Interesting excerpts, referring mainly to Uralic peoples (emphasis mine):
A model-based clustering analysis using ADMIXTURE shows a similar pattern (Fig. 2b and Supplementary Fig. 3). Overall, the proportions of ancestry components associated with Eastern or Western Eurasians are well correlated with longitude in inner Eurasians (Fig. 3). Notable outliers include known historical migrants such as Kalmyks, Nogais and Dungans. The Uralic- and Yeniseian-speaking populations, as well as Russians from multiple locations, derive most of their Eastern Eurasian ancestry from a component most enriched in Nganasans, while Turkic/Mongolic speakers have this component together with another component most enriched in populations from the Russian Far East, such as Ulchi and Nivkh (Supplementary Fig. 3). Turkic/Mongolic speakers comprising the bottom-most cline have a distinct Western Eurasian ancestry profile: they have a high proportion of a component most enriched in Mesolithic Caucasus hunter-gatherers and Neolithic Iranians and frequently harbour another component enriched in present-day South Asians (Supplementary Fig. 4). Based on the PCA and ADMIXTURE results, we heuristically assigned inner Eurasians to three clines: the ‘forest-tundra’ cline includes Russians and all Uralic and Yeniseian speakers; the ‘steppe-forest’ cline includes Turkic- and Mongolic-speaking populations from the Volga and Altai–Sayan regions and Southern Siberia; and the ‘southern steppe’ cline includes the rest of the populations.
For the forest-tundra populations, the Nganasan + Srubnaya model is adequate only for the two Volga region populations, Udmurts and Besermyans (Fig. 5 and Supplementary Table 8).
For the other populations west of the Urals, six from the northeastern corner of Europe are modelled with additional Mesolithic Western European hunter-gatherer (WHG) contribution (8.2–11.4%; Supplementary Table 8), while the rest need both WHG and early Neolithic European farmers (LBK_EN; Supplementary Table 2). Nganasan-related ancestry substantially contributes to their gene pools and cannot be removed from the model without a significant decrease in the model fit (4.1–29.0% contribution; χ2 P ≤ 1.68 × 10−5; Supplementary Table 8).
NOTE. It doesn’t seem like Hungarians can be easily modelled with Nganasan ancestry, though…
For the 4 populations east of the Urals (Enets, Selkups, Kets and Mansi), for which the above models are not adequate, Nganasan + Srubnaya + AG3 provides a good fit (χ2 P ≥ 0.018; Fig. 5 and Supplementary Table 8). Using early Bronze Age populations from the Baikal Lake region (‘Baikal_EBA’; Supplementary Table 2) as a reference instead of Nganasan, the two-way model of Baikal_EBA + Srubnaya provides a reasonable fit (χ2 P ≥ 0.016; Supplementary Table 8) and the three-way model of Baikal_EBA + Srubnaya + AG3 is adequate but with negative AG3 contribution for Enets and Mansi (χ2 P ≥ 0.460; Supplementary Table 8).
Bronze/Iron Age populations from Southern Siberia also show a similar ancestry composition with high ANE affinity (Supplementary Table 9). The additional ANE contribution beyond the Nganasan + Srubnaya model suggests a legacy from ANE-ancestry-rich clines before the Late Bronze Age.
Even among the earliest available inner Eurasian genomes, east–west connectivity is evident. These, too, form a longitudinal cline, characterized by the easterly increase of a distinct ancestry, labelled Ancient North Eurasian (ANE), lowest in western European hunter-gatherers (WHG) and highest in Palaeolithic Siberians from the Baikal region. Flow-through from this ANE cline is seen in steppe populations until at least the Bronze Age, including the world’s earliest known horse herders — the Botai. However, this is eroded over time by migration from west and east, following agricultural adoption on the continental peripheries (Fig. 1b,c).
Strikingly, Jeong et al. model the modern upper steppe cline as a simple two-way mixture between western Late Bronze Age herders and Northeast Asians (Fig. 1c), with no detectable residue from the older ANE cline. They propose modern steppe peoples were established mainly through migrations post-dating the Bronze Age, a sequence for which has been recently outlined using ancient genomes. In contrast, they confirm a substantial ANE legacy in modern Siberians of the northernmost cline, a pattern mirrored in excesses of WHG ancestry west of the Urals (Fig. 1b). This marks the inhospitable biome as a reservoir for older lineages, an indication that longstanding barriers to latitudinal movement may indeed be at work, reducing the penetrance of gene flows further south along the steppe.
