A first look at the DNA of Neolithic inhabitants from Sweden

I was eager to get my hands on the ancient DNA from ancient Swedish Neolithic people, as soon as I became aware of it a few months ago. I was finally able to extract SNPs from the data, and I decided to test the samples with my K7b and K12b calculators from the Dodecad Project, as I had also recently done for the Tyrolean Iceman. You can skip to the Results section if you want.

I extracted the following number of SNPs from the different individuals, that were in common with my main HGDP reference:

• 15,734 SNPs: Gok4 (TRB / farmer)
• 15,385 SNPs: Ajv52 (PWC / hunter-gatherer)
• 25,108 SNPs: Ajv70 (PWC / hunter-gatherer)

I did have some trouble with my code for extracting data from the third hunter-gatherer individual, Ire8. This also happens to have the lower number of SNPs as reported by Skoglund et al. so its non-inclusion is probably not a great loss.

K7b and K12b are based on a set of 166,770 SNPs, so I intersected the SNPs of the ancient individuals with them, resulting in:

• 4,054 SNPs: Gok4
• 4,077 SNPs: Ajv52
• 6,631 SNPs: Ajv70

This is, of course, a small number of SNPs, and if we tried to infer structure within West Eurasia with them, we might fail. Nonetheless, by exploiting the structure inferred by the larger number of SNPs in modern populations, and using these to test the ancestry of the ancient ones, we get results that appear to agree fairly well with Skoglund et al. (2012).

A problem with comparing against extant populations, rather than ancestral components is that relationships are averaged: for example, Turks in Skoglund et al. appear quite distant to both Neolithic groups, but we cannot know to what extent this is due to their small levels of central/east Eurasian ancestry and to what extent this is due to their native Anatolian ancestry.

We now have two ancient autosomal DNA sampling locations, and both turned up unexpected results. The Iceman, a Copper Age inhabitant of the Alps resembled modern Sardinians. A Megalithic Swedish farmer resembled Southern Europeans, while his hunter-gatherer contemporaries were outside the range of modern variation. These results should give us caution about the identity of past populations elsewhere. 

As I have argued elsewhere, the past seems to have been much more dynamic than many had suspected, and the people that walked and rowed to the ends of the Earth during the Upper Paleolithic did not suddenly grow fetters or decided to stay put during the Neolithic as many "Paleolithic continuity" adepts had proposed.

A couple more caveats:

1. Irrespective of their actual origin, these individuals would still be inferred to be some admixture of the ancestral components adding up to 100%. This hints at their affiliations, but does not -in itself- supply evidence of their absolute proximity to the ancestral components.
2. As more and more ancient individuals are sampled, we will be able to generate genuine ancient populations that are ancestral to modern ones. When that happens, we can directly test modern individuals against panels of ancient ones. For the time being, we have to make do with the reverse, i.e., test ancient individuals against panels of modern ones.

RESULTS


In all comparisons with other K7b results, you should take into account the much smaller number of SNPs used on the Neolithic remains from Sweden.

The K7b results are below. Consult the spreadsheet for comparative values in modern populations.

The K12b results are below. Consult the spreadsheet for comparative values in modern populations.

The raw percentages can be seen below; I have also added the results previously calculated for Oetzi, the Tyrolean Iceman:

DISCUSSION


The results for the two hunter-gatherer samples are as expected predominantly "Atlantic_Baltic" at K=7 and "North_European" at K=12. Since these two samples are outside modern variation, it is expected that their mapping may have added noise; see this post about the dangers lurking at the edges of variation.

Nonetheless, the results can be interpreted as reflecting the fact that the "Atlantic_Baltic" and "North_European" components partially reflect the pre-farming population of Europe. At K=12 it is noteworthy that there is a minor "Atlantic_Mediterranean" admixture in the two hunter-gatherer samples. In my opinion, this may reflect either some level of admixture with the incoming farmers and/or the pre-farming component (but of Western European rather than Baltic type) that may also exist in these foragers. On balance, however, the "North_European" component far outweighs the "Atlantic_Mediterranean" one, which is also consistent with their location (Gotland) which ties them to the Baltic rather than Atlantic Europe.

The farmer sample is remarkable in that, like the Tyrolean Iceman, she seems to be made up entirely of "Atlantic_Baltic" and "Southern" at K=7. There is a hint that the order of the two components are reversed in Gok4 relative to Oetzi, which probably makes sense. The third major West Eurasian component at this level of resolution, the "West Asian" one is conspicuously absent. This component -bimodal in the Caucasus and Balochistan, and strongly represented in the highlands of West Asia in between the two- does occur at ~10% in modern Northern Europeans, so its absence in all Neolithic samples so far hints at its later introduction into at least parts of Europe.

The result at K=12 is fascinating, since Gok4 turns out to be 81% "Atlantic_Mediterranean", and, like Oetzi, with a noticeable "Southwest_Asian" strain. The "Atlantic_Mediterranean" component is bimodal in modern Sardinians and Basques, while the "Southwest_Asian" radiates from southern parts of the Near East into Mediterranean Europe.

Gok4 appears to be "even more Atlantic_Mediterranean" than any modern population. As I mention in the original post, the inhabitants of megalithic monuments  of North-Western Europe were believed by Coon to belong to a "Long Barrow type" which he considered ancestral to the modern "Atlanto-Mediterranean" type. Here is his description:

Toward the end of the Neolithic period, the western Mediterranean countries were invaded by seafarers of a tall, exceptionally long-headed Mediterranean variety; some of these invaders passed through the Straits of Gibraltar, whence they also invaded the British Isles and Scandinavia. 

I would say that the evidence is not incompatible with this scenario. We must probably wait to see whether Gok4 was descended from seafarers from the eastern Mediterranean (where the "Southern" component is modal), following the Mediterranean and then Atlantic coasts up to Scandinavia, or whether they are descended from a different group of people who followed the plains and river valleys of the Balkans and Central Europe and arrived to the north via the inland route. The strong Atlantic_Mediterranean result, coupled with high levels of allele sharing in Cyprus, Greece, France, and the Netherlands, but not particularly in the northern Balkans leads me to prefer the maritime colonization scenario, at least for now. Ancient DNA from more European regions will hopefully help us better understand "what really happened in prehistory."