Scientific American has a summary of a new paper in Science which chronicles the oscillating climate of Africa 3 to 1 million years ago. Here's the relevant snip:

Lake sediments in ten Ethiopian, Kenyan and Tanzanian rift basins suggest there were three humid periods at 2.7-2.5 Ma, 1.9-1.7 Ma and 1.1-0.9 Ma before the present superimposed on the longer-term aridification of East Africa. These humid periods correlate with increased aridity in Northwest and Northeast Africa and significant global climate transitions....

The main problem with studies like this is the whole correlation does not equal causation issue...though climate change does usually result in selection and drift working on a population. The fact that great a swath of Africa which is today rainforest was once predominantly scrub with mere islands of verdancy is often held to have resulted in the relative profusion of various species which seem to fill the same niche. As the forests retreated and fragmented the thesis is that the fauna that depended upon these biomes retreated into the refuges and over time were subject to allopatric speciation. When the forests expanded the species comingled, increasing local diversity all throughout the forest zone.

For hominids, the relevance is that the repeated shifts in climate and fragmentation of habitats (whether it be savanna or forest) would likely have resulted in repeated selection events as well as spatial segregation. Looking at this chart which shows the putative hominid speciation events in their timeframe the overlap with the period of climate change alluded to above is clear. Of course, I haven't read the paper, so I don't know how they claim that this was that much different than any other period in the past...though I suppose Stephen Jay Gould would say something about contingencies and smile munificently.