Dienekes points me to a new paper, Searching for the origin of Gagauzes: Inferences from Y-chromosome analysis:
The Gagauzes are a small Turkish-speaking ethnic group living mostly in southern Moldova and northeastern Bulgaria. The origin of the Gagauzes is obscure. They may be descendants of the Turkic nomadic tribes from the Eurasian steppes, as suggested by the "Steppe" hypothesis, or have a complex Anatolian-steppe origin, as postulated by the "Seljuk" or "Anatolian" hypothesis. To distinguish these hypotheses, a sample of 89 Y-chromosomes representing two Gagauz populations from the Republic of Moldova was analyzed for 28 binary and seven STR polymorphisms. In the gene pool of the Gagauzes a total of 15 Y-haplogroups were identified...The present Gagauz populations were compared with other Balkan, Anatolian, and Central Asian populations by means of genetic distances, nonmetric multidimentional scaling and analyses of molecular variance. The analyses showed that Gagauzes belong to the Balkan populations, suggesting that the Gagauz language represents a case of language replacement in southeastern Europe. Interestingly, the detailed study of microsatellite haplotypes revealed some sharing between the Gagauz and Turkish lineages, providing some support of the hypothesis of the "Seljuk origin" of the Gagauzes. The faster evolving microsatellite loci showed that the two Gagauz samples investigated do not represent a homogeneous group. This finding matches the cultural and linguistic heterogeneity of the Gagauzes well, suggesting a crucial role of social factors in shaping the Gagauz Y-chromosome pool and possibly also of effects of genetic drift.
The Turkic peoples have their origins in the Trans-Siberian regions. Until the rise of Genghis Khan the western half of what is today Mongolia was dominated by Turks. Though the Turks of Turkey itself might resemble Europeans both physically and genetically, Turkic populations in Xinjiang and Central Asia exhibit a much stronger exterior affinity to East Asians. The Uyghurs are in fact nearly prefect hybrids insofar as their genetic background is equally balanced between alleles which have presumed origins in western Eurasia and eastern Eurasia. Various genetic and historical data suggest that the original nature of the Turkic speaking peoples in Central Asia was similar to that of eastern, not western, Eurasians. But as the Turkic peoples, united by a common language* and historical memory, expanded from their Trans-Siberian redoubt between the 5th to 15th centuries west and south, the original genetic signal dissipated along the demic wave of advance. When I tried to do some digging I found that very generous estimates suggest that on the order of 10% of the ancestry of Anatolian Turks might be attributed to lineages which roots in greater Mongolia. And yet considering the vast distances and time scales is this really that small of a proportion? Among the Magyars, who were a Finno-Ugric people who seem to have picked up a nomadic lifestyle from Turkic peoples who moved the Volga region, it is very difficult to find a non-European genetic signal. Though the Ugric peoples were likely somewhat closer in genetic distance to Europeans than Turks, during the medieval period several groups of Turks settled in this region at the invitation of the Hungarian monarchy, fleeing from the the wave of the expanding Mongol Hordes. The Stanford geneticist Marcus Feldman recently suggested that common finding of a strong correlation between genetic and linguistic phylogenetic trees is due to the fact that people tend to marry those who speak the same language. This is not a particularly surprising contention. Consider a pre-modern circumstance when choice of marriage partners was relatively constrained along many axes of freedom; spousal choice was not just a matter of personal preference, but was contingent upon the small sample space as well as the input (or determination of!) of one's extended lineage. In normal times gene flow occurs over small distances as men and women move from deme-to-deme. If two populations are on opposite sides of a linguistic-cultural frontier, it stands to reason that there would be a strong spatial bias in terms of exchange of marriage partners; in plain English villagers would be more likely to marry into nearby groups which speak the same language as opposed to those which don't. Though gene flow would certainly occur along these boundaries, the rate of change of allele frequency is liable to be somewhat greater on these frontiers because of somewhat reduced gene flow. With the rise of transportation via horse within the last 5,000 years, as well as trans-regional empires, deviations from conventional constant gene flow dynamics between adjacent demes have likely become more frequent. In other words, if you had a model of gene flow with a few endogenous parameters, I believe it would explain the extant variation better before the rise of horse nomadism than after. The expansion of the Turks, the rise of the Mongols, and likely the bursting forth of the Indo-Europeans, are exogenous shocks to any conventional model based on short-distance constant exchange. These shocks have become more and more frequent as time passes, and they've scrambled the correlation between genetic and linguistic families. Why this would be is rather obvious from a historical and anthropological viewpoint. The Mongol general Subutai led a series of "raids" across Eastern Europe. Though historically most humans might have lived and died with 10 miles of where they were born, men like Subutai ranged across thousands of miles. This was not atypical for tens of thousands of Mongol men during the 13th century. These men left descendants. Though language is a barrier to the marriage of two peasants who have to cooperate to produce on or just above the Malthusian boundary, it is less of a concern for a nomadic warlord who maintains a polygynous lifestyle. Not only are the economic considerations different, but while a peasant farmer must have at a minimum of symmetrical consent, nomadic warriors can acquire mates at the point of a sword. Acquisition of women by force is an important option to consider because across the full expanse of Turkdom the people who were self-styled Turks** formed a military elite. Even today in places like Iran the military has a disproportionate number of Turks. Take home: The magnitude of the correlation between genetic & linguistic phylogenetic trees should be proportional to the number of sedentary peasants historically within the family. Or, more conveniently, the magnitude is inversely proportional to the number of warrior nomads within the group over time. * In Sons of the Conquerors Hugh Pope claims that the Turkic dialect of Xinjiang is intelligible with the Turkic dialect of Turkey, roughly the eastern and western limits of Turkdom. ** Self-styled because the expansion of nomadic warrior generally involves a great deal of assimilation of local elites along the way. There is a fair amount of this attested in the case of the Ottoman Turks where the ghazis often had attested Greek and Armenian backgrounds.