First of all please don't make me a strawman or use my failure for accusing David and G25 of failing. Because I just start using these tools and my mistakes are not his. Also, I never claimed I have the authority to decide things with my quick runs, I just shared results I created and am open to correction and criticism.
I looked at YGorcs results and I quickly realised what went wrong with Sicilians and why it worked for Italians much better: They need additional North African admixture, while other Italians don't need any significant amount of it. This was what ruined my run. If adding ancient North African to the run, it looks like Sicilians are like other Southern Italians, just with some small percentage of North African ancestry. This was present in Ygorcs run and I can repeat it with both prehistoric and modern samples.
Prehistoric North African reference:
Modern North African reference:
This doesn't mean my model is good, but I think it shows that this additional slight NA is there and that was my mistake before. Its not more Levantine, but more North African admixture which was swallowed by Levantine because I didn't include it in the run and it wasn't necessary for other Italians, which was my main objective before. That was my fault.
As for the historical accounts of Levantine admixture into the Imperial Roman population: Yes, we had little hard data, but the study on ancient Romans delivered it with many clearly Near Eastern derived inhabitants. And while we can't be sure how many of those were permanent residents, we know from the historical records many descriptions of neighbourhoods and the people of Rome changing. But this can be, of course, subjective accounts. What else do we have? We have accounts of real people and biographies and nobody can deny that the percentage of people with "Eastern" ancestry in their pedigree was constantly rising from Early Republican (almost all Latin, Italic and Etruscan only) to later Imperial (many with Greek, Greco-Levantine and outright Near Eastern or North African) ancestry. Even in the senatorial class, among politicians, writers, soldiers, but also common people where we have accounts and of course masses of slaves coming in, like private teachers, farmworkers, merchant helpers, which often stayed as freed men and so on. This is of course hard to evaluate statistically, and I'm not a competent authority to make a definitive claim at all, so please keep that in mind when reading what I write. I doubt anybody can right now, but its impossible to deny this influx as if it didn't happen. Its amount can only be evaluated by even more samples and there is a decrease of clearly Near Eastern derived ancestry in Late Antiquity when the urban settlements were in decay and new people from rural areas and the North repopuled the later Medieval agglomerations.This is a very important moment, because it means whatever was there in the Imperial time, was reduced because of the collapse of urban Roman life throughout the imperium, but particularly in Central Italy. We all know the numbers for Rome at its height and after its collapse. I say a lot of these people moved to the East again, others died or sought refuge in the provinces, but some left descendents in the region.
The other reason seems to me therefore panmixture, in which only a slight shift remained, but no clear Levantine profile any more and the Northern influx pulled everything "up again". But that's just my opinion based on the latest study's results and historical accounts, but I have no troubles changing it according to the data anyone can provide. Sorry if anybody felt bad because of my comments, I had no intention to stir anything up, but just shared my opinion and ideas, which might sometimes just be wrong and need correction.
That's why I like honest debates, we all can improve. Ygorcs corrected my model from before, which was just wrong. That was not G25 data's failure, but my fault.