The Sakhtysh micro-region is located in the Volga-Oka interfluve, along the headwaters of the Koyka River in the Ivanovo Region, central European Russia (Fig. 1). The area has evidence of human habitation from the Early Mesolithic to the Iron Age, and includes altogether 11 long-term and seasonal settlements (Sakhtysh I–II, IIa, III–IV, VII–XI, XIV) and four artefact scatters (sites V–VI, XII–XIII), in addition to which burials have been detected at five sites (I–II, IIa, VII, VIII) (Kostyleva and Utkin, 2010). The locations have been known since the 1930s and intensively studied since the 1960s under the leadership of D.A. Kraynov, M.G. Zhilin, E.L. Kostyleva, and A.V. Utkin.
Sakhtysh II and IIa are the most extensively studied sites of the complex, with ca. 1500m2 and around 800m2 excavated, respectively. The burial grounds at both sites are considered as fully investigated.
The AMS dates do not support the previously proposed phasing of the Sakhtysh burials to early (4750–4375 BP/3600–3000 cal BCE), late (or developed; 4375–4000 BP/3000–2500 cal BCE), and final (4000–3750 BP/2500–2200 cal BCE): the early and late burials at Sakhtysh IIa do not stand out as two separate groups, and also the burials and hoards from Sakhtysh II, connected to the final phase, are temporally overlapping with these. Neither the use sequence, where the settlement and burial phases are non-overlapping and also complementary between the sites (Kostyleva and Utkin, 2010, 2014), finds support in the present material.
The AMS datings indicate that the Volosovo people started to bury their dead at Sakhtysh IIa after 3700 cal BCE; dates earlier than this may be affected by FRE or suffer from mixed contexts and poor quality of dates. The present data questions the interpretation that the Sakhtysh IIa cemetery was used without interruptions between 4800 and 4080 BP (Kostyleva and Utkin, 2010), i.e. for a millennium between 3550 and 2600 cal BCE. The AMS dates rather suggest a use period of some centuries only around the mid-4th millennium cal BCE, tentatively 3650–3400 cal BCE. This would also be more realistic considering the number of burials at the site.
The absolute dating of Volosovo culture was for a long time hampered by the small number of radiocarbon dates (see Kraynov, 1987). Today,>100 datings connected with it can be found in literature (Korolev and Shalapinin, 2010; Chernykh et al., 2011; Nikitin, 2012; Mosin et al., 2014). Unfortunately, the available dates do not form solid grounds for dating the cultural phenomenon, as many of them have quality-related issues, large measurement errors, and ambiguous cultural or physical contexts. Consequently, particular datings may be connected to different cultural phases by different scholars. Finally, a large part of the newly-published datings are obtained through direct dating of potsherds (Kovaliukh and Skripkin, 2007; Zaitseva et al., 2009), and therefore, their cogency must be faced with reservation (see Van der Plicht et al., 2016; Dolbunova et al., 2017).
The datings connected with Volosovo cover a wide time range between ca. 5500 BP (4400 cal BCE) and ca. 3700 BP (2100 cal BCE). However, datings from secure contexts, with good quality (error ca. 50 years or below) and no probable FRE, place the beginning of Volosovo culture to the first half of the 4th millennium cal BCE, around 3700–3600 cal BCE. This is also supported by the roughly coeval terminal dates given for the preceding Lyalovo (Zaretskaya and Kostyleva, 2011) and Volga-Kama cultures (Lychagina, 2018), as well as the appearance of related neighbouring cultures, for example, in the Kama region (Nikitin, 2012; Lychagina, 2018), the southern forest steppe area (Korolev and Shalapinin, 2014), and north-western Russia and Finland (Nordqvist, 2018). Still, the dating of many of these cultural phases suffers from the same problems as of Volosovo.
A handful of contested datings place the end of Volosovo culture to the final centuries of the 3rd millennium cal BCE, or even later (Kostyleva and Utkin, 2010; Chernykh et al., 2011; Nikitin, 2012). On the other hand, the new AMS dates indicate that Volosovo activities at Sakhtysh II and IIa ceased before or towards the early 3rd millennium cal BCE; if this reflects the general decline of Volosovo culture must be still confirmed by more dates from Sakhtysh and elsewhere. In this context, the general cultural development must be accounted for. To what extent – if at all – the Volosovo people were present after the arrival of the Corded Ware culture-related Fatyanovo-Balanovo populations? Based on the current, albeit scant and inconclusive radiocarbon data this took place from ca. 2700 cal BCE onwards (Krenke et al., 2013).
Uralicists have come a long way from the 1990s, when the picture of Uralic before Balto-Slavic in the Baltic was already evident, and Uralians were identified with Comb Ware peoples. The linguistic data and relative chronology are still valid, despite the now outdated interpretations of absolute archaeological chronology, as happens with interpretations of Krahe or Villar about Old European.
NOTE. Kallio’s contribution appeared in the book Languages in Prehistoric Europe (2003), which I hold nostalgically close in my Indo-European library (now almost impossible to read fully). It is still one of my preferred books (from those made up of mostly unconnected chunks on European linguistic prehistory), because it contains Oettinger’s essential update of North-West Indo-European common vocabulary, which led us indirectly to our Modern Indo-European project from 2005 on.
In any case, the Uralic arrival in the region east of the Baltic Sea preceded the Indo-European one (…).
This theory that the ancestors of Finno-Saamic speakers arrived in the Baltic Sea region earlier than those of Balto-Slavic speakers is still rejected by some scholars (e.g. Napolskikh 1993: 41-44), who claim, for instance, that Finno-Saamic speakers would not have known salmons before they met Balts because the Finno-Saamic word for ‘salmon’ (i.e. *losi) is a borrowing from Baltic. Similarly, one could claim that English speakers would not have known salmons before they met Frenchmen because English salmon is a borrowing from French. In other words, Worter und Sachen are not necessarily borrowed hand in hand. Otherwise, it would not be so easy to explain how many Finnish names of body parts are borrowings from Baltic (e.g. hammas ‘tooth’, kaula ‘neck’, reisi ‘thigh’) and from Germanic (e.g. hartia ‘shoulder’, lantio ‘loin’, maha ‘stomach’).
A more probative argument is the fact that Balto-Slavic features in Finno-Saamic are mostly lexical ones (i.e. typical superstrate features), where Finno-Saamic features in Balto-Slavic are mostly non-lexical ones (i.e. typical substrate features). Note that there are more Balto-Slavic features in Finnic than in Saamic and more Finno-Saamic features in Baltic than in Slavic. This fact could be explained by presuming that Pre-Saamic was spoken north of the Corded Ware area and Pre-Slavic was spoken south of the Typical Pit-Comb Ware area, whereas Pre-Finnic and Pre-Baltic alone were spoken in the area, where both the Typical Pit-Comb Ware culture (ca. 4000-3600 BC) and the Corded Ware culture (ca. 3200-2300 BC) were situated. This area was most probably bilingual, until Finnic and Baltic won in the north and in the south, respectively.
As is well-known, the idea of Uralic substrate features in Balto-Slavic is not new (cf. e.g. Pokorny 1936/1968: 181-185). As recent studies (e.g. Bednarczuk 1997) have shown, their density is the most remarkable in the four Balto-Slavic languages spoken in the earlier Pit-Comb Ware area (i.e. Latvian, Lithuanian, Belorussian, Russian). On the other hand, occasional Uralisms in the other Balto-Slavic languages spoken west of the Vistula and south of the Pripyat may rather be considered adstrate features spread from the northeast.
The idea of Indo-European superstrate features in Finnic is not new either (cf. e.g. Posti 1953). As Jorma Koivulehto (1983) has recently shown, the earliest Indo-European loanword stratum in the westernmost Uralic branches alone can be considered Northwest Indo-European and connected with the Corded Ware culture (ca. 3200-2300 BC). Since this layer, there have been continuous contacts between Baltic and Finnic. According to Koivulehto (1990), the following stratum can be called Proto-Balt(o-Slav)ic and dated to the Late Neolithic period (ca. 2300-1500 BC). Note that this Proto-Balt(o-Slav)ic dating agrees with the established ones (cf. e.g. Shevelov 1964: 613-614, Kortlandt 1982: 181), when we remember the fact that archaeologists have also moved their datings back by centuries during the last decades.
