodus resulting from gradual desiccation since 5300 B.C.E. Southward shifting of the desert margin helped trigger the emergence of pharaonic civilization along the Nile, influenced the spread of pastoralism throughout the continent, and affects sub-Saharan Africa to the present day.
In some of my work with colleagues (notably Jesse), we've been finding evidence that it's very costly for populations to adapt to their climate, even in the long run (see here, here, here, and here). This implies that populations may will endure large welfare losses to the climate without employing adaptation (eg. migrating away). What I find interesting about the figure above is that even after an abrupt drying event, it takes populations ~1000 years to completely abandon a location. Presumably, over the course of that millenium, the abruptly dried climate exacted a substantial welfare toll on the population. I think this general idea has real implications for how we think about human responses to climatic changes. In general, it's assumed that adaptation is relatively cheap so that we adjust quickly to reduce welfare losses to the climate. However, if the costs of adaptation are high, the adjustments will be slow and the welfare losses will be large. (A more detailed discussion is in this paper.)