Introduction

In the 1990s–early 2000s, studies of Paleolithic diet in Eurasia based on carbon and nitrogen stable isotope ratios, often accompanied by direct radiocarbon (14C) dating of human remains (see review: Kuzmin and Keates 2014), were initiated (Bocherens et al. 1991, 1999; Richards et al. 2000, 2001). They were concentrated mainly in Western and Central Europe with numerous finds of Neanderthals and early anatomically modern humans (AMHs) (see summaries: Richards 2009; Richards and Trinkaus 2009; Trinkaus et al. 2009; Wißing et al. 2016, 2019). Later on, some research was conducted in Eastern Europe and Siberia (Prat et al. 2011; Dobrovolskaya et al. 2012; Dobrovolskaya and Tiunov 2013; Fu et al. 2014; Drucker et al. 2017). However, the small amount of analyzed contemporaneous animals in Eastern and Central Europe to some extent hampers the interpretation of the results obtained.

In this paper, we present all the available data on stable isotopes for Paleolithic hominins and Late Pleistocene mammals in Siberia and Eastern Europe, and compare it to information collected from Central Europe and neighboring regions. Although zooarchaeological data for the Middle and Upper Paleolithic of Siberia and Eastern Europe exist (see, for example, Hoffecker 2002; Vasil’ev 2003; Kuzmin 2011; Turner et al. 2013), they are less relevant to the issue of which mammals were the dominant objects of hunting for ancient populations due to incomplete survival of the remains of prey bones at Paleolithic sites. The carbon and nitrogen stable isotope values provide the best evidence known to date on the protein sources of prehistoric people (e.g., Richards 2019).

Material and methods

The number of Paleolithic hominin remains analyzed for composition of C and N stable isotopes in both Siberia and Eastern Europe is relatively small (Fig. 1; Table 1). For Siberia and the Russian Far East, only eight individuals have stable isotope values, namely, those from the sites of Denisova and Okladnikov caves, Ust’-Ishim, Maly Log 2, and Malta. In Eastern Europe, there are five sites—Kostenki 1, 8, and 14; and Buran-Kaya III and Sungir—with ten AMH individuals for which the stable isotope values were measured. Basic data on these sites can be found in Hoffecker (2002, 2017), Gerasimova et al. (2007), Mednikova (2011), Fu et al. (2014), Zubova et al. (2016), and Douka et al. (2019). For comparison of data available for Siberian and East European Paleolithic hominins, information generated for the Central European finds of AMH derived from pre-Last Glacial Maximum (LGM) times was used (Table 1).

Fig. 1
figure 1

Location of Paleolithic sites with stable isotope values in Siberia and Eastern Europe, and main geographic regions mentioned in this paper

Table 1 Stable isotope values and 14C dates for the pre-LGM and LGM hominins from Siberia, and Eastern and Central Europe (AMHs, unless otherwise indicated)

The overall number of Paleolithic hominin remains in Eastern Europe and Siberia is not large compared to Central and Western Europe. For example, in a catalog compiled by Gerasimova et al. (2007), 49 fossil hominin localities in Eastern Europe and nine in Siberia are mentioned. Turner et al. (2013: 386–390) discussed the very small number of Paleolithic hominin remains in Siberia—keeping in mind the enormous size of this region, ca. 14,000,000 km2—and arrived at eight possible suggestions, none of which can explain this phenomenon.

The analysis of stable isotopes’ ratios for carbon (δ13C) and nitrogen (δ15N) in bone collagen of the Paleolithic hominins was carried out according to standard procedures (e.g., Richards 2019). Unfortunately, not all hominin samples from Siberia, Eastern Europe, and Central Europe have the whole set of parameters to evaluate the reliability of results obtained (van Klinken 1999; Brock et al. 2012). Nevertheless, for most of them (81% of the total dataset), the crucial value of C:Natom ratio, which should be within the interval of 2.9–3.6 (DeNiro 1985; Ambrose 1990), is available (Table 1). This gives us confidence that the δ13C and δ15N data on Paleolithic hominins considered in this paper are reliable. Many C:Natom values are close to “optimal” C:Natom = 3.243 as suggested by Schwarcz and Nahal (2021). All of them (except the Předmostí 1 site) are less than what was recently suggested as the upper limit for this kind of collagen—C:Natom = 3.45–3.50 (Guiry and Szpack 2021).

