No large-scale steppe migration into Anatolia; early Yamna migrations and MLBA brought LPIE dialects in Asia

Another, simultaneous paper with the Eurasian samples from Nature, The first horse herders and the impact of early Bronze Age steppe expansions into Asia, by de Barros Damgaard et al., Science (2018).

A lot of interesting data, I will try to analyse its main implications, if only superficially, in sections.

Anatolian samples

Anatolia_EBA from Ovaören, and Anatolia_MLBA (this including Assyrian and Old Hittite samples), all from Kalehöyük, show almost no change in Y-DNA lineages (three samples J2a, one G2a), and therefore an origin of these people in common with CHG and Iranian Neolithic populations is likely. No EHG ancestry is found. And PCA cluster is just somehow closer to Europe, but not to EHG populations.

NOTE. Hittite is attested only in the late first half of the 2nd millennium, although the authors cite (in the linguistic supplement) potential evidence from the palatial archives of the ancient city of Ebla in Syria to argue that Indo-European languages may have been already spoken in the region in the late 3rd millennium BCE.

Regarding the Assyrian samples (one J2a) from Ovaören:

Layer V of GT-137 was the richest in terms of architectural finds and dates to the Early Bronze Age II. In this layer, 2 different structures and a well were uncovered. The well was filled with stones, pottery, and human skeletons (Figs. S2 and S3). In total, skeletons belonging to 22 individuals, including adults, young adults, and children, must belong to the disturbed Early Bronze Age II graves adjacent to the well (103). Pottery and stones found below the skeletons demonstrate that the water well was consciously filled and closed. The fill consists of dumped stones, sherds and skeletons, and the closing stones demonstrate that the water well was consciously filled and cancelled.

Regarding the site most likely associated with the emergence of Old Hittite (two samples J2a1, one G2a2b1), this is what we know:

The Middle Bronze Age at Kaman-Kalehöyük represented by stratum IIIc yields material remains (seals and ceramics) contemporary with the international trade system managed by expatriate Assyrian merchants evidenced at the nearby site of Kültepe/Kanesh. It is therefore also referred to as belonging to the “Assyrian Colony Period” (98). The stratum has revealed three burned architectural units, and it has been suggested that the seemingly site-wide conflagration might be connected to a destruction event linked with the emergence of the Old Hittite state (99). (…) Omura (100) suggests that the rooms could belong to a public building, and that it might even be a small trade center based on the types of artifacts recovered. Omura (100) has concluded that the evidence from the first complex indicates a battle between 2 groups took place at the site. It is possible that a group died inside the buildings, mostly perishing in the fire, while another group died in the courtyard.

NOTE. For more on the Old Hittite period, you can read this for example.

Regarding PCA:

The PCA (Fig. 2B) indicates that all the Anatolian genome sequences from the Early Bronze Age ( -2200 BCE) and Late Bronze Age (-1600 BCE) cluster with a previously sequenced Copper Age ( -3900- 3700 BCE) individual from Northwestern Anatolia and lie between Anatolian Neolithic (Anatolia_ N) samples and CHG samples but not between Anatolia_N and EHG samples.

(…) we are not able to reject a two-population qpAdm model in which these groups derive -60% of their ancestry from Anatolian farmers and -40% from CHG-related ancestry (p-value = 0.5). This signal is not driven by Neolithic Iranian ancestry.

pca-ancient-modern-eurasians
Principal Component Analysis estimated with ancient and modern Eurasians.

NOTE. Anatolian Iron Age samples, from the Hellenistic period, which was obviously greatly influenced by different, later Indo-European migrations, does show a change in PCA.

Regarding CHG ancestry:

Ancient DNA findings suggest extensive population contact between the Caucasus and the steppe during the Copper Age (-5000-3000 BCE) (1, 2, 42). Particularly, the first identified presence of Caucasian genomic ancestry in steppe populations is through the Khvalynsk burials (2, 47) and that of steppe ancestry in the Caucasus is through Armenian Copper Age individuals (42). These admixture processes likely gave rise to the ancestry that later became typical of the Yamnaya pastoralists (7), whose IE language may have evolved under the influence of a Caucasian language, possibly ‘from the Maykop culture (50, 55). This scenario is consistent with both the “Copper Age steppe” (4) and the “Caucasian” models for the origin of the Proto-Anatolian language (56).

The CHG specific ancestry and the absence of EHG-related ancestry in Bronze Age Anatolia would be in accordance with intense cultural interactions between populations in the Caucasus and Anatolia observed during the late 5th millennium BCE that seem to come to an end in the first half of the 4th millennium BCE with the village-based egalitarian Kura-Araxes society (59, 60), thus preceding the emergence and dispersal of Proto-Anatolian.

