There was a limit to the number of subdivisions by country, also based on size and population.
I do understand that, but there are bound to be some distortions in a country like Italy given its sub-structure and demographic history.
Unfortunately, as the map shows, political boundaries don't necessarily correlate with linguistic or "ethnic" boundaries.
The Lunigiana (the far north-west "neck" of Toscana) speaks a variant of Emiliano (although influenced by both Tuscan and Ligurian), and most of it was part of Emilian fiefs until a hundred or so years ago, but it is now politically part of Tuscany.
The so-called "Piedmont" sample used by academics doing pop gen research was taken from the Val Borbera which is now politically part of Piemonte. However, that's also a recent (last hundred years) development. The area was part of Liguria historically, as the town names indicate, and as the map shows they still speak Ligurian "dialects" there. Genetically, I would think they're actually "Ligurian", although this is an area where four provinces meet: Liguria, Piemonte, Lombardia, and Emilia.
This is what happens when geneticists use samples that were drawn to study disease, not ancestry.
Also, for any areas where there was migration from the south, ideally the person should have all great-grandparents from the location in question, otherwise the results won't be accurate. This is especially important if the testee is rather young. In La Spezia, for example, there was migration from the south starting in the late 1800s because of the shipyards. In the Trentino a lot of southern Italians were encouraged to migrate there quite early.
Even Lazio is tricky. In part of it, the people are descendants of northern Italians who were moved there by Mussolini. On the other hand, part of Lazio used to be part of Campania.
It's complicated.
