Origins of Greek Civilization: Minoans and Mycenaeans

The complex genetic structure of Cyprus
Figure 1 presents the phylogenetic relationships and frequencies of the Y-chromosome lineages detected in the six districts of Cyprus. Like other populations in Anatolia and Lebanon, Cyprus exhibits a high level of haplogroup J2-M172 related diversity. J2a-M410 is the dominant Y-chromosome lineage, constituting 26.0 % of the overall Cypriot samples. J2b-M12/M102 splits into mainly J2b-M205 (5.9 %), frequent in Southern Levant (Additional file 5: Figure S2), and J2b-M241 (0.6 %), most frequent in Greece and the Balkans [20, 35]. Overall, the E-M35 haplogroup totals to 23.1 % and contains various E-M78 sub-haplogroups including E-V13 (7.3 %) that is common in Greece [10, 18, 35] and E-V22 (3.5 %), that is frequent in Egypt [10] and Sudan [49]. Another E-M35 related haplogroup, E-M34, previously reported in Asia Minor [31], Southern Levant [50, 51], and the Balkans [35] also was observed in Cyprus (10.3 %).

Fig. 1



Topological relationship of Y-chromosome binary markers and their observed haplogroup frequencies (absolute and relative) in the six districts of Cyprus. Nomenclature used in that recommended by [89]. Common names of markers are shown along the branches whose lengths are uninformative with respect to time. The asterisk refers to the unresolved status of the paragroup beyond the specific polymorphism. Six markers shown in italic font were not genotyped but provide context. The following 18 binary markers (with their haplogroup affiliation) were also genotyped but displayed no derived alleles: (E) V42, V6, V92, V257, M81, (G) P20, P16.1, P16.2, (I) M26, (J2a) M318, M419, M322, (O) M175, (Q) M25, M3, M378, (R2) M479, and (R1b) U106

Full size image
In Cyprus, haplogroup G2a-U5 (12.9 %) is widely distributed. While only 0.3 % distribute to the U5* paragroup, the PF3147 component (2.5 %) includes lineages like L91 (1.3 %), also seen in Asia Minor and Crete [30] and attributed to reflect the early Neolithic settlement of Sardinia [52]. The G2a-L293 lineage emerges directly from the U5 node and occurs at 2.4 %, a level similar to that observed in Anatolia and the Levant (Additional file 5: Figure S2). Within the L30 defined clade (7.5 %), all lineages with the exception of L497 also occur in Anatolia [30] including the L30* paragroup (0.2 %). Haplogroup G2a-M406 (2.4 %) occurs at similar levels in Asia Minor and also on Crete (1.9 %), one of the earliest (ca. 9000 y BP) known sites outside the Levant colonized by Neolithic peoples [18]. Within the G2a-P303 portion (4.9 %), the as yet unresolved P303* paragroup is seen (3.5 %) with both the Central European and Northern Italian G2a-L497 and the G2a-M527 lineages occurring at 0.3 %. The overall frequency, the haplogroup G2a-P303, especially the U1 branch is highest (20–39 %) in populations of the southern and northwest Caucasus [30]. Lastly, we detected G2a-CTS342 (0.6 %), a lineage that has been reported in Sardinia [52] and as well in ancient DNA from Asia Minor [53].

Unlike samples from the present day interior Levant, such as Palestinians, Bedouins, and Jordanians [17, 54], J1-M267 is less common in Cyprus at 6.5 %. Haplogroup I2a lineages, thought to have arisen from a post-Last Glacial Maximum refuge and now present in Balkans, Sardinia, and Northwestern continental Europe [35, 55], were observed at ca. 3.5 % in our sample. Overall, haplogroup R1 presence was 15.1 %. The total frequency of associated R1a-M449 and R1b-M415 sub-haplogroups was 4.5 % and 10.7 %, respectively. The paragroup R1b-M269*(xL23) lineage is present (2.5 %). Furthermore sub-branches reflecting distinctive European versus Asian divergences similarly occur in both R1a and R1b. Within R1b, the central/west Europe M412 constituent (2.2 %) is offset by the western/central/south Asia Z2105 fraction (5.4 %) that was previously reported as paragroup L23*(xM412) [56]. Similarly in R1a, both the European Z282 component (3.0 %) and the counterpart Asian Z93 clade (1.1 %) occur. Notable is finding that none of the R1a-Z93 Cypriots carried the diagnostic Ashkenazi Levite DYS456 14 repeat YSTR allele [57]. Lastly, traces of geographically remote B-M60, C-M130, L-M20, and Q-M346 lineages were detected, mainly in the Nicosia district.


