Proto-Turkic homeland

This post is part of a draft on South Siberian language homelands and Sprachbünde.

The following text contains a description of Proto-Turkic and its main dialectal split. Contacts with Samoyed, Ob-Ugric, Iranian, Yeniseian, Tocharian, Chinese and Mongolic, as well as palaeolinguistics, hydronymy, and ethnonymy are taken into account to pinpoint the succeeding homelands and expansion territories. The archaeological-archaeogenetic discussion is focused on the Middle and Late Bronze Age Altai Mönkhkhairkhan and Deer Stone-Khirigsuur Complex and related groups, as well as on Ulaanzukh; Early Iron Age “Scytho-Siberian” Pazyryk & Uyuk and Slab Grave cultures; and on the Late Iron Age Xiongnu and its connection with Huns.

1. Turkic
1. Dialectal split
2. Areal linguistics
1. Proto-Samoyed
2. Ob-Ugric
3. Common Yeniseian
4. Proto-Tocharian
5. Pre-Saka Iranian
6. Late Old Chinese
7. Pre-Proto-Mongolic
3. Palaeolinguistics
4. Onomastics
1. Hydronymy
2. Ethnonymy
2. Archaeology and Population Genomics
1. Agriculture
2. Middle/Late Bronze Age (ca. 1900-900 BC)
1. Mönkhkhairkhan / Deer Stone-Khirigsuur Complex
2. Ulaanzuukh/Shape Burial (1450-1150 BC)
3. Early Iron Age (ca. 900-300 BC)
1. Slab Grave
2. Scytho-Siberians
4. Xiongnu Empire
1. Early Xiongnu
2. Late Xiongnu
5. Wrap-up

PLEASE NOTE. Many of the Y-SNP calls from ancient samples referred to below have been analyzed by the FamilyTreeDNA Haplotree team formed by phylogeneticist Michael Sager and Göran Runfeldt from the R&D team. Those ancient samples with validated haplogroup inferences are marked by a hyperlink to the FTDNA Haplotree. Occasionally, though, such hyperlinks are also used in the text when discussing Y-SNP branches in general, without referring to specific ancient samples. For a quick reference of ancient samples, you can check out the Ancient DNA Dataset, also visually in an Online Web Map, in SNP Tracker, or in AncientDNA.info. T_MRCA and formation dates have been checked from YFull.

1. Turkic

Old Turkic is the oldest actually attested form of (Orkhon) Turkic from Göktürk and Uyghur inscriptions (dated ca. AD 7^th-13^th c.), whereas Proto-Turkic is estimated to date to ca. 500 BC (Dybo 2007). On the other hand, Early Proto-Turkic is described as a set of non-stabilised dialects, which could have ranged ca. 3000-500 BC, and which are represented by the Late Proto-Turkic ‘ancestor dialects’ from which later Turkic languages emerged, first with the formation of the isoglosses r/l and z/š (Róna-Tas 1998).

NOTE. It is disputable whether (1) Oghuric and Common Turkic are parallel branches of Archaic Turkic, or (2) Oghuric (or Common Turkic?) is the most archaic branch, representing an earlier split that broke away before a series of phonetic changes occurred (e. g. *z → r, *lš or *lč → l), perhaps ca. 100-400 AD, even (3) potentially as a sort of ‘Para-Turkic’ branch.

Traditionally, the first known Turkic expansion is identified with the polyglot Xiongnu (Hsiung-nu) empire ca. 3^rd c. BC – AD 48, mostly based on the interpretation the language of the so-called Jie couplet included in Jin Shu, a history of the Jin dynasty period (Shiratori, 1900; Benzing, 1959; Tenišev, 1997; Schönig, 1997–1998; Dybo, 2007; Janhunen, 2010). Interactions among its highly heterogeneous populations probably included Mongolic and Tungusic – apart from Iranian and even Chinese – elements, and the Sprachbund that is often identified as the “Micro-Altaic” core must have happened within (or in contact with) it, based on their earliest lexical parallels.

An up-to-date interpretation of the 56 most transparent Xiongnu etymologies (Dybo 2007, 2014) shows that the lexicon of written history is partly Turkic and partly Eastern Iranian in origin. The two lexical strata are distributed differently in chronology and semantics: Eastern Iranian occurs in earlier – mostly Western Han – sources and refer almost entirely to titles and terms of dairy products, whereas Turkic words occur in both early and late sources, and show a wider semantic reach. Isolated Tocharian and Mongolic forms may have been present, but these etymologies are the least transparent part of the Xiongnu vocabulary (Savelyev & Jeong 2020).

NOTE. Puleleyblank (1062) & Vovin (2000, 2002, 2016) assume that the empire had a Yeniseian core tribal elite, which ruled over various Iranian and Altaic (Turco-Mongol) groups (Brosseder 2018), but they assessed mostly personal names, which allow for a broad range of interpretations. In fact, the most plausible ones may be part of shared cultural vocabulary of non-native origin in either Xiongnu or Yeniseian (Savelyev & Jeong 2020).

The Huns are traditionally identified with the Xiongnu based on similar ethnonyms (cf. Atwood 2012), early proto-historical accounts, and archaeological parallels – such as similar cauldron and sword cult traditions – but it is unclear to what extent the ethnic composition of both coincided. It is assumed that Turkic peoples were heavily concentrated in the western half of the old Xiongnu Empire.

If the names of Hunnic tribes and rulers spreading from Kazakhstan ca. AD 4^th c. are a rough guide to ethnicity, they seem to have a core Turkic element ruling over an initially large Turco-Iranian population, and then – after their conquests in Europe – over a largely Germanic population in the western half of their empire (Brosseder 2018).

1.1. Dialectal split

West Old Turkic irregularities and dialectal loans from Chinese (different from East Old Turkic), Tocharian, and Iranian, as well as those parallel Oghuz vs. Oghur loans apparently present in Proto-Mongolic, suggest that the Proto-Turkic split that gave rise to the Oghuz dialect was a long procedure of dialects in loose contact (Róna-Tas 1998: 67-80).

The early dialectal division seems to stem from special phonetic environments where the opposition r vs. z & l vs. š was neutralized, appearing new sounds, an with a later generalization that divided that pronunciation in two areas. Following the dialectal splits, different isoglosses were shared by (probably originally neighbouring) branches, which is also attested by the presence of lexical isoglosses not strictly following phonological ones; cf. Old Uyghur tül ~ tüš. While Oghur Turkic is believed to have been part of the first westward waves of nomadic Turkic peoples, the possible presence of Oghuric forms in Mongolic borrowings suggest a community that also reached the Trans-Baikal area.

The Dingling (2^nd-1^st c. BC) are frequently mentioned as the first people with a reliable Turkic ethnonym to appear in proto-history (see below). They were brought by force into the Xiongnu union and remained recalcitrant vassals of the Xiongnu of the Altai (the later Huns of Central Asia and Europe) some time during the so-called two-hundred-year gap (ca. AD 2^nd-4^th c.). They are probably to be identified as (Pre-)Proto-Oghuric Turks neighbouring the Kangju to the north – in northern and central Kazakhstan – like the succeeding and likely hononymous Tiele (Brosseder 2018).

They were probably ruled by an elite called the Asiani/Asi (Yuezhi confederacy), possessing state level organization. The Dingling/Tiele were largely in possession of the Kazakh steppes by AD 350, supplanting Iranian nomads in the region, before spreading west into the north Pontic steppes (after AD 460).

1.2. Areal linguistics

Mutual contacts between Turkic and Central Asian, West Siberian, and East Asian languages are witness to the most likely spread (and known dialectal split) of the Proto-Turkic community in the 1^st millennium BC.

1.2.1. Proto-Samoyed

Intense Proto-Samoyedic ~ Proto-Turkic contacts dating to the 1^st millennium BC are attested by the increasing number of parallels (ca. 30) found in Proto-Samoyedic reconstructions, witnessing a likely elite dominance of Proto-Turkic over a Proto-Samoyedic-speaking community (see above Turkic loans on Proto-Samoyed).

Probable Samoyedic loans in Proto-Turkic include (Dybo 2014):

• PSmy. *käsa ‘bark’ → CTk. *kās ‘bark’ (OUygh. qas, qasuq, Krh.-Uygh. qas, qasuq (MK), Halaj qās). The PSmy. word has a standard Uralic etymology, cf. PFU *ko(ń)ćkз ‘bark’;
• (?) PSmy. *kaəti̮ə ‘fir tree’ → CTk. *Kadi ‘pine tree’;
• (?) PSmy. *ti̮te̮ŋ ‘Siberian cedar’ → CTk. *Tīt ‘larch’;
• PSmy. *koəjə ‘mountain, watershed’ → CTk. *K(i)aja ‘rock, mountain’. PSmy. ← PU. *kaδ’a (cf. Hung. hegy ‘top, mountain’), CTk. has no good Altaic etymology;
• PSmy. *we̮ńз ‘meat or fish soup, broth’ → (?) CTk. *bün (*büjn?) ‘meat soup, broth’.

This expected asymmetry in loanwords seems to be tightly complemented by the potential correlative Samoyed substrate found in the shared “western” palatal vowel harmony [contrast between palatal/front vs. velar/back vowels], a system reconstructed with certainty for Proto-Turkic and Proto-Samoyed – with some controversy regarding the rules for Proto-Uralic.

In contrast with this group, an “eastern” system of retracted tongue root harmony [“retracted” vs. “non-retracted” or “advanced” tongue root] is reconstructed for Proto-Mongolic, Proto-Tungusic – with a controversial reconstruction without vowel harmony proposed by Starostin – and reaching up to Korean, Amuric, and Chukotic (cf. Janhunen 1981; Barrere 2016; Ko, Joseph, & Whitman 2016; Barrere & Janhunen 2019; etc.).

NOTE. Much more complex is the potential underlying relationship of the language of Siberian herders (Samoyed, Turkic, Tungusic) displaying non-tonal, strongly suffixal-agglutinating nominal and verbal inflectional morphology, relative to the widespread Yeniseic language family (also stemming from the Circum-Baikal area) showing prefixing verb morphology, pronominal possessive prefixes, and phonemic distinctions in prosody (Vajda 2009, 2020).

There are also Turkic borrowings into individual Samoyed languages, including Mongolic borrowings through a Turkic intermediate (cf. Róna-Tas 1988: 746-747).

1.2.2. Ob-Ugric

There are 6 potential loanwords from Turkic into Ob-Ugric, including some with reconstructible Oghuric traits (see above Turkic influence on Ob-Ugric).

Loans from Ob-Ugric into Proto-Turkic, attesting to bilateral contacts, include (Dybo 2014):

• PTk. *kunduř ‘beaver’ ← POUg. *kuntз-l’ ← PUg. *kuntз ‘beaver’;
• PTk. *Kutan ‘heron, pelican’ ← POUg. *koti̮ŋ ← PUg. *kottVŋV ‘swan’;
• PTk. *iŋir ‘pack saddle’ ← *iŋgri ← POUg. *näγrä ← PUg. *närkV ‘saddle’.

Further support for Early Proto-Turkic contacts with the Ob-Ugric-speaking community are further supported by the traditionally noted – albeit simplistic and never fully investigated – similarities between PT š → Oghur l and the similar evolution in Mansi.

NOTE. The geographical space and guesstimates for Proto-Samoyed and Proto-Turkic before the spread of their dialects can be said thus to coincide in West Siberia during the first millennium BC as their most likely common period. Of note, there was apparently no Pre-PSmy.~POUg. or even PSmy.~PKh. interaction after the short-lived Proto-East Uralic community, supporting the independent development of both Eastern Uralic branches separated from each other probably some centuries after the Proto-Uralic split, and until well into their respective dialectal periods: the Ugric Sprachbund developing close to the Southern Urals, and the Pre-Proto-Samoyedic community closer to the Altai.