Finally, there is also a Baltic loanword stratum which was not borrowed from the ancestral stage of Latvian, Lithuanian and/or Old Prussian but from some extinct Baltic language or dialect (Nieminen 1957). However, as these words still go back to the early Proto-Finnic stage, they can hardly be dated later than Bronze-Age ( ca. 1500-500 BC). Therefore, we may conclude that they were probably borrowed from a Baltic superstrate, which arrived in the Finnish Gulf area during the Corded Ware period and survived there until the Bronze Age, when it was no longer identical with other Baltic dialects. In any case, as later Baltic loanword strata concern southern Finnic languages alone, we may presume that this ‘North Baltic’superstrate had become extinct.
The traditional association of Uralic with Volosovo hunter-gatherers doesn’t make sense, since they neither miraculously survived for thousands of years nor mixed for hundreds of years with Corded Ware peoples, so we can now more confidently reject the recent assumption by Carpelan & Parpola that their language was adopted by incoming Fatyanovo, Balanovo and Abashevo groups, to develop into the known Uralic languages (more here). This includes one of the many models of the the Copenhagen group, who simplistically follow “Steppe ancestry” for Indo-Europeannes.
In this study, we present new genomic data from Estonian Late Bronze Age stone-cist graves (1200–400 BC) (EstBA) and Pre-Roman Iron Age tarand cemeteries (800/500 BC–50 AD) (EstIA). The cultural background of stone-cist graves indicates strong connections both to the west and the east [20, 21]. The Iron Age (IA) tarands have been proposed to mirror “houses of the dead” found among Uralic peoples of the Volga-Kama region .
(…) The 33 individuals included 15 from EstBA, 6 from EstIA, 5 from Pre-Roman to Roman Iron Age Ingria (500 BC–450 AD) (IngIA), and 7 from Middle Age Estonia (1200–1600 AD) (EstMA) and yielded endogenous DNA ∼4%–88%, average genomic coverages ∼0.017–0.734×, and contamination estimates <4% (Table S1). We analyzed the data in the context of modern and other ancient individuals, including from Neolithic Estonia .
We identified chrY hgs for 30 male individuals (Tables 1 and S2; STAR Methods). All 16 successfully haplogrouped EstBA males belonged to hg R1a, showing no change from the CWC period, when this was also the only chrY lineage detected in the Eastern Baltic [11, 13, 30, 31]. Three EstIA and two IngIA individuals also belonged to hg R1a, but three EstIA males belonged to hg N3a, the earliest so far observed in the Eastern Baltic. Three EstMA individuals belonged to hg N3a, two to hg R1a, and one to hg J2b. ChrY lineages found in the Baltic Sea region before the CWC belong to hgs I, R1b, R1a5, and Q [10, 11, 12, 13, 17, 32]. Thus, it appears that these lineages were substantially replaced in the Eastern Baltic by hg R1a [10, 11, 12, 13], most likely through steppe migrations from the east [30, 31]. (…) Our results enable us to conclude that, although the expansion time for R1a1 and N3a3′5 in Eastern Europe is similar , hg N3a likely reached Estonia or at least became comparably frequent to modern Estonia  only during the BA-IA transition.
A clear shift toward West Eurasian hunter-gatherers is visible between European LN and BA (including Baltic CWC) and EstBA individuals, the latter clustering together with Latvian and Lithuanian BA individuals . EstIA, IngIA, and EstMA individuals project between BA individuals and modern Estonians, partially overlapping with both.
(…) EstBA individuals are clearly distinguishable from Estonian CWC individuals as the former have more of the blue component most frequent in WHGs and less of the brown and yellow components maximized in Caucasus hunter-gatherers and modern Khanty, respectively. The individuals of EstBA, EstIA, IngIA, EstMA, and modern Estonia are quite similar to each other on average, indicating that the relatively high proportion of WHG ancestry in modern Eastern Baltic populations compared to other present-day Europeans  traces back to the BA.
When comparing Estonian CWC and EstBA using autosomal outgroup f3 and Patterson’s D statistics (Table S3), the latter is more similar to other Baltic BA populations, to Baltic IA and Middle Age (MA) populations, and also to populations similar to WHGs and Scandinavian hunter-gatherers (SHGs), but not to Estonian CCC (Figures 2A and S2A; Data S1). The increase in WHG or SHG ancestry could be connected to western influences seen in material culture [20, 21] and facilitated by a decline in local population after the CCC-CWC period . A slight trend of bigger similarity of Estonian CWC to forest or steppe zone populations and of EstBA to European early farmer populations can also be seen.
(…) When comparing to modern populations, Estonian CWC is slightly more similar to Caucasus individuals but EstBA to Baltic populations and Finnic speakers (Figure 2B; Data S1). Outgroup f3 and D statistics do not reveal apparent differences when comparing EstBA to EstIA, EstIA to IngIA, and EstIA to EstMA (Data S1).
These results highlight how uniparental and autosomal data can lead to different demographic inferences—the genetic change between CWC and BA not seen in uniparental lineages is clear in autosomal data and the appearance of chrY hg N in the IA is not matched by a clear shift in autosomal profiles.
EstBA individuals have no Nganasan-related ancestry and EstIA, IngIA, and EstMA individuals on average have 2% or 4% (Figure 3; Data S1). The differentiation remains when using BA or IA Fennoscandian populations  instead of Nganasans (Data S1). Notably, the proportion of Nganasan-related ancestry varies between 0% and 12% among sampled EstIA, IngIA, and EstMA individuals (Data S1), which may suggest its relatively recent admixture into the target population. Moreover, two individuals from Kunda (0LS10 and V10) have the highest proportions of Nganasan ancestry among EstIA (6% and 8%), one of them has chrY hg N3a, and isotopic analysis suggests neither individual being born in Kunda .
About these two males from Tarand-graves, ‘foreign’ to Kunda:
0LS10: Male from tarand III (burial 9; TÜ 1325: L777), age 17–25 years . He had a fragment of a sheep/goat bone and ceramics as grave goods. This burial has two radiocarbon dates: 2430 ± 35 BP (Poz-10801; 760–400 cal BC) and 2530 ± 41 BP (UBA-26114; 800–530 cal BC) . According to the isotopic analysis, the person was not born in the vicinity of Kunda; his place of birth is still unknown (but south-western Finland and Sweden are excluded) . Sampled tooth r P1.
V10: Male from tarand XI (burial 24; TÜ 1325: L1925), age 25–35 years , date 2484 ± 40 BP (UBA-26115; 790–430 cal BC) . He had a few potsherds near the skull. Likewise, this person was not locally born . Sampled tooth l P1.
The paper shows thus:
Major continuity of ancestry from Corded Ware to modern Estonians, with only slight changes in different periods. In fact, one of the best fits for the Late Bronze Age ancestry is Gyvakarai1, one of the Corded Ware “outliers” described as “closer to Yamna”, which I already said may be closer to Sredni Stog/EHG populations instead. Another interesting take is that the change from Bronze Age to Iron Age corresponds to an increase in Baltic Corded Ware-related ancestry, rather than being driven by Siberian ancestry.
A Volosovo-related migration of hg. N1c with Netted Ware into the area seems to be discarded, based on the full replacement of paternal lines and continuity of R1a-Z283. It is only during the Tarand-grave period when a system of chiefdoms (spread from Ananyino/Akozino) brings haplogroup N1c to the Gulf of Finland. During the Iron Age, the proportion of paternal lineages is still clearly in favour of R1a (50% in the coast, 100% in Ostrobothnia), which indicates a gradual replacement led by elites, likely because of the incorporation of Akozino warrior-traders spreading all over the Baltic, bringing the described shared Mordvinic traits in Fennic.
