Italian Genetics: Rome as a Genetic Melting Pot

Y-Chromosome perspective
The 326 unrelated individuals from 8 different locations of SSI have been assigned to 33 different haplogroups whose frequencies, for both the whole dataset as well as for each of the 8 sampling points, are detailed in Table S2. Y-STR haplotypes for the 119 newly-typed individuals are provided in Table S3. Haplogroups G-P15 (12.3%), E-V13 and J-M410* (both 9.5%), together with R-M269* (7.4%) represent the most frequent lineages found in Sicily and Southern Italy (SSI). These are followed by five R1-sublineages (R-M17, R-L2, R-P312, R-U152, R-U106), whose frequencies range from 5.2% to 3.7%, and by J-M267 which embraces almost the 5% of total variability. All these paternal lineages reportedly originated in Europe or in the Near East, whereas much lower it seems to be the African paternal contribution, mainly represented by haplogroups belonging to HG-E sub-lineages (E-V12, 2.76%; E-V22, 2.15%; E-M81, 1.53%). Contrary to what previously reported in literature [8], no differential distribution of Y-chromosome lineages has been found in our dataset. Fisher exact tests performed on HG frequencies between Southern Italy and Sicily (P-value: 0.4765), as well as between Eastern and West Sicily (P-value: 0.2998), indeed do not reveal any significant differentiation. No significant percentage of variance among groups of populations (FCT) has been detected by regional AMOVAs (Table S4). In the same way, when our Sicilian populations were grouped with those of Di Gaetano et al. 2009 following their East-West subdivision scheme and by using the same HG resolution level, both AMOVA (variation among groups 0.30%, P-value 0.091) and Fst index (P-value 0.094), failed to reveal any significant difference in Y-chromosome HGs composition, thus pointing out a substantial homogeneous pattern of genetic variation within the island.

Moreover, when the distribution of Y-chromosome lineages in the present-day Sicilian and Southern-Italian population has been compared with the one of the surname-based selected subset, no significant differentiation appeared (P-value: 0.9551).


Figure 1. Spatial Principal Component Analysis (sPCA) based on Y-chromosome haplogroups frequencies.

The first two global components, sPC1 (a) and sPC2 (b), are depicted. Positive values are represented by black squares; negative values are represented by white squares; the size of the square is proportional to the absolute value of sPC scores.

Interestingly, the second sPC (Figure 1b), despite being much less representative compared to the first one in terms of both variance and spatial autocorrelation, identifies a subdivision between the two Mediterranean coastlines, which seems to involve the Eastern and Western parts of Sicily. The first group (black squares) is indeed represented by populations from the South-Eastern Mediterranean shore (Levant and North-Africa), including also the most western Sicilian provinces (Trapani and Agrigento) and the Iberian populations. Conversely, the second cluster (white squares) is mainly a North-Eastern Mediterranean centred group, encompassing the Balkans, South-Italy and East-Sicily, together with the other central European populations. When the reliability of the sPCA-identified structures was tested by means of an AMOVA based on haplogroup frequencies, the proportion of genetic variation between groups (FCT) results however two times higher when grouping according to the sPC1 (8.31%, P-value<0.001) than sPC2 (4.31%, P-value = 0.004). The sPCA-suggested pattern of genetic relationships among the different Mediterranean populations, has been confirmed in the classical PCA plots reported in Figure S2a

The two high-structured Mediterranean clusters identified with sPC1, were further tested by means of DAPC analysis. Membership probabilities, represented with a structure-like plot (Figure 2), highlight the intermediate position of the Italian samples between the two Mediterranean clusters. In this context, Sicily and Southern Italy show clearly their stronger affinity with the populations from the South-Eastern Mediterranean side (with the partial exception of Catania - CT).


Figure 2. Discriminant Analysis of Principal Components (DAPC) based on Y-chromosome sPC1-identified structure.

The barplot represents DAPC-based posterior membership probabilities for each of the considered populations to belong at each of the two sPC1-identified groups (white = South-Eastern Mediterranean; black = North-Western Mediterranean). Population codes as in Table S1.