Given the findings as reported in the paper, I think it should be much easier to describe different subclines in the “northernmost cline” than in the much more recent “Turkic/Mongolic cline”, which is nevertheless subdivided in this paper in two clines. As an example, there are at least two obvious clines with “Nganasan-related meta-populations” among Uralians, which converge in a common Steppe MLBA (i.e. Corded Ware) ancestry – one with Palaeo-Laplandic peoples, and another one with different Palaeo-Siberian populations:
The inclusion of certain Eurasian groups (or lack thereof) in the PCA doesn’t help to distinguish these subclines visually, and I guess the tiny “Naganasan-related” ancestral components found in some western populations (e.g. the famous ~5% among Estonians) probably don’t lend themselves easily to further subdivisions. Notice, nevertheless, the different components of the Eastern Eurasian source populations among Finno-Ugrians:
Also remarkable is the lack of comparison of Uralic populations with other neighbouring ones, since the described Uralic-like ancestry of Russians was already known, and is most likely due to the recent acculturation of Uralic-speaking peoples in the cradle of Russians, right before their eastward expansions.
A comparison of Estonians and Finns with Balts, Scandinavians, and Eastern Europeans would have been more informative for the division of the different so-called “Nganasan-like meta-populations”, and to ascertain which one of these ancestral peoples along the ancient WHG:ANE cline could actually be connected (if at all) to the Cis-Urals.
Because, after all, based on linguistics and archaeology, geneticists are not supposed to be looking for populations from the North Asian Arctic region, for “Siberian ancestry”, or for haplogroup N1c – despite previous works by their peers – , but for the Bronze Age Volga-Kama region…
An interesting aspect of the paper, hidden among so many relevant details, is a clearer picture of how the so-called Yamnaya or steppe ancestry evolved from Samara hunter-gatherers to Yamna nomadic pastoralists, and how this ancestry appeared among Proto-Corded Ware populations.
Please note: arrows of “ancestry movement” in the following PCAs do not necessarily represent physical population movements, or even ethnolinguistic change. To avoid misinterpretations, I have depicted arrows with Y-DNA haplogroup migrations to represent the most likely true ethnolinguistic movements. Admixture graphics shown are from Wang et al. (2018), and also (the K12) from Mathieson et al. (2018).
1. Samara to Early Khvalynsk
The so-called steppe ancestry was born during the Khvalynsk expansion through the steppes, probably through exogamy of expanding elite clans (eventually all R1b-M269 lineages) originally of Samara_HG ancestry. The nearest group to the ANE-like ghost population with which Samara hunter-gatherers admixed is represented by the Steppe_Eneolithic / Steppe_Maykop cluster (from the Northern Caucasus Piedmont).
Steppe_Eneolithic samples, of R1b1 lineages, are probably expanded Khvalynsk peoples, showing thus a proximate ancestry of an Early Eneolithic ghost population of the Northern Caucasus. Steppe_Maykop samples represent a later replacement of this Steppe_Eneolithic population – and/or a similar population with further contribution of ANE-like ancestry – in the area some 1,000 years later.
This is what Steppe_Maykop looks like, different from Steppe_Eneolithic:
NOTE. This admixture shows how different Steppe_Maykop is from Steppe_Eneolithic, but in the different supervised ADMIXTURE graphics below Maykop_Eneolithic is roughly equivalent to Eneolithic_Steppe (see orange arrow in ADMIXTURE graphic above). This is useful for a simplified analysis, but actual differences between Khvalynsk, Sredni Stog, Afanasevo, Yamna and Corded Ware are probably underestimated in the analyses below, and will become clearer in the future when more ancestral hunter-gatherer populations are added to the analysis.
2. Early Khvalynsk expansion
We have direct data of Khvalynsk-Novodanilovka-like populations thanks to Khvalynsk and Steppe_Eneolithic samples (although I’ve used the latter above to represent the ghost Caucasus population with which Samara_HG admixed).
We also have indirect data. First, there is the PCA with outliers:
Second, we have data from north Pontic Ukraine_Eneolithic samples (see next section).
Third, there is the continuity of late Repin / Afanasevo with Steppe_Eneolithic (see below).
3. Proto-Corded Ware expansion
It is unclear if R1a-M459 subclades were continuously in the steppe and resurged after the Khvalynsk expansion, or (the most likely option) they came from the forested region of the Upper Dnieper area, possibly from previous expansions there with hunter-gatherer pottery.
Supporting the latter is the millennia-long continuity of R1b-V88 and I2a2 subclades in the north Pontic Mesolithic, Neolithic, and Early Eneolithic Sredni Stog culture, until ca. 4500 BC (and even later, during the second half).