1.2.3. Common Yeniseian

The earliest contacts of Proto-Turkic with Yeniseic date to Common Yeniseian times (Dybo 2014):

• PYen. *χɔtyr ‘felt, cloth’ ← PTk. *kidiř ‘felt’ → CTk. *kidiz;
• PYen. *ʔiʔn ‘needle’ ← PTk. → CTk. *(j)igne, jiŋne;
• PYen. *ʔVʔr₁ ‘song’ ← PTk. → CTk. *ir;
• ? PYen. *dam- in *dam-χuχ ‘window’ (-χuχ “hole”) ← PT → CTk. *dām ‘wall’;
• PYen. *KVlpV ‘ladle’ ← PTk. **kaλbuk → PT *kaλuk → CTk. *kašuk;
• PYen. *dəĺi ‘willow’ ← PTk. → CTk. *dal;
• PYen. *χɔpVr ‘foam’ ← PTk. *köp- → PT *köpük ‘foam’, CTk. *köpür- ‘to foam’;
• ? PYen. *bət ‘lenok’ ← CTk. *bińit.

The influence of Yeniseic on Turkic is probably found in PTk. *gejik ‘wild, game’ (no Altaic etymology) ← PYen. *gəʔj ‘hunt; wild animal’.

The lack of (any described) correlative substratal influence of Yeniseic on Turkic suggests that their contacts were rather indirect, with the Common Yeniseian homeland likely entering in the border sphere of influence of the Proto-Turkic community.

NOTE. This lack of superstrate or substrate Yeniseian influence on Proto-Turkic gives indirect support to reject the validity of a Yeniseian interpretation of Hsiung-nu names, which suggests that the Xiongnu represented neither a majority of acculturated Yeniseian speakers under Turkic rule nor a Turkic-speaking majority under Yeniseian elites.

1.2.4. Proto-Tocharian

The following Tocharian lexemes have been proposed to derive from Turkic (Lubotsky 2011):

• Toch. A koṃ, Toch. B kauṃ ‘sun, day’ ← PToch. *kaun(V)- ← PTk. *gün(eĺ) / *guńaĺ (OUygh. kün ‘sun, day’, Turkm. gün ‘id.’, etc.). Its likeliness depends on the potential derivation from PIE.
• Toch. A āle, Toch. B alyiye* ‘palm (of the hand)’ ← PToch. *āl’ye ← PTk. *āja ‘id.’ (OUygh. aja, Turkm. āja, etc.) ← **ālja. Borrowed before the evolution of a hypothetical Pre-PTk. **lj → PTk. *l. Its likeliness depends on the Turkic reconstruction as well as on the potential derivation from the PIE etymon from ‘elbow’ and semantic shift to ‘palm of the hand’.
• Toch. A tor, Toch. B taur ‘dust’ ← PToch. *taur ← PTk. *tṓr ‘dust’ (OUygh. toz, Turkm. tōz). Borrowed before the evolution of Pre-PTk. **ŕ to PT *z. Its likeliness depends on the Turkic reconstruction with *ŕ (cf. Yakut tor vs. a Mongolic origin) and the potential derivation from PIE.
• Toch. B ām* ‘silence’, adv. ‘quietly, still’ ← PTk. *am- ‘to be gentle, quiet’ (Old Turkic amul, amɨl ‘gentle, quiet’). Suspicious because of the borrowing of a monosyllabic verbal root as noun.
• Toch. A kanak, B kenek ← PToch. *kenek ‘cotton cloth’ ← PTk. *köjŋe-lek, *köjŋek ‘shirt’ (Karakh. köŋlek, Turkm. köjnek). Pinault considers the Tocharian word to be borrowed from Iranian (cf. Sogd. knc’k ‘fabric’, Khor. knc(y)k [kancək] ‘shirt’), although the connection seems weaker phonologically and semantically.
• Toch. B olya ‘more’ ← PTk. *ulug ‘big, great’ (OUygh. uluɣ, Turkm. ulu etc.).
• Toch. A tmāṃ, Toch. B t(ᵤ)māne ‘ten thousand, a myriad’ ← PToch. *t(ə)māne ← PTk. *Tümen ‘ten thousand; very many’ (OUygh. tümen, Turkm. tümen).
• Toch. B päŕseri* ‘(head-)louse’ ← PTk. *bürče ‘flea’ (Tat. börčɛ, Kum. bürče, Chuv. pъʷŕza, etc.).
• Toch. B yase* ‘shame’ ← PTk. *jās ‘loss, damage, shame’ (OUygh. jas ‘loss, damage’, Yak. sāt ‘shame’ etc.). The alternative PIE etymology is unsatisfactory.
• Toch. AB kärk- ‘rob, steal’ ← PTk. *Kar-ak ‘bandit’ (OUygh. qaraq-čɨ, Turkm. Garak etc.).

Ana Dybo (2014) considers these borrowings unlikely, except for two attested in Hsiung-nu inscriptions. Further, a few Turkic etyma have been proposed to be of Tocharian origin (Róna-Tas 1961), but Dybo rejects them on a phonological basis, or attributes them to later Turkic → Tocharian contacts (cf. Toch. B kwaṣo ‘country’ ← OTk. *koš ‘cottage’; Toch. B peṣke ‘clarified butter, ghee’ ← *biši- ← PTk. *biλči- ‘to churn’).

This early (including disputable Pre-PTk./Oghuric reconstructions) and unbalanced language-induced contact might suggest that (Oghur?) Turkic became an adstrate/superstrate over Tocharian locals to the south of the Altai Mountains, at a time where no sizeable Tocharian substrate was encountered by the expanding Proto-Turkic population. Further, contacts of Tocharian with Eastern Iranian and Chinese show a similar agricultural terminology (see e.g. Peyrot 2019), suggesting thus a similar relative chronological layer of interregional language influences found as the one found in Turkic.

1.2.5 Pre-Saka Iranian

Early East Iranian borrowings in Proto-Turkic date to the pre-Khotanese Saka, with a likely origin potentially predating the split of the Saka-Wakhi branch ca. 5^th c. BC. These loans include those from a nomadic pastoralist culture (Dybo 2014, Savelyev 2017):

• PTk. *ečkü ‘tame goat’ ←? PIr. *aź-ya-k-ā, from *aźa- ‘goat’;
• PTk. *dạ̄na ‘heifer’ ← PIr. *dainu-kā, Kh.Saka dīnū ‘cow’, a stem also borrowed from Alanic into Pre-PHu. and possibly behind Finnic and Mari forms (see here);
• PTk. *dorak ‘a k. of cheese’ ← MIr. *tura-ka, cf. Av. tūiri- ‘curdled milk’, Kh.Saka (?) ttūra ‘cheese’;
• PTk. *arpa ‘barley’ ←? PIr. *arba- ‘barley’, East Iranian *arbasyā- (or *arpasyā), cf. Ishk. úrvьs, Sangl. vərvəs, Yidgha yaršīo, Pashto orbəˊši, urbəˊši; possibly borrowed further into PMo. *arbai ‘barley’ and PTun. (cf. Manchu arfa ‘barley, oats’).

Other agropastoralist terms show a straightforward development from Proto-Turkic roots, such as PTk. *Kūrï-t ‘a k. of dried quark, cheese’, a common derivative of *Kūr(ï)- ‘to dry’, or PTk. *yogurt ‘curdled milk’, presumably derived from yogur- ‘to knead’ or a homonymous verb meaning ‘to thicken, condense’ (Savelyev 2017).

Further cultural borrowings include (Dybo 2014):

• PTk. *kumlak ‘hop’ ← PIr. *hauma-aryaka “Aryan haoma”, Osset. x°ymællæg ‘hop’ (*hauma ‘ephedra’);
• PTk. *bütnük ‘mint’ ← MIr. *bodina-ka (Osset. bit‛na / bet‛ina ‘mint’, Shugn., Bartang wiδn, Rushan wuδn, Wakhi waδn, Munji wālən, Pashto welə́na);
• PTk. (?)*dura ‘tower, fortification’ ← MIr., cf. Kh.Saka ttaura ‘wall’, Osset. tyrg ‘courtyard’;
• PTk. *gẹl ‘house, home, family’ ← PIr. *gr̥da- ‘house’, MPers. gilistak [glsty, glystk] ‘Dev house’, Zor. Pehl. *gil-šāh ‘house host’, Kh.Saka ggalū ‘family’;
• PTk. *dām ‘wall of a stationary building’ ← PIr. *dam- ‘house’ (Av. dam, Sogd. -dam ‘world’, Yazg. -dom in toponyms, cf. also Kh.Saka damänu ‘house’);
• PTk. *darkan ‘title or post; in names’ ← PIr. (← PIIr. tark- ‘to decide, to judge’ ← PIE *tlk-), cf. Sogd. trγ’n, trχ’n [tarxān] ‘title’, Kh.Saka ttarkana- ‘title’, Osset. tærxon ‘justice’;
• PTk. *qaγan: ‘head of tribal confederation’, ‘lord’ ← MIr. *hva-kama- ‘autokrator’ (hva- ‘self-’ and kam- ‘to wish’), cf. Sogd. xutkame [xwt-k’m’-k’] with the same meaning, Avestan hvata-dataandhva-data.

Iranian loans among Xiongnu Chinese inscriptions (from Shi zi and Han shu) include dairy husbandry words and titles, attesting to the close contacts of Proto-Turks with early Saka (Dybo 2014):

• Hsiung-nu 酪 *rāk ‘koumiss’ ← PIr. *ranka- (+ -aka), Kh. Saka ragai ‘fermented liquor’; Osset. rong ‘fermented honey liquor’;
• Hsiung-nu 湩 *ṭoŋh ‘milk, koumiss’ (Han epoch, about a liquor from whipped horse milk) ← PIr. *dauγ-na (from *duž- ‘to milk’) → Osset. donq/donγ ‘yield of milk’, Wakhi δingí ‘milk products’;
• Hsiung-nu 酥酪 *sā rāk ‘cream, butter’ – MIr. **sara-ka ← PIr. **sarah-, OInd. śaras- ‘milk skin, cream’;
• Hsiung-nu 醍醐 *tē-g(h)ā ‘light koumiss or clarified oil’ ← MIr. *doga-(ya) ← PIr. *dauga- → Cl.Pers. dōγ ‘buttermilk’, Pers. duγ, Kurd. dō ‘buttermilk’, Pashto lwaγ ‘yield of milk’, Shugn. důγ, Yazg. dəγ, Wakhi δi ‘buttermilk’;
• Hsiung-nu 單于 *tān-wa ‘Shan-yu (title)’. Pulleyblank: = Turk. *darxan, cf. *kraś-pin ‘Kashmir’. Cf. a commentary in Han Shu: “this title means “vast” and demonstrates that this person is as vast as the sky”. Cf. OUygh. tarqan- ‘to spread’, Caus. tarqar-;
• Hsiung-nu 閼氐 *γāt-tə̄́j ‘Shan-yu spouse (title)’ ← MIr. cf. Sogd. xuten ← *xwatāyn ← *hva-tāvyaini (f. from *hva-tāvya-) ‘lady’; → Early Saka *hvatuń → Hsiung-nu *γāt-tə̄́j → Sogd. xātūn [x’twn] ‘lady’ → OTk. qatun;
• Hsiung-nu 攣鞮 *r(h)wan dē ‘royal clan of Hsiung-nu’ ← MIr. cf. Kh.Saka runde ‘kings’ (NPl of rre ‘king’ ← *rwant-);
• Hsiung-nu 自次 *ʒ(h)jəś shjəś ‘second title after Shanyu’ – Chinese compound ‘self’ + ‘following’, can be a loan-translation of a well-known MIr. title – MPers. pasāgrīv, Parf. *pašāgrīv, Sogd. pš’γryw, pas/š- ‘after’ and grīv ‘body, self’;
• Hsiung-nu 日逐 *njət Łhəuk a title of a left-hand dajiang who, against the old custom could not be an heir and thus, was given this title. Cf. Av. negation noit, + Ir. *(v)rau-ka- ‘not-ruling’;
• Hsiung-nu 居次 *ka shjəś ‘princess’ ← MIr. *kan-čača (Sogd. knčy, knčyg, Pehl. kanīčag, Munji kinčäkä ‘girl’ ← PIr. *kan- ‘little, young’;

1.2.6. Late Old Chinese

No processes later than Early Postclassical Chinese can be observed in these contacts, dating thus to the 3^rd c. BC – AD 3^rd c.

