Nitpick: Barcelona-Madrid-Lisbon is a poor way to sample the Iberian peninsula. Geographical features and historical events divide the territory in bands along a NW-SE axis (roughly Atlantic vs Mediterranean).
Two interesting sample spots that are lost in this NE-SW sample would be the Basque country (a pre-indoeuropean culture) and Galicia-Asturias (remains of celtic culture, colonized by Swabs[1], as opposed to West Goths, after the fall of the Roman Empire, and never conquered by the Muslims).
Another reason why these would be interesting would be that Irish celtic legends say that their people come from the north of the Iberian peninsula.
[1] The Swabs also occupied territories in northern Italy. This might explain why Portugal (originally a Galician county) seems closer to Italy than (West Goth) Madrid and Barcelona.
Canadian cities like Vancouver, Montreal or Toronto would be insane! So many different groups of people marry and their kids marry too creating an amazing mix of people.
We're lucky we developed genetics when we did. The study of human gene migration depends on people staying relatively immobile over time, and population groups remaining genetically separate. In a couple generations, if populations continue to mix (which, in all likelyhood, they will), much of this information may have been lost.
But with what kind of error band around that inference? Who has conducted a validation study on this?