The arrival of Akozino warrior-traders (bringing N1c and R1a lineages) was probably linked to this minimal “Nganasan-like” ancestry of some samples in the transition to the Iron Age. This arrival is supported by samples 0LS10 (the earliest hg. N1c) and V10 (of hg. R1a), both dated to ca. 800-400 BC, with V10 showing the highest “Nganasan-like” ancestry with 4.8%, both of them neighbouring samples showing 0%. This variable admixture among local and foreign paternal lineages might support the described social system of family alliances with intermarriages. In fact, a medieval sample, 0LS03_1 (hg. R1a) also shows a recent “Nganasan-like” ancestry, which probably points to the integration of different Arctic-related ancestry components among Modern Estonians, in this case related to Finnish expansions and thus integration of Levänluhta-related ancestry, as per the supplementary data.
NOTE. Such minimal proportions of “Nganasan-like” ancestry evidence the process of admixture of Volga Finns in Akozino territory through their close interactions with Permians of Ananyino, who in turn acquired this Palaeo-Arctic admixture most likely during the expansion of the linguistic community to hunter-gatherer territories, to the north of the Cis-Urals. This process of stepped infiltration and expansion without language change is not dissimilar to the one seen among Indo-Iranians and Balto-Slavs of hg. R1b, or Vasconic speakers of hg. I2a, although in the case of Baltic Finns of hg. R1a the process of infiltration and expansion of hg. N1c is much less dramatic, with no radical replacement anywhere before the huge bottlenecks observable in Finns.
The expansion of haplogroup N1c among Finnic populations, as we are going to see in samples from the Middle Ages such as Luistari, is the consequence of late founder effects after huge bottlenecks expected based on the analysis of modern populations. The expansion of N1c-VL29 is different in origin from that of N1c-Z1936 among Samic (later integrated into Finnish populations), most likely from the east and originally associated with Lovozero Ware.
In spite of all this, the conclusion of the paper is (surprise!) that Siberian ancestry and hg. N heralded the arrival of Finnic to the Gulf of Finland in the Iron Age… However, this conclusion is supposedly* supported, not by their previous papers, but by a recent phylogenetic study by Honkola et al. (2013), which doesn’t actually argue for such a late ‘arrival’: it argues for the split of Balto-Finnic around 1500 BC.
NOTE. I say ‘supposedly’ because Kristiina Tambets, for example, has been following the link of Uralic with haplogroup N since the 2000s, so this is not some conclusion they just happened to misread from some random paper they Googled. In those initial assessments, she argued that the “ancient homeland” of the Tat C mutation suggested that Finno-Ugrians were in Fennoscandia before Indo-Europeans. Apparently, since haplogroup N appears later and from the east, it is now more important to follow this haplogroup than what is established in archaeology and linguistics.
Even in the referred paper, this split is considered an in situ development, since the phylogenetic study takes the information – among others – 1) from Parpola and Carpelan, who consider Netted Ware, a culture derived from Fatyanovo/Abashevo and Volosovo, as the culprit of the Finno-Ugric expansion; and 2) from Kallio (2006), who clearly states that Proto-Balto-Finnic (like Proto-Finno-Samic) was spoken around the Gulf of Finland during the Bronze Age. Both of them set the terminus ante quem of the language presence in the Baltic ca. 1900 BC.
Anyways, as a consequence of geneticists keeping these untenable pre-ancient DNA haplogroup-based arguments today, I expect to see this “Finnic” language expansion also described for the Western Baltic, Scandinavia or northern Europe, when this same proportion of hg. N1c and “Nganasan” ancestry is observed in Iron Age samples around the Baltic Sea. The nativist trends that this domination of “Finns” all over Northern Europe 2,500 years ago will create will be even more fun to read than the current ones…
EDIT (10 May 2019) How I see the reaction of many to ancient DNA, in keeping their old theories:
So, the Bronze Age results for Iberia I2a, Yamna/BBC R1b, Baikalic/Palaeo-Arctic N1c, and now Corded Ware/Fennoscandia R1a (https://t.co/AKPnqpQHsb …), mean that the simplistic associations of haplogroup-language by geneticists during the pre-ancient DNA era was *right*? Mmmm… pic.twitter.com/pq9tde2QiU
I read from time to time that “we have not sampled Uralic speakers yet”, and “we are waiting to see when Uralic-speaking peoples are sampled”. Are we, though?
Proto-language homelands are based on linguistic data, such as guesstimates for dialectal evolution, loanwords and phonetic changes for language contacts, toponymy for ancient territories, etc. depending on the available information. The trace is then followed back, using available archaeological data, from the known historic speakers and territory to the appropriate potential prehistoric cultures. Only then can genetic analyses help us clarify the precise prehistoric population movements that better fit the models.
The linguistic homeland
We thought – using linguistic guesstimates and fitting prehistoric cultures and their expansion – that Yamna was the Late Proto-Indo-European culture, so when Yamna was sampled, we had Late Proto-Indo-Europeans sampled. Simple deduction.
We thought that north-eastern Europe was a Uralic-speaking area during the Neolithic:
For those supporting a western continuity (and assuming CWC was Indo-European), the language was present at least since the Comb Ware culture, potentially since the Mesolithic.
For those supporting a late introduction into Finland, Uralic expanded the latest with Abashevo-related movements after its incorporation of Volosovo and related hunter-gatherers.
The expansion to the east must have happened through progressive infiltrations with Seima-Turbino / Andronovo-related expansions.
Finding the linguistic homeland going backwards can be described today as follows:
I. Proto-Fennic homeland
Based on the number of Baltic loanwords, not attested in the more eastern Uralic branches (and reaching only partially Mordvinic), the following can be said about western Finno-Permic languages (Junttila 2014):
The Volga-Kama Basin lies still too far east to be included in a list of possible contact locations. Instead, we could look for the contact area somewhere between Estonia in the west and the surroundings of Moscow in the east, a zone with evidence of Uralic settlement in the north and Baltic on the south side.
The only linguistically well-grounded version of the Stone Age continuation theory was presented by Mikko Korhonen in 1976. Its validity, however, became heavily threatened when Koivulehto 1983a-b proved the existence of a Late Proto-Indo-European or Pre-Baltic loanword layer in Saami, Finnic, and Mordvinic. Since this layer must precede the Baltic one and it was presumably acquired in the Baltic Sea region, Koivulehto posited it on the horizon of the Battle Axe period. This forces a later dating for the Baltic–Finnic contacts.
Today the Battle Axe culture is dated at 3200 to 3000 BC, a period far too remote to correspond linguistically with Proto-Baltic (Kallio 1998a).
Since the Baltic contacts began at a very initial phase of Proto-Finnic, the language must have been relatively uniform at that time. Hence, if we consider that the layer of Baltic loanwords may have spread over the Gulf of Finland at that time, we could also insist that the whole of the Proto-Finnic language did so.
II. Proto-Finno-Saamic homeland
The evidence of continued Palaeo-Germanic loanwords (from Pre- to Proto-Germanic stages) is certainly the most important data to locate the Finno-Saamic homeland, and from there backwards into the true Uralic homeland. Following Kallio (2017):
(…) the loanword evidence furthermore suggests that the ancestors of Finnic and Saamic had at least phonologically remained very close to Proto-Uralic as late as the Bronze Age (ca. 1700–500 BC). In particular, certain loanwords, whose Baltic and Germanic sources point to the first millennium BC, after all go back to the Finno-Saamic proto-stage, which is phonologically almost identical to the Uralic proto-stage (see especially the table in Sammallahti 1998: 198–202). This being the case, Dahl’s wave model could perhaps have some use in Uralic linguistics, too.
The presence of Pre-Germanic loanwords points rather to the centuries around the turn of the 2nd – 1st millennium BC or earlier. Proto-Germanic words must have been borrowed before the end of Germanic influence in the eastern Baltic at the beginning of the Iron Age, which sets a clear terminus ante quem ca. 800 BC.
(…) the earliest Indo-European loanwords in the Uralic languages (…) show that Proto-Uralic cannot have been spoken much earlier than Proto-Indo-European dated about 3500 BC (Koivulehto 2001: 235, 257). As the same loanword evidence naturally also shows that the Uralic and Indo-European homelands were not located far from one another, the Uralic homeland can most likely be located in the Middle and Upper Volga region, right north of the Indo-European homeland*. From the beginning of the Subneolithic period about 5900 BC onwards, this region was an important innovation centre, from where several cultural waves spread to the Finnish Gulf area, such as the Sperrings Ware wave about 4900 BC, the Combed Ware wave about 3900 BC, and the Netted Ware wave about 1900 BC (Carpelan & Parpola 2001: 78–90).