Late Old Chinese (“cultural”) borrowings in Proto-Turkic (Dybo 2014):

• PTk. *(a)laču-k ‘cottage, small yurt’ ← LOCh. la-λiaʔ 廬 舍 ‘cottage’;
• PTk. (?)*ạltun, Chuv. ïltăn ‘gold’ ← LOCh. dōŋ 銅 ‘copper, bronze’;
• PTk. *gümüλ ‘silver’ ← LOCh. *kəm-liw 金 鐐 ‘bright silver’;
• PTk. *Tẹmür ‘iron’ ← LOCh. *tiēt-mhwit (= mwut) ‘iron thing’, dial. Tiēr-mwur 鐵 物;
• PTk. *könüg suv ‘quicksilver’ ← LOCh. *köuŋ 汞 ‘quicksilver’;
• PTk. *bẹk ‘a nobility rank’ ← LOCh. pẹ̄k 伯 ‘to be elder’;
• PTk. *čerig ‘army’, originally ‘army formation’ ← LOCh cjə̄t, dial. cjə̄r 節 ‘knee; detachement’;
• PTk. *sü ‘army’ ← LOCh. *śwò 戍 ‘frontier guards’;
• PTk. *biti- ‘write’ ← LOCh. pit: 筆 ‘writing brush’;
• PTk. *kujn ‘scroll, book’ ← LOCh. kwén 卷 ‘reel, coil, volume’;
• PTk. *bengü ‘eternal’ ← LOCh. mwə̄n 萬 ‘be ten thousand, myriad’ + LOCh 古 kṓ ‘to be ancient’;
• PTk. *čin ‘verity’ ← LOCh. 貞 tʽeŋ ‘to test, try out, correct’;
• PTk. *deng ‘equal’ ← LOCh. 等 tə̄ŋʔ ‘equal’;
• PTk. *dōn ‘clothes’ ← LOCh. tōn ‘black straight robe’;
• PTk. *kög ‘tune’ ← LOCh. 曲 khok ‘melody’;
• PTk. *sïr ‘color, dye, lacque’ ← LOCh. 漆 shjit ‘lacquer tree, lacquer (Rhus vernicuflua)’;
• PTk. *jinčü ‘pearls’ ← LOCh. 真 ćin ‘true’ + 珠 ćwo ‘pearls’;
• PTk. *čavlï ‘a species of falcon’ ← LOCh. 鷂 źawh ‘sparrow hawk (Accipiter nisus)’;
• PTk. *Turma ‘radish, horseradish’ ← LOCh. 土卵 thārhwān ‘yam’ (lit. “earthen egg”).

Proto-Turkic loanwords in Chinese Han transcriptions – Shi zi, Han shu (Dybo 2014):

• LOCh. 撐黎 *ṭhāŋ-rə̄j ‘sky’ ← PTk. *taŋrï;
• LOCh. 瀧 *roŋ ‘headquarters’ ← PTk. *orun;
• LOCh. 徑路 *kēŋh-rāh ‘sword’ ← PTk. *Kïŋrak;
• LOCh. 廓洛 *k(h)wā(k)-r(h)āk ‘belt’ ← PTk. *Kur-γak;
• LOCh. 服匿 *bwək-ṇək ‘a k. of vessel’ ← PTk. *bök-lüg ‘having a cork’;
• LOCh. 駃騠 *kwjāt-d(h)ē ‘a k. of horse’ ← PTk. *Kạtïr ‘mule’;
• LOCh. 騊駼 *Łhə̄w-Łhā ‘a small horse’ ← PTk. *ïlaλa ‘a bad horse’;
• LOCh. 驒 *d(h)ān-gēh/ kēh ‘wild horse’ ← PTk. *Takï;
• LOCh. 蛩蛩 *g(h)oŋ-g(h)oŋ ‘a k. of horse’ ← PTk. *Koŋur ‘brown’;
• LOCh. 橐駝 *thāk-lhāj ‘camel’ ← PTk. *tạj-lag ‘young camel’;
• LOCh. 毆脫 *γwā́-lwāt ‘nomad settlements’ ← PTk. *Koλ-ut;
• LOCh. 隔昆 *krēk-kwə̄n ‘Kirgiz’ ← PTk. *Kïrkïř;
• LOCh. 匈奴 *ŋōŋ-nhā ‘Hsiung-nu’ ← PTk. *hunga;
• LOCh. 屠耆 dā-grjəj ‘right’ ← PTk. *dogro ‘right’;
• LOCh. 谷蠡 *kōk-r(h)ə ‘patrimony’ ← PTk. *Kor(ï)γï;
• LOCh. 且居 *chiá-ka a title ← PTk. *čïka-n;
• LOCh. 稽粥 *kjə̄j-təuk N. pr. ← PTk. *Kạtïk ‘hard’;
• LOCh. 呼廚泉 *wā-ḍwa-ʒjwan N.pr. ← PTk. *otočï-n ‘healer’.

1.2.7. Pre-Proto-Mongolic

The close association of Turkic with the developments of Mongolic and Tungusic – whether as areal phenomena or as genetic relationship – suggests that their common location was intertwined, if not in the earliest stage, at least in the Late Proto-Turkic period.

Janhunen (1996: 242, 252), for example, sees a complex network of areal contacts between Pre-Proto-Bulgaric (i.e. Oghuric) and Pre-Proto-Mongolic in the pre-Xiongnu era, in support of his theory of an early split of the Oghur branch. In that sense, the earliest source of Turkic loanwords in early Mongolic seems to be Oghuric (cf. Róna-Tas 1998; Janhunen 1999; Schönig 2003, 2005; Golden 2011; etc.):

• Mong. ikere ‘twins’ ← Pre-PBul. *ikir (→ Hu. iker) vs. CTk. ikiz);
• Mong. hüker ‘ox’ ← Pre-PBul. *hekür vs. CTk. öküz;
• Mong. jer ‘weapon’ ← Pre-PBul. *jer vs. CTk. yäz;
• Mong. biragu ‘calf’ ← Pre-PBul., vs. CTk. buzagu;
• Mong. siri- ‘to smelt ore’ ← Pre-PBul., vs. CTk. siz- ‘to melt’.

Janhunen (2005: 140-141) suggests that these contacts with Pre-Proto-Mongolic peoples can be traced back to the 2^nd century AD in the steppes north of the Chinese borders, under control of the Xianbei, traditionally viewed – like the Khitan – as Para-Mongolic speakers (Janhunen 2006: 405-406). In any case, the intermediate position of Turkic between Mongolic-Tungusic on one hand and Chinese, Iranian, and Tocharian on the other constrains its geographical location to the westernmost area of the “Micro-Altaic” community.

Loans into Early Proto-Mongolic (before the 3^rd c. AD) also include (Dybo 2014):

• PT *ařïg ‘fang’ → PMo. ariγa ‘fang’ by PMo. aral ‘fang’, ara-tai ‘predator’;
• PT *bōř ‘grey’ → PMo. *boro ‘grey’ by PMo. *buγurul ‘grey-haired’.

NOTE. Among the many other (later) Old Turkic loanwords in Proto-Mongolic, omitted here for concision, there are many Chinese loans in Proto-Turkic, which are dated before AD 3^rd century (Dybo 2014).

A possible early Mongolic loan in Turkic is found in the Hsiung-nu language (Dybo 2014):

• Hsiung-nu 犀毗 slə̄j b(h)jəj ‘buckle’ ← Mo. *silbi ‘buttonhole’.

1.3. Palaeolinguistics

Pastoral nomadism very likely evolved in agricultural communities in which animal husbandry became the dominant economic activity. On the basis of this and other aspects of daily life reflected in this ancient (and sometimes reconstructed) vocabulary, the argument has been made for placing the Proto-Turkic homeland in the southern, taiga-steppe zone of the Altai-Sayan region.

Lexical material of Turkic provides clues about the topography, flora, and fauna of the Proto-Turkic territory that survived into its known branches (Golden 2011: 35-38):

It was located in a cold, northerly climate, subject to snow (qar), hail (tolı), ice (buz), fog (tuman) and rain (yağ- “to rain”), one in which “whirlwinds” of snow (or sand, qasurqa, qasırqu) were not unknown. There were “snow storms” (tipü or tüpi, qâḏ → qay, borağan) as well as other forms of extreme, inclement weather.

It was a land of mountains (tağ, qır), massive rocks or rock piles (qorum), cliffs (qaya), forests and dense thickets (orman, yıš), groves of woods or thickets (bük) around flowing water valleys (öz), ravines (yar), flatlands and plains or steppes (yazı, qayır), with sand (qum) in some places, swamps or saltmarshes (qaq) in others and traversed by rivers (ögüz, yırmaq “big river,” özen “brook”) and lakes (köl). Larger bodies of water were not unknown (teŋiz/tengiz “sea,” talay “ocean, sea”).

The area had an abundance of “wild game” (keyik), e.g. elig (“roe”), buğu (“[male] deer”), sığın (“stag”), buŋğaq/muyğaq/muŋgaq (female of sığın), ıvıq (“gazelle”), yegeren (“antelope”), bulan (“elk”), qulan (“wild ass/onager”), toŋguz (“wild boar”), arqar (mountain sheep/ram”), teyiŋ (“squirrel”), qoḏan, tabušqan (“hare”), kiš (“sable”), qama (“beaver”), tilkü (“fox”), as well as predatory animals such as lions (arslan), tigers, (bars) and panthers (irbiš, yalbars).

There is an extensive vocabulary for domesticated animals, closely tied to the pastoral nomadic economy followed by the Proto-Turks: horses (at as well as aḏğır “stallion,” bé, biye “mare,” qısraq “young mare,” baytal “barren mare,” qulun “colt up to two years of age” etc.), cattle (uḏ), cows (iŋek, sığır “milk-cow”), oxen (öküz), rams and sheep (qoč, qoyn, qazı “lamb”), camels (teve, buğra “male camel,” ingen “female camel”), asses (eškek), swine (čučqa), dogs (ıyt, köpek, qančıq “bitch,” eker “wolfhound”), cats (pišik, četük, mačı).

There are also numerous words for different kinds of falcons (toyğan [toğan], lačın, toğrıl), some of which were probably trained by humans for hunting and other birds (e. g. bürküt “golden eagle”), reptiles, fish and insects.

There is also a rich vocabulary for trees (terek “poplar,” tağaraq “white poplar,” emen “oak,” qaḏın “birch,” qaḏı “pine,” bȫš “cedar,” kebrüč “ash tree,” *yerük “alder” as well as berry-bearing trees, plants etc, cf yımurt “cherry-tree”), grains (tarığ “cereals, millet,” yögür “millet,” qoñaq “coarse millet,” buğday “wheat,” arpa “barley” etc. ) attesting a familiarity with agriculture (tarlağ “ploughed field,” azal “wooden plough,” sarpan “plough,” orğaq “scythe” etc.).

There is a conspicuous lack of Bulgar cognates for Common Turkic agriculture-related derivatives from non-agricultural roots, which might suggest that Oghuric peoples lost contact with a developing agricultural late Proto-Turkic community among millet-oriented centers close to the Altai. On the other hand, a few very archaic, non-derived verbs can be reconstructed for the parent language, such as *ek- ‘to sow’, *ôr- ‘to reap, to harvest (a crop)’ and *Tarї- ‘to cultivate (ground)’. This includes two names for cereals, definitely pointing to an agricultural tradition in the Proto-Turkic community which might have been lost early among expanding Oghuric steppe pastoralists.

In fact, two roots meet formal requirements to be regarded as Proto-Turkic (Savelyev 2017): *ügür (cf. Chuv. vir ‘millet’), and *dạrїg ‘cereal’ ← ‘that which is cultivated’, from *TArї- ‘to cultivate (ground)’. Interestingly, there are no examples of a semantic shift ‘millet’ → ‘wheat’, but there are two in the opposite direction – both of non-Turkic origin. This is paralleled by the recent mix or unification of names for ‘millet’ with names for ‘corn’, which probably arose from corn becoming the most important cereal (Stachowski 2008).

NOTE. Proto-Mongolic and Proto-Tungusic cognates might be reconstructible for some shared pastoralist loanwords, whereas some non-borrowed agricultural terms have also parallels. This suggests at least a shared ancestral community, if not a genetic relationship between these languages. Recent attempts to locate the “Proto-Altaic” homeland tracing the expansion of millet cultivation are based on an often hypothetical linguistic basis (cf. Savelyev 2017 cited above), and on an incomplete archaeological-archaeobotanical picture combined with highly complex temporal and geographic transect of genomic data. For more on this question, see e.g. the open source Evol. Hum. Sci. (vol. 2) 2020 collection Transeurasian millets and beans languages and genes.

1.4. Onomastics

1.4.1. Hydronymy

Turkic hydrotoponymy in the Eurasian steppes seems not to be well studied. Due to the mobile nature of Asian (forest-)steppe pastoralists, this field is in dire need of a comprehensive reappraisal of the works of Matveev, especially regarding the relative chronology of the assumed earlier (Yeniseian, Uralic, Iranian) as well as later (Mongolic) layers in medieval and modern Turkic-speaking areas of Central Asia. Further, there seems to be little information on the described formants from Central Asia, and even less so from China or Mongolia.

The more recent summary by Maloletko suggests that formants in -ba (-be) can be generally regarded as Samoyedic, with those in -as having known parallels in Ugric hydronymy, whereas those in -rap, -rop, -rep, -djan, or -man have obvious parallels in Iranic. Variants of chu(j-), katun, idil’, or’, ugun, ozëk, or özö are found in Turkic dialects, but the use of many of them as whole denominators in the Upper Ob’ and Yenisei river basins (indicating a recent use, much like Selkup hydronymic area shows recent Samoyedic formants) does not help establish an accurate chronology. Interestingly, some are used as second formants closer to Lake Baikal and to the south of the Samoyed area (cf. -su/sukh, -katun), so they they appear to have been later renamed by incoming Samoyed, Turkic, or Mongolic speakers.

An example of conflicting interpretation of place and river names might be found in the first component of the river Irtysh, potentially traceable to O.Tk. iyir ‘wandering’, with a meaning ‘meandering river’ (Saparov, Chlachula, Yeginbayeva 2018). This would support a quite recent use of the Yeniseic formant ‘river’ to refer to the Irtysh before its renaming by Turkic peoples. This is, however, disputable, since an also questionable Yeniseic origin for the first component can be proposed, based on other examples from the Yeniseic hydronymic area.

1.4.2. Ethnonymy

The Dingling to the north of the Xiongnu, extending from the Irtysh region to Lake Baikal and the Middle Yenisei (2^nd c. BC) are frequently mentioned as the first tribe to display reliable Oghur Turkic ethnonyms. This is due to their likely association (including its name) with the later Oghuric confederation of nomadic tribes (AD 460s), the T’ieh-lê ← Tk. *tegrek ‘rim, ring’ hence ‘wheel; cart’; and also with the Ke-k’un (also Chien-k’un = Qïrghïz) and the Hsin-li (also Hsüeh = Sir?), which were the first recorded Turkic peoples or, in the case of the Qïrghïz, possibly a Palaeo-Siberian people under Turkic leadership and in the process of Turkicization. Other early Turkic peoples probably include the Hu-Chieh or Wu-chieh (*Hagaŕ = Oghur? Oghuz). (Golden 2006).

Ethnonyms of Turkic origin might also lie behind some of the few recorded names of the Xiongnu. In particular, the succeeding Hunnic tribes probably display an Oghuric Turkic or “Pre-Proto-Bulgharic” form (cf. Janhunen 1996: 185-189). Even western Hunnic kings and nobles before and after Attila show highly probable Turkic etymologies, despite the generalized use of Germanic languages by their subjects, attesting to the continued language use among elite families (cf. Brosseder 2018):

From Mundzuk (Attila’s father, from Turkic Munčuq = ‘pearl/jewel’), Oktar/Uptar (Attila’s uncle, Öktär = ‘brave/powerful’), Oebarsius (another of Attila’s paternal uncles, Aïbârs = ‘leopard of the moon’), Karaton (Hunnic supreme king before Ruga, Qarâton = ‘blackcloak’), Basik (Hunnic noble of royal blood, early fifth century, Bârsiğ = ‘governor’), Kursik (Hunnic noble of royal blood, from either KÜrsiğ, meaning ‘brave or noble’, or Quršiq meaning ‘belt-bearer’). All three of Attila’s known sons have probable Turkic names: Ellac, Dengizich, Hernak, and Attila’s principal wife, the mother of the first son Ellac, has the Turkic name Herekan, as does another wife named Eskam (Ešqam = ‘companion of the Shaman’).

2. Archaeology & Population Genomics

2.1. Agriculture

Domesticated broomcorn and foxtail millet grains from Inner Mongolia directly dated to the mid-6^th millennium cal. BC suggests that millet flowed relatively rapidly out of China to the northeast. However, the Xinjiang region and Eurasian Steppe remained devoid of broomcorn millet until the early 2^nd millennium cal. BC. Broomcorn millet was present on the western edges of the Eurasian Steppe in southeastern Kazakhstan by 2200 cal. BC, somewhat later than wheat, and in Turkmenistan and Afghanistan by the mid-2^nd millennium cal. BC. Overall, evidence for broomcorn millet in the Eastern Steppe at this time is sparse, limited to late 2^nd to early 1^st millennium BC paleobotanical remains and impressions visible in ceramics from the Trans-Urals, and early 1^st millennium BC to 1^st cent. AD evidence of cultivation from the Minusinsk basin (Tagar culture), and later in the northern foothills of the Altai (Hunnic Maima culture), as well as late 1^st millennium BC from Mongolia (Miller & Makarewicz 2019).

Modern archaeology shows that quintessential nomads from Central Asian steppe confederations such as Scythians or Xiongnu practiced a combination of semi-nomadic pastoralism and irrigated agriculture in sedentary, agrarian elements of their polities. In fact, LBA cultures from Western Siberia (Barkhatovo, Irmen, and Alekseyevskoye-Sargary cultures) show cultivation of wheat, barley, and rye at the turn of the 2^nd and late 1^st millennia BC. The appearance of grain bruisers, hoes, and sickles, as well as the presence of irrigation structures make it possible to include the forest-steppe Tobol region, Northern Kazakhstan, Baraba, and the forest-steppe Ob region into a tentative area of agriculture (Ryabogina & Ivanov 2011).

NOTE. The traditional picture of highly specialized nomadic cultures in conflict with neighbouring sedentary populations is thus incorrect for most ancient cultures, since many of these complex polities show mixed and complex economies. See the recent Spengler et al. (2021) for a full discussion of this topic.

Intensification of wheat and barley consumption in the Trans-Urals occurred during the Iron Age when Sauro-Sarmatian and Sargat interaction spheres spread across large swaths of the Eurasian steppe, evident in shared prestige good assemblages, warrior equipment, and mortuary rituals. This region was deeply involved in exchange with areas in central and southeastern Kazakhstan that consumed millet, yet Trans-Ural groups opted instead to grow wheat and/or barley. Early evidence for a dietary focus on wheat and/or barley, rather than millet, suggests that Trans-Urals populations may have been trading these cultivars to other areas of the steppe that lacked the water resources to cultivate these crops. Investments in different farming technique or a higher value placed on wheat and/or barley may also have been factors affecting the decision to focus on the production of these grains (Miller & Makarewicz 2019), and the formation of genetic varities also informs of their evolution (cf. Matuzeviciute, Mir-Makhamad, & Spengler 2021).

2.2. Middle/Late Bronze Age (ca. 1900-900 BC)

The Middle/Late Bronze Age (MLBA) in Mongolia is characterized by the sudden and widespread appearance of monumental mortuary architecture across the region. Primarily taking the form of stone mounds, but also including stone stelae and other features, possibly ultimately stemming from the Afanasievo tradition continued by Chemurchek and Okunevo.

This section is divided according to the two main ancestry groups that can be distinguished during the MLBA in Mongolia, excluding outliers. The following texts on archaeology and population genomics are modified from Jeong et al. (2020) supplementary materials (under a CC-BY license):

2.2.1. Mönkhkhairkhan / Deer Stone-Khirigsuur Complex

Mönkhkhairkhan (1850-1350 BC). Dating to after the Chemurchek period, ca. 1850-1350 cal. BC (Taylor et al. 2019), Mönkhkhairkhan burials are like their predecessors concentrated in northern and western areas of Mongolia, and characterized by a flexed leg burial position. Like Afanasievo, these features have a flat surface morphology and a central pit burial, and they may be either circular or rectangular in overall shape – although unlike Afanasievo graves they are not constructed with a perimeter of upright slabs. Artifacts recovered from within these features include animal bones (wild, and perhaps domestic), bone jewelry, copper or bronze tools, and bone tools (Clark 2015; Eregzen 2016).

Burials of this type represent the end of the flexed-leg burial tradition, which was replaced by prone (facedown) and supine traditions in the mid-2^nd millennium BC. Because of the scarcity of these features, very little can be said about the economy of this culture or period, although the presence of wild animal fauna in at least one burial indicates a role for hunting, while connections with Chemurchek and Afanasievo would suggest the presence of domestic livestock.

Baitag (1050-900 BC). Found in a restricted region of western Mongolia, Baitag burials consist of nonmounded, small stone rings constructed from a single layer of small flat stone slabs. A central burial pit oriented west-east contains a single individual oriented in a supine position with knees flexed.

Unlike Deer Stone/Khirigsuur burials but similar to preceding Altai groups, such as the Mönkhkhairkhan and Chemurchek, the Baitag burials contain various small grave goods, including bronze and stone jewelry. These artifacts share similarities with those included in Karasuk culture graves from the Minusinsk Basin, as well as in burials of Xinjiang and Gansu (Sibu culture) in northwestern China (Kovalev & Erdenebaatar 2009).

Deer Stone-Khirigsuur Complex (DSKC) (1350-900 BC). This culture comprises three different monumental features – khirigsuurs, deer stones, and sagsai-style graves – and is tightly associated with the emergence of horsemanship in the Mongolian Steppe during the late 2^nd millennium BC. In general, DSKC sites are concentrated in the western, northern, and central parts of Mongolia, with only a small number of sites further east (Honeychurch 2015a). Radiocarbon modeling dates khirigsuurs to between ca. 1350-900 cal BC, deer stones to ca. 1150-750 cal. BC, and places the emergence of DSKC horse ritual at ca. 1200 cal. BC (Taylor et al. 2019, 2017).