Only at the end of the Early Eneolithic with the disappearance of Novodanilovka (and beginning of the steppe ‘hiatus’ of Rassamakin) is R1a to be found in Ukraine again (after disappearing from the record some 2,000 years earlier), related to complex population movements in the north Pontic area.
NOTE. In the PCA, a tentative position of Novodanilovka closer to Anatolia_Neolithic / Dzudzuana ancestry is selected, based on the apparent cline formed by Ukraine_Eneolithic samples, and on the position and ancestry of Sredni Stog, Yamna, and Corded Ware later. A good alternative would be to place Novodanilovka still closer to the Balkan outliers (i.e. Suvorovo), and a source closer to EHG as the ancestry driven by the migration of R1a-M417.
The first sample with steppe ancestry appears only after 4250 BC in the forest-steppe, centuries after the samples with steppe ancestry from the Northern Caucasus and the Balkans, which points to exogamy of expanding R1a-M417 lineages with the remnants of the Novodanilovka population.
4. Repin / Early Yamna expansion
We don’t have direct data on early Repin settlers. But we do have a very close representative: Afanasevo, a population we know comes directly from the Repin/late Khvalynsk expansion ca. 3500/3300 BC (just before the emergence of Early Yamna), and which shows fully Steppe_Eneolithic-like ancestry.
Compared to this eastern Repin expansion that gave Afanasevo, the late Repin expansion to the west ca. 3300 BC that gave rise to the Yamna culture was one of colonization, evidenced by the admixture with north Pontic (Sredni Stog-like) populations, no doubt through exogamy:
This admixture is also found (in lesser proportion) in east Yamna groups, which supports the high mobility and exogamy practices among western and eastern Yamna clans, not only with locals:
We don’t have a comparison with Ukraine_Eneolithic or Corded Ware samples in Wang et al. (2018), but we do have proximate sources for Abashevo, when compared to the Poltavka population (with which it admixed in the Volga-Ural steppes): Sintashta, Potapovka, Srubna (with further Abashevo contribution), and Andronovo:
The two CWC outliers from the Baltic show what I thought was an admixture with Yamna. However, given the previous mixture of Eneolithic_Steppe in north Pontic steppe-forest populations, this elevated “steppe ancestry” found in Baltic_LN (similar to west Yamna) seems rather an admixture of Baltic sub-Neolithic peoples with a north Pontic Eneolithic_Steppe-like population. Late Repin settlers also admixed with a similar population during its colonization of the north Pontic area, hence the Baltic_LN – west Yamna similarities.
NOTE. A direct admixture with west Yamna populations through exogamy by the ancestors of this Baltic population cannot be ruled out yet (without direct access to more samples), though, because of the contacts of Corded Ware with west Yamna settlers in the forest-steppe regions.
A similar case is found in the Yamna outlier from Mednikarovo south of the Danube. It would be absurd to think that Yamna from the Balkans comes from Corded Ware (or vice versa), just because the former is closer in the PCA to the latter than other Yamna samples. The same error is also found e.g. in the Corded Ware → Bell Beaker theory, because of their proximity in the PCA and their shared “steppe ancestry”. All those theories have been proven already wrong.
NOTE. A similar fallacy is found in potential Sintashta→Mycenaean connections, where we should distinguish statistically that result from an East/West Yamna + Balkans_BA admixture. In fact, genetic links of Mycenaeans with west Yamna settlers prove this (there are some related analyses in Anthrogenica, but the site is down at this moment). To try to relate these two populations (separated more than 1,000 years before Sintashta) is like comparing ancient populations to modern ones, without the intermediate samples to trace the real anthropological trail of what is found…Pure numbers and wishful thinking.
Interesting excerpts (emphasis mine; most internal references removed):
The earliest, most secure archaeological evidence of human occupation of the region comes from the artefact-rich, high-latitude (~70° N) Yana RHS site dated to ~31.6 kya (…)
The Yana RHS human remains represent the earliest direct evidence of human presence in northeastern Siberia, a population we refer to as “Ancient North Siberians” (ANS). Both Yana RHS individuals were unrelated males, and belong to mitochondrial haplogroup U, predominant among ancient West Eurasian hunter-gatherers, and to Y chromosome haplogroup P1, ancestral to haplogroups Q and R, which are widespread among present-day Eurasians and Native Americans.