Our results indicate that the early spread of IE languages into Anatolia was not associated with any large-scale steppe-related migration, as previously suggested (61). Additionally, and in agreement with the later historical record of the region (62), we find no correlation between genetic ancestry and exclusive ethnic or political identities among the populations of Bronze Age Central Anatolia, as has previously been hypothesized ( 63).

The Anatolian question

There is no steppe ancestry or R1b-M269 lineages near early historic Hittites. Yet.

Nevertheless, we already know about potentially similar cases:

So there seems to be thus no theoretical problem in accepting:

  • That neither steppe ancestry nor R1b-M269 subclades, already diminished in Bulgaria in the mid-5th millennium, did reach Anatolia, but only those Common Anatolian-speaking Aegean groups over whose ancestors Proto-Anatolians (marked by incoming EHG ancestry) would have previously dominated in the Balkans.
  • That steppe ancestry and R1b-M269 subclades did in fact arrive in the Aegean, but EHG was further diluted among the CHG-related population by the time of the historic Anatolian-speaking peoples in central Anatolia. Or, the most likely option, that their trace have not been yet found. Probably the western Luwian peoples, near Troy, were genetically closer to Common Anatolians.

Both of these scenarios are interesting, in that they show potential links between Pre-Greek peoples of Hellas (related to Anatolians) and the Pelasgian substrate of early Greek dialects, since they show a similar recent CHG-related wave from the East.

What we can assert right now is that Proto-Anatolian must have separated quite early for this kind of data to show up. This should mean an end to the Late PIE origin of Anatolian, if there was some lost soul from the mid-20th century still rooting for this.

As I said in my review of Lazaridis’ latest preprint, we will have to wait for the appropriate potential routes of expansion of Proto-Anatolian to be investigated. As he answered, the lack of EHG poses a problem for steppe expansion into Anatolia, but there is still no better alternative model proposed.

Eurasian-CHG-ancestry
Model-based clustering analysis of present-day and ancient individuals assuming K = 6 ancestral components. The main ancestry components at K = 6 correlate well with CHG (turquoise), a major component of Iran_N, Namazga_CA and South Asian dines; EHG (pale blue), a component of the steppe dine and present in South Asia; East Asia (yellow ochre), the other component of the steppe d ine also in Tibeto-Burman South Asian populations; South Indian (pink), a core component of South Asian populations; Anatolian_N (purple), an important component of Anatolian Bronze Age and Steppe_MLBA; Onge (dark pink) forms its own component.

This is what the authors have to say:

Our findings are thus consistent with historical models of cultural hybridity and “Middle Ground” in a multi-cultural and multi-lingual but genetically homogeneous Bronze Age Anatolia (68, 69). Current linguistic estimations converge on dating the Proto-Anatolian split from residual PIE to the late 5th or early 4th millennia BCE (58, 70) and place the breakup of Anatolian IE inside Turkey prior to the mid-3rd millennium (53, 71,72).

We cannot at this point reject a scenario in which the introduction of the Anatolian IE languages into Anatolia was coupled with the CHG-derived admixture prior to 3700 BCE, but note that this is contrary to the standard view that PIE arose in the steppe north of the Caucasus (4) and that CHG ancestry is also associated with several non-IE-speaking groups, historical and current. Indeed, our data are also consistent with the first speakers of Anatolian IE coming to the region by way of commercial contacts and small-scale movement during the Bronze Age. Among comparative linguists, a Balkan route for the introduction of Anatolian IE is generally considered more likely than a passage through the Caucasus, due, for example, to greater Anatolian IE presence and language diversity in the west (73). Further discussion of these options is given in the archaeological and linguistic supplementary discussions (48, 49).

If you are asking yourselves why the Danish school (of Allentoft, Kristiansen, and Kroonen, co-authors of this paper) was not so fast to explain the findings the same way the proposed their infamous Indo-European – steppe ancestry association (i.e. ancestry = language, ergo CHG = PIE in this case), and resorted to mainstream anthropological models instead to explain the incongruence, I can think of two main reasons:

The possibility of having an early PIE around the Caucasus, potentially closely related not only to Uralic to the north, but also to Caucasian languages, Sumerian, Afroasiatic, Elamo-Dravidian, etc. could be a good reason for those excited with these few samples to begin dealing with macro-language proposals, such as Eurasiatic and Nostratic. If demonstrated to be true, a Northern Iranian origin of Middle PIE would also help relieve a little bit the pressure that some are feeling about the potentially male-driven Indo-European continuity (even if not “autochthonous”) associated with the expansion of R1b-L23 subclades.