Fig. 2




Island substructure
Spatial analysis of Y-STR variance among the six Cypriot districts barely showed significant geographical structure at K = 4 with northern, eastern, and southern regions separating (4.68 %, p < 0.05, Table 1). The nearly featureless geographic structure of the haplotype data is reiterated by the non-significant spatial autocorrelation (Additional file 6: Figure S3) as well as a lack of genetic affinity with district affiliation displayed in the MDS plot (Additional file 7: Figure S4). On the contrary, dispersal of Y-chromosome haplogroups (Fig. 3) reached significance at K = 3 to 5 (range of percentage of variation, 0.39–0.66 %, p values <0.01, Table 1).

We then searched for shared STR haplotypes .While the majority of our 574 STR haplotypes are unique we observed 197 perfect matches, most of which are shared within the same district and to a lesser extent between adjacent districts, indicative of recent demographic growth. To assess the degree of possible local patterns of genetic diversity we conducted a SAMOVA using the distributions of 574 hts within a uniform grid of 38 areas across the island. Mainly single areas separated first (Additional file 8: Table S4). However at the K= 3 level (Fig. 3) we detected two clusters that separate from the remaining samples, one of which is composed of 3 coastal grid areas that is characterized by 17.8 % of J2a-M67 derived chromosomes. The other cluster, comprised of two grid areas in the center of the island, entirely lacks any J2a-M67. At higher levels, K=6 and above, significant zones of reduced variance are restricted to single grid areas indicative of recent growth.

To assess the degree of possible local patterns of genetic diversity shaped by recent demographic forces, we conducted a SAMOVA using the distributions of 574 haplotypes within a uniform grid across the island. From K to K + 1, single areas stood out from the rest of Cyprus (Additional file 8: Table S4). Wherever we detected statistically significant zones of reduced variance (red dots in Fig. 2, Additional file 8: Table S4), the amount of within-group variance made them split at K + 1. Small geographical patterns of reduced genetic variation remained in the Southwest and in the East, attributable to recent demography processes associated with genetic drift. Before we tested for any possible signals of correspondences between settlement history and predefined groups of districts, we first confirmed that male lineage distribution among the six Cypriot districts (Table 1) showed an uneven percentage of variation (0.61 %, p value <0.001). The inland/coast partition model did not significantly pull apart the variance (0.09 %, p value >0.05). Neither the Central South West versus East partition (Pottery Neolithic to Early Bronze Age versus Late Bronze Age occupation site model) nor the East versus the rest of Cyprus (one of the Philia submodels) were found significant (0.18 and 0.19, respectively, p > 0.05). However, the best clustering with a district was found for one of the two submodels concerning the arrival of the Philia phase, namely the Pottery Neolithic to Early Bronze versus Late Bronze Age occupation site model, when Kyreneia is considered separately (0.39 %, p value <0.05).

Discussion
Early and subsequent haplogroup dispersals
The strong correspondence between geography and Y-chromosome binary haplogroups is well known [59]. This feature is consistent with a link between the distribution of haplogroups and past human movements. However, the task of deconvoluting prehistoric gene flows from subsequent transformations owing to ensuing migrations, local differentiations, and recent demographic growth (e.g., [60]) overlaid upon previous ones is complicated. For example, more recent migrations may also contain older haplogroups. Despite rigorous geographically targeted sampling and time parallels with cultural traits [20], current attempts to link of modern Y-chromosome patterns to prehistoric events are preliminary and best viewed with prudence. Such interpretations will be reappraised in the future using a combination of approaches, including simulation modeling [61], ancient DNA (e.g., [53, 58, 62]), and assessment of haplogroups that coalesce near the time frames of interest. This latter strategy is plausible due to the development of elaborately branched SNP dense phylogenies with branches proportional to time [63–67]. While, for reasons summarized in [63], ambiguity currently exists regarding an established mutation rate to use for calibration, this uncertainty will narrow as additional pedigree and clan based “whole” single-copy X-degenerate sequences are analyzed [46, 68] and cross-checked by ancient DNA-based rate estimates (e.g., [69]). With these caveats in mind, we proceed to define the putative prehistoric roots of Cypriot male genetic diversity by: (a) identifying lineages representative of non-Greek genetic influences, (b) reporting statistical support for correlations between settlement zones and haplogroup frequencies, (c) taking guidance from preliminary temporal estimates reported in vanguard studies of Y-chromosome phylogenies with meaningful branch lengths (e.g., [63, 64, 66]), and (d) noting coherence of ancient DNA data with pertinent archeological context.