The mainstream position is nowadays trying to hold together the traditional views of Corded Ware as Indo-European, and a Uralic Fennoscandia during the Bronze Age.
The following is an example of how this “Volosovo/Forest Zone hunter-gatherer theory” of Uralic origins looks like, as a ‘mixture’ of cultures and languages that benefits from the lack of genetic data for certain regions and periods (taken from Parpola 2018):
The Corded Ware (or Battle Axe) culture intruded into the Eastern Baltic and coastal Finland already around 3100 BCE. The continuity hypothesis maintains that the early Proto-Finnic speakers of the coastal regions, who had come to Finland in the 4th millennium BCE with the Comb-Pitted Ware, coexisted with the Corded Ware newcomers, gradually adopting their pastoral culture and with it a number of NW-IE loanwords, but assimilating the immigrants linguistically.
The fusion of the Corded Ware and the local Comb-Pitted Ware culture resulted into the formation of the Kiukais culture (c. 2300–1500) of southwestern Finland, which around 2300 received some cultural impulses from Estonia, manifested in the appearance of the Western Textile Ceramic (which is different from the more easterly Textile Ceramic or Netted Ware, and which is first attested in Estonia c. 2700 BCE, cf. Kriiska & Tvauri 2007: 88), and supposed to have been accompanied by an influx of loanwords coming from Proto-Baltic. At the same time, the Kiukais culture is supposed to have spread the custom of burying chiefs in stone cairns to Estonia.
The coming of the Corded Ware people and their assimilation created a cultural and supposedly also a linguistic split in Finland, which the continuity hypothesis has interpreted to mean dividing Proto-Saami-Finnic unity into its two branches. Baltic Finnic, or simply Finnic, would have emerged in the coastal regions of Finland and in the northern East Baltic, while preforms of Saami would have been spoken in the inland parts of Finland.
The Nordic Bronze Age culture, correlated above with early Proto-Germanic, exerted a strong influence upon coastal Finland and Estonia 1600–700 BCE. Due to this, the Kiukais culture was transformed into the culture of Paimio ceramics (c. 1600–700 BCE), later continued by Morby ceramics (c. 700 BCE – 200 CE). The assumption is that clear cultural continuity was accompanied by linguistic continuity. Having assimilated the language of the Germanic traders and relatively few settlers of the Bronze Age, the language of coastal Finland is assumed to have reached the stage of Proto-Finnish at the beginning of the Christian era. In Estonia, the Paimio ceramics have a close counterpart in the contemporaneous Asva ceramics.
I will not comment on Siberian or Central Asian homeland proposals, because they are obviously not mainstream, still less today when we know that Uralic was certainly in contact with Proto-Indo-European, and then with Pre- and Proto-Indo-Iranian, as supported even by the Copenhagen group in Damgaard et al. (2018).
This is what Kallio (2017) has to say about the agendas behind such proposals:
Interestingly, the only Uralicists who generally reject the Central Russian homeland are the Russian ones who prefer the Siberian homeland instead. Some Russians even advocate that the Central Russian homeland is only due to Finnish nationalism or, as one of them put it a bit more tactfully, “the political and ideological situation in Finland in the first decades of the 20th century” (Napolskikh 1995: 4).
Still, some Finns (and especially those who also belong to the “school who wants it large and wants it early”) simultaneously advocate that exactly the same Central Russian homeland is due to Finnlandisierung (Wiik 2001: 466).
Hence, for those of you willing to learn about fringe theories not related to North-Eastern Europe, you also have then the large and early version of the Uralic homeland, with Wiik’s Palaeolithic continuity of Uralic peoples spread over all of eastern and central Europe (hence EHG and R1a included):
These fringe Finnish theories look a lot like the Corded Ware expansion… Better not go the Russian or Finnish nationalist ways? Agreed then, let’s discuss only rational proposals based on current data.
The archaeological homeland
For a detailed account of the Corded Ware expansion with Battle Axe, Fatyanovo-Balanovo, and Abashevo groups into the area, you can read my recent post on the origin of R1a-Z645.
1. Textile ceramics
During the 2nd millennium BC, textile impressions appear in pottery as a feature across a wide region, from the Baltic area through the Volga to the Urals, in communities that evolve from late Corded Ware groups without much external influence.
While it has been held that this style represents a north-west expansion from the Volga region (with the “Netted Ware” expansion), there are actually at least two original textile styles, one (earlier) in the Gulf of Finland, common in the Kiukainen pottery, which evolves into the Textile ware culture proper, and another which seems to have an origin in the Middle Volga region to the south-east.
The Netted ware culture is the one that apparently expands into inner Finland – a region not densely occupied by Corded Ware groups until then. There are, however, no clear boundaries between groups of both styles; textile impressions can be easily copied without much interaction or population movement; and the oldest textile ornamentation appeared on the Gulf of Finland. Hence the tradition of naming all as groups of Textile ceramics.
The fact that different adjacent groups from the Gulf of Finland and Forest Zone share similar patterns making it very difficult to differentiate between ‘Netted Ware’ or ‘Textile Ware’ groups points to:
close cultural connections that are maintained through the Gulf of Finland and the Forest Zone after the evolution of late Corded Ware groups; and
no gross population movements in the original Battle Axe / Fatyanovo regions, except for the expansion of Netted Ware to inner Finland, Karelia, and the east, where the scattered Battle Axe finds and worsening climatic conditions suggest most CWC settlements disappeared at the end of the 3rd millennium BC and recovered only later.
NOTE. This lack of population movement – or at least significant replacement by external, non-CWC groups – is confirmed in genetic investigation by continuity of CWC-related lineages (see below).
The technology present in Textile ceramics is in clear contrast to local traditions of sub-Neolithic Lovozero and Pasvik cultures of asbestos-tempered pottery to the north and east, which point to a different tradition of knowledge and learning network – showing partial continuity with previous asbestos ware, since these territories host the main sources of asbestos. We have to assume that these cultures of northern and eastern Fennoscandia represent Palaeo-European (eventually also Palaeo-Siberian) groups clearly differentiated from the south.
The Chirkovo culture (ca. 1800-700 BC) forms on the middle Volga – at roughly the same time as Netted Ware formed to the west – from the fusion of Abashevo and Balanovo elites on Volosovo territory, and is also related (like Abashevo) to materials of the Seima-Turbino phenomenon.
Bronze Age ethnolinguistic groups
In the Gulf of Finland, Kiukainen evolves into the Paimio ceramics (in Finland) — Asva Ware (in Estonia) culture, which lasts from ca. 1600 to ca. 700 BC, probably representing an evolving Finno-Saamic community, while the Netted Ware from inner Finland (the Sarsa and Tomitsa groups) and the groups from the Forest Zone possibly represent a Volga-Finnic community.
NOTE. Nevertheless, the boundaries between Textile ceramic groups are far from clear, and inner Finland Netted Ware groups seem to follow a history different from Netted Ware groups from the Middle and Upper Volga, hence they could possibly be identified as an evolving Pre-Saamic community.
Based on language contacts, with Early Baltic – Early Finnic contacts starting during the Iron Age (ca. 500 BC onwards), this is a potential picture of the situation at the end of this period, when Germanic influence on the coast starts to fade, and Lusatian culture influence is stronger:
The whole Finno-Permic community remains thus in close contact, allowing for the complicated picture that Kallio mentions as potentially showing Dahl’s wave model for Uralic languages.
Genetic data shows a uniform picture of these communities, with exclusively CWC-derived ancestry and haplogroups. So in Mittnik et al. (2018) all Baltic samples show R1a-Z645 subclades, while the recent session on Estonian populations in ISBA 8 (see programme in PDF) clearly states that:
[Of the 24 Bronze Age samples from stone-cist graves] all 18 Bronze Age males belong to R1a.
