Most of the genetic markers used in human phylogeographic studies have been dated to the prehistoric period, and the majority of them are of Upper Paleolithic origin.
Lately, subclades identified within some human lineages on the Y-chromosome have crossed the Neolithic barrier, and in even rarer cases, "signatures" of historical events, such as the dominance of the Mongols, the Manchu, or the Ui Neill.
As a result, most markers are suitable for examining events of human prehistory, and not of historical ethnic groups.
Of course, scientists have tried to apply genetic information to historical processes, e.g., in the case of Jewish origins, but it turns out that the "Jewish gene" or Cohen Modal Haplotype actually turns out to to be much older and not particularly Jewish after all.
Even with old markers, it is still possible to reason about historical events. For example, the theories of white nationalist Arthur Kemp about the widespread prevalence of black African slavery in the classical world have been squarely defeated by the near-complete absence of Sub-Saharan African markers in the Italian and Balkan peninsulas. Similar theories propagated by Gustav Kossina and the Aryan-Nordic camp about the Northern European origin of the Indo-Europeans of India have similarly been defeated, since Indians completely lack haplogroup I chromosomes that are frequent in European Nordic populations.
So, even though the markers in question are very old (I is of Upper Paleolithic age), we can still reason historically with them.
Often, this historical reasoning can be shaky. For example, Spencer Wells has made tall claims about the Phoenicians, the Sea Peoples, and the Carthaginians in a National Geographic article which were based on the analysis of haplogroup J and E distribution in the Levant and North Africa.
For example, he found that there was little impact of Phoenicians on Carthage, but his conclusions are based on the paucity of haplogroup J in modern North African populations, who are a much broader-group than the socially and geographically constrained group of the ancient Carthaginians. Similar claims were made regarding the non-impact of the Sea Peoples in the Levant, but again, this is based on the similarity between coastal and non-coastal populations.
But, for all we know subdivisions of haplogroup J and other Near Eastern markers may differ between coastal and non-coastal populations, or perhaps, the Sea Peoples did initially affect the coastal peoples, but later their genes diffused into non-coastal populations, removing the distinctiveness of the two.
Let us take a further example of Sicily. The island of Sicily was colonized initially by farmers, and later by Greeks and Phoenicians. All three groups are believed to have contained some "Neolithic" markers, such as haplogroups J, E3b, and G, so any inferences about the relative contributions of the three groups are on very shaky ground.
For example, Semino et al. proposed that only 7% of Calabrian Y-chromosomes are of Greek with the assumption that J2a and E3b represent Anatolian and Greek lineages respectively. But, the frequency of E3b in modern Peloponnesians is not necessarily representative of its frequency in the very specific ancient city states and medieval Greek populations that colonized Southern Italy, and J2a may have arrived in Calabria either from Anatolia, e.g., during the Neolithic, or from Greece, during the age of colonization.
Things become even more complex when we turn to the Balkans or to Anatolia. For example, I playfully recounted some random facts about Phrygo-Armenians, but these hardly scratch the surface of the problem. Hittites, themselves either native or intrusive, were unseated by Phrygians, who were conquered by Persians, who were conquered by Macedonian Greeks, who were conquered by Romans, who were conquered by Turks. Not to mention the Galatians of Ancyra, or the ubuiquitous Armenians of the Byzantine Empire, or even the Jews of both the ancient and more recent origin, and of course the Turks themselves as well as imported Muslims from former provinces or vassals of the Ottoman Empire. And, of course, we should not forget that present-day Anatolians are only a subset of very recent Anatolians, several million of who were liquidated or deported following World War I.
These remarks underscore the near hopelessness of untangling historical patterns on the basis of phylogeography. Is there a way out?
Part of the solution will consist of performing huge studies with large sample sizes and very recently derived genetic markers, augmented by separate genome-wide autosomal clustering methods that may unmask latent genetic components that may be correlated with historical groups. Such studies will be very costly, even though the price of DNA testing is likely to go down, because ultimately the hard work of sample collection has to be done and paid for.
The ultimate solution, would be some significant progress in ancient DNA extraction. At present, mtDNA is the only game in town, and inferences from mtDNA are always up for grabs, due to the potential for contamination, uncertainties about selection, and of course the simple fact that ancient civilizations were largely patriarchal.
An even more exciting development would be the discovery -in modern human populations- of the genetics underlying common human variation in metric and morphological traits. Then, by examing ancient skeletal remains, we will be able to estimate the genetic identity of populations even if DNA cannot be directly observed.
The technical challenges are enormous, but -in my opinion- are not the main challenges at all. As hinted in Genetic vs. Mythical Origins, the study of the past forces us to question our ideas of descent and ethnicity. In the end, will it lead to an erosion of ethnic identity, or to its reinforcement along genetic and hence "objective" lines?