Fisher exact tests were carried out among groups of populations in order to identify significantly over- or under-represented HGs in any of the geographic areas analysed, against a background of all the other Mediterranean populations (Table S6). Haplogroup G-M201 appears significantly over-represented in the SSI genetic pool. Haplogroup R-M269, has been found significantly over-represented in Western-Mediterranean populations (IBE, GER and NCI), and under-represented in the South-Eastern Mediterranean ones (BALK, LEV and NAFR). By contrast, haplogroup J-M304(xM172) is significantly over-represented in the non-European Mediterranean shore (LEV and NAFR), being instead under-represented in European Mediterranean populations. In order to investigate further, we then performed a set of Bonferroni-corrected Chi-square tests by comparing frequencies of single lineages in SSI with those of each reference Mediterranean population group, this time exploiting the highest Y-SNP level of resolution available for each pairwise populations comparison (and considering only those lineages with absolute frequency of at least 10 individuals in SSI). Being aware that migration processes cannot be linked only with single specific haplogroups, it is however known that signals of migration should be more easily detected in more highly differentiated lineages [44]. Different haplogroups have shown significantly higher frequency in specific comparison groups than in SSI: R1b-sublineages in the western European samples (R-U152 for North-Central Italy, P-value<0.001; R-P312 for Iberian Peninsula, P-value<0.001; and R-U106 for German region, P-value<0.001), R-M17 in the Balkan Peninsula and Germany (both P-values<0.05), and J1-M267 in both Levant and North-Africa (both P-values<0.001).


Table 1. Age estimates (in YBP) of STR and HVS variation for the most frequent haplogroups in Sicily and Southern Italy.

As for TMRCA estimates, STR variation within the most frequent haplogroups of SSI suggests that most of them (with the exception of haplogroup G2a-P15: 9339±3302 YBP) date back to relatively recent times (Table 1), in some cases falling into time periods compatible with specific documented historical events occurred in SSI. Despite the fact that these time estimates must be taken with caution, as they might be affected by the choice of both STRs markers and their mutation rates, overall our results agree in suggesting that most of the Y-chromosomal diversity in modern day Southern Italians originated during late Neolithic and Post-Neolithic times (∼2,300 YBP for E-V13; from ∼3,200 to ∼3,700 YBP for J sub-lineages; ∼4,300 YBP for R-M17 and R-P312; and ∼2,000 YBP for R-U106 and R-U152).

Discussion and Conclusions
Sicily and Southern Italy have long represented a natural hub for the expansion of human genes and cultures within the Mediterranean Basin [1]. Accordingly, the genetic pool of current populations inhabiting this area can be interpreted as the result of complex interplays and superimpositions between different prehistoric and more recent demographic events, ranging from the Neolithic expansion and the proto-historic Greek and Phoenician colonisations, up to the post-Roman invasions by Byzantines, Arabs and Normans. The real demographic impacts of these settlements on the population structure remain still largely uncertain based on the study of material culture and the available historical sources, and different hypotheses about the relative contributions of these events to the current gene pool composition have been proposed from a genetic point of view [7]–[9].

As a contribution to the human history of such a key area of the Mediterranean we surveyed, by means of a comprehensive evaluation of both maternal and paternal genetic landscapes, the genetic variability of a wide number of populations settled in a broad transect encompassing Sicily and Southern Italy (Figure S1). Previous reconstructions of the genetic structure of Sicily [7]–[9] focused their attention mainly on two points in the attempt to clarify its genetic history: a) the presence or absence of internal genetic differentiation along an east-west axis, and b) the extent of the genetic relationship with other populations of the Mediterranean Basin.



Population structure and genetic history of Sicily and Southern-Italy
In contrast with previous investigations on the distribution pattern of genetic variation in Sicily [7]–[8], our results point to a substantially homogeneous composition of maternal and paternal genetic pools both within Sicily (East vs. West) as well as between Sicily and Southern Italy (Table S4). The absence of significant differences in the distribution of HG frequencies along the east-west axis of the island, as observed not only among our Sicilian populations, but also when including the samples from Di Gaetano et al. (2009) [8], provides further support to these conclusions. The comparison of the whole SSI dataset with a subset based on founder surnames, moreover suggests that the observed homogeneity in Y-chromosome composition is not the result of recent events (e.g. increased population mobility related to the social and economic changes of the 19th and 20th centuries); on the contrary it has been preserved at least since the initial founding and spreading of surnames in Italy. In addition, and consistently with the complex history of migration pathways and cultural exchanges characterizing the peopling history of the area, high levels of Y-chromosome and mtDNA genetic variability at both SNP and haplotype (STRs or sequence) data, have been observed in all the SSI populations here examined (Table S5).