Regarding non-Uralic substrates found in Saami, supposedly absorbed during the expansion to the north (and thus representing languages spoken in northern Fennoscandia during the Bronze Age) this is what Aikio (2012) has to say:
The Saami substrate in the Finnish dialects thus reveals that also Lakeland Saami languages had a large number of vocabulary items of obscure origin. Most likely many of these words were substrate in Lakeland Saami, too, and ultimately derive from languages spoken in the region before Saami. In some cases the loan origin of these words is obvious due to their secondary Proto-Saami vowel combinations such as *ā–ë in *kāvë ‘bend; small bay’ and *šāpšë ‘whitefish’. This substrate can be called ‘Palaeo-Lakelandic’, in contrast to the ‘Palaeo-Laplandic’ substrate that is prominent in the lexicon of Lapland Saami. As the Lakeland Saami languages became extinct and only fragments of their lexicon can be reconstructed via elements preserved in Finnish place-names and dialectal vocabulary, we are not in a position to actually study the features of this Palaeo-Lakelandic substrate. Its existence, however, appears evident from the material above.
If we wanted to speculate further, based on the data we have now, it is very likely that two opposing groups will be found in the region:
A) The central Finnish group, in this hypothesis the Palaeo-Lakelandic group, made up of the descendants of the Mesolithic pioneers of the Komsa and Suomusjärvi cultures, and thus mainly Baltic HG / Scandinavian HG ancestry and haplogroups I / R1b(xM269) (see more on Scandinavian HG).
B) Lapland and Kola were probably also inhabited by similar Mesolithic populations, until it was eventually assimilated by expanding Siberian groups (of Siberian ancestry and N1c-L392 lineages) from the east – entering the region likely through the Kola peninsula – , forming the Palaeo-Laplandic group, which was in turn later replaced by expanding Proto-Saamic groups.
Siberian ancestry appears first in Fennoscandia at Bolshoy Oleni Ostrov ca. 1520 BC, with haplogroup N1c-L392 (2 samples, BOO002 and BOO004), and with Siberian ancestry. This is their likely movement in north-eastern Europe, from Lamnidis et al (2018):
The large Siberian component in the Bolshoy individuals from the Kola Peninsula provides the earliest direct genetic evidence for an eastern migration into this region. Such contact is well documented in archaeology, with the introduction of asbestos-mixed Lovozero ceramics during the second millenium BC, and the spread of even-based arrowheads in Lapland from 1,900 BCE. Additionally, the nearest counterparts of Vardøy ceramics, appearing in the area around 1,600-1,300 BCE, can be found on the Taymyr peninsula, much further to the east. Finally, the Imiyakhtakhskaya culture from Yakutia spread to the Kola Peninsula during the same period.
Obviously, these groups of asbestos-tempered ware are not connected to the Uralic expansion. From the same paper:
The fact that the Siberian genetic component is consistently shared among Uralic-speaking populations, with the exceptions of Hungarians and the non-Uralic speaking Russians, would make it tempting to equate this component with the spread of Uralic languages in the area. However, such a model may be overly simplistic. First, the presence of the Siberian component on the Kola Peninsula at ca. 4000 yBP predates most linguistic estimates of the spread of Uralic languages to the area. Second, as shown in our analyses, the admixture patterns found in historic and modern Uralic speakers are complex and in fact inconsistent with a single admixture event. Therefore, even if the Siberian genetic component partly spread alongside Uralic languages, it likely presented only an addition to populations carrying this component from earlier.
2. The Early Iron Age
The Ananino culture appears in the Vyatka-Kama area, famed for its metallurgy, with traditions similar to the North Pontic area, by this time developing Pre-Sauromatian traditions. It expanded to the north in the first half of the first millennium BC, remaining in contact with the steppes, as shown by the ‘Scythian’ nature of its material culture.
NOTE. The Ananino culture can be later followed through its zoomorphic styles into Iron Age Pjanoborskoi and Gljadenovskoi cultures, later to Ural-Siberian Middle Age cultures – Itkuska, Ust’-Poluiska, Kulaiska cultures –, which in turn can be related as prototypes of medieval Permian styles.
At the same time as the Ananino culture begins to expand ca. 1000 BC, the Netted Ware tradition from the middle Oka expanded eastwards into the Oka-Vyatka interfluve of the middle Volga region, until then occupied by the Chirkovo culture. Eventually the Akozino or Akhmylovo group (ca. 800-300 BC) emerged from the area, showing a strong cultural influence from the Ananino culture, by that time already expanding into the Cis-Urals region.
The Akozino culture remains nevertheless linked to the western Forest Zone traditions, with long-ranging influences from as far as the Lusatian culture in Poland (in metallurgical techniques), which at this point is also closely related with cultures from Scandinavia (read more on genetics of the Tollense Valley).
Different materials from Akozino reach Fennoscandia late, at the end of the Bronze Age and beginning of the Early Iron Age, precisely when the influence of the Nordic Bronze Age culture on the Gulf of Finland was declining.
This is a period when Textile ceramic cultures in north-eastern Europe evolve into well-armed chiefdom-based groups, with each chiefdom including thousands or tens of thousands, with the main settlements being hill forts, and those in Fennoscandia starting ca. 1000-400 BC.
Mälar-type celts and Ananino-type celts appear simultaneously in Fennoscandia and the Forest Zone, with higher concentrations in south-eastern Sweden (Mälaren) and the Volga-Kama region, supporting the existence of a revived international trade network.
The Paimio—Asva Ware culture evolves (ca. 700-200 BC) into the Morby (in Finland) — Ilmandu syle (in Estonia, Latvia, and Mälaren) culture. The old Paimio—Asva tradition continues side by side with the new one, showing a clear technical continuity with it, but with ornamentation compared to the Early Iron Age cultures of the Upper Volga area. This new south-eastern influence is seen especially in:
Akozino-Mälar axes (ca. 800-500 BC): introduced into the Baltic area in so great numbers – especially south-western Finland, the Åland islands, and the Mälaren area of eastern Sweden – that it is believed to be accompanied by a movement of warrior-traders of the Akozino-Akhmylovo culture, following the waterways that Vikings used more than a thousand years later. Rather than imports, they represent a copy made with local iron sources.
Tarand graves (ca. 500 BC – AD 400): these ‘mortuary houses’ appear in the coastal areas of northern and western Estonia and the islands, at the same time as similar graves in south-western Finland, eastern Sweden, northern Latvia and Courland. Similar burials are found in Akozino-Akhmylovo, with grave goods also from the upper and middle Volga region, while grave goods show continuity with Textile ware.
The use of asbestos increases in mainland Finnish wares with Kjelmøy Ware (ca. 700 BC – AD 300), which replaced the Lovozero Ware; and in the east in inner Finland and Karelia with the Luukonsaari and Sirnihta wares (ca. 700-500 BC – AD 200), where they replaced the previous Sarsa-Tomitsa ceramics.
The Gorodets culture appears during the Scythian period in the forest-steppe zone north and west of the Volga, shows fortified settlements, and there are documented incursions of Gorodets iron makers into the Samara valley, evidenced by deposits of their typical pottery and a bloom or iron in the region.
Iron Age ethnolinguistic groups
According to (Koryakova and Epimakhov 2007):
It is commonly accepted by archaeology, ethnography, and linguistics that the ancestors of the Permian peoples (the Udmurts, Komi-Permians, and Komi-Zyryans) left the sites of Ananyino cultural intercommunity.
Certain innovations shared between Proto-Fennic (identified with the Gulf of Finland) and Proto-Mordvinic (from the Gorodets culture) point to their close contact before the Proto-Fennic expansion, and thus to the identification of Gorodets as Proto-Mordvinic, hence Akozino as Volgaic (Parpola 2018):
the noun paradigms and the form and function of individual cases,
the geminate *mm (foreign to Proto-Uralic before the development of Fennic under Germanic influence) and other non-Uralic consonant clusters.
the change of numeral *luka ‘ten’ with *kümmen.
The presence of loanwords of non-Uralic origin, related to farming and trees, potentially Palaeo-European in nature (hence possibly from Siberian influence in north-eastern Europe).
