In Table 1, the frequencies of the nine different haplogroups defined in 457 Croatian males divided into five subpopulations are compared. Haplogroup I, defined by M170, is the prevailing Y chromosome group through Croatian mainland and islands with the highest frequency (49% on average) reported in Europe so far. This haplogroup is spread in all European populations with variable frequencies, occuring at >40% in Saami, and at <10% in south (Greeks, Italians) and at approx20% in west Europe (Dutch, French).33,41 Haplogroup I occurs significantly more frequently in three southern islands – Brac , Hvar and Korcula than in the mainland and Krk in the north (chi2=25.62, 1 d.f., P=0.000). When mainland sample is subdivided by geographic areas higher frequency of I is observed in eastern and southern parts of Croatia while being also the most frequent haplogroup in Central/Western Balkan (Bosnia and Herzegovina – our unpublished data). Although the samples from Yugoslavia were not typed for M170, Semino et al.41 data indicate that most if not all of the unclassified derivates of M89 mutation in Europe are expected to have either M170 or M201 mutation. STR analyses of Croatian haplogroup I-M170 lineages defined 110 haplotypes, 32 of which occurred more than once. The most frequent haplotype, typed by eight STR loci, is 16–13–10–28–24–11–11–13 observed in 31 individuals. A phylogenetic reconstruction of haplogroup I haplotypes occurring three or more times is shown in Figure 2. Many of them are shared between populations because of either parallel mutation and deep time depth or a common origin from a relatively large number of founder haplotypes. Most Croatian haplogroup I chromosomes (126/221) share a similar STR haplotype from here on called the Dinaric Modal Haplotype (DMH: 16–24–11–11–13, defined by DYS19–390–391–392–393, respectively) or its one-step derivatives. Interestingly, DMH was not present among 102 Nordic Jobling's haplogroup 2 chromosomes (corresponds to haplogroup I42) reported by Helgason et al.43 Instead, the North Europeans share another modal haplotype 14–23–10–11–13 which is three steps away from the DMH and together with its one step derivatives covers again more than half of haplogroup 2 chromosomes (54/102) in the region of its spread.
Haplogroups R1a and R1b showed the opposite frequency distribution to I-M170 and were more common in the northwest among the mainland population and in the northern island Krk. The frequency of R1b in Croatian sample (7.9%) is comparable to some other Balkan populations, for example, Macedonian (10%),41 Greek (11%) and Yugoslavian (11%),33 but it is considerably lower than in western European populations where its frequency ranges between 40% in Germans and 81% in Irish.33 Unusual for European populations,41,44 14% of Y chromosomes from Hvar carried haplogroup P* without M173 mutation associated with a relatively high number of repeats (n=15) in the DYS392 locus. A detailed description of the occurrence of this haplogroup will be reported elsewhere.
In contrast to R1b, its sister clade R1a shows eastern distribution in Europe being as low as 10–12% in neighboring Balkan populations, Greeks and Albanians, lower in Western Europe, in Italians, Germans and the Dutch (4–6%), but as high as 50–60% in eastern and northern Europe, for example, Hungarians, Ukrainians and Poles.41 The average frequency of R1a in Croatian mainland population (34%) is similar to the value of other Slavs, for example, Macedonians (35%), Czechs and Slovaks (27%).41 Figure 2 presents a phylogenetic reconstruction of haplogroup R1a haplotypes occurring three or more times. The most frequent one (16–12–10–27–25–10–11–13) has been observed in all populations except in Hvar. Among all R1a haplotypes samples from the northern island of Krk and mainland Croatia share the highest number of haplotypes, whereas the southern island of Hvar shows the lowest number of shared haplotypes.
SNP-based branching pattern of clades I, G and unresolved derivatives of haplogroup F did not recapitulate in MJ using eight STRs (data not shown). Three haplotypes were shared between haplogroups I and G, and a number of M170 haplotypes were reconstructed as descending from M201 haplotypes, in addition, M89* haplotypes were dispersed over the network lacking apparent phylogenetic structure.
