@Thorbjorn,
One of the gurus of population genetics put it this way to paraphrase him roughly: over the last tens of thousands of years, three "tribes" of people moved into the "European" part of Eurasia. First to arrive were hunter gatherers, WHG, perhaps from the Near East originally, although some may have arrived from due east through eastern Europe, maybe some directly from Anatolia, perhaps a few through Gibraltar. There are differences of opinion which hopefully more ancient dna will resolve.
Around 9000 years ago, some hunter-gatherers, having developed farming, moved far and wide in a rather star burst fashion, including into Europe. The ones in Anatolia are usually labeled ENF. The ones in Europe are EEF. There were probably always more of them in southern Europe, mostly because that's where they first arrived, but also because their "agricultural" package of plants and animals was more suited to that climate and those soil conditions. There was some intermarriage between the two groups, although the original "farmer" component predominated, and it took thousands of years for the admixture to take place.
Another group of hunter-gatherers, from Northern Eurasia this time, moved into parts of far eastern and far northeastern "Europe" (although that's a later quasi-political term) around roughly 10-12,000 years ago. They're called the ANE, and probably, according to some people, they admixed with the WHG (the first hunter-gatherers) to form the EHG or eastern hunter-gatherers, although there are differences of opinion about this as well.
Then, around 5,000 or so years ago, in an area north of the Black Sea called the Pontic Caspian steppe, a group of people, half EHG and half what is called CHG, which is a group, according to some people, related to the early farmers from Anatolia but perhaps with some ANE, developed a culture based on various types of technology learned from others, like agriculture, animal herding, metallurgy, pottery, carts, maybe the wheel, and added to it the horse and a patriarchal culture. Their language is Indo-European. They moved from the steppes into Europe (among other places), admixing with the prior inhabitants along the way.
Everybody in Europe is descended from these groups, but in different proportions. In far northeastern and northwestern Europe perhaps there were pockets of WHG people remaining, which didn't exist in the south. Perhaps when the Indo-European speakers got to Central and Northwestern Europe there weren't all that many mixed ENF/WHG people left because there had been a climate or environmentally caused population crash, perhaps because the newcomers carried plague, perhaps because they killed a lot of the males there. In southern Europe,perhaps the population density was higher, so more of the prior inhabitants remained.
The differences are probably also due to subsequent migrations in Europe which scrambled things up again, particularly just before and after the Roman Era; Celts moved into Northern Italy, Germanic tribes went west and south after the fall of Rome, Germanic tribes also moved into Britain as Angles, Saxons and Jutes. Vikings did the same, and then there were the Slavic migrations west and into southeastern Europe, and the Moorish invasions in Spain, Sicily and southern Italy. Each of these groups had their own particular mix of the ancient "tribes".
The basic picture is one of stasis interrupted by punctuated burst of large folk migrations. From about 1000 AD to the late 1800s in Italy, there's been stasis in most cases.
If you want an academic paper discussing the ancient migrations, Haak et al is a good place to start. It will tell you in general terms how much of each group is in each European "national" group.
http://biorxiv.org/content/early/2015/02/10/013433
If you're using the "calculators" created by various hobbyists and put on gedmatch, you have to be aware that they're not showing you your WHG/ANE/ENF percentages. Neither are they giving you how much "Italian" you have, the way 23andme is attempting to do, except as an approximation in the Oracle function. The calculators are looking at "components". Once again, it's a sort of "cluster". "Northern European" just means the genetic signature that is most common in northern Europe. That has ENF genes in it, and WHG, and EHG etc. Likewise, "Southern European" has ENF, and WHG, and EHG. Or substitute EEF, WHG, and ANE. It's only the proportions that are different.
What those calculator results can't tell you is WHEN those different elements arrived in southern Italy or with whom, which is what I think you want to know. Yes? You want to know if that "Northern European" arrived in the historical era, i.e. in the last 1000 years, with Normans, perhaps, or Angevins etc.? I don't think those calculators can tell you that. It might be from the Italici, or other Indo-European migrants into Italy for all we know. The only way I think we could tell in general terms is if we had an ancient sample from, say, the Roman or post Roman era and then one from, say, 1600 or so from your area.
What the calculators are good for, in my opinion, is so that you can compare yourself with other people in your area, because it's indeed true that people in certain regions of Italy get similar scores for each of these components. If your scores are in the same ballpark, then you are indeed "typical" for your area of Italy. It got to the point, after seeing dozens of those scores, that I could tell someone's ethnicity by just scanning them, I didn't even have to look at the Oracle results.
You might want to take a look at this graph for the Dodecad calculators. If your scores on that calculator approximate those scores for southern Italian/Sicilian, then you're indeed "typical" for your area.
http://4.bp.blogspot.com/_UOHFTxL-bOA/TOYpHMIrsVI/AAAAAAAAAOE/on63ho681uI/s1600/ADMIXTURE10.jpeg
It would be best, however, to compare against other people from your area. Then you'd know if you're an "outlier" for some reason.
One other way of telling if it's accurate for you is, however, to run the Oracle function on these things. The lower the number, the better the "fit". Now, Italians have a lot more variation than people in northern Europe, so you're not going to get the fits way below 1 that some of them get. However, there is a difference in terms of calculators. The more representative the samples, the better the results. For me, the MDLP 23 is by far the best at pinpointing my regional ethnicity in Italy, and that's because it uses lots of academic samples for northern Italy. For example, one of the samples is from the border area between Piemonte, Liguria and Emilia, and, predictably enough, that's a very good match for me. In other words, I'm pretty "typical" for my area.
If you already knew a lot of this I apology for droning on, but perhaps it will help some newbies.