Genetic legacy/substratum of the aceramic Neolithic
The Neolithic transition has diffused a wide array of culture, economic strategies, and social changes spanning the Levant, the Caucasus, and Europe [70–72] including Mediterranean islands that served as both way stations and terminal settlements [3, 14, 72–74]. Previous Y-chromosome studies [30, 31, 34, 37, 75, 76] hypothesized that lineages from haplogroups G, J, E, and R1b-M269 would have accompanied this cultural expansion although high levels of I2 and E-V13 versus low levels of extant G and J in the Balkans raise the possibility that only vestiges of pioneering agriculturalists to southeast Europe remain [35]. In Europe, certain sub-haplogroups of G and specifically E-V13 were detected in ancient DNA, including Linear Band Keramik (LBK) remains from Central Europe (ca. 8000 y BP), Epicardial skeletons from Iberia (7000 y BP), South of France Late Neolithic (5000 y BP), and a Tyrol specimen (5300 y BP) [77–80].

In Cyprus, the G2a-U5 assemblage has an overall frequency of 12.9 %. Notably, PF3147 and related lineages occur mostly in the insular Mediterranean and display frequencies consistent with a relic distribution (e.g., [80]). Haplogroup G-M406 (2.4 %) is widely distributed across the island, but highest near Khirokitia, an aceramic Neolithic site (Table 1) located on the southern coast of Larnaka district. Interestingly, the more deeply rooted sub-haplogroup G2a-L293 also occurs in Anatolia and northern Levant (Additional file 5: Figure S2), consistent with the PPNB crescent including Syrian areas whose maternal genetic legacy is coherent with the maritime movements of early farmers [14]. In addition, recent archeological studies have demonstrated that Epipaleolithic hunter-fisher-foragers colonized the island first (11,000–13,000 y BP) [81, 82], and that these forays are located on the southern coast as well (Aetokremnos, Amathus, Klimonas, and Asprokremnos), on the opposite slope of the Troodos Mountains [6, 83, 84]. Considered together the lithic industry, chronology and locations of these first human settlements match the PPNA tradition from Levant [6]. Cyprus would thus represent one of the first stops of this diffusion, bringing probably some G male lineages to continental and insular Europe [6, 81]. As far as J, E, and R are concerned, refined investigation of their derived lineages could match with further events, characterized by the intensification of commercial exchanges throughout the Levantine Sea and other regions discussed below.



Centripetal gene flows during pottery Neolithic
It has been hypothesized that J2b-M12 may have been associated with the Neolithic immigration of farmers to Greece [18]. Haplogroup J2b-M12 splits into J2b-M205 and J2b-M241. Since J2b-M241 frequency distribution was already well characterized [35], we mapped the frequency distribution of J2b-M205 (Additional file 5: Figure S2).

Both genetic data from the literature and Syria (Chiaroni J, unpublished results) show a frequency peak of J2b-M205 in the Southern Levant in which the frequency decreases northwards with latitude (Pearson’s R 2 = 0.282, p value = 0.011). The J2b-M205 distribution coincides with the Neolithic crop package dissemination found in common between the Levant and Cyprus [5]. Also J2b-M12 Td estimates (Table 3) coupled with the J2b-M205 distribution overlap significantly with the Pottery Neolithic to Early Bronze Age pattern of settlements in Nicosia, Pafos, Limassol, and Kyreneia (chi-square = 11.29, p < .00084). This suggests the possibility that Cyprus experienced a later (Pottery Neolithic) immigration from the Southern Levant.

Regarding J2a-M410, the most common M530 sublineages are J2a-Z387 (4.9 %) linked with the distinctive six repeat allele at the DYS445 short tandem repeat locus proposed [18] to represent a Neolithic expansion from Anatolia to Greece and Italy, a pattern similar to G2a-M406 [30], at 4.9 % and J2a-Page55*(xM67,M319,Z387) at 5.2 %. The J2a-M319 lineage, previously observed in Crete and the Levant [18, 85] is also present in Cyprus at 1.1 %. However, its Y-STR haplotype diversity (Additional file 9: Table S5) is considerably higher from that in Crete (variance, 0.279 versus 0.121).