Haplogroups G, E and J have been associated with the contribution of Neolithic farmers to the European gene pool.41,45 Decreasing frequencies of haplogroup G from the Middle-East to Europe have been observed with the highest frequency noticed in Georgia. Its frequency in the investigated Croatian population is low with the exception of the most southern and the most distant investigated island Korc caronula (10%), suggesting that the Neolithic spread may have come by sea rather than by land. Haplogroup J was found at low frequencies (0–4%) over all Croatia except of Krk where 11% of the lineages had an insertion at 12f2 locus. This low frequency of J is similar to Croatian northern neighbors Slovenians,33 but it is sharply in contrast to high frequencies of the cluster, often associated with M172,45 in other Mediterranean populations where >20% frequencies are common.33,41,46 Haplogroup E was found in Croatian populations in generally low frequencies as well, ranging from 4% in the south to 7% in Krk. This frequency is even lower than reported for other south European populations (Calabrian 16.2%, Sardinian 10.4%, Greek 24%, Macedonian 15%), but comparable or even higher than its 2–9% frequency in Czechs, Hungarians, Poles, Germans, northern Italians and Ukrainians.41
Top of page
Discussion
The southern eastern Adriatic islands of Brac caron, Hvar and Korc caronula (Figure 1) had the highest frequencies reported in Europe to date (54–66%) of haplogroup I defined by M170. According to Underhill et al45 this mutation originated in Europe before the Last Glacial Maximum (LGM) and might have spread together with the Gravettian culture. Although, the frequency of its ancestral F* lineages is very low in Europe,45 their traces have been found in the investigated Croatian population (1.3%). During the LGM the Balkans,47 probably along with the northern Adriatic plain then connected with the present-day islands, could have been among a few refugia of Europe (eg Franco-Cantabria and Ukraine). High frequencies of mutation M170 noticed in the Croatian population investigated in this study and Bosnian and Herzegovian population (our unpublished data) imply that the Western Balkan Peninsula could have been an LGM reservoir of M170, a starting point of an expansion that spread M170 around the neighboring populations. Postglacial re-expansion from a different refugia might explain the differences of the northern European STR founder types from those characteristic to Croats.
The second most frequent haplogroup in mainland and island populations was R1a (frequency ranging from 9 to 38%). This haplogroup is almost completely absent in western Europe but widely present in eastern Europe and central Asia.33,41,44 The same geographic pattern can be seen in Croatia. A striking north–south difference implies that at least some of the founding ancestral groups of the investigated islands are of different origin. Possibly, continental Croatia was inhabited by Indo-European speaking populations who had migrated from southern Russia 2000 BP48,49,50 and probably had carried this mutation. It is logical that Dinarides are a natural mountain barrier to migrations between the southern Croatia and north-eastern Europe and possibly explaining the lower frequency of R1a lineages in the southern Adriatic islands of Brac caron, Hvar and Korc caronula.
Haplogroups R1b and I are likely to have been present in Europe since the Palaeolithic age.41 The frequency of R1b in Europe shows a major west-to-east decline being most frequent in the Basque. The observed frequency of R1b in Croatian sample (7.4%) harmonizes with its frequency in neighboring populations, although general frequency pattern runs also through the northwest to southeast directional layout of the Croatian populations showing a marked difference between mainland and northern Adriatic island of Krk and three southern Adriatic island groups (Brac caron, Hvar and Korc caronula). This indicates that migrations carrying M173 mutation were more pronounced in the northern part of Croatia than in the southern coastal area.
Worthy of note is the finding of considerable frequency of haplogroup P* (xM173) in the population of the island of Hvar. According to Wells et al44 this lineage displays a maximum in Central Asia while being rare in Europe, Middle East and East Asia. Its presence in Hvar recapitulates our finding of mtDNA haplogroup F on the island of Hvar and in mainland Croatian population that is virtually absent in Europe but, again, common in populations from Central and Eastern Asia.51 There are several possibilities for the occurrence of the ancestral lineage of M173. One is the well-documented alliance of Avars (a Mongol people) and Slavs (Croatians) that followed Avar arrival to the eastern Adriatic in 6th century AD. The other is the expansion of the Ottoman Empire from the 16th to 18th century AD when refugees from the western Balkan frequently immigrated to the islands. Lastly, the ancient Silk Road linking China with western Asia and Europe could be a possible path of P(xM173) lineage, too. Any of these migratory patterns could have introduced this mutation to the investigated population.