The introduction of a strongly hierarchical chiefdom system can quickly change the pre-existing social order and lead to a major genetic shift within generations, without a radical change in languages, as shown in Sintashta-Potapovka compared to the preceding Poltavka society (read more about Sintashta).
Fortified settlements in the region represented in part visiting warrior-traders settled through matrimonial relationships with local chiefs, eager to get access to coveted goods and become members of a distribution network that could guarantee them even military assistance. Such a system is also seen synchronously in other cultures of the region, like the Nordic Bronze Age and Lusatian cultures (Parpola 2013).
The most likely situation is that N1c subclades were incorporated from the Circum-Artic region during the Anonino (Permic) expansion to the north, later emerged during the formation of the Akozino group (Volgaic, under Anonino influence), and these subclades in turn infiltrated among the warrior traders that spread all over Fennoscandia and the eastern Baltic (mainly among Fennic, Saamic, Germanic, and Balto-Slavic peoples), during the age of hill forts, creating alliances partially based on exogamy strategies (Parpola 2013).
Over the course of these events, no language change is necessary in any of the cultures involved, since the centre of gravity is on the expanding culture incorporating new lineages:
first on the Middle Volga, when Ananino expands to the north, incorporatinig N1c lineages from the Circum-Artic region.
then with the expansion of the Akozino-Akhmylovo culture into Ananino territory, admixing with part of its population;
then on the Baltic region, when materials are imported from Akozino into Fennoscandia and the eastern Baltic (and vice versa), with local cultures being infiltrated by foreign (Akozino) warrior-traders and their materials;
and later with the different population movements that led eventually to a greater or lesser relevance of N1c in modern Finno-Permic populations.
To argue that this infiltration and later expansion of lineages changed the language in one culture in one of these events seems unlikely. To use this argument of “opposite movement of ethnic and language change” for different successive events, and only on selected regions and cultures (and not those where the greatest genetic and cultural impact is seen, like e.g. Sweden for Akozino materials) is illogical.
NOTE. Notice how I write here about “infiltration” and “lineages”, not “migration” or “populations”. To understand that, see below the next section on autosomal studies to compare Bronze Age, Iron Age, Medieval and Modern Estonians, and see how little the population of Estonia (homeland of Proto-Fennic and partially of Proto-Finno-Saamic) has changed since the Corded Ware migrations, suggesting genetic continuity and thus mostly close inter-regional and intra-regional contacts in the Forest Zone, hence a very limited impact of the absorbed N1c lineages (originally at some point incorporated from the Circum-Artic region). You can also check on the most recent assessment of R1a vs. N1c in modern Uralic populations.
Iron Age and later populations
From the session on Estonian samples on ISBA 8, by Tambets et al.:
[Of the 13 samples from the Iron Age tarand-graves] We found that the Iron Age individuals do in fact carry chrY hg N3 (…) Furthermore, based on their autosomal data, all of the studied individuals appear closer to hunter-gatherers and modern Estonians than Estonian CWC individuals do.
EDIT (16 OCT) A recent abstract with Saag as main author (Tambets second) cites 3 out of 5 sampled Iron Age individuals as having haplogroup N3.
EDIT (28 OCT): Notice also the appearance of N1a1a1a1a1a1a1-L1025 in Lithuania (ca. 300 AD), from Damgaard (Nature 2018); the N1c sample of the Krivichi Pskov Long Barrows culture (ca. 8th-10th c. AD), and N1a1a1a1a1a1a7-Y4341 among late Vikings from Sigtuna (ca. 10th-12th c. AD) in Krzewinska (2018).
Looking at the plot, the genetic inflow marking the change from the Bronze Age to the Iron Age looks like an obvious expansion of nearby peoples with CWC-related ancestry, i.e. likely from the south-east, near the Middle Volga, where influence of steppe peoples is greater (hence likely Akozino) into a Proto-Fennic population already admixed (since the arrival of Corded Ware groups) with Comb Ware-like populations.
All of these groups were probably R1a-Z645 (likely R1a-Z283) since the expansion of Corded Ware peoples, with an introduction of some N1c lineages precisely during this Iron Age period. This infiltration of N1c-L392 with Akozino is obviously not directly related to Siberian cultures, given what we know about the autosomal description of Estonian samples.
Rather, N1c-L392 lineages were likely part of the incoming (Volgaic) Akozino warrior-traders, who settled among developing chiefdoms based on hill fort settlements of cultures all over the Baltic area, and began to appear thus in some of the new tarand graves associated with the Iron Age in north-eastern Europe.f
A good way to look at this is to realize that no new cluster appears compared to the data we already have from Baltic LN and BA samples from Mittnik et al. (2018), so the Estonian BA and IA clusters must be located (in a proper PCA) in the cline from Pit-Comb Ware culture through Baltic BA to Corded Ware groups:
This genetic continuity from Corded Ware (the most likely Proto-Uralic homeland) to the Proto-Fennic and Proto-Saamic communities in the Gulf of Finland correlates very well with the known conservatism of Finno-Saamic phonology, quite similar to Finno-Ugric, and both to Proto-Uralic (Kallio 2017): The most isolated region after the expansion of Corded Ware peoples, the Gulf of Finland, shielded against migrations for almost 1,500 years, is then the most conservative – until the arrival of Akozino influence.
Only later would certain regions (like Finland or Lappland) suffer Y-DNA bottlenecks and further admixture events associated with population displacements and expansions, such as the spread of Fennic peoples from their Estonian homeland (evidenced by the earlier separation of South Estonian) to the north and east:
The initial Proto-Fennic expansion was probably coupled with the expansion of Proto-Saami to the north, with the Kjelmøy Ware absorbing the Siberian population of Lovozero Ware, and potentially in inner Finland and Karelia with the Luukonsaari and Sirnihta wares (Carpelan and Parpola 2017).
This Proto-Saami population expansion from the mainland to the north, admixing with Lovozero-related peoples, is clearly reflected in the late Iron Age Saamic samples from Levänluhta (ca. 400-800 AD), as a shift (of 2 out of 3 samples) to Siberian-like ancestry from their original CWC_Baltic-like situation (see PCA from Lamnidis et al. 2018 above).
Also, Volgaic and Permic populations from inner Finland and the Forest Zone to the Cis-Urals and Circum-Artic regions probably incorporate Siberian ancestry and N1c-L392 lineages during these and later population movements, while the westernmost populations – Estonian, Mordvinic – remain less admixed (see PCA from Tambets et al. 2018 below).
We also have data of N1c-L392 in Nordic territory in the Middle Ages, proving its likely strong presence in the Mälaren area since the Iron Age, with the arrival of Akozino warrior traders. Similarly, it is found among Balto-Slavic groups along the eastern Baltic area. Obviously, no language change is seen in Nordic Bronze Age and Lusatian territory, and none is expected in Estonian or Finnish territory, either.
Therefore, no “N1c-L392 + Siberian ancestry” can be seen expanding Finno-Ugric dialects, but rather different infiltrations and population movements with limited effects on ancestry and Y-DNA composition, depending on the specific period and region.
An issue never resolved
Because N1c-L392 subclades & Siberian ancestry, which appear in different proportions and with different origins among some modern Uralic peoples, do not appear in cultures supposed to host Uralic-speaking populations until the Iron Age, people keep looking into any direction to find the ‘true’ homeland of those ‘Uralic N1c peoples’? Kind of a full circular reasoning, anyone? The same is valid for R1a & steppe ancestry being followed for ‘Indo-Europeans’, or R1b-P312 & Neolithic farmer ancestry being traced for ‘Basques’, because of their distribution in modern populations.
I understand the caution of many pointing to the need to wait and see how samples after 2000 BC are like, in every single period, from the middle and upper Volga, Kama, southern Finland, and the Forest Zone between Fennoscandia and the steppe. It’s like waiting to see how people from Western Yamna and the Carpathian Basin after 3000 BC look like, to fill in what is lacking between East Yamna and Bell Beakers, and then between them and every single Late PIE dialect.