Altogether, the high levels of within-population variability and the lack of significant genetic sub-structures fit well with the historic role of Sicily and Southern Italy as a major migration crossroad within the Mediterranean Basin. Anyway, differential contributions from the considered Euro-Mediterranean areas were observed. For instance, if the Near East, the Balkans, and – at a lesser extent – North-Western Italy probably had a relevant role in the genetic make-up of SSI, Northern African contributions seem to be almost negligible. As for the Iberian Peninsula, at present its specific genetic contribution cannot be distinguished from that of North-Western Italy, given their observed genetic similarity. These multiple migration events have probably favoured the reduction of genetic differentiation across the region, by increasing the rates of gene flows between different ethnic groups and in some cases mixing up the different genetic strata. Interestingly, the presence of massive migratory phenomena not necessarily yields genetic homogeneity in a given region. For instance, recent studies [46]–[47] showed how ethno-linguistic minorities from Sicily and Southern Italy - such as the Albanian-speaking Arbereshe - may conserve a significant genetic diversification from the rest of the population. In general, such features are more easily observed in isolated populations, thanks to their reduced population size and their cultural distinctiveness, if compared to open populations.

The patterns of genetic variability observed in our SSI sample are in agreement with the general statement that Southern European populations tend to show higher levels of genetic diversity when compared with those located at more northern latitudes [48] by virtue of the several past demographic events that affected their genetic composition over time. Additionally to the postglacial re-expansion and the demic diffusion of agriculture from Near East, also more recent events (e.g. gene flows from North Africa [48]) have been recently advocated as other possible explanations for the increased genetic diversity in the Southern European populations. Among the several historical occupations of Sicily and Southern Italy, the Pre-Roman colonisation by Greeks and Phoenicians as well as the subsequent invasions from North Africa (including the Muslim conquest, that, at least in part, was conducted by Berber forces) have been previously suggested as putative contributors to the gene pool of current Sicilian population (at least from a male perspective [8]). At this respect, the distribution of Y-chromosome haplogroup E-M81 is widely associated in literature with recent gene flows from North-Africa [49]. Besides the low frequency (1.5%) of E-M81 lineages in general observed in our SSI dataset, the typical Maghrebin core haplotype 13-14-30-24-9-11-13 [8] has been found in only two out of the five E-M81 individuals. These results, along with the negligible contribution from North-African populations revealed by the admixture-like plot analysis, suggest only a marginal impact of trans-Mediterranean gene flows on the current SSI genetic pool. Together with the Berber E-M81, the occurrence of the Near-Eastern J1-M267 in Southern-European populations has been linked to population movements from the Near East through North-Africa, and particularly as a marker of the Islamic expansion over Southern-Europe (started approximately in the 8th century AD and lasted for more than 500 years). Fisher exact tests based on HGs frequencies have revealed the presence of haplogroup J1-M267 at significantly higher frequencies in both North-Africa and the Levant than in Sicily and Southern Italy (both P-values<0.001). However, the estimated age for Sicilian and Southern-Italian J1 haplotypes refers to the end of the Bronze Age (3261±1345 YBP), thus suggesting more ancient contributions from the East. Nevertheless, our time estimate does not necessarily coincide with the time of arrival of J1 in SSI; in fact a pre-existing differentiation could potentially backdate the time estimate here obtained.

By the collapse of the Late Bronze Age societies (approximately 3200 YBP), the Mediterranean Basin underwent different waves of invasion, particularly by the Greeks of the Aegean Sea and, to a lower extent, by Levantine (Phoenicians) groups [50]. Both of them established a set of different colonies along the Mediterranean coasts of Southern Europe and North Africa. The Phoenician colony of Carthage (present-day Tunisia), given its geographic proximity to Sicily, may have played an important role in the colonization of this region. Previous Y-chromosome genetic studies on the Phoenician colonization demonstrated that haplogroup J2 in general, and six haplotypes in particular (PCS1+ through PCS6+), may potentially have represented lineages linked with the spread of the Phoenicians (“Phoenician Colonization Signal”) into the Mediterranean [51]. At this respect, it is worth noting the presence of 4 PCS+ haplotypes (namely PCS1+, PCS2+, PCS4+, PCS5+; [51]) in 9 samples of our Sicilian and Southern Italian dataset, particularly belonging to haplogroups J1-M267 (n = 2), J2-M410* (n = 1), J2-M67 (n = 5), and J2-M12 (n = 2). However, sub-lineages of haplogroup J2 have been also associated with the Neolithic colonization of mainland Greece, Crete and Southern Italy [52], and our TMRCA estimates for J2-subhaplogroups (ranging from 3271±1157 YBP to 3767±1332 YBP) cannot exclude an earlier arrival of at least some of the J2 chromosomes in Sicily and Southern-Italy during Neolithic times.



