Dairying enabled Early Bronze Age Yamnaya steppe expansions

During the Early Bronze Age, populations of the western Eurasian steppe expanded across an immense area of northern Eurasia. Combined archaeological and genetic evidence supports widespread Early Bronze Age population movements out of the Pontic–Caspian steppe that resulted in gene flow across vast distances, linking populations of Yamnaya pastoralists in Scandinavia with pastoral populations (known as the Afanasievo) far to the east in the Altai Mountains1,2 and Mongolia3. Although some models hold that this expansion was the outcome of a newly mobile pastoral economy characterized by horse traction, bulk wagon transport4–6 and regular dietary dependence on meat and milk5, hard evidence for these economic features has not been found. Here we draw on proteomic analysis of dental calculus from individuals from the western Eurasian steppe to demonstrate a major transition in dairying at the start of the Bronze Age. The rapid onset of ubiquitous dairying at a point in time when steppe populations are known to have begun dispersing offers critical insight into a key catalyst of steppe mobility. The identification of horse milk proteins also indicates horse domestication by the Early Bronze Age, which provides support for its role in steppe dispersals. Our results point to a potential epicentre for horse domestication in the Pontic–Caspian steppe by the third millennium bc, and offer strong support for the notion that the novel exploitation of secondary animal products was a key driver of the expansions of Eurasian steppe pastoralists by the Early Bronze Age.


Archaeological information for sites and individuals included in this study
We selected samples for this study from sites and individuals spanning the Eneolithic through to the Late Bronze Age in the Pontic Caspian region. Individuals were sampled from three different institutions: All individuals except those from Kammenyi Ambar 5 and Botai are curated at Samara State University's Department of Archaeology. The three individuals from the site of Kammenyi Ambar 5 belong to the scientific collections of the Museum at the Institute of Plant and Animal Ecology (Ural Branch of the Russian Academy of Sciences). The two samples collected from individuals from the site of Botai were sampled by Alan Outram and delivered to MPI-SHH, the individuals are stored by Victor F. Zaibert at the Botai Research Station. Sampled individuals were dated using both established archaeological material culture associations and radiocarbon dating. While many of the radiocarbon dates had been previously published, we also obtained unpublished new dates issued by the PSUAMD lab at Penn State University through the cooperation of the David Reich Laboratory at Harvard University. We present all cultural assignments and absolute dates here.
We also provide archaeological information on the mortuary sites where the samples were collected for this study; in each case we include site name, date(s), number of individuals, and GPS coordinates; see summary in Supplementary Table S1. All individuals included in this study are adults, and the protein identifications for each are included in Supplementary Table S3. We classified the sites following three date ranges: 1) Eneolithic (ca. 4600 to 3300 BCE): 2) Early Bronze Age (ca. 3300 to 2500 BCE; EBA); and 3) Middle/Late Bronze Age (ca. 2500 to 1700 BCE; MLBA). Dates obtained from direct radiocarbon analysis of human skeletal remains are listed as cal BCE, and those estimated from either archaeological artifact seriation or direct dates on associated material are listed as ca. BCE.

