Allele frequencies of the eight Alu insertions in the three groups are shown in Table 2. In all the loci, two alleles were revealed with the exception of ApoA1 in Armenians, which is fixed in the presence of an Alu insertion. Two of 24 tests performed for Hardy-Weinberg equilibrium indicated significant departure from equilibrium (ApoA1 in Georgians and Armenians; p <0.05). Thus, these deviations most likely represent random statistical fluctuation.

      The average expected heterozygosity for each group varied from 0.359 to 0.398 (Table 2). The expected average heterozygosity for each marker in all populations was 0.047 for ApoA1 to 0.492 for TPA25.

      The standard test for heterogeneity of allele frequencies between the groups showed substantial heterogeneity. While there are only two alleles in the locus, Gst value is equivalent to Wright’s Gst [15]. The Gst values ranged from 0.001 for the NBC27 locus to 0.023 for NBC182. The average Gst for the whole dataset was 0.010, which is almost one half that for European populations (0.018) [16]. This result is most similar to that obtained for the North Caucasus populations by Kutuev et al. [17], whereas the level of genetic differentiation only in the populations of Dagestan, for the same loci (except B65, NBC123 and NBC51) was 0.045 [18]. The Gst value that was obtained for both North and South Caucasus populations and for different Alu markers (ACE, TPA25, PV92, APO, FXIIIB, D1, A25, B65) was 0.113 [4].

      The level of genetic differentiation for each pair of our three groups has been estimated. The Gst value for Abkhazian and Georgian population was the same as for Georgian and Armenian populations, i.e., 0.007. The differentiation between Abkhazian and Georgian populations was 0.009.

      To assess the genetic relationships of the three groups, a principal component analysis was performed (Figure 1). The first two components together account for 55.4% of the observed variance. The first principal component accounts for 36.3%. It separates the studied Caucasian populations from populations of Central Asia and Siberia.           

      The studied Abkhazian group is closer to the European rather than to Asian populations along the first axis. It is clearly separated from Siberian and Asian populations, but there are no visible correlations on plot with Volga-Ural and Dagestan populations according to their linguistic or geographic attributes. The Abkhazian population from the studied group appears closer to populations of Dagestan, who are placed together except Karanogays, positioned with Kazakhs and Uzbeks. This can be explained by an apparent Asian origin of Karanogays [19]. With the exception of that, Dagestan, Central Asia and Siberian populations are clearly separated from each other. Armenian and Georgian populations cluster together with Tatars, Mordvinians and Finns. Regarding the second principal component, the latter are located close to Uralic speaking groups such as Mari, Udmirts and Komis. This concurs with the idea of Finno-Ugric Turkish interaction during past millennia.