The genetic clusters on that PCA are isolated hunter-gatherer populations (e.g. Hazda). I've never denied there is genetic discontinuity between some local breeding populations; my issue is with the old idea there are genetic macro-groups such as "Caucasoid", "Negroid", "Mongoloid". Terms I stopped using years ago after learning why they are invalid. What modern population genetics has shown is there is a smooth genetic gradient between broad geographical regions - so there are no racial genetic clusters:
"The genetic exchanges occurred in the course of the frequent contacts have resulted in a smooth, continuous variation of many genetic parameters. As a consequence,
zones of sharp genetic change are not the rule, but the exception; most human populations are not surrounded by clear genetic boundaries" (
Barbujani & Colonna, 2011)
As Barbujani clarifies, those few zones of "sharp genetic change" (i.e. genetic discontinuity): "do not surround large clusters of populations, on a continental or nearly continental scale. On the contrary, they occur irregularly, within continents and even within single countries." Its possible to argue multi-generational isolated ethnic groups are races, but these races are not what most people have in mind - so there is a mismatch objection:
"Such genetic isolation might refer to the Amish in America (Appiah 1996, 73) or to Irish Protestants (Zack 2002, 69), but they clearly cannot refer to those groupings of people presently subsumed under American racial census categories. Because the concept “race” can only apply to groups not typically deemed races (Amish, Irish Protestants), and because this concept cannot apply to groups typically deemed races (African Americans, Whites, Asians, Native Americans), a mismatch occurs between the concept and its typical referent. Thus, the concept of race must be eliminated due to its logical incoherence (Mallon 2006, 526, 533)."
http://plato.stanford.edu/entries/race/