Eneolithic sites
During the Eneolithic (ca. 4500 to 3300 BCE), the area that today covers western Russia was occupied primarily by sedentary hunter-gatherer-fisher (HGF) populations. Archaeological remains of domesticated ruminants appear at low prevalence in occasional burials, but played an uneven role in the diet, varying from site to site. Stable isotope analysis was used to establish that the typical Eneolithic diet included significant amounts of freshwater fish 1 . Freshwater foods are typically depleted in 14 C content when compared to the contemporaneous atmosphere, and their consumption, accordingly, results in radiocarbon dates that appear older, an effect known as the dietary freshwater radiocarbon reservoir effect 2,3 . We employed a Bayesian mixing model and estimates of local freshwater radiocarbon reservoir effects to produce a more accurate age estimate for each Eneolithic individual (the method is described in two papers by Fernandes et al. 2,3 and our main text methods section). Such age estimates result in wider chronological ranges than direct calibration of human radiocarbon measurements and show that actual chronologies of human remains may be younger by several centuries. However, in every Eneolithic sample in the present study, these are still within chronological ranges that do not alter their cultural assignment. This Eneolithic non-kurgan cemetery contains more than 100 burials and is located in the southern part of the Middle Volga region near the modern city of Samara. Grave goods included items similar to those at Khvalynsk (described below) and domesticated ruminant bones, but, unlike Khvalynsk, the each female and few children, and seems to have been a burial place primarily for adult males with greater abundances of copper offerings. The Khvalynsk I and II cemeteries collectively contained more than 370 copper items, probably made of imported Balkan copper, the largest number of copper objects from any fifth-millennium BCE site in the Pontic-Caspian steppes. Archaeological and isotopic data suggest that people maintained a diet focused on fish and wild resources, as well as a smaller percentage of domesticated animals. The human remains generally date between 400-700 years older than the terrestrial animals included in the same graves, indicating a reservoir effect from the consumption of aquatic resources (See Supplementary Table S4). Sheep and cattle bones from the site have been dated to between ca. 4500 to 4300 BCE. While there are a small number of clearly domesticated animals in these burials, archaeologists have assumed that these animals were used in more of a ritual manner rather than as a consistent feature of daily subsistence 6,8,9 .

Ekaterinovsky Mys
A date of 4450-4355 cal BCE was obtained on a ring made of sheep bone (GrA-29178, 5565±40 BP) from grave 147 at Khvalynsk I. The human female buried with this bone ring was dated to 4789-4618 cal BCE (PSUAMS 2886, 5845±25 BP); this suggests that there was a 280±147 14 C year discrepancy, due to the reservoir effect. A date of 4448-4362 cal BCE (GrA-34100, 5570± 40 BP) was obtained on a cow bone from grave 10 at Khvalynsk II. These ruminant grazers would not have been subject to the reservoir effect, with dates of 4450-4355 cal BCE from Khvalynsk I and 4448-4362 cal BCE from Khvalynsk II. These two dates provide the best estimate currently available for the true age of the two cemeteries.
If we compare the radiocarbon ages of the ruminants to those of the humans, the resulting offsets range between 55±47 and 405±47 14 C years. The multi-period burial ground (n=19 burials) is located in the northern forest region of the Volga-Kama River. The Eneolithic burials are related to the Ust-Kama culture, and funeral rituals appear to share some similarities to those from Ekaterinovsky Mys. While the Khvalynsk burials also contained domesticated animal bones, the Murzikha II burials did not. However, at a similar and nearby cemetery, Gulkinsky II, a Bos sp. scapula (probably a tool) was included in grave number 124, and some bones of domesticated calves were also recovered from the grave cluster. Several graves at Murzikha II contained small rings made of copper, a cultural feature also present at Khvalynsk. Ten burials from the site have been dated, including three that are included in this study. The calibrated dates range from 4599 to 3526 cal BCE. The isotopic composition of the bones has not been studied, but the associated artifacts indicate that the diet was likely hunting and fishing-based; hence, we can assume that the reservoir effect has influenced the dates 10 . The earliest radiocarbon dates associated with graves that have artifacts assigned to the Yamnaya culture fall between ca. 3300-3000 BCE. The Yamnaya culture is normally equated with the beginning of the EBA in the chronology of the Pontic Caspian region. This region is seen as a focal one where a number of key genetic and cultural traits critical to the development of the Yamnaya were assembled, with potential links to the oldest phase of Proto-Indo-European 6 . The Yamnaya culture introduced marked changes in funeral rites and settlement patterns, and stable isotopes from human bones suggest dietary changes as well. We can now confirm the dietary shift towards a complementary pastoralist component to the existing hunting and fishing economy with new protein evidence for dairy consumption from dental calculus, as presented in this paper. The shift from riverine (Eneolithic) to grassland and riverbank resources (EBA), seen isotopically, was accompanied by the widespread abandonment of riverine Eneolithic residential sites and by the initial appearance of kurgan cemeteries in the grassland plateaus of major river valleys. While zooarchaeological remains are limited, they nonetheless support the evidence for a shift to pastoral resources. Early Bronze Age and Yamnaya zooarchaeological remains in the Volga-Ural study area are limited to occasional sacrifices in graves, 70-90% of which were sheep-goat 52 . In the Dnieper Valley, two Yamnaya settlements had primarily cattle, followed by sheep-goat and horse 45,48,49 . Furthermore, a majority of the sites from which the individuals in the present study derive fall across a landscape that would have been ideal for large ruminant herds 11 , and archaeologists have suggested that regional Bronze Age populations relied on semi-sedentary cattle, sheep, and goat pastoralism [12][13][14][15] .