On the other hand, I am a firm supporter of solid anthropological models of migration, and of “late and small” language expansions, usually accompanied by demic diffusion, which has been demonstrated to be linked with haplogroup expansion and reduction in variability.

Therefore, for the moment, even if it is weak – as weak as it always was (but still stronger than Gimbutas’ Maykop route) – the Balkan route seems like the best fit for all the data combined.

In fact, we already have steppe ancestry moving into the Lower Danube and Bulgaria in the mid-5th millennium. Let’s not forget that.

Yamna expansion to the East

Interesting data from an early East Yamna offshoot at Karagash, ca. 3018-2887 BC, of R1b-Z2106 lineage, which shows some ancestry, lineage, and cultural continuity in Sholpan, ca. 2620-2468 BC, in Kazakhstan.

This sample might be part of another descendant group from the migration waves that reached Afanasevo, and can thus be related to other early Asian R1b-L23 samples found in Narasimhan et al. (2018).

On the formation of Yamna and its CHG contribution, from the supplementary material:

  1. An admixture event, where Yamnaya is formed from a CHG population related to KK1 [=Kotias, dated ca. 7800 BC] and an ANE population related to Sidelkino and Botai. We inferred 54% of the Yamnaya ancestry to come from CHG and the remaining 46% to come from ANE.
  2. A split event, where the CHG component of Yamnaya splits from KK1. The model inferred this time at 27 kya (though we note the larger models in Sections S2.12.4 and S2.12.5 inferred a more recent split time [see below graphic]).
  3. A split event, where the ANE component of Yamnaya splits from Sidelkino. This was inferred at about about 11 kya.
  4. A split event, where the ANE component of Yamnaya splits from Botai. We inferred this to occur 17 kya. Note that this is above the Sidelkino split time, so our model infers Yamnaya to be more closely related to the EHG Sidelkino, as expected.
  5. An ancestral split event between the CHG and ANE ancestral populations. This was inferred to occur around 40 kya.
yamnaya-chg-ancestry
A 10-leaf model based on combining the models in Fig. S16 and Fig. S19 and re-estimating the model parameters.

On the expansion of domestication

CHG is not found in Botai, no gene flow from Yamna is found in its samples, and they are more related to East Asians, while Yamna is related to West Eurasians:

The lack of evidence of admixture between Botai horse herders and western steppe pastoralists is consistent with these latter migrating through the central steppe but not settling until they reached the Altai to the east (4). More significantly, this lack of admixture suggests that horses were domesticated by hunter-gatherers not previously familiar with farming, as were the cases for dogs (38) and reindeer (39). Domestication of the horse thus may best parallel that of the reindeer, a food animal that can be milked and ridden, which has been proposed to be domesticated by hunters via the “prey path” (40); indeed anthropologists note similarities in cosmological beliefs between hunters and reindeer herders (41). In contrast, most animal domestications were achieved by settled agriculturalists (5).

NOTE. I am not sure, but they seem to hint that there were separate events of horse domestication and horse-riding technique by the Botai and Yamna populations due to their lack of genetic contribution from the latter to the former. I guess they did not take into account farming spreading to the steppe without genetic contribution beyond the Dnieper… In fact, the superiority in horse-riding shown by the expanding Yamna peoples – as they state – should also serve to suggest from where the original technique expanded.

Indo-Iranian migrations

On the expansion of Yamna, and the different expansion of Steppe MLBA (with Indo-Iranian speakers) into Asia, further supporting Narasimhan et al. (2018), they have this to say:

However, direct influence of Yamnaya or related cultures of that period is not visible in the archaeological record, except perhaps for a single burial mound in Sarazm in present-day Tajikistan of contested age (44, 45). Additionally, linguistic reconstruction of proto-culture coupled with the archaeological chronology evidences a Late (-2300-1200 BCE) rather than Early Bronze Age (-3000-2500 BCE) arrival of the Indo-Iranian languages into South Asia (16, 45, 46). Thus, debate persists as to how and when Western Eurasian genetic signatures and IE languages reached South Asia.

Samples from the Namazga region (current Turkmenistan) from the Iron Age show an obvious influence from steppe MLBA (ca. 2300-1200 BC), and not steppe EBA (i.e. Yamna), population, in contrast with samples from the Chalcolithic (ca. 3300 BC), which don’t show this influence. This helps distinguish prior contacts with Iran Neolithic from the actual steppe population that expanded Indo-Iranian into Asia.

