What evolutionary theory seems to say now is that humanity, for all its outward differences, isn't very genetically diverse at all. By most molecular measures we're the fruits of a rather recent radiation out of Africa. Our ancestry was genetically bottlenecked some 100-200K years back. Thus, differences between subgroups in average intelligence might exist but should not be large. There hasn't been time.
I'm not sure that's correct. Leaky's Australopithecus findings date back only about 3 million years. If he's correct about Australopithecus being an evolutionary precursor (or at least representative of pre-human evolution), then the timescale you provided represents a significant portion of the very short time over which human intelligence evolved.
Note also that current theories of evolution seem to require a very small genetic pool, such that mutations would have a greater relative effect on the population. (Large populations would tend to bury any but the very strongest mutations.) Thus, the "genetic bottleneck" of which you speak might well argue for greater variability in "intelligence."
BTW, in the discussion w/ Dr. Stochastic, I suggested that it might be reasonable to move away from "intelligence" as a measure -- it can mean too many different things -- toward something more along the lines of "task oriented" mental abilities, and one could see how the difference between, say, the Gobi Desert and equatorial Africa might well select for different mental abilities. Thus, one may well explain the preponderance of Asians in college Electrical Engineering Classes, or the preponderance of West Africans in the NBA, at least partly in terms of divergent evolutionary paths.