But the answer for Yamna-Bell Beaker-Poltavka peoples during the Late PIE expansion is always going to be “R1b-L23, but with R1a-Z645 nearby” (we already have a pretty good idea about that); and the answer for the Forest Zone and northern Cis- and Trans-Urals area – during the time when Uralic languages are known to have already been spoken there – is always going to be “R1a-Z645, but with haplogroup N nearby”, as is already clear from the data on the eastern Baltic region.
So, without a previously proposed model as to where those amateurs expressing concern about ‘not having enough data’ expect to find those ‘Uralic peoples’, all this waiting for the right data looks more like a waiting for N1c and Siberian ancestry to pop up somewhere in the historic Uralic-speaking area, to be able to say “There! A Uralic-speaking male!”. Not a very reasonable framework to deal with prehistoric peoples and their languages, I should think.
But, for those who want to do that, let me break the news to you already:
And here it is, an appropriate fantasy description of the ethnolinguistic groups from the region. You are welcome:
During the Bronze Age, late Corded Ware groups evolve as the western Textile ware Fennic Balto-Slavic group in the Gulf of Finland; the Netted Ware Saamic Balto-Slavic group of inner Finland; the south Netted Ware / Akozino Volgaic Balto-Slavic groups of the Middle Volga; and the Anonino Permic Balto-Slavic group in the north-eastern Forest Zone; all developing still in close contact with each other, allowing for common traits to permeate dialects.
These Balto-Slavic groups would then incorporate west of the Urals during and after the Iron Age (ca. 800-500 BC first, and also later during their expansion to the north) limited ancestry and lineages from eastern European hunter-gatherer groups of Palaeo-European Fennic and Palaeo-Siberian Volgaic and Permic languages from the Circum-Artic region, but they adopted nevertheless the language of the newcomers in every single infiltration of N1c lineages and/or admixture with Siberian ancestry. Oh and don’t forget the Saamic peoples from central Sweden, of course, the famous N1c-L392 ‘Rurikid’ lineages expanding Saamic to the north and replacing Proto-Germanic…
The current model for those obsessed with modern Y-DNA is, therefore, that expanding Neolithic, Bronze Age and Iron Age cultures from north-eastern Europe adopted the languages of certain lineages originally from sub-Neolithic (Scandinavian and Siberian) hunter-gatherer populations of the Circum-Artic region; lineages that these cultures incorporated unevenly during their expansions. Hmmmm… Sounds like an inverse Western movie, where expanding Americans end up speaking Apache, and the eastern coast speaks Spanish until Italian migrants arrive and make everyone speak English… or something. A logic, no-nonsense approach to ethnolinguistic identification.
I kid you not, this is the kind of models we are going to see very soon. In 2018 and 2019, with ancient DNA able to confirm or reject archaeological hypotheses based on linguistic data, people will keep instead creating new pet theories to support preconceived ideas based on the Y-DNA prevalent among modern populations. That is, information available in the 2000s.
So what’s (so much published) ancient DNA useful for, exactly?
[Next post on the subject: Corded Ware—Uralic (III): Seima-Turbino and the Ugric and Samoyedic expansion]
Experienced researchers, particularly those interested in population structure and historical inference, typically present STRUCTURE results alongside other methods that make different modelling assumptions. These include TreeMix, ADMIXTUREGRAPH, fineSTRUCTURE, GLOBETROTTER, f3 and D statistics, amongst many others. These models can be used both to probe whether assumptions of the model are likely to hold and to validate specific features of the results. Each also comes with its own pitfalls and difficulties of interpretation. It is not obvious that any single approach represents a direct replacement as a data summary tool. Here we build more directly on the results of STRUCTURE/ADMIXTURE by developing a new approach, badMIXTURE, to examine which features of the data are poorly fit by the model. Rather than intending to replace more specific or sophisticated analyses, we hope to encourage their use by making the limitations of the initial analysis clearer.
The default interpretation protocol
Most researchers are cautious but literal in their interpretation of STRUCTURE and ADMIXTURE results, as caricatured in Fig. 1, as it is difficult to interpret the results at all without making several of these assumptions. Here we use simulated and real data to illustrate how following this protocol can lead to inference of false histories, and how badMIXTURE can be used to examine model fit and avoid common pitfalls.
STRUCTURE and ADMIXTURE are popular because they give the user a broad-brush view of variation in genetic data, while allowing the possibility of zooming down on details about specific individuals or labelled groups. Unfortunately it is rarely the case that sampled data follows a simple history comprising a differentiation phase followed by a mixture phase, as assumed in an ADMIXTURE model and highlighted by case study 1. Naïve inferences based on this model (the Protocol of Fig. 1) can be misleading if sampling strategy or the inferred value of the number of populations K is inappropriate, or if recent bottlenecks or unobserved ancient structure appear in the data. It is therefore useful when interpreting the results obtained from real data to think of STRUCTURE and ADMIXTURE as algorithms that parsimoniously explain variation between individuals rather than as parametric models of divergence and admixture.
For example, if admixture events or genetic drift affect all members of the sample equally, then there is no variation between individuals for the model to explain. Non-African humans have a few percent Neanderthal ancestry, but this is invisible to STRUCTURE or ADMIXTURE since it does not result in differences in ancestry profiles between individuals. The same reasoning helps to explain why for most data sets—even in species such as humans where mixing is commonplace—each of the K populations is inferred by STRUCTURE/ADMIXTURE to have non-admixed representatives in the sample. If every individual in a group is in fact admixed, then (with some exceptions) the model simply shifts the allele frequencies of the inferred ancestral population to reflect the fraction of admixture that is shared by all individuals.
Several methods have been developed to estimate K, but for real data, the assumption that there is a true value is always incorrect; the question rather being whether the model is a good enough approximation to be practically useful. First, there may be close relatives in the sample which violates model assumptions. Second, there might be “isolation by distance”, meaning that there are no discrete populations at all. Third, population structure may be hierarchical, with subtle subdivisions nested within diverged groups. This kind of structure can be hard for the algorithms to detect and can lead to underestimation of K. Fourth, population structure may be fluid between historical epochs, with multiple events and structures leaving signals in the data. Many users examine the results of multiple K simultaneously but this makes interpretation more complex, especially because it makes it easier for users to find support for preconceptions about the data somewhere in the results.
In practice, the best that can be expected is that the algorithms choose the smallest number of ancestral populations that can explain the most salient variation in the data. Unless the demographic history of the sample is particularly simple, the value of K inferred according to any statistically sensible criterion is likely to be smaller than the number of distinct drift events that have practically impacted the sample. The algorithm uses variation in admixture proportions between individuals to approximately mimic the effect of more than K distinct drift events without estimating ancestral populations corresponding to each one. In other words, an admixture model is almost always “wrong” (Assumption 2 of the Core protocol, Fig. 1) and should not be interpreted without examining whether this lack of fit matters for a given question.
Because STRUCTURE/ADMIXTURE accounts for the most salient variation, results are greatly affected by sample size in common with other methods. Specifically, groups that contain fewer samples or have undergone little population-specific drift of their own are likely to be fit as mixes of multiple drifted groups, rather than assigned to their own ancestral population. Indeed, if an ancient sample is put into a data set of modern individuals, the ancient sample is typically represented as an admixture of the modern populations (e.g., ref. 28,29), which can happen even if the individual sample is older than the split date of the modern populations and thus cannot be admixed.
This paper was already available as a preprint in bioRxiv (first published in 2016) and it is incredible that it needed to wait all this time to be published. I found it weird how reviewers focused on the “tone” of the paper. I think it is great to see files from the peer review process published, but we need to know who these reviewers were, to understand their whiny remarks… A lot of geneticists out there need to develop a thick skin, or else we are going to see more and more delays based on a perceived incorrect tone towards the field, which seems a rather subjective reason to force researchers to correct a paper.
A potential hindrance to our advice to upgrade from PCA graphs to PCA biplots is that the SNPs are often so numerous that they would obscure the Items if both were graphed together. One way to reduce clutter, which is used in several figures in this article, is to present a biplot in two side-by-side panels, one for Items and one for SNPs. Another stratagem is to focus on a manageable subset of SNPs of particular interest and show only them in a biplot in order to avoid obscuring the Items. A later section on causal exploration by current methods mentions several procedures for identifying particularly relevant SNPs.