Very interesting therefore the Namazga CA sample (ca. 855 BC), of R1a-Z93 subclade, showing the sign of immigrant Indo-Aryans in the region. For more on this we will need an evaluation in common with the corrected data from Narasimhan et al. (2018), and all, including de Barros (Nature 2018), in combination with statistical methods to ascertain differences between early Indo-Aryans and Iranians.

namazga-expansion-south-asia
A summary of the four qpAdm models fitted for South Asian populations. For each modern South Asian population. we fit different models with qpAdm to explain their ancestry composition using ancient groups and present the f irst model that we could not reject in the following priority order: 1. Namazga_CA + Onge, 2. Namazga_CA + Onge + Late Bronze Age Steppe, 3. Namazga_CA + Onge + Xiongnu_lA (East Asian proxy). and 4. Turkmenistan_lA + Xiongnu_lA. Xiongnu_lA were used here to represent East Asian ancestry. We observe that while South Asian Dravidian speakers can be modeled as a mixture of Onge and Namazga_CA. an additional source related to Late Bronze Age steppe groups is required for IE speakers. In Tibeto-Burman and Austro-Asiatic speakers. an East Asian rather than a Steppe_MLBA source is required.

Siberian peoples and N1c lineages

We have already seen how the paper on Eurasian steppe samples tries to assign Uralic to Neolithic peoples east of the Urals. The association with Okunevo is unlikely, since most are of haplogroup Q1a2, but they seem to suggest (combining both papers) that they accompanied N lineages from Siberian hunter-gatherers (present e.g. in Botai or Shamanka II, during the Early Neolithic), and formed part of (or suffered from) different demic diffusion waves:

These serial changes in the Baikal populations are reflected in Y-chromosome lineages (Fig. SA; figs. S24 to S27, and tables S13 and SI4). MAI carries the R haplogroup, whereas the majority of Baikal_EN males belong to N lineages, which were widely distributed across Northern Eurasia (29), and the Baikal_LNBA males all carry Q haplogroups, as do most of the Okunevo_EMBA as well as some present-day Central Asians and Siberians.

NOTE. Also interesting to see no R1a in Baikal hunter-gatherers after ca. 3500 BC, and a prevalence of N lineages as supported in a previous paper on the Kitoi culture, which some had questioned in the past.

In fact, the only N1c1 sample comes from Ust’Ida Late Neolithic, 180km to the north of Lake Baikal, apparently before the expansion of Q1a2a lineages during the EBA period. While this sample may be related to those expanded later in Finno-Ugric territory (although it may only be related to those expanded much later with Yakuts), other samples are not clearly from those found widely distributed among North-East Europeans only after the Iron Age, or – as in the case of Shamanka II (N1c2), they are clearly not of the same haplogroup.

eurasian-n-subclades
Geographical location of ancient samples belonging to major clade N of the Y-chromosome.

Wrap-up

It is great to see the paper and the supplementary material deal with Y-DNA haplogroups and their relevance for migrations with such detail. Especially because this paper comes from the same Copenhagen-based research group that originally associated ancestry with language, creating thus today’s mess based on steppe ancestry.

Regarding Y-DNA data, once again almost 100% of samples from late Khvalynsk/Yamna and derived cultures (like Afanasevo and Bell Beaker) are R1b-L23, no single R1a-M417 lineage found, and few expected by now, if any, within Late Proto-Indo-European territory.

While they claim to take Y-DNA into account to assess migrations – as they do for example with Asian cultures – , their previous model of a Yamna “R1a-R1b community” remains oddly unchanged, and they even insist on it in the supplementary materials, as they do in their parallel Nature paper.

They have also expressly mitigated the use of ancestral components to assess populations, citing the ancestral and modern association of CHG ancestry with different ethnolinguistic groups in the Middle East, to dismiss any rushed conclusions on the origin of Anatolian, and consequently of Middle PIE. And they did so evidently because it did not fit the anthropological data that is mainstream today (supporting a Balkan route), which is the right thing to do.

However, they have apparently not stopped to reconsider the links of CWC and steppe ancestry to ancestral and modern Uralic peoples – although they expressly mention the strong connection with modern Karelians in the supplementary material.

Also, after Narasimhan et al. (2018), there is a clear genetic continuity with East Yamna (in ancestry as in R1b-L23 subclades), so their interpretations about Indo-Iranian in this paper and especially de Barros (Nature 2018) – regarding the Abashevo -> Sintashta/Srunba/Andronovo connection – come, again, too late.

Related: