Caught in the act: lactose intolerant and lactose tolerant populations together but mostly unmixed in Chalcolithic Upper Ebro

There is a new paper on ancient DNA from the southernmost Basque Country that has found the oldest known lactose tolerance alleles in prehistoric Europe:

Theo S. Plantiga et al., Low prevalence of lactase persistence in Neolithic South-West Europe. European Journal of Human Genetics 2012. Pay per view.

Context

The samples studied are the same (in part) as those from Izagirre and de la Rúa 1999, specifically 19 individuals from San Juan Ante Porta Latinam (Araba) and 7 from Longar  (Navarre). In spite of both sites being inside Euskal Herria (Basque Country) as usually acknowledged, the are both very close to Logroño, capital of La Rioja autonomous region. I mention because this area of the Upper Ebro is at least somewhat distinct from the higher reaches of the core Basque Country and may not be representative of ancient (proto-)Basque genetics but rather of Ebro Valley populations.

It is easy to appreciate in genetic research as that of Adams 2008, that Aragon (I presume representative of the overall Ebro Valley) is markedly distinct from sub-Pyrenean neighbors like Basque Country or Catalonia, holding a much larger apportion of presumably Neolithic haplogroups of Transmediterranean origin (surely E, G, J and T, as well as maybe most of I), being more similar in this to Valencia, Ibiza or Western Iberia.

Paleo- and Epipaleolithic Basque Country

Similarly the Ebro Valley is a region of very early Neolithic settlement in which often no pre-Neolithic habitation is known to have existed. However the data I know for the Basque part of the Upper Ebro basin (X. Peñalver 1996), suggests an Epipaleolithic (epi-Magdalenian) settlement of most of Araba (but not the Ebro banks as such) and only the highland parts of Navarre (right). 

It is therefore only in the Neolithic and later on in the Chalcolithic (term used sometimes loosely to indicate the late part of the Neolithic, when social complexity increases, regardless of the presence of copper, gold or silver), when the Upper Ebro, including that within conventional Basque borders is colonized. 

These colonists may or not have been of aboriginal (proto-Basque) stock. In fact anthropometric estimates emphasize the presence of Gracile Mediterranean types in devastated La Hoya town, contrasting with the native Pyrenean type of apparent Paleolithic stock.

Chalcolithic and Early Bronze sites

Whatever the case, it is clear that the Upper Ebro valley is not the core of the Basque Country, located further north, but a border area which may have been colonized partly by people of exotic origins upon the arrival of Neolithic. Of course all this is open for discussion and further analysis of the available data. 

[See also: Wikipedia: Basque Prehistory].

Findings

Luckily several readers have sent me copies of the paper (you people are awesome, thanks), and even some ideas on what is most interesting (not always the most obvious).

The findings are detailed in table 3:

The most important finding is also the most obvious: the lactose tolerance allele (or more specifically rs4988235 in its T variant, as there are others less famous such alleles) is found at the oldest known site in Europe and that is (in spite of all my previous caveats) in the Basque Country. 

There is a similar aged site in Götland, Sweden (Mälstrom 2010), but the apportions are even lower (1/19 alleles or 1/10 persons had the CT combo). All other older of contemporary sites in Central Europe or nearby Occitania have yielded negative for what is nowadays a dominant allele in Western and Northern Europe. However there are many blanks: for example no Neolithic (never mind Paleolithic) British or Atlantic French or Portuguese samples have ever been tested for. In fact no Basque samples other than these either... so the question marks are still many.

The second finding is that the apportions are low, even if it is enough for lactose tolerance to kick in that a single T allele exists (classical Mendelian dominance), only 31% and 14% of the people sampled were lactose tolerant, meaning that (most probably) non-processed milk was rare in their diet. This is contradictory with the idea that the allele became common thanks to positive selection, after all these people had been herding livestock of various types for 2000-2500 years already (just in situ, i.e. not counting their likely ancestors in Italy, the Balcans or West Asia: add other 3000 or 4000 years maybe) and yet most were lactose intolerant (apparently). 

This is something that the authors also find surprising, pondering that maybe its frequency levels may have risen more recently just because of cultural pressure. What?! Drinking milk became fashionable and you were stoned to death if you found it nasty? Hard to believe, honestly, and if you did not die from not drinking milk, there is no (or extremely low) selective pressure we can take seriously.

Two populations?

But the most striking finding, not even mentioned by the authors is the following: there are almost no heterozygous individuals in these samples: with two exceptions in the San Juan APL  sample, people fall either to the CC (lactose intolerant) or the TT genotype. The percentage of heterozygous is much lower than that of the allele itself (11% vs. 26% in SJAPL and 0% vs. 14% in Longar).

I don't know how can you explain this but for me it is crying two populations that are still at the early stages of blending: one almost homogenously TT and the other almost homogenously CC. 

Otherwise, if the admixture process would be advanced, there would be more heterozygous (CT) individuals than TT homozygous ones. 

Which can be these two populations? Hard to say for sure but my natural intuition, knowing what I know about Basque Prehistory, is that the CC population would be made up of colonists of Mediterranean stock (possibly trans-Mediterranean: Balcanic, Italian or West Asian, as suggested by recent studies of mitochondrial DNA from Epicardial sites) who were quite obviously lacking the T allele altogether even if they were the ones to have the goats, sheep and cows initially.

These peoples left a legacy in the Ebro Valley and elsewhere in Iberia and Europe but they ultimately must have failed to become fully dominant and today their ancestry seems to be just a fraction of modern Western Europeans, seldom looking dominant even in the patrilineages. 

The TT population must therefore be some other. Which one? In this context I can only think of the local aborigines rooted in the Paleolithic cultures of the area and described by anthropometric usages as of Pyrenean stock (I think Heraus would say Dinaromorphic Atlanto-Mediterranean but that's too long - incidentally he just posted on the phenotypes of Arabako Errioxa). 

So this brings me to my fallback theory: the lactose tolerance allele is a mere fluke which became fixated randomly in pre-Neolithic populations just because (or maybe, speculatively, related to motherly lactation practices), eventually showing an unexpected use in drinking milk and eating desserts, what was nice for the development of Basque cuisine but mostly unrelated to survival. 

Being able to enjoy rice pudding (as I'm doing right now) is probably the most clear advantage of the T allele

See also: Leherensuge: Actual lactase persistance more common than genes predict.

Update (Jan 19): a scanned paper by Belén Márquez et al., describes in detail the burials of SJAPL and Longar, both looking rather militarized (mostly adult males, abundance of arrow points and arrow injuries...)