Trying to confirm previous runs showing "Amerindian" in North/West Europeans and in the Caucasus, I ran a more conventional supervised analysis. I set up the following poles:
1. Egyptians once again representing the Fertile Crescent.
2. Nganasans, comprising the 5 less admixed individuals, but keeping the rest in the analysis (but out of the pole)
3. Totonac+Surui+Karitiana: I included 3 especially recently unadmixed populations. Amerindians suffered from severe founding effects. The analysis is not too different using only one of them as the pole, but it's more chunky since some original diversity is lost if only one group is considered..
Lets assume the Solutrean Hypothesis is correct, as suggested by ADMIXTURE's results (except the Solutreans can't be properly called the ancestors of modern Europeans).
It's somewhat strange that Amerindians fall 100% in their own pole, since I suspect they also received significant gene flow from Siberians proper (Y and mit haplogroups seem to indicate such a scenario). However those severe founding effects during the Atlantic crossing by the Ice Cap led to a highly distinctive population reaching the other side. It is indeed likely that Northern Europeans and Caucasus valley populations harbour gene variants present in the original "Western Siberian" or Cro-Magnon populations and which were lost due to drift in the Atlantic crossing by the ice cap. Thus ADMIXTURE tend to assume the Amerindians are the "unadmixed" population (similar to what it did to Sardinians in unsupervised mode). This may be why the Amerindian modal component can be found in strange places in Eurasia, since that may really be about Eastern Siberian less common variants that also suffered from drift during the crossing into America via Beringia.
So I suspect Amerindians may really be the admixed population, whose parent populations are the West "Siberians"/Solutreans/Native Europeans going by the Atlantic ice cap and on the other hand East or proper Siberians entering through Beringia. Both populations having undergone drift, but especially the ones coming from Europe, and producing an "offspring" population that can be related to ADMIXTURE to both groups but, being so distinctive, is assumed to be the "parent".
This may be why the "Amerindian" component is quite chunky when analysed with the Amerindian proper populations, and quite uniform and more widespread when analysed without them, with a "blind pole" instead fishing them out independently. In addition, maybe Spaniards and Basques don't present so much "Forager A" component in Amerindian containing runs due to founding effects in Northern Europe, when Solutrians expanded there from their Iberian refugium at the end of the Ice Age >10.000 years ago. So Northern Europeans would share the first founding effects with Amerindians, unlike Iberian Solutrians and those distinctive signatures make those segments easier to spot in an analysis including Amerindians?
Still affinities of "Forager A" segments in Europeans and those in Amerindians are sufficiently clear for them to be grouped together every run, more or less chunkily.
Yesterday's blind analysis spreadsheet.