Symmetry tests using f4 statistics reject tree-like clade relationships with both Early West Eurasians (EWE; Sunghir) and Early East Asians (EEA; Tianyuan); however, Yana is genetically closer to EWE, despite its geographic location in northeastern Siberia
Using admixture graphs (qpGraph) and outgroup-based estimation of mixture proportions (qpAdm), we find that Yana can be modelled as EWE with ~25% contribution from EEA
Among all ancient individuals, Yana shares the most genetic drift with Mal’ta, and f4 statistics show that Mal’ta shares more alleles with Yana than with EWE (e.g. f4(Mbuti,Mal’ta;Sunghir,Yana) = 0.0019, Z = 3.99). Mal’ta and Yana also exhibit a similar pattern of genetic affinities to both EWE and EEA, consistent with previous studies.The ANE lineage can thus be considered a descendant of the ANS lineage, demonstrating that by 31.6 kya early representatives of this lineage were widespread across northern Eurasia, including far northeastern Siberia.
(…) the 9.8 kya Kolyma1 individual, representing a group we term “Ancient Paleosiberians” (AP). Our results indicate that AP are derived from a first major genetic shift observed in the region. Principal component analysis (PCA), outgroup f3-statistics and mtDNA and Y chromosome haplogroups (G1b and Q1a1a, respectively) demonstrate a close affinity between AP and present-day Koryaks, Itelmen and Chukchis, as well as with Native Americans.
For both AP and Native Americans, ANS ancestry appears more closely related to Mal’ta than Yana, therefore rejecting a direct contribution of Yana to later AP or Native American groups.
Lake Baikal Neolithic – Bronze Age
(…) the newly reported genomes from Ust’Belaya and recently published neighbouring Neolithic and Bronze Age sites show a succession of three distinct genetic ancestries over a ~6 ky time span. The earliest individuals show predominantly East Asian ancestry, closely related to the ancient individuals from DGC. In the early Bronze Age (BA), we observe a resurgence of AP ancestry (up to ~50% ancestry fraction), as well as influence of West Eurasian Steppe ANE ancestry represented by the early BA individuals from Afanasievo in the Altai region (~10%) This is consistent with previous reports of gene flow from an unknown ANE-related source into Lake Baikal hunter-gatherers.
Our results suggest a southward expansion of AP as a possible source, which is also consistent with the replacement of Y chromosome lineages observed at Lake Baikal, from predominantly haplogroup N in the Neolithic to haplogroup Q in the BA. Finally, the most recent individual from Ust’Belaya, dated to ~600 years ago, falls along the Neosiberian cline, similar to the ~760 year-old ‘Young Yana’ individual from northeastern Siberia, demonstrating the widespread distribution of Neosiberian ancestry in the most recent epoch.
At the western edge of northern Eurasia, genetic and strontium isotope data from ancient individuals at the Levänluhta site documents the presence of Saami ancestry in Southern Finland in the Late Holocene 1.5 kya. This ancestry component is currently limited to the northern fringes of the region, mirroring the pattern observed for AP ancestry in northeastern Siberia. However, while the ancient Saami individuals harbour East Asian ancestry, we find that this is better modelled by DGC rather than AP, suggesting that AP influence was likely restricted to the eastern side of the Urals. Comparison of ancient Finns and Saami with their present-day counterparts reveals additional gene flow over the past 1.6 kya, with evidence for West Eurasian admixture into modern Saami. The ancient Finn from Levänluhta shows lower Siberian ancestry than modern Finns .
EDIT (27 OCT 2018): By comparing the three, I see these are samples published already (at least two) in Lamnidis et al. (2018), but here with added (1) specific radiocarbon dates, (2) comparison with Neosiberian populations and (3) strontium isotope analyses.
Finnish_IA (ca. 350 AD) is probably a Saami-speaking individual, just like the Saami_IA with newly reported radiocarbon dates from Levänluhta ca. 400-600 AD (since Fennic peoples were then likely around the Gulf of Finland).
The conflicting strontium isotope data on marine dietary resources on certain samples from the supplementary material hint at possible external origin of the diet of some of the previously reported (and possibly one newly reported) Saami Iron Age individuals, from some 25-30 km. to the northwest through the river up to hundreds of km. to the southwest of Levänluhta (i.e. the whole coast of the Bothnian Sea). It is unclear why they would prefer an origin of the dietary source in southern Baltic regions instead of some km. to the west, though, unless that’s what they want to propose based on the sample’s admixture…
The coast of the Bothnian Sea (=the northern part of the Baltic Sea, between Sweden and Finland) lay only 25-30 km to the northwest, and accessible to the Iron Age people of the Levänluhta region via the Kyrönjoki river. (…) For individual JA2065/DA236, the low 87Sr/86Sr value (0.71078) would imply an exceptionally heavy reliance on Baltic Sea resources. The δ13C and δ15N values of the individual are near comparable (especially considering within-Baltic latitudinal gradients in δ13C; Torniainen et al. 2017) to the δ13C and δ15N values of a Middle Neolithic population on the Baltic island of Gotland (Eriksson, 2004) interpreted to have subsisted primarily on seals.