From this data as well as the modeled Bayesian radiocarbon age estimates discussed in the section above, it is evident that there was a dramatic increase in the use of domesticated ruminant animal products in the Early Bronze Age. In comparison to the Eneolithic period, freshwater fish consumption was lower in the Early Bronze Age; and stable isotope studies on Bronze Age populations from the region show that an increase in nitrogen values in humans corresponded to higher values in local herbivores as well rather than from freshwater fish consumption 1,16 . While we acknowledge that fish may have been consumed at a small scale, it likely did not affect AMS dates from this period to the extent of individuals living during the Eneolithic period. Therefore, radiocarbon date ranges for individuals from the Bronze Age were not adjusted (Supplementary Table  S5).
Krasnoholm III (Number of Individuals: 1; Individual Archaeology Codes: KRA3 K.1 N-1) This burial mound was discovered on the left bank of the Ural River in the Pre-Urals region. It is one of the three mounds that contained artifacts ranging in age from the EBA to the Early Iron Age. The individual in our study was buried in a grave in the style of the Yamnaya culture. It was a single burial, sprinkled with ochre, and the grave included charcoal and four polished bone tubes. The complex has not yet been radiocarbon dated 17 .
Krasikovskyi I (Number of Individuals: 2; Individual Archaeology Code: KRS K.2 N-1; KRS K.1 N-1) This burial mound cluster is located in the Volga-Ural region in the forest-steppe zone of the Tok River, a tributary of the Samara River. The cluster consists of five kurgans, which contained five Yamnaya culture graves (EBA) and one grave associated with the Abashevo culture (MLBA). Early graves with individual burials were disturbed, but accompanying artefacts included a pectoral ornament made of a wild boar tusk and the stone tip of a dart 18  This burial ground is located between three small rivers in the Don basin, 30 km northeast of Rostovon-Don. It consists of seven kurgans that in total contained 92 burials. The site was used from the Eneolithic through the LBA, and also included some later household pits of the medieval Khazar period. The analyzed archaeological remains come from four mounds; the earliest phase of mound construction dates to the EBA Yamnaya culture (No. 1, 2, 4) and MBA (No. 5). The EBA burials were in pit graves, the MBA burials were interred in catacombs. Individual and collective burials (up to five skeletons), with traces of ochre, were characteristic of both periods. There were few funeral offerings or artifacts in any of these graves; and most of the deceased had no accompanying artefacts. Exceptions include a single ceramic vessel in one burial from the Yamnaya culture, as well as a stone flake and small cattle bones in one of the MBA Catacomb burials 19 .
Radiocarbon dates from individuals sampled in the present study: KRI9 K.4 N-21A: 3345 to 3096 cal BCE (4495±25 BP, PSUAMS-7979) This was a collective grave containing five individuals, two adults and three immatures. The dental calculus sample was taken from adult A, pictured below. A second date on adult C was significantly different: 2904-2701 calBCE (4225±25 BP, PSUAMS-7980). This mound cluster was discovered on the left bank of the Samara River, a tributary of the Volga in the Middle Volga forest-steppe region, and two out of the eight mounds in the cluster have been excavated. The excavators believed that the burials belonged to the Poltavka and Srubnaya cultures; however, the ceramic vessel that was used to assign a phase to the burial was not found at the bottom of the grave (which contained only the skeleton), but rather higher up. Judging by the AMS date for the individual in our study, the burial belongs to the Yamnaya culture. The location of the skeleton, its orientation, and the use of ochre do not contradict this assumption.
