A paper in the December issue of Geology, by Ted Maxwell and colleagues
We believe that the middle and late Pleistocene drainage was influenced by repeated Nile flooding, following on the working hypothesis of Haynes (1985), who suggested a large Pleistocene lake that drained into the Nile from what he termed the Kiseiba-Dungul depression. Using the elevation of the fossil (Middle Paleolithic) Nilotic fish found at Bir Tarfawi (Van Neer, 1993) as a base level, the SRTM data indicate that a paleolake at that level (247 m) would have flooded the entire Kiseiba-Tushka depression (Fig. 3), and is the same elevation at which the Selima paleochannels and other channel remnants to the west blend into the terrain (Fig. 2). We interpret the combination of topographic coincidence and ages of Middle Paleolithic occupations at Selima and Tarfawi as evidence of at least one lake level at that elevation, forming a local base level, reducing the competence of inflowing streams, and inhibiting channel incision below ?247 m. Such a lake would have covered an area of 68,200 km2, and would have extended from the Sudan border (22N) north to the Kharga and Dakhla Oases, until dammed by the limestone plateau at 26N.
They believe that the lake would have been filled by Nile outflow. The paper does not commit to any chronology, except to point out that a few late Acheulean sites are present in the basin near a presumed lower lake level of 190 m, which may represent a relatively stable size, flooded once or multiple times to the higher level of 247 m. Wired has a nice short description of the paper, which includes some dates that are not actually discussed in the paper.
A better understanding of the Nile corridor is of course very important to the issue of human movement into and out of Africa during the Late Pleistocene. More recent Late Pleistocene and Holocene paleolakes are known up and down the Nile valley, from the Fayum to Darfur.
I wonder if a Nile corridor that was ostensibly more habitable may have actually excluded gene flow back into Africa. A denser and more stable human population in this area would have been a relative population source much of the time, sending migrants out into adjacent regions. These regions would have been much less habitable at some times, but displacement of the large Nile valley population may have been impossible. Furthermore, a larger Nile corridor population would have been a reservoir for endemic parasites and diseases that would have posed challenges for migrants into the region.
The problem on an evolutionary timescale is not getting people out of Africa, but explaining the level of population structure between regions that constantly shared an overland and shoreline connection.