Khirigsuurs are large stone mounds, surrounded by an exterior fence that is either circular or rectangular in shape. Although their exclusive function as burials is a subject of contention (Wright 2012), khirigsuurs often contain a supine human body (Littleton et al. 2012) and do not typically yield other kinds of artifacts. Deer stones are anthropomorphic standing stones found either independently or co-occurring with khirigsuurs. Deriving their name from the common motif of stylized deer, carvings on these stelae also depict belts, weapons, and tools – and occasionally even a human face.

Many of the weapons depicted on deer stones are of recognizably Karasuk style, bearing a strong resemblance to bronzes found in tombs in the Minusinsk Basin more than 500 km to the northwest (Honeychurch 2015b), and the presence of deer stones in nearby Tuva further support the possibility of long-distance interaction between the Karasuk and the DSKC (Honeychurch 2015b).

At many Mongolian khirigsuurs and deer stones, smaller stone mounds containing the head, jaw, neck, and hooves of individual horses are found surrounding the eastern perimeter of the monument transport and likely riding, as well as their sophisticated management as herd animals (Taylor et al., 2015, 2018). Another kind of satellite feature found at DSKC sites, open stone circles, often yield partial remains of sheep, goat, or cattle.

Sagsai-style graves (1350-1050 BC) are often associated with the DSKC culture. These burials are also referred to as “slope burials” because of their common occurrence on the edge of hillslopes. Alternate names include Munguntaiga and even khirigsuur. Sagsai burials are similar to khirigsuurs in that they also have a supine burial position. They are sometimes (although not always) mounded, and can occasionally have a small external fence. Sagsai burials are either rectangular or circular; however, unlike khirigsuurs, these features have four upright corner posts and do not have external satellite mounds Similar to other DSKC features, they are concentrated in western and northern areas of Mongolia.

Altai_MLBA cluster: contains seven individuals* from the Altai-Sayan region, who show a gene pool associated with Khövsgöl_LBA/Baikal_EBA and incorporating a substantial genetic influx from Steppe_MLBA-related populations, although ancestry proportions estimates within this group vary along a cline (Jeong et al. 2020).
*BER002, BIL001, ULZ001, ARS026, SBG001, ULI001, ULI003

NOTE. The genetic profile of one outlier from Khövsgöl_LBA (ARS026) also falls within the Altai_MLBA group (see above Proto-Samoyed Homeland). Of note, one member of this group, ULZ001, is found not in the Altai, but in far eastern Mongolia..

Regarding external contacts, three outliers are representative of them:

• UAA001 (Mönkkhairkhan) from the Altai is well-fitted with 3-way admixture model using Afanasievo, Baikal_EBA and Gonur1_BA, despite the fact that they date to ~1500 years after the Afanasievo culture.
• KHI001 (unclassified culture) from the Altai, is well-fitted with 3-way admixture model using Steppe_MLBA, Baikal_EBA and Gonur1_BA (p-value=0.056), presenting minor genetic component from Gonur1_BA, although it can also be modeled as a 2-way admixture between Afanasievo and Khövsgöl_LBA (p-value=0.117).
• UUS001 (DSKC) from Khövsgöl province is well-fitted with 3-way model using Steppe_MLBA, eastMongolia_preBA and Gonur1_BA.

NOTE. For the previously published Khövsgöl_LBA (seventeen individuals), see above Proto-Samoyed.

In terms of haplogroups, Altai_MLBA and related samples* show:
*Especially Wang et al. (2019) has many samples with broad, general labels due to the lack of proper archaeological context. Here, Y-DNA is used to make tentative genealogical connections.

• Q1b-L54 in Sagsai (4), Mönkhkhairkhan (3), Khövsgöl (9), most of them L330, probably all within Q1b-BZ180, as are 3 “Mongolia_LBA_CenterWest_4” and 1 “Mongolia_BA_1” (see map of ancient Q-BZ180 and timeline slideshow):
• Basal Q-FT142440 branch in I12973 (“Mongolia_BA_1”) and basal subclade Q-Y146631 in Khövsgöl ARS007, ARS008, and ARS016. This haplogroup is also found later in Tian Shan Saka DA47, of subclade Q-BZ366.
• Q-BZ181 branch as Q-FT421589 in Mönkkhairkhan I6348, and its subclade Q-FTA37655 in I7039 (“Mongolia_LBA_CenterWest_4”).
• Basal Q-F15008 branch in I13767 (“Mongolia_LBA_CenterWest_4”), and later in one Pazyryk sample from Berel (BRE002) and two Huns from the Korgantas phase (DA20, KBO001).
• R1a-Z93 in Sagsai (6) and a Khövsgöl individual:
• Basal R-Z2124(*/Z2125*) is found in ARS026, the Khövsgöl Steppe_MLBA outlier (see above Proto-Samoyed Homeland).
• R-S23592 in ULI001, I6363, I7033, the latter with downstream basal R-YP1456, found earlier in a Krasnoyarsk MLBA individual (see above Proto-Samoyed Homeland), and later widespread among early “Scythian” nomads, including Saka, Tasmola, Uyuk, and also among the Xiongnu and Turks (see map of ancient R-YP1456 and timeline slideshow).
• N-M231 is found in Mönkhkhairkhan (5), DSKC (1) & Sagsai (1), and Khövsgöl (1):
• N2-Y147658 (pre-Y148364?) in Mönkhkhairkan Afanasievo-related outlier UAA001 (ca. 1770 BC), an N2 branch also found as N-Y148364 in I13768 (“Mongolia_LBA_CenterWest_4” ca. 1125 BC), and as basal N-Y147658 in a “Pre-Scythian” Cimmerian from the Mezőcsát culture in Hungary (ca. 900 BC) and later in a Wusun individual from the Tian Shan. The parent N-MF52704 clearly stems from Cis-Baikal Neolithic, and it is very likely that the potential N2 found in Sargat’s Mt. Bitiya belongs to this subclade that spread with Iron Age nomads (see map of ancient N-MF52704 and timeline slideshow).
• N1a-L729>F1360>pre-F4309 in Sagsai I12976, which links him distantly to the Gonur1_BA-related outlier (of unclassified culture) KHI001 of hg. N-CTS11713. This haplogroup is found widespread among likely Turkic and Mongolic speakers, including Pazyryk, Xiongnu, Yakutia IA and MA, a Mongolia Türk and Mongols, and has an ultimate origin in the Trans-Baikal area, based on the finding of sister clade N-F4309>FT210118, and on an Iron Age N-pre-F1360 sample from the Amur region (see map of ancient N-F1419 and timeline slideshow). This suggests that its nature as outlier is due to recent female exogamy.
• Basal N1a-L729>Tat>F1419>Y24317>Z35159 in Mönkhkhairkan I12955 (ca. 1875 BC) and DSKC BER002 (ca. 1230 BC), a haplogroup connected to the Cis-Baikal area from the Eneolithic I14460_I14461 (ca. 4500 BC) and Ust’-Ida LN DA345 (ca. 3500 BC). It is unclear which branch is behind the N1a-F1419 (xM2005>M2019,xZ35159>B187) found in Khövsgöl ARS003.

NOTE. Weirdly enough, sample I12955 (1488-1308 calBC) is supposed to be a 1^st or 2^nd degree relative of the distant I12506, classified as Ulaanzukh (1438-1299 calBC). Their cultural attribution and actual relationship is difficult to interpret, due to the warnings about the date and different skeletal sample of I12506, and its low coverage. Its inferred Y-DNA haplogroup BT also does not help clarify the situation.

• Q1a-F4743 in Khövsgöl ARS015, with an ancestry in common with Altai_MLBA peoples despite sharing Y-DNA with Ulaanzuukh-related individuals (see below).
• Undefined C2a-F6301 is reported for I12975 (“Mongolia_LBA_CenterWest_4” ca. 1160 BC), belonging to a branch widespread among Circum-Baikal Neolithic individuals and populations from neighbouring regions (see map of ancient C-F6301 and timeline slideshow).
• Undefined J1a-P58 in Gonur1_BA-related DKSC outlier UUS001 from Khirigsuur, a branch in common with Iran Neolithic populations that is later found spreading with nomads, hence probably with an origin in population movements through the Inner Asia Mountain Corridor (see map of ancient J-P58 and timeline slideshow).

2.2.2. Ulaanzuukh/Shape Burial (1450-1150 BC)

Beginning in the mid-2^nd millennium BC, a number of different burial traditions emerged in the southern and eastern regions of Mongolia. United by a common prone or face-down burial position, these groups are sometimes considered a single cultural unit, and other times classified separately as discrete burial types (Honeychurch 2015b).

• Ulaanzuukh features, named after the type site in southeastern Mongolia, are non-mounded square or rectangular features with a wall of upright slabs or layered stone and a central pit.
• Shape burials, also called Tevsh, ant-shaped, hourglass-shaped, and other names, are similar, but with a waisted hourglass-style edge construction. These features are typically made of layered stone, and sometimes with a single edge ringed with upright slabs.
• Other variations often included within this culture group include D-shaped or stirrup shaped graves with a prone body position.

Burials of this culture often contain apparently domestic livestock remains, including sheep, goat, horse, and cattle (Nelson et al. 2009), although the earliest horses from these features date to only ca. 1250 BC (Taylor et al. 2017). Recent analysis of proteins in human dental calculus has confirmed the utilization of ruminant dairy products and the presence of domestic animals in the Ulaanzuukh economy (Wilkin et al. 2019).

A few bronze knives of Karasuk origin have been found in Ulaanzuukh-Tevsh graves, indicating possible long-distance connections to the Minusinsk basin (Honeychurch 2015b).

Ulaanzuukh_SlabGrave cluster: contains 11 individuals* with Ulaanzuukh burial type and 5 individuals with Slab Grave burials (see below), from eastern Mongolia. They all are classified into one single genetic group given their strong genetic homogeneity with ANA and the geographic links between the two.
*BUL001, BUL002, ULN001, ULN002, ULN003, ULN005, ULN006, ULN007, ULN009, ULN010, ULN015

This clustering of Ulaanzuukh and Slab Grave confirms previous archaeological hypotheses that the Slab Grave culture likely emerged out of the Ulaanzuukh gene pool. This genetic cluster also explains another Khövsgöl_LBA outlier – ARS017, who now genetically falls within the Ulaanzuukh_SlabGrave group, as well as a single individual with unknown burial type from central Mongolia, TSI001, who also falls into this cluster. Of note, one male Mönkhkhairkhan individual (KHU001) also has a large proportion of ancestry from Ulaanzuukh_SlabGrave in addition to his main genetic component from Baikal_EBA.

Together, two females (ARS017, TSI001) and KHU001 of hg. Q1b-L330 suggest contact with the Ulaanzuukh_SlabGrave group in northern, central Mongolia, even though these individuals were buried according to local burial customs. Overall, this Ulaanzuukh_SlabGrave genetic cluster is a continuation of the ANA easternMongolia_preBA gene pool (represented by SOU001) of 3,000 years earlier (Jeong et al. 2020).

In terms of haplogroups, Ulaanzuukh samples (ca. 1400-1200 BC) show hg. Q1a-F4743 (formed ca. 4000 BC, T_MRCA ca. 2700 BC) in BUL001, BUL002, I12960, I14037, and also its subclade Q-FT394800 in ULN001, ULN007. Other samples from this branch of Q-M120 include an earlier YangShao MN XW-M1R18 from the Yellow River (ca. 5000 BC), as well as a later Cimmerian from Glinoe Sad (ca. 870 BC) and a central steppe nomad from Alai Nura (ca. AD 300), with Q-M120 becoming later widespread among nomadic populations connected to the Trans-Baikal area (see map of ancient Q-F4743 and timeline slideshow).