These new data on the samples give us some more information than what we already had, because the early date of Finnish_IA implies that there was few East Asian admixture (if any at all) in west Finland during the Roman Iron Age, which pushes still farther forward in time the expected appearance of Siberian ancestry among Saamic (first) and Fennic populations (later). It is unclear whether this East Asian ancestry found in Finnish_IA is actually related to DGC, or it is rather related to the ENA-like ancestry found already in Baltic hunter-gatherers (i.e. in some EHG samples from Karelia), for which Baikal_EN is a good proxy in Lazaridis et al. (2018).
The paper finds thus increased (probably the actual) Siberian ancestry in modern Finns compared to this Iron Age Saami individual. Coupled with the later Saami Iron Age samples, from between one to three centuries later – showing the start of Siberian ancestry influx – , we can begin to establish when the expansion of Siberian ancestry happened in central Finland, and thus quite likely when the Saami began to expand to the north and east and admix with Palaeo-Laplandic peoples.
One sample of haplogroup N1a1a1a1a4a1-M1982, Yana_MED, is found in the Arctic region (north-eastern Yakutia) ca. 1100 AD. Since it is derived from N1a1a1a1a-L392, it might be a surprise for some to find it in a clearly non-Uralic speaking environment at the same time other subclades of this haplogroup were admixing in the west with well-established Finno-Saamic, Volga-Finnic, Ugric, and Samoyedic populations…
On the growing doubts that these data – contradicting the CWC=IE theory – are creating among geneticists (from the supplementary materials):
The Proto-Saami language evolved in southern Finland and Karelia in the Early Iron Age, an area now host to Finnish and the closely related Karelian, but with Saami toponyms showing that the latter two languages are intrusive here (Saarikivi 2004). Saami-speaking populations are thought to have retreated to Lapland during the Middle Iron Age (300–800 AD), where it diverged into the modern Saami dialects. Genetically, the northward retreat of the Saami language correlates with the documented decrease of Saami ancestry in Southern Finland between the Iron Age and the modern period (cf. Lamnidis et al. 2018).
On the way to Lapland, the Saami replaced at least two linguistically obscure groups. This can be inferred from 1) an influx of non-Uralic loanwords into Proto-Saami in the Finnish Lakeland area, and 2) an influx of non-Uralic, non-Germanic words into Saami dialects in Lapland (Aikio 2012). Both of these borrowing events imply contact with non-Saami-speaking groups, e.g. non-Uralic-speaking hunter-gatherers that may have left a genetic and linguistic footprint on modern Saami populations.
The linguistic prehistory of Finland thus does not allow for a straightforward interpretation of the genetic data. The detection of East Asian ancestry in the genetically Saami individual is indicative of a population movement from the east (cf. Lamnidis et al. 2018, Rootsi et al. 2007), one that given the affinities with the ~7.6 ky old individuals from the Devil’s Gate Cave may have been a western extension of the Neosiberian turnover. However, it remains unclear whether this gene flow should be associated with the arrival of Uralic speakers, thus providing further support for a Uralic homeland in Eastern Eurasia, or with an earlier immigration of pre-Uralic, so-called “Paleo-Lakelandic” groups.
I think the genetic interpretation is already straightforward, though. We had a sneak peek at how this late admixture with non-Uralians (mainly Palaeo-Lakelandic and Palaeo-Laplandic peoples from Lovozero and related asbestos ware cultures) is going to unfold among expanding Saami-speaking populations thanks to Lamnidis et al. (2018):
Also, still no trace of R1a in far East Asia (reported as M17 ca. 5300 BC near Lake Baikal by Moussa et al. 2016), so I still have doubts about my previous assessment that R1a split into M17 (and thus also M417) in Siberia, with those expanding hunter-gatherer pottery.
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:
Thus, our data favor a process of Near Eastern animal domestication that is dispersed in space and time, rather than radiating from a central core (3, 11). This resonates with archaeozoological evidence for disparate early management strategies from early Anatolian, Iranian, and Levantine Neolithic sites (12, 13). Interestingly, our finding of divergent goat genomes within the Neolithic echoes genetic investigation of early farmers. Northwestern Anatolian and Iranian human Neolithic genomes are also divergent (14–16), which suggests the sharing of techniques rather than large-scale migrations of populations across Southwest Asia in the period of early domestication. Several crop plants also show evidence of parallel domestication processes in the region (17).
