A new paper in The American Journal of Human Genetics, Estimates of Continental Ancestry Vary Widely among Individuals with the Same mtDNA Haplogroup, tells you something which should be obvious:one marker tells you only so much about individual ancestry. In other words, the history of one gene can only tell you so much about the whole genome. Because mtDNA and Y chromosome* does not recombine you can treat it as one long genetic marker. On the coarse grain they can tell us a great deal. Both mtDNA and Y confirmed that it seems the modern human populations seem to be diverged from a group with an African origin. Genomics, even using the whole genome, has confirmed this. Additionally, non-recombining regions of the genome are more tractable for a coalescent framework. They are actually trees.
But Richard Lewontin’s insight that a great deal of human genetic variation is not partitioned across populations, but within them, applies to mtDNA and the Y chromosomes as well. Where Lewontin’s insight misleads is that using just a few more markers one can obtain relatively robust phylogenetic trees which reflect well the population structure and history of a given species. One can see this when one considers mtDNA and Y chromosomal lineages jointly. If someone tells you that their Y chromosomal lineage is R1a1a you can infer that their ancestry is anywhere from Central Europe all the way to South Asia. But if they add that their mtDNA is U2b you can be confident that they are South Asian. U2b spans South Asia, as well as parts of the Middle East. But in the latter zones R1a1a is very rare.
Though the phylogeographic import of mtDNA for a given individual is often questionable, that’s true of any marker. Because mtDNA is amenable to phylogenetic modelling it is still quite useful, especially when making very geographically coarse inferences. But the paper’s argument that the geographic inferences of DTC tests is questionable based on the fact autosomal SNP-chips are the same framework used within the paper to test the informativeness of mtDNA.**
* The NRY
** Disclosure, I have done consulting for Family Tree DNA on their autosomal tests.