Radiocarbon date from the individual sampled in the present study: POD K.3 N-3: 3330 to 3028 cal BCE (4465±20 BP, PSUAMS-4412); Pyatiletka (Number of Individuals: 1; Individual Archaeology Code: PYA K.6 N-2) Three mounds were found and excavated on the left bank of the Ural River in the northern Pre-Urals region. The main burials of each barrow belong to the Yamnaya culture, as determined by the specific grave features (ochre, burial position, and inhumation style). Some noteworthy features of these burials include rare examples of paired burials, a ditch around a mound, and metal tools, all of which differentiate this site from other contemporaneous regional sites.
Trudovoy II (Number of Individuals: 1; Individual Archaeology Code: TRU K.5 N-1) A large burial mound was found on the right bank of the Ural River in the Pre-Urals of the northern steppe. Two of the three largest excavated mounds were erected over the graves of Yamnaya burials. The grave of the individual analyzed in the present study was constructed after the original kurgan monument was completed. It was identified as a Yamnaya burial based on structural details, body position, the presence of a ceramic vessel-censer, and other artifacts. In addition, the mandible of a sheep was found in the burial. The site has no associated radiocarbon dates 24 .

Middle and Late Bronze Age Sites
The Middle/Late Bronze Age (ca. 2500 to 1700 BCE) of the Pontic Caspian region was dominated by Poltavka, Catacombnaya, Abashevo, Sintashta, Srubnaya, and Andronovo culture groups who relied heavily on horses and ruminant animals 26,27 . Originating from the earlier Yamnaya, these populations shared similar ceramic styles, metal tools, kurgan burials, and chariots 13,20,28 . As opposed to the EBA populations, the Abashevo, Sintashta, Srubnaya and Andronovo groups became more sedentary and began building larger, more permanent settlements 29 . Zooarchaeological evidence combined with stable isotope analysis of Sintashta individuals from western Kazakhstan's Kamennyi Ambar 5 and Bestamak indicate that while diets included both horses and ruminants, they were weighted towards the latter 26,30 .  Four kurgans (26 graves and more than 100 individuals) contain graves of the Sintashta culture [34][35][36] . Subsistence at the site is assumed to have been based primarily on pastoralism, but was likely supplemented with wild plants and animals. There is no direct evidence for use of domesticated crops, but the possibility for cultivation cannot be completely discounted. At the site, other material artifacts were also dated to between ca. 2000 to 1700 BCE, indicating date ranges closer to the unmodelled human individual dates that did not take the freshwater reservoir effect into account. The faunal assemblage at KA5 is composed of 98.8% domesticated animals and 1.2% wild fauna, and of the domesticated animals, ruminants make up 91.6%, with horses at 6.3% 30 .

Bolshekaraganskyi
Krasikovskyi I (Number of Individuals: 1; Individual Archaeology Code: KRS K.3 N-1) This burial mound cluster is located in the Volga-Ural region of the forest zone in the Tok River basin. Five kurgans contained five Yamnaya-type graves (EBA) and one Abashevo culture grave (MLBA). The sample for the present study derives from the Abashevo individual, whose grave also contained a ceramic vessel with traces of repair using bronze staples 37 Radiocarbon date from the individual sampled in the present study: KRS K.3 N-1: 2145 to 1897 cal BCE (3632±55 BP, SPb-2224) 37 Krivyanskyi IX (Number of Individuals: 2; Individual Archaeology Codes: MBA: K.1 N-30; KRI9 K.5 N-6) We presented the details of the Krivyanskyi IX cemetery in the EBA section above, as individuals from the site sampled in the current study come from both the EBA and M/LBA. The mound cluster consists of four mounds and 92 burials located between three small tributaries of the Don River; the earliest phase of their building dates to the EBA (No. 1, 2, 4) and MBA (No. 5). The EBA burials were pit graves, and the MBA burials were placed within catacombs. Individual and collective burials (including up to five skeletons) and the use of ochre were characteristic in both periods. Grave goods were rare in all burials, with the exception of a single ceramic vessel in one of the Yamnaya culture burials, as well as a stone flake of small cattle bones in one of the burials of the Catacomb culture. There are no radiocarbon dates for the MBA individuals 19 .