2.3. Early Iron Age (ca. 900-300 BC)

The following texts on archaeology and population genomics are modified from Jeong et al. (2020) supplementary materials (under a CC-BY license):

2.3.1. Slab Grave

Beginning around 1000 BC, a new burial style known as Slab Grave began appearing in eastern Mongolia. Slab graves are so called because of the large stone slabs used to mark the surface of the burial and to contain the rectangular burial space (hence in Mongolian they are called “square burials”) wherein single individuals are interred (Tsybiktarov 1998). Although occasionally found singly, Slab Grave burials are more typically grouped into small cemeteries (Honeychurch 2015a). Stone slabs are set upright in the ground, and are thus prominent grave markers.

The burial pits are quite shallow, and human remains are rarely found complete or in good preservation. Over time, the Slab Grave culture expands northwards into eastern Baikal and westwards into central Mongolia, where it intrudes into former DSKC territory. Some slab graves tear apart the stone structures of khirigsuurs to construct the graves, and some even reuse deer stones for standing corner stones or laid-down slabs within the burial pit (Honeychurch 2015a).

Unlike earlier Bronze Age burials, grave goods become more common in Slab Grave burials, consisting primarily of bronze beads, buttons, and small ornaments, as well as horse gear, arrowheads, axes, and knives. Stone, ceramic, and bone artifacts are also found in slab graves, and a few burials contained tripod-shaped pottery similar to those from Inner Mongolia and Manchuria or other non-local grave goods such as turquoise and carnelian beads from Central or South Asia (Honeychurch 2015a).

Portions of livestock are often set at the edge or just outside of the rectangular burial space. In addition to faunal remains demonstrating the presence of domestic animals in the Slab Grave economy, recent analysis of proteins in human dental calculus has confirmed the utilization of ruminant and horse dairy products (Wilkin et al. 2019).

Although the Slab Grave phenomenon emerges out of the former territory of the Ulaanzuukh culture, archaeological evidence for the relationship between these two groups has been ambiguous. Nevertheless, the similarity of bronze artifacts, especially relating to horse gear and weaponry, found at Slab Grave sites to similar artifacts found in the Altai, Tuva, and Minusinsk regions may indicate a continuation of previously established long-distance relationships between these regions (Honeychurch 2015b).

In addition to the eleven Ulaanzuukh burials described above, four Slab Grave individuals (BOR001, DAR001, MIT001, SHU001) from eastern Mongolia also presented a homogeneous genetic profile with Ulaanzuukh (Ulaanzuukh_SlabGrave cluster). This includes another Slab Grave individual from Trans-Baikal, PTO001, which is consistent with an archaeologically described expansion of the Slab Grave culture into the Baikal region during EIA (Losey et al. 2017).

Haplogroups of the Slab Grave include:

• Q1a-(pre-?)M120 in I12969, I6352, I6349, I6353, and possibly also in MIT001, although in some cases there are no calls for Q-FT9308. All of them likely represent the spread of a branch that split before the full-fledged Q-M120. Basal Q1a-FT9308(or Q-pre-F11089), derived from Q-M120, is found in I6359, suggesting that Ulaanzuukh/Slab-Grave-related groups were the ultimate source of expansion of Q-M120 lineages found later among Iron Age nomads, including Tasmola and Xiongnu, but also Zhou Dynasty vassal Peng Clan, or the post-Yanbulake Shirenzigou M15-2 (see map of ancient Q-M120 and timeline slideshow).
• Q2b-Z19128 in DAR001, found earlier in BMAC (ca. 1800 BC) and later among SPGT Loenbar IA and Saidu Sharif, as well as in a Tian Shan Hun, suggesting a potential connection with the Inner Asia Mountain Corridor (see map of ancient Q-Z19128 and timeline slideshow).
• N1a-pre-M2019 in I6365, probably ancestral to N4a1 (in turn probably splitting the earlier N-M2019 trunk, see above Ymyyakhtakh & Yukaghir), with later samples found among Khazars, Early Magyars, and Yakuts connected to the medieval Turkic expansion (see map of ancient N-M2019 and timeline slideshow).
• Q1b-BZ180 in I6359, a haplogroup common in “western” Cis-Baikal BA- and Altai MLBA-related groups (see above).

2.3.2. Scytho-Siberians

The Uyuk culture (ca. 700-200 BC) is centered in the Upper Yenisei River area, in modern day Tuva, with some extensions into northwestern Mongolia (Murphy 2003; Savinov 2002), referred to as the Aldy-Bel or Sagly-Bazhy culture, and is known best in Mongolia by the thoroughly excavated site of Chandman Mountain (Tseveendorj 1980).

Graves were marked by a round pile of stones and are often found in cemeteries of one to two dozen graves. Beneath the stone mounds are large log chambers containing several individuals (often assumed to be kin as they include men, women and children) all laid in partially flexed positions on their sides. Portions of sheep are also often placed in the graves. The Uyuk log chambers resemble similar log architecture constructed by the contemporaneous Pazyryk culture in the Russian Altai and surrounding areas, and both the Uyuk and Pazyryk have been associated with the broader Saka culture (Parzinger 2006).

Similar to Slab Graves, recent analysis of proteins in human dental calculus has confirmed the utilization of ruminant and horse milk among those at Chandman Mountain (Wilkin et al. 2019). Isotopic studies have also shown that some Uyuk communities, including at Chandman Mountain, had a significant amount of millet in their diet (Murphy et al. 2013; Ventresca Miller & Makarewicz 2019; Wilkin et al. 2019). This links them to agropastoralist cultures of the southern steppe and Central Asia, where millet cultivation was widely adopted during the westward spread of the cereal from China to the Caucasus during the 2^nd millennium BCE (Ventresca Miller & Makarewicz 2019).

Majemir (from ca. 900 BC) is followed by the Pazyryk culture (ca. 500-200 BC), known mainly for its type site of Pazyryk, whose large tombs contain numerous exotic imports, including silks from China and textiles from Achaemenid Persia (Rudenko 1970). Pazryk burials are found mostly within the northern Altai areas of Russia, far eastern Kazakhstan (Samashev 2011) and northwestern Mongolia (Törbat et al. 2009). Similar to Uyuk and other ‘Saka’ style graves, Pazyryk burials are marked by round piles of stones. Beneath these stone piles, however, most Pazyryk graves have smaller wooden chambers with only one or two persons; their size and burial goods vary greatly, though many of them are accompanied by whole horses laid beside the burial chamber (Kubarev and Shul’ga 2007).

Pazyryk is important to consider because the northern Altai practice of whole horse burials later appears in scattered central Mongolia cemeteries of the subsequent Xiongnu period. Genome-wide data from Pazyryk individuals have been previously reported from site of Berel in the Altai region of Kazakhstan (Unterländer et al. 2017).

Chandman_IA cluster: samples from the Altai-Sayan region in western Mongolia have a genetic profile that matches the preceding Altai_MLBA cline [Krasnoyarsk_MLBA is used as the representative for Central Steppe_MLBA group for admixture modelling because it is geographically closest to the test EIA groups]. Krasnoyarsk_MLBA + Baikal_EBA fails to adequately model the Chandmand_IA cluster, as does Karsnoyarsk_MLBA + Khövsgöl_LBA. Further changing the steppe_MLBA source from Karsnoyarsk_MLBA to Sintashta_MLBA did not rescue the 2-way admixture model. A 3-way admixture model by adding Iranian-related ancestry as the third source, using a BMAC group from the Gonur Tepe site (Gonur1_BA) as a proxy, Jeong et al. (2020) observe 51.3% of Steppe, 42.2% of Baikal_EBA and 6.5% of Iranian ancestry in Chandman_IA.

This ancestry is found among the Tagar from the Minusinsk Basin, Central Saka from central Kazakhstan, Kazakhstan_Berel_IA from eastern Kazakhstan, and Tian Shan Saka from Kyrgyzstan. Iranian-related ancestry proportions range from ~7-28% in the tested Iron Age groups, while it is not required to model for Final Bronze Age Karasuk. In particular, the Tian Shan Saka, geographically closest to the Gonur Tepe site, has the highest amount of estimated Iranian-related ancestry.

NOTE. For the Iron Age Eastern Steppe, genetic data alone can only narrow down the source of the Iranian ancestry to a broad region east of the Caspian Sea. Taken in context, the authors of Jeong et al. (2020) propose that this ancestry likely arrived via a local contact around the Transoxiana/Sogdiana region (i.e., the border between Kazakhstan, Uzbekistan and Kyrgyzstan).

Estimates of admixture dates (DATES) corroborate the different chronological and geographical development of these cultures: the estimated admixture date between Sintashta and Baikal_EBA for the Karasuk and Tagar is consistent with the admixture date observed in Altai_MLBA – at around 3,500 BP, whereas for Central Saka, Pazyryk (Kazakstan_Berel_IA) and Uyuk (Chandman_IA), the admixture date (including also Gonur1_BA) is estimated to be a few centuries later, with the most recent admixture date estimated for the Saka from the Tian Shan (Jeong et al. 2020).

Haplogroups among Uyuk samples include:

• R1a (19, including 7 STR-based), at least 12 (probably all) R-Z93:
• R1a-Z2125 (at least 10, probably 11), and among them R1a-S23592 is found at least in 8, with some ancestral (CHN008, CHN012, I7030, I6233, I6224) and three derived for R-YP1456 (formed ca. 2400 BC, T_MRCA ca. 2200 BC), one basal (I0577) and two (CHN001, I7027) showing basal subclade R-PH1397 (T_MRCA ca. 1500 BC). This specific subclade is found widespread among the Early Iron Age Sakas, although its R-FGC56408 branch is found earlier in Krasnoyarsk MLBA, and the immediate upstream R-YP1456 in Karasuk and Altai_MLBA (see map of ancient R-FGC56408 and timeline slideshow).
• Basal R1a-Y41571 (formed ca. 1800 BC, T_MRCA ca. 1500 BC), found to date also in Wusun DA223 from the Tian Shan (ca. 300 BC) and as subclade R-Y195965 in a “Hun-Sarmatian” nomad from the central steppe (ca. AD 450). As a subclade of R-FGC82884, it seems to be more directly related to Iranic-speaking peoples (see map of ancient R-FGC82884 and timeline slideshow).
• Q1b, basal Q1b-L53 in CHN003, but most (at least 13, probably 15, counding 5 STR-based) are Q-L330:
• Q1b-YP771>BZ433>L332 (fomred ca. 3200 BC, T_MRCA ca. 1600 BC) in I12970, I6231, I6230, CHN007, a subclade also found in Pazyryk, in a Western Scythian and in a Sarmatian sample, as well as among Huns. Probably all are Q-Y145452, like I7029 and CHN016 – a subclade also found in an early and a late Xiongnu – with the highest coverage one (I6232) showing subclade Q-FTA37701 (see map of ancient Q-Y145452 and timeline slideshow of Q-L332).
• Q1a-FT414149 in I7022, shared with one Okunevo EBA and one Karasuk individual. Sister clade Q-L715 is found spreading later with Huns and Turks (see map of parent Q-YP787 and timeline slideshow).
• N-M231 in ARZ-T15 (STR-based), without further reported subclade.

