Re-evaluating Scythian lifeways: Isotopic analysis of diet and mobility in Iron Age Ukraine

The Scythians are frequently presented, in popular and academic thought alike, as highly mobile warrior nomads who posed a great economic risk to growing Mediterranean empires from the Iron Age into the Classical period. Archaeological studies provide evidence of first millennium BCE urbanism in the steppe while historical texts reference steppe agriculture, challenging traditional characterizations of Scythians as nomads. However, there have been few direct studies of the diet and mobility of populations living in the Pontic steppe and forest-steppe during the Scythian era. Here, we analyse strontium, oxygen, and carbon isotope data from human tooth enamel samples, as well as nitrogen and carbon isotope data of bone collagen, at several Iron Age sites across Ukraine commonly associated with ‘Scythian’ era communities. Our multi-isotopic approach demonstrates generally low levels of human mobility in the vicinity of urban locales, where populations engaged in agro-pastoralism focused primarily on millet agriculture. Some individuals show evidence for long-distance mobility, likely associated with significant inter-regional connections. We argue that this pattern supports economic diversity of urban locales and complex trading networks, rather than a homogeneous nomadic population.

with temperatures ranging from −3.5 to 17.7°C respectively. Only a single locale has published surface water δ 18 O values, which are from the Seversky Donets River near Kharkiv and range from −6.2 to −11.7‰ from May and September, respectively. Nearby, groundwater δ 18 O values range from −10.1 to −12.5‰ for the same period [18,19].
The direct comparison of ancient δ 18 Odw(VSMOW) values and modern precipitation values is not recommended as the transformation of human δ 18 Oc(VPDB) to drinking water values δ 18 Odw(VSMOW) include large error ranges of up to 3.5‰ [20][21][22]. Furthermore, the extent of environmental change is unknown for many regions, leading to inaccurate interpretations of mobility. Therefore, we have not transformed human δ 18 Oc(VPDB) to drinking water values δ 18 Odw(VSMOW) instead comparing a range of values for each site. At a single locale, the range of variation in human δ 18 O values has been determined as 0.5 to 3.0‰ [23][24][25][26]. Seasonal variation in stable oxygen isotope measurements of precipitation for the region, based on modern measured values, is extensive at c. 8.0‰, suggesting that in strongly seasonal environments we might expect a wider range of human δ 18 O values for individuals who resided in a single locale during tissue formation. Accounting for strong seasonal variation, we calculated the local range for individuals drinking from the same water source as the median value ±1.5‰ [26].
We investigated the stable carbon isotope measurements of terrestrial vegetation in the vicinity of the sites of Medvin and Mamai-Gora to inform the dietary intake of Scythian era populations. Modern environmental zones are on a north-south gradient shifting from mixed forests, to foreststeppe, to open steppe vegetation in the south, with the Black Sea littoral along the southern border (Fig 1). The site of Medvin is located in central Ukraine within the forest-steppe zone (49°24' 26.9"N 30°50'45.0"E). This region is characterized by deciduous woodlands interspersed with meadow steppe, while to the north there are mixed coniferous and deciduous forests [27,28]. Heading south, the forests gradually decrease in density and open steppe predominates. Mamai-Gora (47°26'07.7"N 34°16'31.1"E) is located just south of the Dnieper River in the open steppe. Central and southern Ukraine consists of highland and lowland plains, with floral biomes supporting mainly C3 vegetation. Terrestrial plants using a C3 pathway exhibit average δ 13 C values of −27‰, while those following a C4 pathway have higher average δ 13 C values of −14‰ [29].
To clarify human enamel apatite δ 13 C values, we determined the best offset between enamel apatite and diet, which varies by taxa and by tissue type. The offset between diet and apatite is ~14.0‰ for herbivore bone carbonate and ~9.5‰ for carnivore bone carbonate [30][31][32][33]. For human enamel apatite, an offset of ~9.0‰ has been suggested to be appropriate [31,[34][35][36]. We use the following equation from Ambrose and Norr (1993): δ 13 Cdiet = 1.04 * δ 13 Capatite −9.2 for all humans in this study. Average herbivore δ 13 C values (of enamel) for 100% C3 and C4 reliance in pre-industrial CO2 conditions are approximately −12.0‰ and 0.0‰, respectively [37,38]. Tooth enamel isotope values of humans with a 100% reliance on C3 plants are estimated at −13.0 and −17.0‰, while humans with a high reliance on C4 plants are estimated at −5.0 to −1.0‰ [30], however, it should be borne in mind that a lack of experimental work means these values are poorly defined for humans.

Faunal and plant isotope references to understand human diet
The carbon isotopic values for animals at Bel'sk range from −19.8 to −16.3‰, indicating a mixed diet of both C3 and C4 plants, therefore, it is possible that livestock at Bel'sk ingested small to moderate amounts of C4 plants such as millet that were being farmed in the vicinity of the site. The observed range of δ 13 C for humans at Medvin  only overlap slightly with the faunal values, with the majority exceeding them more than the expected 1‰ trophic increase. The average faunal δ 13 C values of −18.6‰, were 3.1‰ lower than the average human values of Bel'sk (−15.5‰). This suggests that the dietary intake of humans relied partially on livestock and C3 plants, alongside C4 food sources such as millet. Measurements of modern millets (in controlled environments and across Eurasia) produce a range of δ 13 C values from −10.0 to −14.1‰ and δ 15 N values that range from 1.8 to 7.8‰ [39-42]. Assuming a diet-consumer enrichment of 3-5‰, the range of human collagen δ 15 N values for humans (9.6 to 14.6‰) fits well within the adjusted range of values (+3-5‰) for both modern millets (4.8 to 12.8‰) and ancient herbivores from Bel'sk (10.8 to 14.3‰). When plotted together, δ 13 C and δ 15 N measurements of bone collagen suggest that millet was an important part of dietary intake in the Scythian era. In addition, slightly higher δ 15 N values at the site of Mamai-Gora might indicate that fish were a dietary component.
We identified a similar pattern in the carbon and nitrogen isotopic composition of humans and livestock from the Pontic steppe as was previously discussed in Ventresca Miller and Makarewicz [43] for much of Central and Inner Asia. At Bel'sk there is a positive correlation between δ 13 C and δ 15 N values for livestock, but a breakdown in this correlation for humans, suggesting a departure from Pontic steppe dietomes. Livestock with a positive correlation between δ 13 C and δ 15 N values could have ingested small amounts of naturally available C4 plants or arid impacted C3 vegetation, rather than feeding on fallow fields. At the sites of Bel'sk, Medvin and Mamai-Gora there was a shift in the average δ 13 C values relative to reference livestock. Pastoral societies (lacking domesticated grains) have isotopic data that demonstrate a shared carbon isospace between humans and livestock, from the same region and time period [43]. As humans and fauna at the Iron Age sites under study share very little carbon isospace, their dietary intake included a partial reliance on foodstuffs atypical for Pontic steppe floral biomes, such as domesticated millet.