Previous lipid study by Mileto et al., 2017
While the present study points to a clear shift in milk consumption patterns between the Eneolithic and EBA in the Pontic-Caspian Steppe, a previous study of ceramic lipid residues from sites located in the steppe zone of Ukraine 40 found some evidence for milk consumption prior to the EBA. Lipids from ruminant dairy products were recovered from ceramics at one site classified as Mid-Eneolithic and three sites considered to be Late Eneolithic/Early Bronze Age. While these findings seem to contradict our results from the Eneolithic Volga steppes, where no dairy proteins were identified for the same time period, the three sites identified as Late Eneolithic/Early Bronze Age sites in the lipid study are in fact normally regarded as EBA Yamnaya [41][42][43][44][45] . While some scholars disagree on the timing of Yamnaya origins with early (ca. 3800/3900 BCE 23,24,46 ), middle (ca. 3500/3300 BCE 1,13,47-49 ), and late (ca. 3000 BCE 40,50-52 ) onsets, using any of these chronologies, ubiquitous dairy consumption only occurs during the tenure of the Yamnaya (Supplementary Table S2). The only evidence for dairy lipids in steppe pottery that precedes the EBA Yamnaya when the more widely accepted middle chronology (or the late chronology) is employed occurs at Mikhailovka I (ca. 3800-3000 BCE), which may reflect an initial stage in the production of dairy foods in the steppe, just prior to the rapid spread of the EBA Yamnaya culture 40 . If small-scale dairy use did occur in this period, it could explain the casein peptides found in our single Eneolithic steppe individual. Our results suggest that whatever the date of arrival of dairying, milk was probably not broadly consumed across steppe communities until ca. 3300 BCE.

Reassignment of period designations from lipid study of Mileto et al., 2017
The time period designations employed in the Mileto et al. 40 study are debated by regional experts. Here, we outline the long, middle and short chronologies that exist for the transition from the Eneolithic into the Early Bronze Age Yamnaya culture, and explain why we support the widely accepted middle chronology that would reassign most of Mileto et al.'s milk findings 40 to the Early Bronze Age (see also Supplementary Table S2).
The Yamnaya chronology has been established for some decades. Initially, in the 1950s, Yamnaya culture sites were designated as Late Neolithic 53 . In the 1960s, however, radiocarbon dates, a growing number of metal artifacts discovered in Yamnaya graves, and typological links with the Maikop culture combined to prompt the widespread acceptance of the Yamnaya culture as dating to the beginning of the EBA. N. I. Merpert's classic synthesis of the Yamnaya culture 42 included two important Yamnaya settlement sites, the Repin site on the Don River and Mikhailovka II on the Dnieper River, as chronological and typological anchors for what was otherwise a chronology based entirely on graves, few of which presented stratified sequences. Repin and layer II at Mikhailovka, which contained Repin-style pottery, are accepted as defining early Yamnaya artifact assemblages and the start of the EBA by prominent experts on the Yamnaya culture in Ukraine and Russia 54-56 . The published dates for Repin and Mikhailovka II (from bone samples) range between 3400-3000 cal BCE. Layer III at Mikhailovka was defined as late Yamnaya by the archaeologists who excavated the site 45 . Rassamakin's proposal to make the EBA and early Yamnaya begin with Mikhailovka layer III 51 -a late chronology for the EBA -therefore, requires complex adjustments. In the Mileto et al. 40 chronology, following Rassamakin 52 , Mikhailovka II and other Yamnaya sites like it are re-assigned to the preceding Eneolithic phase. While Rassamakin and Mileto support a late start of the Yamnaya cultural horizon, others, such as Morgunova 23,24,46 , maintain an earlier Yamnaya origin (ca. 3800/3900 BCE). Here we use a more conservative and widely accepted "middle" chronology, mid-way between the more occasionally referenced early and late chronologies. When interpreting the Mileto et al. 40 results within the more standard middle chronology, there is only one Late Eneolithic site prior to the onset of the Yamnaya cultural horizon, Mikhailovka level I, that contained possible ruminant dairy lipid signatures. Importantly, no matter which chronology is utilized, and whether the material studied is milk lipids or proteins, it is clear that dairy use did not become widely used until the Yamnaya period (ca. 3300 BCE).