Forty-two Jomon genomes put a date on Ice Age isolation and a frequency on cold alleles

Researchers reconstructing the ancestry of Japan's Jomon population sequenced DNA from 42 ancient individuals, 25 of them newly sequenced, and date the lineage's separation from mainland East Eurasians to between 27,000 and 19,000 years ago ^[3][4][6]. The consequential part is not the date but what sits next to it: three genes involved in fat storage and use, carried at frequencies that no living reference population in the comparison matches ^[10][12][13].

That split window falls entirely inside the Last Glacial Maximum, which the study places between 29,000 and 19,000 years ago, the period of most extensive global ice cover ^[1][7]. According to the paper, published in Science Advances, warming at the end of the Ice Age submerged the land bridges and isolated the ancestors of the Jomon on the archipelago, fixing their genetic trajectory ^[6][8]. The Jomon themselves occupied the islands from roughly 16,000 to 3,000 years ago, before farming arrived ^[2].

The design matters here because earlier work on this question was starved of data ^[4]. The team worked from the petrous bone, the dense inner-ear bone that preserves ancient DNA unusually well, and compared the Jomon against ancient and modern populations from Asia, Siberia, Oceania and the Americas ^[17]. Relate and TreeMix were used to reconstruct the trees, and the same data were scanned for selection signals ^[18]. Twenty-five new genomes on top of 17 already available is a different statistical object than the handful previous studies had ^[5].

The selection results are where the argument gets specific. UCP1, FTO and APOA5 all appear at unusually high frequency, and the study proposes they helped brown fat convert fat into heat, keeping people warm without shivering ^[10]. Strong selection also favoured variants in APOA5 and ALDH1A2-LIPC that sustain higher blood triglyceride levels, the fuel for that heat production ^[20]. On the reported frequencies, nearly three in four Jomon carried the FTO variants against one in five people in present-day Central Asia and Siberia, a gap of about 55 percentage points ^[12][15]. For APOA5 the contrast is 89 percent of Jomon against 29 percent of modern East Asians, roughly a threefold difference ^[13][14]. Note that the two comparisons use different modern reference groups, so they are not two readings of the same ruler ^[12][13].

There is a cost attached. One FTO variant tied to heat production is also strongly associated with higher BMI in modern populations, which is the familiar shape of an allele that paid off under glacial conditions and does something else now ^[11].

Two other findings tidy up long-running disputes. Northern Siberian populations contributed to the Jomon gene pool, which speaks to how East Asia was first peopled ^[9]. And the team found no Denisovan ancestry unique to the Jomon, implying any such ancestry entered before the split rather than after isolation ^[19].

What to watch: whether the selection scans hold when other groups add genomes, since the frequency contrasts do the load-bearing work and rest on one dataset ^[18][12]. Also worth tracking is the internal structure, with the Hokkaido and Hondo lineages diverging 10,000 to 8,800 years ago, some 6,000 to 7,200 years into the Jomon period ^[16][23]. The older framing asked whether Ice Age survival was clothing and fire or biology ^[22]. This study does not settle that, but it now has named alleles on one side of the ledger ^[10][20].