Haplogroups among the few sampled Pazyryk show:

• R1a-S23201 in I0563 – a subclade of R-FGC82884 and sister clade of R-Y41571 found in Uyuk – later found in a Kangju (ca. AD 250), and also in three Huns (ca. AD 300-400) under a specific subclade R-S10885.
• Q1b-L330, with different unrelated subclades:
• Q-B287>Y11938>Y11236>BZ94>BZ93 in BRE005. This direct connection with Y-DNA of modern Kets (see above Proto-Yeniseian Homeland) is probably the best indirect proof of the potential validity of names etymologizable as Yeniseian, assuming that the elites were not demographically strong enough to have an impact on the development of Proto-Turkic (see above Yeniseian influence) On the other hand, this haplogroup stems from Cis-Baikal EBA (Glazkovo-related) groups, and modern Kets are arguably not a reliable proxy for Yeniseic peoples.
• Q-YP771>BZ180>F15008 (formed ca. 1800 BC, T_MRCA ca. 300 BC) in BRE002, a haplogroup found earlier in Altai_MLBA and later in two Huns from Korgantas.
• Basal Q-YP771>BZ433>L332* in I3752, a haplogroup found in neighbouring Uyuk and widespread among Iron Age nomads.
• R1b-Y13202 in I0562, an R1b-M73 subclade that shows partial continuity in the region with the previous period (see below).
• N-M231 (STR-based) in both samples from the Al Alakha burial, with the best coverage AlAlakha2 suggesting an N1a-CTS6380 subclade, found also in a late Xiongnu (see below).

There are probably regional population replacement events in the region compared to the preceding EBA-LBA populations (from Narasimhan, Patterson et al. 2019):

• A Zevakinskiy_BA sample (2134-1936 calBC) of Central_Steppe_MLBA ancestry shows R1a-Z2124(xS23592).
• A Zevakinskiy_MLBA sample (1599-1436 calBC) shows hg. R1a-Y20746, a subclade of hg. R1a-Z2123, and a Steppe_LBA ancestry (see map of ancient R-Z2123 and timeline slideshow).
• Zevakinskiy_LBA samples (ca. 1200-1000 BC) from the Final Bronze Age Dongal culture, continuing the Sargary tradition (ca. 1200-900 BC), also of Steppe_LBA ancestry, show:
• R1a-Z93 in three samples, two R1a-Z2124 (xZ2122;xS23592), and one negative for Z2124.
• Q-L332 in a Baikal Neolithic-related outlier.
• Basal R1b-M478 (formed ca. 6400 BC, T_MRCA ca. 5300 BC) in I3977, a haplogroup also found during the Eneolithic across the Eurasian steppes, and probably closer chronologically and geographically in the Bolschemys culture (as undefined R-P297), and later in a Pazyryk sample. Two main R-Y13202 subclades appear spreading with nomads: R-BY40441, found in an unclassified Central Steppe IA nomad (ca. AD 50) and a Mongol; and R-L1432, appearing first in Botai and (as WSHG-outlier) in Sintashta, and later in two EMA Turkic samples (see map of ancient R-Y13202 and timeline slideshow of R-M73).

Cherkaskul- and Karasuk-related R1a-S23592 can be seen as an important part of the Uyuk population, pointing to the most likely origin of the Samoyed substrate of an Early Proto-Turkic population in immediate contact with Tagar (see above Proto-Samoyed homeland).

The expected presence of a recent Eastern Iranian-speaking superstratal/adstratal population influencing a local Early Proto-Turkic one is therefore difficult to see among “Scytho-Siberian” (Pazyryk and Uyuk) elites, despite an apparent further influx of Sintashta-related ancestry, and the linguistic data showing a clear influx of Eastern Iranian terms predating the Saka-Wakhi split (ca. 5^th century BC). The Iranic hydronymic area partially overlapping the late Pazyryk cultural region, with Pazyryk showing strong cultural influence from the Saka to the west – and influencing the development of the later Xiongnu (see below) – also suggests that this culture was a polyglot area. It is possible that the detected BMAC-related ancestry among “Scytho-Siberians” shows the clearest intrusion of Iranian-speaking populations.

The genetic picture is further obscured by the incorporation of “eastern” (Pazyryk-like) lineages and ancestry among Eastern Iranian-speaking populations of the Early Iron Age (compared to the previous LBA period), including Sakas and Sargat, possibly mediated by earlier “Pre-Scythian” expansions, as recently shown in Gnecchi-Ruscone et al. (2021). The Proto-Turkic loanwords found in Ob-Ugric support that Turkic-speaking elites probably also spread with some of these Early Iron Age nomadic groups as far west as the Trans-Urals.

NOTE. The correlative Ob-Ugric loans in Turkic are difficult to justify given the separation between Sargat and Uyuk, but long-distance contacts among nomadic groups as well as far-reaching Wanderwörter cannot be discarded for the domain of animals and horse-related vocabulary. For example, all Mongolic loans on Yeniseian were mediated through Turkic (cf. Khabtagaeva 2019).

2.4. Xiongnu Empire

The following texts on archaeology and population genomics are modified from Jeong et al. (2020) supplementary materials (under a CC-BY license):

The Xiongnu Empire (ca. 200 BC – 100 CE) was a multi-regional political entity formed in Mongolia, and attested not only by historical records but also by ample archaeological remains throughout Inner Asia (Brosseder & Miller 2011; Honeychurch 2015a). For roughly three centuries the Xiongnu ruled from their core realms in central and eastern Mongolia, expanding into western Mongolia, northern China and eastern Baikal, as well as making inroads into more distant regions in Central Asia.

Most graves of the Xiongnu period were shaft pits set beneath thick rings of stones on the surface. These burials represent the vast network of regional and local elites and not the “commoner” people of Xiongnu society, whose burials are far less conspicuous, lying under small piles of stones or in unmarked pits. The graves of the uppermost ruling elites of the empire, on the other hand, were constructed on a far grander scale than that of ring graves.

While ring grave structures are found throughout the entire Xiongnu era, prestige accoutrements (and to some degree burial rituals) changed during the course of the empire. According to these changes, we can discern a general division between Early (200-50 BC) and Late (50 BC – AD 100) Xiongnu periods (Miller 2014).

Overall, Xiongnu graves are marked by a dramatic increase in grave goods and furnishings as compared to previous time periods and cultures in Mongolia. As the Xiongnu expanded their empire, they conquered numerous neighboring groups to their east and west as well as subduing their Han Chinese neighbors to the south (Di Cosmo 2002). They continually traded and warred with Han China, defying the Great Wall boundaries, and held significant sway over the Silk Road kingdoms of Central Asia (Hulsewé 1979). The findings of exotic items from China, Persia and the Mediterranean attest to these far-flung interactions, with Egyptian-style faience beads in graves of local elites and Roman glass bowls in the tombs of the rulers (Miller and Brosseder 2017; 2011).

The end of the Xiongnu period ca. 100 CE is marked by the widespread decline of Xiongnu power and influence following defeats by the Xianbei in northeastern China and the Han Dynasty of China, although isolated groups from the Xiongnu empire continued to exist in northern China until the 5^th century CE.

NOTE. A detailed historical atlas of the territorial evolution of the Xiongnu and its neighbours is found in the site Empires, Barbarians, & Barbarian Empires.

Iron Age Chandman_IA is a good Steppe ancestry proxy for many Xiongnu individuals, but there are also many who have western Eurasian ancestry in higher proportion than that of Chandman_IA. These individuals with high western Eurasian ancestry proportion show strong affinity to the Iranian-related ancestry that cannot be explained by the earlier Late Bronze Age steppe groups (e.g. Krasnoyarsk_MLBA, Sintashta_MLBA or Srubnaya). Instead, Gonur1_BA or Iron Age Sarmatian fit better with the genetic profile required. Also, a few individuals fall into the eastern Eurasian cline along PC2 and are explained as a combination of the eastern Eurasian gene pools, Ulaanzuukh_SlabGrave and present-day Han Chinese, without contribution from western Eurasian sources (Jeong et al. 2020).

2.4.1. Early Xiongnu (200-50 BC)

Prestige items during the Early Xiongnu period are dominated by large bronze belt pieces; however, burial customs within graves of the Early period varied to a great degree between regions. One example of this occurs at Salkhityn Am cemetery, where rituals of ring graves show a high degree of variation, even including offerings of whole horses that are more typical of Altai elites such as those in Pazyryk graves (Ölziibayar et al. 2019). It can be genetically split into two subgroups:

• earlyXiongnu_west (SKT010, SKT001, SKT003, SKT009, SKT008, AST001). In terms of haplogroups:
• Basal R1a-PH1397 in SKT009 (showing family connections with SKT010, and indirectly with females SKT001, SKT003, SKT010, & AST001), probably stemming from Uyuk, since it is a haplogroup found widely distributed among Early Iron Age nomads (see above).

NOTE. Another 18 STR-based R1a samples are inferred for late Xiongnu, with 13 of them showing haplotype matches with R1a-Z2125.

• Q1b-Y145452 in SKT008, found later in DEL001, and also stemming most likely from Uyuk (see above).
• earlyXiongnu_rest (JAG001, SKT002, SKT004, SKT005, SKT006, SKT012), including the previously published samples (SKT and AST):
• R1b-PH155 in SKT005, further defined as basal R1b-Y33147* in DA41 and brothers SKT002, SKT006. Another subclade, R-BY86775, is found in two Shirenzigou nomads (ca. 300 BC), and later in a Hun outlier from the Tian Shan (ca. AD 275), as well as in a Xianbei individual, and a different subclade R-M335 is found in a “Gepid” (ca. AD 550) of likely Hunnic origin based on its ancestry, whereas an R-pre-PH155 sample appears in a medieval sample from Mongolia. The earliest parent R-BY14355 sample appeared in a BMAC-related individual from Dzharkutan (ca. 1500 BC), which suggests a common origin of this lineage’s spread in the Inner Asian Mountain Corridor (see map of ancient R-BY14355 and timeline slideshow).
• Basal C2a-Y11606 in JAG001, possibly splitting the Y11606 branch – with an SNP pattern matching late Xiongnu CHN010 – and full-fledged C2a-Y11606 in KHO006 and I6228. Immediately upstream basal C2a-Y10428 is found in TEV003. Another C2a-F1756 is found in low coverage BRL003, with this haplogroup likely originating ultimately in Trans-Baikal Bronze Age populations, and spread in the Late Iron Age / Early Middle Ages clearly associated with Pre-Proto-Mongolic and Proto-Mongolic peoples, as well as with Turks and Mongols, mostly linked to sister clade C2a-F3830 (see map of ancient C-F1756 and timeline slideshow).
• J2a-pre-FGC61855 in SKT012, with a different J-Z7700 subclade found in late Xiongnu. The parent haplogroups seems to have been widespread among Central Asian BMAC-related populations at least since the expansion of Andronovo (see map of ancient J-Z7700 and timeline slideshow).
• One individual outlier, SKT007 (Khövsgöl_LBA-like), of basal Q1b-BZ180, possibly continued in late Xiongnu BTO001, of hg. Q1b-BZ180(xY146631). This haplogroup shows thus continuity with Uyuk and Pazyryk Y-DNA (see above), but is probably directly connected to Altai_MLBA populations not affected by the influx of Eastern Iranian ancestry.

2.4.2. Late Xiongnu (50 BC – AD 100)

Prestige items in the Late Xiongnu period shift to more iron items, often covered with gold foil or even inlaid with precious stones, and increasingly focused on long-distance exotic materials. At the same time, burial customs in ring graves throughout the empire become more regularized. Most elites were buried in wooden coffins in shaft pits with livestock portions and ceramic vessels set beside the coffin.

During the Late period, the high ruling Xiongnu elites adopted a radically new form of burial structure. These square tombs were marked on the surface by rectangular stone structures with trapezoidal ‘ramp’ entryways, their burial pits were extremely deep, and wooden coffins were decorated and nested within larger wooden chambers is split into three subgroups (Jeong et al. 2020):

• main lateXiongnu group*: well modeled as a mixture of two main Iron Age clusters, Chandman_IA + Ulaanzuuk_SlabGrave (p=0.316; 76.6±0.8% from Ulaanzuuk_SlabGrave).