PCA affinity (Fig. 2), supported by qpGraph and outgroup f3 analyses, suggests that modern European goats derive from a source close to the western Neolithic; Far Eastern goats derive from early eastern Neolithic domesticates; and African goats have a contribution from the Levant, but in this case with considerable admixture from the other sources (figs. S11, S16, and S17 and tables S26 and 27). The latter may be in part a result of admixture that is discernible in the same analyses extended to ancient genomes within the Fertile Crescent after the Neolithic (figs. S18 and S19 and tables S20, S27, and S31) when the spread of metallurgy and other developments likely resulted in an expansion of inter-regional trade networks and livestock movement.
Our results imply a domestication process carried out by humans in dispersed, divergent, but communicating communities across the Fertile Crescent who selected animals in early millennia, including for pigmentation, the most visible of domestic traits.
Native Americans from the Amazon, Andes, and coastal geographic regions of South America have a rich cultural heritage but are genetically understudied, therefore leading to gaps in our knowledge of their genomic architecture and demographic history. In this study, we sequence 150 genomes to high coverage combined with an additional 130 genotype array samples from Native American and mestizo populations in Peru. The majority of our samples possess greater than 90% Native American ancestry, which makes this the most extensive Native American sequencing project to date. Demographic modeling reveals that the peopling of Peru began ∼12,000 y ago, consistent with the hypothesis of the rapid peopling of the Americas and Peruvian archeological data. We find that the Native American populations possess distinct ancestral divisions, whereas the mestizo groups were admixtures of multiple Native American communities that occurred before and during the Inca Empire and Spanish rule. In addition, the mestizo communities also show Spanish introgression largely following Peruvian Independence, nearly 300 y after Spain conquered Peru. Further, we estimate migration events between Peruvian populations from all three geographic regions with the majority of between-region migration moving from the high Andes to the low-altitude Amazon and coast. As such, we present a detailed model of the evolutionary dynamics which impacted the genomes of modern-day Peruvians and a Native American ancestry dataset that will serve as a beneficial resource to addressing the underrepresentation of Native American ancestry in sequencing studies.
The high frequency of Native American mitochondrial haplotypes suggests that European males were the primary source of European admixture with Native Americans, as previously found (23, 24, 41, 42). The only Peruvian populations that have a proportion of the Central American component are in the Amazon (Fig. 2A). This is supported by Homburger et al. (4), who also found Central American admixture in other Amazonian populations and could represent ancient shared ancestry or a recent migration between Central America and the Amazon.
Following the peopling of Peru, we find a complex history of admixture between Native American populations from multiple geographic regions (Figs. 2B and 3 A and C). This likely began before the Inca Empire due to Native American and mestizo groups sharing IBD segments that correspond to the time before the Inca Empire. However, the Inca Empire likely influenced this pattern due to their policy of forced migrations, known as “mitma” (mitmay in Quechua) (28, 31, 37), which moved large numbers of individuals to incorporate them into the Inca Empire. We can clearly see the influence of the Inca through IBD sharing where the center of dominance in Peru is in the Andes during the Inca Empire (Fig. 3C).
A similar policy of large-scale consolidation of multiple Native American populations was continued during Spanish rule through their program of reducciones, or reductions (31, 32), which is consistent with the hypothesis that the Inca and Spanish had a profound impact on Peruvian demography (25). The result of these movements of people created early New World cosmopolitan communities with genetic diversity from the Andes, Amazon, and coast regions as is evidenced by mestizo populations’ ancestry proportions (Fig. 3A). Following Peruvian independence, these cosmopolitan populations were those same ones that predominantly admixed with the Spanish (Fig. 3B). Therefore, this supports our model that the Inca Empire and Spanish colonial rule created these diverse populations as a result of admixture between multiple Native American ancestries, which would then go on to become the modern mestizo populations by admixing with the Spanish after Peruvian independence.
