According to "The Role of Recent Admixture in Forming the Contemporary West Eurasian Genomic Landscape", something like a third of Southern Italian and Tuscan genetic ancestry appears to derive from the Levant in Roman times:
Moorjani et al [S?], who use a method based on allele frequency comparisons, and not haplotypes (ROLLOFF), found evidence for sub-Saharan African admixture in Sardinia 71±28 generations ago, at a proportion of 3%. These are the same Sardinians included in our analysis. In the largest Sardinian (sardi13) cluster in our analysis we infer West African admixture 66 (53-82) generations ago at a proportion of 2%.
S5.2 Continuous low level African admixture in the Mediterranean and Anatolia
We infer West African admixture across broad date ranges, but at low admixture proportions (admixture α < 0.07; Figs. 2 and S3) in several Mediterranean groups, consistent with a long term movement be- tween sub-Saharan Africa and southern Europe [S?,S?]. Specific West African admixture dating to the Arabic conquest of the Mediterranean [S?] is seen in Spanish (spani27: 1042 (740-1201CE)), Southern Italian and Sicilian (sicil30: 1105 (882-1250CE)), and Basque (basqu24: 886 (283-1162CE)) clusters. Earlier African admixture at low admixture proportion is inferred in the Cypriots (cypri12: 427(107- 734CE)), and a Sardinian cluster (sardi13: 36 (458BCE-430CE); α = 0.02). This latter event is con- sistent with the occurrence of A3b2-M13 (0.6%) and E1a-M44 (0.4%) African Y chromosome lineages in Sardinia [S?]. and the dating is more compatible with documented exchanges between the island and Mauretania Cesariensis in Roman times (2 nd century BCE to 2 nd century CE) than later displacements of northern-African males to Sardinia at the time of the Vandals rule (5 th century CE) [S?]. [. . .]
S5.3 A key role for the Levant in the genetic history of the Mediterranean
Early admixture involving source groups most similar to contemporary populations from in and around the Levant (which we define as the World Region containing individuals from Syria, Palestine, Lebanon, Jordan, Saudi, Yemen and Egypt) is seen at high proportions in several clusters from Italy dating to the first half of the first millennium CE, from Southern Italy (itali8: 295CE (72BCE-604CE); α = 0.34), Tuscany (tsi23: 400CE(30BCE-686); α = 0.29), and Sardinia, as well as in a large cluster from Armenia at an early date (armen27: 363BCE(1085BCE-383CE)). [. . .] these events loosely coincide with the formation of the pan-Mediterranean Roman Empire [S?], which may also have allowed increased gene flow from east to west Mediterranean. [. . .] We infer more recent Levant admixture in the French (frenc24: 728(424-1011CE)) and in a complex multiway event in a Spanish cluster (spani9: 668 (286-876CE)). The dates and sources of admixture in these cases are consistent with movements of Middle Eastern and North African individuals during the Islamic Conquest of Spain [S?], and suggest a legacy of this key moment in southern European history in the genomes of French as well as Spanish populations.