The observed frequency of the haplogroups E, G and J in Croatian sample is low, suggesting a minor genetic impact from Middle East. The minor Neolithic input in Croatia from the Near East as seen in frequency distribution of Y chromosome haplogroups is in concordance with the archaeological findings in this region. Chapman and Müller52 perceive the absence of late Mesolithic Q1sites in the area of the spread of early Neolithic sites in Greece and southern Italy; however, late Mesolithic sites do occur in Eastern Adriatic coast (known as Dalmatia) and further up in north and east parts of Croatia. Also, King and Underhill53 noticed in the Croatian region absence of distribution of Neolithic figurines and painted pottery that are associated with early agriculture period. This finding could imply that in this region transition to agriculture occurs later and less because of a replacement than diffusion. Highest frequency of haplogroup G in Korc caronula could be either because of drift or reflect a different source of migration, likely over the sea. A markedly higher frequency of haplogroup J noted at the northern island Krk might be a consequence of the fact that the island was most exposed to the immigrations from the mainland. During the Ottoman invasions to the Balkans population groups from the Adriatic hinterland (Dinarides) and Western Balkan area sought refuge at the eastern Adriatic islands. These populations often referred to as 'Vlachs' or 'Morlachs' included a few families of Romanian origin (sometimes described as descendants of the Roman army veterans as well) who left numerous traces of their presence at Krk (eg Skok54).
The lack of haplogroup N and its Tat C daughter clade in Croatians suggests that proto-Slavic males probably did not carry this lineage in substantial frequency and that Russians and Poles have likely obtained it through a recent admixture with Finno-Ugric people.34,55
In conclusion, the investigated Croatian populations show the presence of Y chromosomal haplogroups specific to Western, Southern and Eastern Europe. Moreover, Croatian Y chromosomal lineages testify to different migrational movements carrying mostly Palaeolithic European ancestry, a minor Neolithic impact from the Near East, as well as a Slavic (Croatian) influence which is today clearly expressed in the Croatian language which belongs to the Southern Slav linguistic group. Haplogroup I, one of the few haplogroups of Palaeolithic European origin, present in Croatians in the highest frequency noticed in Europe so far, could potentially classify this area as a birth place of this mutation as well as a source of its post-LGM spread in Europe. Our data (12.5%) do not support the estimation of more than 82% of Neolithic contribution in Croatian population suggested by Chikhi et al.56 Chikhi et al56 assume homogeneity of Palaeolithic gene pool in Europe by taking only the Basques as representatives of Palaeolithic Europeans thus defining all other lineages by default as Neolithic. This assumption is likely not to hold, at least for haplogroup I, because it is restricted to Europe and almost absent in the Middle-East.
Having in mind a gradual geographic appearance of haplogroups R1a and R1b and domination of I in southern Adriatic islands of Brac caron, Hvar and Korc caronula one could expect that the ancestral founding groups of investigated populations could be of different origins. The genetic legacy of Adriatic islands corresponds to the proximal mainland and not to the neighboring Mediterranean populations, although the archaeological evidence suggests a different pattern. Apparently, observed genetic background of the studied southern islands area might be primarily structured by a population dispersal directed from the mainland towards the islands, whereas a cultural exchange was transferred mainly via Mediterranean maritime routes.
A general conclusion of previous anthropological studies was that the islands more distant from the coast (Hvar and Korc caronula) were the ones exhibiting higher degree of isolation, endogamy and inbreeding.13,57 This study showed that Y chromosome haplotype diversity on Korc caronula island was lower than in other islands (0.9749plusminus0.0059 in comparison with the northern island Krk 0.9882plusminus0.0047, and southern islands of Brac caron 0.9787plusminus0.0099 and Hvar 0.09839plusminus0.0049). This is consistent not only with the findings based on endogamy and inbreeding, but also with the implications of our previous mtDNA studies,16 stressing the influence of drift in small isolated Adriatic island populations.
Finally, clear and meaningful in terms of ethnogenesis pattern and gradients of Y chromosome distribution variants in Croatia, uniparentally inherited genetic loci do not only prove their usefulness in understanding demographic history of human populations, but also indicate the need for their evaluation within the context of isonymic analyses.
:D