One of several data transformations is ordinarily applied to SNP data prior to PCA computations, such as centering by SNPs. These transformations make a huge difference in the appearance of PCA graphs or biplots. A SNPs-by-Items data matrix constitutes a two-way factorial design, so analysis of variance (ANOVA) recognizes three sources of variation: SNP main effects, Item main effects, and SNP-by-Item (S×I) interaction effects. Double-Centered PCA (DC-PCA) removes both main effects in order to focus on the remaining S×I interaction effects. The resulting PCs are called interaction principal components (IPCs), and are denoted by IPC1, IPC2, and so on. By way of preview, a later section on PCA variants argues that DC-PCA is best for SNP data. Surprisingly, our literature survey did not encounter even a single analysis identified as DC-PCA.
The axes in PCA graphs or biplots are often scaled to obtain a convenient shape, but actually the axes should have the same scale for many reasons emphasized recently by Malik and Piepho . However, our literature survey found a correct ratio of 1 in only 10% of the articles, a slightly faulty ratio of the larger scale over the shorter scale within 1.1 in 12%, and a substantially faulty ratio above 2 in 16% with the worst cases being ratios of 31 and 44. Especially when the scale along one PCA axis is stretched by a factor of 2 or more relative to the other axis, the relationships among various points or clusters of points are distorted and easily misinterpreted. Also, 7% of the articles failed to show the scale on one or both PCA axes, which leaves readers with an impressionistic graph that cannot be reproduced without effort. The contemporary literature on PCA of SNP data mostly violates the prohibition against stretching axes.
The percentage of variation captured by each PC is often included in the axis labels of PCA graphs or biplots. In general this information is worth including, but there are two qualifications. First, these percentages need to be interpreted relative to the size of the data matrix because large datasets can capture a small percentage and yet still be effective. For example, for a large dataset with over 107,000 SNPs for over 6,000 persons, the first two components capture only 0.3693% and 0.117% of the variation, and yet the PCA graph shows clear structure (Fig 1A in ). Contrariwise, a PCA graph could capture a large percentage of the total variation, even 50% or more, but that would not guarantee that it will show evident structure in the data. Second, the interpretation of these percentages depends on exactly how the PCA analysis was conducted, as explained in a later section on PCA variants. Readers cannot meaningfully interpret the percentages of variation captured by PCA axes when authors fail to communicate which variant of PCA was used.
Five simple recommendations for effective PCA analysis of SNP data emerge from this investigation.
Use the SNP coding 1 for the rare or minor allele and 0 for the common or major allele.
Use DC-PCA; for any other PCA variant, examine its augmented ANOVA table.
Report which SNP coding and PCA variant were selected, as required by contemporary standards in science for transparency and reproducibility, so that readers can interpret PCA results properly and reproduce PCA analyses reliably.
Produce PCA biplots of both Items and SNPs, rather than merely PCA graphs of only Items, in order to display the joint structure of Items and SNPs and thereby to facilitate causal explanations. Be aware of the arch distortion when interpreting PCA graphs or biplots.
Produce PCA biplots and graphs that have the same scale on every axis.
I read the referenced paper Biplots: Do Not Stretch Them!, by Malik and Piepho (2018), and even though it is not directly applicable to the most commonly available PCA graphs out there, it is a good reminder of the distorting effects of stretching. So for example quite recently in Krause-Kyora et al. (2018), where you can see Corded Ware and BBC samples from Central Europe clustering with samples from Yamna:
NOTE. This is related to a vertical distorsion (i.e. horizontal stretching), but possibly also to the addition of some distant outlier sample/s.
The so-called ‘Yamnaya’ ancestry
Every time I read papers like these, I remember commenters who kept swearing that genetics was the ultimate science that would solve anthropological problems, where unscientific archaeology and linguistics could not. Well, it seems that, like radiocarbon analysis, these promising developing methods need still a lot of refinement to achieve something meaningful, and that they mean nothing without traditional linguistics and archaeology… But we already knew that.
Also, if this is happening in most peer-reviewed publications, made by professional geneticists, in journals of high impact factor, you can only wonder how many more errors and misinterpretations can be found in the obscure market of so many amateur geneticists out there. Because amateur geneticist is a commonly used misnomer for people who are not geneticists (since they don’t have the most basic education in genetics), and some of them are not even ‘amateurs’ (because they are selling the outputs of bioinformatic tools)… It’s like calling healers ‘amateur doctors’.
NOTE. While everyone involved in population genetics is interested in knowing the truth, and we all have our confirmation (and other kinds of) biases, for those who get paid to tell people what they want to hear, and who have sold lots of wrong interpretations already, the incentives of ‘being right’ – and thus getting involved in crooked and paranoid behaviour regarding different interpretations – are as strong as the money they can win or loose by promoting themselves and selling more ‘product’.
As a reminder of how badly these wrong interpretations of genetic results – and the influence of the so-called ‘amateurs’ – can reflect on research groups, yet another turn of the screw by the Copenhagen group, in the oral presentations at Languages and migrations in pre-historic Europe (7-12 Aug 2018), organized by the Copenhagen University. The common theme seems to be that Bell Beaker and thus R1b-L23 subclades do represent a direct expansion from Yamna now, as opposed to being derived from Corded Ware migrants, as they supported before.
NOTE. Yes, the “Yamna → Corded Ware → Únětice / Bell Beaker” migration model is still commonplace in the Copenhagen workgroup. Yes, in 2018. Guus Kroonen had already admitted they were wrong, and it was already changed in the graphic representation accompanying a recent interview to Willerslev. However, since there is still no official retraction by anyone, it seems that each member has to reject the previous model in their own way, and at their own pace. I don’t think we can expect anyone at this point to accept responsibility for their wrong statements.
I love the newly invented arrows of migration from Yamna to the north to distinguish among dialects attributed by them to CWC groups, and the intensive use of materials from Heyd’s publications in the presentation, which means they understand he was right – except for the fact that they are used to support a completely different theory, radically opposed to those defended in Heyd’s model…
Now added to the Copenhagen’s unending proposals of language expansions, some pearls from the oral presentation:
Corded Ware north of the Carpathians of R1a lineages developed Germanic;
R1b borugh [?] Italo-Celtic;
the increase in steppe ancestry on north European Bell Beakers mean that they “were a continuation of the Yamnaya/Corded Ware expansion”;
“Corded Ware groups  stopped their expansion and took over the Bell Beaker package before migrating to England” [yep, it literally says that];
Italo-Celtic expanded to the UK and Iberia with Bell Beakers [I guess that included Lusitanian in Iberia, but not Messapian in Italy; or the opposite; or nothing like that, who knows];
2nd millennium BC Bronze Age Atlantic trade systems expanded Proto-Celtic [yep, trade systems expanded the language]
1st millennium BC expanded Gaulish with La Tène, including a “Gaulish version of Celtic to Ireland/UK” [hmmm, datBritish Gaulish indeed].
You know, because, why the hell not? A logical, stable, consequential, no-nonsense approach to Indo-European migrations, as always.
Also, compare still more invented arrows of migrations, from Mikkel Nørtoft’s Introducing the Homeland Timeline Map, going against Kristiansen’s multiple arrows, and even against the own recent fantasy map series in showing Bell Beakers stem from Yamna instead of CWC (or not, you never truly know what arrows actually mean):
I really, really loved that perennial arrow of migration from Volosovo, ca. 4000-800 BC (3000+ years, no less!), representing Uralic?, like that, without specifics – which is like saying, “somebody from the eastern forest zone, somehow, at some time, expanded something that was not Indo-European to Finland, and we couldn’t care less, except for the fact that they were certainly not R1a“.
This and Kristiansen’s arrows are the most comical invented migration routes of 2018; and that is saying something, given the dozens of similar maps that people publish in forums and blogs each week.
It’s hard to accept that this is a series of presentations made by professional linguists, archaeologists, and geneticists, as stated by the official website, and still harder to imagine that they collaborate within the same professional workgroup, which includes experienced geneticists and academics.
I propose the following video to close future presentations introducing innovative ideas like those above, to help the audience find the appropriate mood: