I'd bet the labels are where those components are highest today, which has little to do with their origin. Because Ancestry's done such a poor job explaining their bio-geographical ancestry (BGA) analysis, I'll use a third party BGA project as an example.

The Dodecad Ancestry Project (http://dodecad.blogspot.com/) has several different analyses. One of the earliest was a 12 component model, which one of the users mapped the percentages in Europe (http://bergep.clanteam.com/DOD_Maps/dod_map.html). You'll also notice the component labels are broader and more abstract. Using narrower, more specific labels like Ancestry's done is simply wrong and confusing.

And if you had access to your own raw data, you could get the results from any of Dodecad's various analyses, or other third party BGA projects.