* Including BTO001, CHN010, DEL001, DOL001, IMA001-IMA008, JAA001, KHO006, KHO007, SAN001, SOL001, TEV002, TEV003, TUK003, UGU004, UGU011, ULN004, UVG001

• Q1b-BZ180 in BTO001, shared with an earlyXiongnu_west.
• Q1b-Y145452 in DEL001, shared with an earlyXiongnu_west.
• Basal C2a-Y11606* in CHN010 and KHO006, shared with an earlyXiongnu_rest.
• R1a-Z94(xZ2123) in IMA003, possibly R1a-FT129490 (1xG->A), shared with an earlier Sarmatian from the Southern Urals (ca. 375 BC) and with a later Turk from Kazakhstan (ca. AD 720). (map of ancient R-FT197453). R-Z94 is also shared with (STR-based) TUK08, TUK37, TUK18, TUK23, and others from late Xiongnu samples below.
• Q1b-L330(xB287, xBZ433) in IMA004, a haplogroup an ultimate “eastern” origin but difficult to pinpoint among Early Iron Age nomads without further subclade.
• Basal N1a-L1026>CTS6967>CTS3103>Y6058>B197>P89 (formed ca. 2600 BC, T_MRCA ca. 400 BC) in IMA005, and subclade N-F22331 is found in a central steppe nomad DA95 from Gregorievka, Kazakhstan (ca. AD 350). The same haplogroup is later found predominantly among (19) Early Avar elites from the 7^th-8^th c. (see map of ancient N-P89 and timeline slideshow of N-B197). This genetic link supports an expected direct connection of Avars with the Oghuric-speaking Xiongnu instead of the Mongolic-speaking Rouran (see e.g. the discussion in Savelyev & Jeong 2020).

NOTE. This haplogroup is sister clade of N-CTS2929, found at roughly the same time as the Pre-Scythian and Scythian expansions spreading through the Baltic. Both are distantly related to the N-Z1936 found (to date) first in the Sargat horizon, and a parent N-CTS3103 (negative for most downstream clades) has been found in at least two Sargat elite individuals, BIY003 and SMV001. This increasingly detailed phylogeographic picture coupled with their “eastern” ancestry supports an ultimate connection of these specific N-L1026 lineages to a recent spread of nomads with an origin in admixed South Siberian communities, the earliest of them showing a direct connection with the likely Proto-Turkic homeland.

• N1a-Tat(xBY1233;xCTS6380;xP89;xM2019>M2058,xB187;xP83)in KHO007, which could belong to N-M2019, N-L1026, or a more basal N-Tat clade like N-L708, N-L708>M2005, etc. Other four N samples are inferred based on STRs, with at least three of them N1a-Tat (including one confirmed with Y-SNP capture).
• Q1a-FT9308 (T_MRCA ca. 1400 BC) in IMA006, found earlier in Slab Grave SHU001 (ca. 980 BC) and in Tasmola BIR001 (ca. 600 BC). The sister clade Q-F4743 is found in Ulaanzuukh and Khövsgöl (see above), so it is unclear which specific subclades lie behind the (STR-based) Q1a-M120 reported for TUK43, TUK26, TUK01, TUK20, TUK44 (see map of ancient Q-M120 and timeline slideshow).
• Basal C2a-Y10428 in TEV003, CHN010, KHO006, and I6228, in common with the earlyXiongnu_rest JAG001.



• lateXiongnu_sarmatian: Two of them (NAI002, BUR001) are explained by Chandman_IA+Gonur1_BA, a model for earlyXiongnu_west, but the remaining 11 need a major contribution from a Sarmatian-related source, including three that are cladal to Sarmatian (BUR003, TM001, UGU010). Taken together, the 13 show ca. 75.7±2.8% Sarmatian-related ancestry.

* Including BRL002, BUR001-BUR004, DUU001, HUD001, NAI001, NAI002, TMI001, UGU005, UGU006, UGU010

• E1b-BY7758 in BUR002, a rare subclade of E-V22, a haplogroup probably spread with the expansion of Iran Neolithic ancestry. A different E-V22 subclade, E-L1250, is found in a Hun, but it is negative for the E-PH2818 subclade found in an earlier BMAC sample (see map of ancient E-V22 and timeline slideshow).
• J1a-YSC0000076 in UGU006, a haplogroup also found in Han-like BRU001, and probably also spread ultimately with Iran Neolithic ancestry through the Inner Asian Mountain Corridor (see map of ancient J-YSC0000076 and timeline slideshow).
• J2a-pre-Y26741 in NAI001, with SNPs shared with an earlier BMAC (ca. 3150 BC) and an Aigyrzhal_BA individual (ca. 1990 BC), both of Iran Neolithic-related ancestry. A derived J2a-Y234666 subclade is found in a central steppe nomad KNT003 of the Otrar-Karatau culture. Their parent haplogroup J-Z32167 is sister of the J-Z7706 found in an early Xiongnu (see above).
• R1a-pre-Y190991 in NAI002, a subclade of R1a-Z282>Y17491, showing that rare R1a-Z282 subclades also accompanied the Andronovo and Andronovo-like horizons. Another case is found in the basal R1a-Z280 of the “Srubnaya-Alakul” brothers from the Urals.
• R1a-Z2124(xBY48287,xF1345,xY57;xS23592) in UGU005.
• R1a-Y52 (formed ca. 2100 BC, T_MRCA ca. 1300 BC) in Sarmatian-like BUR003. Sarmatians DA134, KBU003, chy002 (ca. 800 BC – AD 100), as do two likely early medieval Turks from the central steppe (see map of ancient R-Y52).
• R1a-Z2124, probably R1a-BY30764 (formed ca. 1900 BC, T_MRCA ca. 900 BC), in Sarmatian-like UGU010. Its tentative subclade is found earlier in Srubnaya kzb007 (ca. 1720 BC) and later in a Georgievsky Late Iron Age I11540 (ca. AD 220).
• lateXiongnu_han group*: eight individuals with affinity to Han Chinese and other East Asian populations that Ulaanzuuk_SlabGrave cannot explain (37.2±10.6% from Han.DG), with the previously published Xiongnu_royal individual showing substantial Han-related ancestry.

* Including ATS001, BAM001, BRU001, EME002, SON001, TUH001, TUH002, YUR001

• J1a-YSC0000076 in BRU001, a haplogroup also found in UGU006 (see above).
• J2a-pre-PH358 in TUH002, a specific splitting haplogroup found among early Turks and Mongols, but also as derived subclades earlier in BMAC (ca. 2300 BC) and geographically and chronologically closer in the Sary-Bel kurgan of the Bulan-Kobin culture (ca. 500 BC), succeeding Pazyryk in the region. The ultimate origin of the haplogrop probably lies in Iran Neolithic-like groups from Turan (see map of parent J-PF5040 and timeline slideshow).
• Basal N1a-CTS6380 in SON001, a haplogroup shared with earlier Pazyryk and later Turks and Mongols (see above).
• C2a-F6379(xF5485) in TUH001, of a C-M48 haplogroup shared with a Boisman Neolithic and later with Turks and Mongols (see map of ancient C-M48).
• Further, the late Xiongnu individual YUR001, of hg. N1b-Y125475(xF1437,xY24190,xSK1507,xFT258887), is an extreme East Asian outlier, who genetically resembles “Han_2000BP”, two Han empire soldiers recovered from a mass grave near a Han fortress in the southern Gobi (Damgaard et al., 2018). This makes sense given the distribution of ancient N-F2905 subclades (see map of ancients).

These last two groups, lateXiongnu_sarmatian and lateXiongnu_han, robustly support influxes of new ancestries both from the west and the east that were not previously observed in early Xiongnu or earlier populations.

Two individuals show an unusual ancestry profile:

• TAK001 mostly resembles Khövsgöl_LBA.
• TUK002, of hg. O2a-Y16154, is modeled as Chandman_IA + Ulaanzuuk_SlabGrave_Gonur1_BA. In contrast to the strong east-west genetic division among Bronze Age Eastern Steppe populations through the end of the Early Iron Age, the Xiongnu period is characterized by an extreme degree of genetic diversity and heterogeneity that does not have any obvious geographic correlation. The same haplogroup is later found in Mongol KNN001, and O2a-MF2463 is found in DA43, from the same O-F2180 branch as TUK002. Other O2 samples include O2a-MF76414 in DA45; and O2a-CTS11192 (low coverage) in EME003.

Wrap-up

It seems from the current linguistic, archaeological and genetic picture that the “Scytho-Siberian” Uyuk culture is the most likely representative of the Early Proto-Turkic homeland, i.e. the community speaking the language ancestral to all known Turkic languages.

The similar cultural and genetic profile of Pazyryk despite a greater “Scythian” cultural influx and its coincidence with Iranian hydronymic territory in South Siberia suggests that it was probably a community of Turkic elites dominating over an Iranian-speaking population. They were thus the most likely vector of Eastern Iranian loanwords in Proto-Turkic.

A Proto-Turkic language spoken around the Sayan mountains during the 1^st millennium BC is also assumed to be behind the Proto-Turkic superstrate on Samoyed (cf. Piispanen 2018), hence probably represented by Tagar elites – likely spreading at the same time as the formation of Uyuk – although information about them is still scarce.

Proto-Turkic loans in the coeval (i.e. 1^st millennium BC) Common Yeniseian also supports a northward expansion of Turkic elites at roughly the same time, and possible loans in Proto-Tocharian further constrain the homeland to a more westerly location in Mongolia.

Finally, the finding of the “eastern” ancestry and lineages (proper of Altai_MLBA groups) spreading to the west with “Scythian” groups like Tian Shan Sakas, Tasmola, and Sargat, suggests that these admixed groups with elites stemming from the Altai-Sayan area might have been the source of the few Proto-Turkic loanwords – among the many Eastern Iranian ones – found in Ob-Ugric.

Xiongnu is believed to represent the Late Proto-Turkic homeland, and the formation of its early community shows strong cultural and genetic influx from to the preceding “Scytho-Siberians” to the west, before becoming heavily admixed with populations from the Tian Shan, Mongolia, and China in the late period during its expansion, in a pattern similar to that seen e.g. during the expansion of Rome and admixture with Middle Eastern populations, including the spread of intrusive Y-DNA (cf. Antonio et al. 2019)

Despite the assertion of Savelyev & Jeong (2020) that “evidence for a continuity between the Xiongnu of Inner Asia and the Huns of Europe is very weak, largely because of the overall scarcity of an eastern Eurasian component in the interdisciplinary profile of the Huns”, there are already clearly visible strong links in terms of shared patrilineages and ancestry (cf. Gnecchi-Ruscone et al. 2021).

The influence of Pre-Proto-Oghuric on Pre-Proto-Mongolic roughly coinciding with the incorporation of Slab-Grave-related populations into the polyglot Xiongnu empire strongly suggests that populations to the east of Altai_MLBA+Steppe_MLBA groups spoke Mongolic varieties by the Late Iron Age. Further, Oghuric traits found in Proto-Khanty borrowings – but not in borrowings in Ob-Ugric or in Proto-Samoyed – support that they should be attributed to the Hunnic expansion, or closely related westward expansion of Xiongnu-related populations.

Assuming that Steppe_MLBA-related R1a-Z2125-rich populations from the MBA (Fëdorovo-)Cherkaskul groups represented the Eastern Uralic expansion – continued in LBA Ob-Ugric-speaking cultures of the Andronovo-like horizon and Pre-Proto-Samoyed-speaking Karasuk – leaves the Altai_MLBA groups as the most likely candidates for a Pre-Proto-Turkic-speaking community. Their close contacts with Ulaanzuukh could potentially justify ancestral similarities shared among “Micro-Altaic” languages, if any of them withstands proper scrutiny.