On the other hand, Y-chromosome lineage E-V13 is thought to have originated in southern Balkans [53]–[54] and then to have spread in Sicily at high frequencies with the Greek colonization of the island [8]. The E-V13 core haplotype 13-13-30-24-10-11-13 (DYS19-DYS389I-DYS389II-DYS390-DYS391-DYS392-DYS393), which define the southern Balkan Modal Haplotype and reaches frequencies of ∼12% in continental Greece [52], has been found in 10 out of the 31 E-V13 samples of Sicily and Southern Italy. This result, along with the high frequency of E-V13 lineages generally observed in our dataset (the second most frequent haplogroup after G2a), confirms the presence of gene flows into Sicily from the Balkans as previously observed by Di Gaetano et al. (2009) [8]. Accordingly, our TMRCA estimate for E-V13 (2354±832 YBP) agrees with the results previously reported in literature for the Sicilian population (2380 YBP, [8]). Altogether, these results do not exclude the possible introduction of some of these Y-lineages with migration processes originated in the Balkans and particularly associated with the Greek colonisation of Southern Italy.

Y-chromosome haplogroup G2a-P15 turn out to be of particular interest in the paternal genetic make-up of Sicily and Southern Italy. Its older age estimate (9339±3302 YBP) – if compared to those of other haplogroups – along with its significantly over-represented frequency in SSI, are consistent with the hypothesis recently suggested by Boattini et al. (2013) [12] according to whom this lineage could be a possible candidate for a pre-Neolithic ancestry in Italy. However the CIs of our time estimate cannot exclude alternative hypotheses such as a diffusion of its major sub-clades during Neolithic and Post-Neolithic times, as recently discussed by Rootsi et al. 2012 [55].

Contrarily to Y-chromosome results, age estimates for mtDNA haplogroups suggest that most of the maternal diversity of the current Sicilian and Southern Italian population is composed by lineages present in Europe as early as the LGM (Table 1). The Late Glacial and Postglacial re-occupation of Europe from refugial areas located in the Mediterranean Peninsulas, has played a major role in shaping the gene pool of modern Europeans [56] and some of the differences in genetic diversity of current European populations have been attributed also to this process [48]. Consistently, the geographic distribution and ages of some mtDNA haplogroups, such as V, H1 and H3, have been associated to events of postglacial re-colonisation from Southern European glacial refugia, and particularly from the Franco-Cantabrian area [57]–[60]. Further evidences of post-glacial resettlement from Southern refugia have been recently suggested also for the mtDNA haplogroup H5 (the third most common European H-sublineage after H1 and H3), if considering its higher occurrence in southern European populations (particularly Italy) and its evolutionary age ranging approximately between 11,500 and 16,000 YBP [61].

Together with the Iberian and Balkan peninsulas, also Italy and particularly SSI might have played an important role during the post-glacial re-expansion, as widely attested by several animal and plant species [62]–[68]. As in the case of Iberia and the Balkans, the presence of numerous Epigravettian sites suggests that Italy could have acted as such also for humans [69], despite the fact that strong genetic evidences are still missing (except for mtDNA haplogroup U5b3 [70]).

Haplogroups H1 and H5 appeared to represent the most frequent H-sublineages in SSI, and their age estimates (Table 1) are consistent with post-glacial time periods, as previously observed for both Southern Italy [11] and the entire Peninsula [12]. Nevertheless, a significant (P-value 0.045) over-representation of H1 haplotypes and an older age (17295±5119 YBP) has been obtained for the Iberian population (as represented by the considered reference samples) than in our SSI datatset, thus suggesting, at least for H1,a post-glacial re-expansion presumptively originated in the Franco-Cantabrian area.

Interestingly, mtDNA haplogroup HV confirmed to be the most ancient lineage in Sicily and Southern Italy, predating the LGM (32242±12595 YBP) and thus representing a possible candidate for the Palaeolithic ancestry of Southern Italy, even though possible post-LGM expansions of its major sub-branches should be taken into account as potentially affecting the time estimates here obtained. Further analyses, involving the complete sequencing of mtDNA genomes and the analysis of ancient DNA samples, are therefore needed in order to more deeply address this point and to confirm the relevance of this haplogroup in the first peopling of Sicily by moderns humans, as recently suggested by some Palaeogenetic researches [5].

Sarno S, Boattini A, Carta M, Ferri G, Alù M, Yao DY, et al. (2014) An Ancient Mediterranean Melting Pot: Investigating the Uniparental Genetic Structure and Population History of Sicily and Southern Italy. PLoS ONE 9(4): e96074.