Recent modeling (deMenocal et al 2000 Quaternary Science Reviews) of the Sahara's transition from a Serengeti-like grassland to sand dunes indicates that the transition, while forced by overall climate change, happened suddenly, probably as a result of positive feedback from vegetation changes:
As vegetation declined, a critical point occurred around 5500 years ago and the Sahara was born, in possibly only 100-200 years. The top chart shows overall climate change, the two middle charts show models without, and with, vegetation feedback, and the bottom chart shows the paleoecologic record, where "Terrigenuous Flux" is a measure of erosion and sand dune formation.
An analogous rapid vegetation change has been observed in the American Southwest, prompting some to ask if Arizona and New Mexico could be the next Sahara...
(image composite from Santa Rita Experimental Range, University of Arizona)
As landcover transitions from grasslands to sparse shrublands, erosion can increase (Breshears et al 2003 Earth Surface Processes and Landforms). Vegetation cover can also influence rainfall (Kurc et al 2003 Water Resources Research), initiating a positive feedback.
The long-range outlook for the American Southwest is not good: “Model Projections of an Imminent Transition to a More Arid Climate in Southwestern North America” — yes, “imminent” — and reports “a broad consensus among climate models” that a permanent drought, bringing Dust Bowl-type conditions, “will become the new climatology of the American Southwest within a time frame of years to decades.” (Seager et al 2007 Science)
Four representative climate models showing Precipitation minus Evapotranspiration over the entire Southwest. The second model, GFDL, is arguably the worst scenario. However, because the Southwest is so dependent on the Summer monsoon, whose dynamics are not well understood, these models should be taken with a grain of salt. Interestingly, El Nino events in the Southern Pacific, which are marked by increased sea-surface temperatures, often increase the ammount of rainfall in the Southwest, especially during the Winter. Indeed, we have been experiencing a strong El Nino since April which has resulted in a nice wet summer, and hopefully will continue with a wet winter as well. Stong El Nino's also correlate with decreased precipitation in the Northwest, which explains why Washington had a very dry summer this year. It is interesting to note that this relationship has changed over time.
However, the vegetative response to El Nino isn't always simple, either. While summer rains definately benefit native C4 grasses, winter rains tend to benefit invasive C3 shrubs. So, even if the total ammount of precipitation doesn't change, a changed timing or frequency/intensity could continue to drive massive vegetative change, which could in turn continue to influence the climate.