I.2. Upper Palaeolithic

The Aurignacian lithic cultural complex appeared around 38000 BC in central Europe, and ca. 33000 BC in western Europe, replacing the earlier Middle and Upper Palaeolithic stone toolmaking styles under a unifying trend. The Goyet cluster (defined by GoyetQ116-1, ca. 33330 BC) appeared associated with the Aurignacian cultural complex, showing that a more homogeneous population accompanied the expanding culture. This genetic homogeneity is probably the result of inbreeding due to small population sizes, as can be inferred from the current estimates for the mean 1,500 (ca. 800–3,300) persons for western and central Europe (Schmidt and Zimmermann 2019).

The GoyetQ116-1 individual is more closely related to the EEA and ANS than any other subsequent European population. Its mitochondrial haplogroup M (found mainly among East Eurasian, Oceanian, and Native American populations) is probably related to this ancestral link (Yang et al. 2017). The ancestry related to Aurignacian samples would continue to contribute to the European stock for thousands of years after this culture’s demise.

The Gravettian complex (ca. 31000–15000 BC) succeeded the Aurignacian. This culture is known for its Venus figurines, typically made from ivory or limestone carvings. The Věstonice genomic cluster (represented by Vestonice16, ca. 28060 BC) was also genetically quite homogeneous in samples studied from Italy, Austria, the Czech Republic, and Belgium. This population was related to an ancient hunter-gatherer population sampled in far eastern Europe (Kostenki and Sunghir) ca. 35500–30000 BC (Sikora et al. 2017). Samples from Goyet shared no ancestry with Aurignacian samples from the same site, supporting that this cultural change was also associated with a population replacement.

Although it is likely that this population came originally from the east, the primary centre of expansion of the Gravettian culture (and thus the Věstonice cluster), based on archaeological data, was located in the Middle Danube Basin, spreading to the Upper Danube Basin and into the mid–Atlantic France (Middle Rhine Group, Maisierian group) and south-western Europe (western Gravettian), as well as to eastern Europe (Dniester and Prut Basins) and the Russian Plains (Kostenki–Avdeyevo group).

Chronologically coincident with the Gravettian were the Ancient North Siberians, represented by two Yana RHS samples (ca. 29600 BC), of P1-M45 lineage, ancestral to haplogroups Q-M242 and R-M207 (Sikora et al. 2018). They share the most genetic drift with Ancient North Eurasians (ANE), represented by three individuals from Upper Palaeolithic south-central Siberia (from Mal’ta–Buret’ ca. 22350 BC to Afontova Gora ca. 16000 BC), spanning the duration of the Late Glacial Maximum (LGM). Even though Mal’ta–Buret had Venus figurines with potential similarities to Gravettian ones, the ANE ancestry proper of this cluster is not found among Europeans until the Eneolithic steppe expansions.

The Mal’ta boy’s paternal lineage diverged from haplogroup R-M207* shortly before its split into R1-M173 and R2-M479 subclades (Raghavan et al. 2014), which is estimated to have happened ca. 26200 BC. His so-called Ancient North Eurasian (ANE) ancestry contributed substantially to the genetic ancestry of Siberians, Native Americans, and Bronze Age Yamna individuals (Lazaridis et al. 2014),  being close to modern-day Native Americans, Kets, Mansi, Nganasans, and Yukaghirs (Flegontov et al. 2016).

The Magdalenian culture spread after the LGM (ca. 23000–17000 BC) from a refuge in southern Iberia, chasing the retreating ice sheet expanding in a northeast direction with El Mirón genomic cluster. From the Late Magdalenian assemblages of Germany and Poland, the expansion of the Hamburgian and Creswellian technocomplexes ca. 13000 BC spread across the Northern European Lowlands south of the Scandinavian Ice sheet.

Relevant known populations before the Final Upper Palaeolithic period may be simplistically divided into geographic regionswithout care for sub-structured populations and gene flowas (Suppl. Fig. 2):

·         Goyet cluster: stemming from an EWE source, they expand first with the Aurignacian, and then again from an Iberian refuge during the Magdalenian. During this initial expansion, and also later during the Gravettian, hg. C1a2-V20 (formed ca. 43000 BC, TMRCA ca. 41500 BC) dominates the ancient DNA record.

·         Common West Eurasians (CWE): ghost population stemming from an EWE source, probably spread somewhere around the Black Sea. It contributes to different West Eurasian populations to the north (eastern Europe) and south of the Caucasus (Fertile Crescent) during the Palaeolithic.

·         Ancient Middle Easterners (AME): Represented by two samples from the Dzudzuana cave in the southern Caucasus (ca. 25000–22000 BC), they show contribution from CWE (ca. 72%) and a BE or earlier BE-like population (ca. 28%). Their ancestral population contributed to Epipalaeolithic Levantine and Anatolian populations, so they are a likely proxy for a contemporary population spanning the whole Fertile Crescent (Lazaridis et al. 2018).

·         Věstonice cluster: formed probably in far eastern Europe by an admixture of an EWE source close to CWE (ca. 62%), and from another EWE source close to the Goyet cluster (ca. 38%), it expanded west with the Gravettian expansion. The spread of haplogroup I-M170 (formed ca. 40900 BC, TMRCA ca. 25500 BC) in Europe is probably associated with the expansion of certain groups of the Věstonice cluster.

·         Ancient North Siberians (ANS):  It diverged ca. 36000 BC from EWE (soon after this population split from East Asians), with further contributions from EEA (ca. 25%) also soon after the split EEA–EWE.

·         Ancient North Eurasians (ANE): contributed to by ANS, an EWE source close to the Goyet cluster (ca. 75%) and an EEA population (ca. 25%). This component contributed substantially to the genetic ancestry of Siberians, Native Americans, and Bronze Age Yamna individuals (Lazaridis et al. 2014),  being close to modern-day Native Americans, Kets, Mansi, Nganasans, and Yukaghirs (Flegontov et al. 2016). Their wide distribution of a consistent Initial Upper Palaeolithic technology in Siberia and neighbouring territories up to north Mongolia seems to be coincident with the eastward expansion of peoples through the steppe belt (Zwyns and Lbova 2018).

·         El Mirón cluster: it derives most of its ancestry from Goyet (63%), and it has not been found in the previous Gravettian period. This is thus the first described case of a ‘resurgence’ of a population from an (as of yet) unsampled pocket that survived a previous population turnover, and persisted for a long time after their assumed disappearance. The main haplogroup associated with its expansion is I-M170 also found later in the Villabruna cluster, which suggests that Gravettian lineages admixed with a population of the Goyet cluster, and remained in isolation (most likely in Iberia) until the Magdalenian expansion.