Further, it is interesting that this admixture began before the urbanization of Peru (26) because others suspected the urbanization process would greatly impact the ancestry patterns in these urban centers (25). (…)
Animal domestication gives rise to gradual changes at the genomic level through selection in populations. Selective sweeps have been traced in the genomes of many animal species, including humans, cattle, and dogs. However, little is known regarding positional candidate genes and genomic regions that exhibit signatures of selection in domestic horses. In addition, an understanding of the genetic processes underlying horse domestication, especially the origin of Chinese native populations, is still lacking. In our study, we generated whole genome sequences from 4 Chinese native horses and combined them with 48 publicly available full genome sequences, from which 15 341 213 high-quality unique single-nucleotide polymorphism variants were identified. Kazakh and Lichuan horses are 2 typical Asian native breeds that were formed in Kazakh or Northwest China and South China, respectively. We detected 1390 loss-of-function (LoF) variants in protein-coding genes, and gene ontology (GO) enrichment analysis revealed that some LoF-affected genes were overrepresented in GO terms related to the immune response. Bayesian clustering, distance analysis, and principal component analysis demonstrated that the population structure of these breeds largely reflected weak geographic patterns. Kazakh and Lichuan horses were assigned to the same lineage with other Asian native breeds, in agreement with previous studies on the genetic origin of Chinese domestic horses. We applied the composite likelihood ratio method to scan for genomic regions showing signals of recent selection in the horse genome. A total of 1052 genomic windows of 10 kB, corresponding to 933 distinct core regions, significantly exceeded neutral simulations. The GO enrichment analysis revealed that the genes under selective sweeps were overrepresented with GO terms, including “negative regulation of canonical Wnt signaling pathway,” “muscle contraction,” and “axon guidance.” Frequent exercise training in domestic horses may have resulted in changes in the expression of genes related to metabolism, muscle structure, and the nervous system.
Admixture proportions were assessed without user-defined population information to infer the presence of distinct populations among the samples (Figure 2). At K = 3 or K = 4, Franches-Montagnes and Arabian forms one unique cluster; at K = 5, Jeju pony forms one unique cluster. For other breeds, comparatively strong population structure exists among breeds, and they can be assigned to 2 (or 3) alternate clusters from K = 3 to K = 5 including group A (Duelmener, Fjord, Icelandic, Kazakh, Lichuan, and Mongolian) and group B (Hanoverian, Morgan, Quarter, Sorraia, and Standardbred). For group A, geographically this was unexpected, where Nordic breeds (Norwegian Fjord, Icelandic, and Duelmener) clustered with Asian breeds including the Mongolian.Previous results of mitochondrial DNA have revealed links between the Mongolian horse and breeds in Iceland, Scandinavia, Central Europe, and the British Isles. The Mongol horses are believed to have been originally imported from Russia subsequently became the basis for the Norwegian Fjord horse.31 At K = 6, Sorraia forms one unique cluster. The Sorraia horse has no long history as a domestic breed but is considered to be of a nearly ancestral type in the southern part of the Iberian Peninsula.32 However, our result did not support Sorraia as an independent ancestral type based on result from K = 2 to K = 5, and the unique cluster in K = 6 may be explained by the small population size and recently inbreeding programs. Genetic admixture of Morgan reveals that these breeds are currently or traditionally continually crossed with other breeds from K = 2 to K = 8. The Morgan horse has been a largely closed breed for 200 years or more but there has been some unreported crossbreeding in recent times.33
Bayesian clustering and PCA demonstrated the relationships among the horse breeds with weak geographic patterns. The tight grouping within most native breeds and looser grouping of individuals in admixed breeds have been reported previously in modern horses using data from a 54K SNP chip.33,34 Cluster analysis reveals that Arabian or Franches-Montagnes forms one unique cluster with relatively low K value, which is consistent with former study using 50K SNP chip 33,34 Interestingly, Standardbred forms a unique cluster with relatively high K value in this study, different from previous study.33 To date, no footprints are available to describe how the earliest domestic horses spread into China in ancient times. Our study found that Kazakh and Lichuan were assigned to the same lineage as other native Asian breeds, in agreement with previous studies on the origin of Chinese domestic horses.4,5,35,36 The strong genetic relationship between Asian native breeds and European native breeds have made it more difficult to understand the population history of the horse across Eurasia. Low levels of population differentiation observed between breeds might be explained by historical admixture. Unlike the domestic pig in China,8we suggest that in China, Northern/Southern distinct groups could not be used to genetically distinct native Chinese horse breeds. We consider that during domestication process of horse, gene flow continued among Chinese-domesticated horses.
There are large populations of indigenous horse (Equus caballus) in China and some other parts of East Asia. However, their matrilineal genetic diversity and origin remained poorly understood. Using a combination of mitochondrial DNA (mtDNA) and hypervariable region (HVR-1) sequences, we aim to investigate the origin of matrilineal inheritance in these domestic horses.
To investigate patterns of matrilineal inheritance in domestic horses, we conducted a phylogenetic study using 31 de novo mtDNA genomes together with 317 others from the GenBank. In terms of the updated phylogeny, a total of 5,180 horse mitochondrial HVR-1 sequences were analyzed.
Eighteen haplogroups (Aw-Rw) were uncovered from the analysis of the whole mitochondrial genomes. Most of which have a divergence time before the earliest domestication of wild horses (about 5,800 years ago) and during the Upper Paleolithic (35–10 KYA). The distribution of some haplogroups shows geographic patterns. The Lw haplogroup contained a significantly higher proportion of European horses than the horses from other regions, while haplogroups Jw, Rw, and some maternal lineages of Cw, have a higher frequency in the horses from East Asia. The 5,180 sequences of horse mitochondrial HVR-1 form nine major haplogroups (A-I). We revealed a corresponding relationship between the haplotypes of HVR-1 and those of whole mitochondrial DNA sequences. The data of the HVR-1 sequences also suggests that Jw, Rw, and some haplotypes of Cw may have originated in East Asia while Lw probably formed in Europe.
Our study supports the hypothesis of the multiple origins of the maternal lineage of domestic horses and some maternal lineages of domestic horses may have originated from East Asia.
Geographic distributions of horse mtDNA haplogroups
The analysis of geographic distribution of the mitochondrial genome haplogroups showed that horse populations in Europe or East Asia included all haplogroups defined from the mtDNA genome sequences. The lineage Fw comprised entirely of Przewalskii horses. The two haplogroups Iw and Lw displayed frequency peaks in Europe (14.08% and 37.32%, respectively) and a decline to the east (9.33% and 8.00% in the West Asia, and 6.45% and 12.90% in East Asia, respectively), especially for Lw, which contained the largest number of European horses (Table 2). However, an opposite distribution pattern was observed for haplogroups Aw, Hw, Jw, and Rw, which were harbored by more horses from East Asia than those from other regions. The proportions of horses from East Asia for the four haplogroups were 38%, 88%, 62%, and 54%, respectively.
Visualization and exploration of high-dimensional data is a ubiquitous challenge across disciplines. Widely used techniques such as principal component analysis (PCA) aim to identify dominant trends in one dataset. However, in many settings we have datasets collected under different conditions, e.g., a treatment and a control experiment, and we are interested in visualizing and exploring patterns that are specific to one dataset. This paper proposes a method, contrastive principal component analysis (cPCA), which identifies low-dimensional structures that are enriched in a dataset relative to comparison data. In a wide variety of experiments, we demonstrate that cPCA with a background dataset enables us to visualize dataset-specific patterns missed by PCA and other standard methods. We further provide a geometric interpretation of cPCA and strong mathematical guarantees. An implementation of cPCA is publicly available, and can be used for exploratory data analysis in many applications where PCA is currently used.
The Mexican example caught my attention:
Relationship between ancestral groups in Mexico
In previous examples, we have seen that cPCA allows the user to discover subclasses within a target dataset that are not labeled a priori. However, even when subclasses are known ahead of time, dimensionality reduction can be a useful way to visualize the relationship within groups. For example, PCA is often used to visualize the relationship between ethnic populations based on genetic variants, because projecting the genetic variants onto two dimensions often produces maps that offer striking visualizations of geographic and historic trends26,27. But again, PCA is limited to identifying the most dominant structure; when this represents universal or uninteresting variation, cPCA can be more effective at visualizing trends.
The dataset that we use for this example consists of single nucleotide polymorphisms (SNPs) from the genomes of individuals from five states in Mexico, collected in a previous study28. Mexican ancestry is challenging to analyze using PCA since the PCs usually do not reflect geographic origin within Mexico; instead, they reflect the proportion of European/Native American heritage of each Mexican individual, which dominates and obscures differences due to geographic origin within Mexico (see Fig. 4a). To overcome this problem, population geneticists manually prune SNPs, removing those known to derive from Europeans ancestry, before applying PCA. However, this procedure is of limited applicability since it requires knowing the origin of the SNPs and that the source of background variation to be very different from the variation of interest, which are often not the case.
As an alternative, we use cPCA with a background dataset that consists of individuals from Mexico and from Europe. This background is dominated by Native American/European variation, allowing us to isolate the intra-Mexican variation in the target dataset. The results of applying cPCA are shown in Fig. 4b. We find that individuals from the same state in Mexico are embedded closer together. Furthermore, the two groups that are the most divergent are the Sonorans and the Mayans from Yucatan, which are also the most geographically distant within Mexico, while Mexicans from the other three states are close to each other, both geographically as well as in the embedding captured by cPCA (see Fig. 4c). See also Supplementary Fig. 6 for more details.
So, by using a background dataset, it discovers patterns in a single target dataset via dimensionality reduction, that standard dimensionality reduction techniques do not discover. Maybe useful for some prehistoric populations, too…