The stable isotope data for pre-LGM mammals from the Altai Mountains of southern Siberia are presented in Table 2. The δ13C and δ15N values for Late Pleistocene mammals from the Ust’-Ishim area of Western Siberia are based on samples collected by us (Table 3). These bones have come from alluvial deposits of the lower course of the Irtysh River that were formed during the Marine Isotope Stage 3 interstadial, with 14C dates on wood in the range of ca. 24,500–46,600 BP (Krivonogov 1988). There are a number of 14C dates from this region run on animal bones of woolly rhinoceros, Pleistocene bison and horse, red deer, and musk ox, in the range from ca. 24,600 BP to more than 48,120 BP (see Stuart and Lister 2012; Soubrier et al. 2016; Rabanus-Wallace et al. 2017; Doan et al. 2021). Some of our samples with stable isotope values are directly 14C-dated from ca. 22,460 BP to more than 48,120 BP (Table S1).

Table 2 Composition of stable isotopes for the pre-LGM mammals from the Denisova and Chagyrskaya caves, Altai Mountains, southern Siberia
Table 3 Composition of stable isotopes for the pre-LGM mammals from the Ust’-Ishim area and neighboring regions of Western Siberia (this study; see complete records in Table S1)

The measurements of δ13C and δ15N for West Siberian samples (Table 3) were conducted at the Laboratory of Radiocarbon Dating and Electronic Microscopy, Institute of Geography, Russian Academy of Sciences (Moscow, Russia). Collagen extraction was performed by slow dissolution of the mineral part of the bone in 0.2 M solution of HCl, rinsing, treatment with 0.1 M NaOH, rinsing again, dissolution of the extract in weak HCl at 80° C, and centrifuging. The material obtained was dried and used for analysis. The measurements of δ13C and δ15N values were conducted on an Isoprime PrecisION IRMS coupled on an Elementar varioIsotope Cube (Elementar, Germany and UK), against IAEA-600, B2155, and B2159 (Elemental Microanalysis Ltd.) standards. The analytical precision is 0.2‰ for both δ13C and δ15N. The C:Natom ratios for all animal collagen samples are within the range of 2.9–3.6 (Table S1).

For comparison of hominin stable isotope values with baseline data on the pre-LGM animals, we used the latest summary publications (Rey-Iglesia et al. 2019, 2021; Raghavan et al. 2014; see also Szpak et al. 2010; Kirillova et al. 2015), mainly for Siberia and to a lesser degree for Eastern Europe and the Urals (Table 4). Only samples with direct 14C dates older than ca. 24,000 BP are included into the dataset.

Table 4 Composition of stable isotopes for the 14C-dated pre-LGM mammals from Western Beringia (a.k.a. Northeastern Siberia and Yakutia), Eastern Siberia (a.k.a. South/Central Siberia), and Eastern Europe/Urals (see Fig. 4)

The 14C dating of most Paleolithic hominins mentioned in this study was performed using different protocols of collagen extraction, namely, bulk collagen (Longin 1971; Arslanov and Svezhentsev 1993), ultrafiltered collagen (Higham et al. 2006; Douka et al. 2019), and individual amino acids (mainly hydroxyproline) (Marom et al. 2012). Details can be found in references indicated in Table 1. Uncalibrated 14C ages are given as “BP,” and calendar dates are presented as “cal BP” (mainly as the median values of calibrated age intervals).

The relatively stable composition of isotopic values for mammoths in various parts of northern Asia throughout the last ca. 50,000 years (Kuitems et al. 2019; Rey-Iglesia et al. 2019) allows us to assume that data from these areas are comparable to each other (see also Table 4). We are fully aware that comparison of stable isotope data for hominins and animals from Siberia and Eastern Europe is to some extent limited due to the vast geographic area where the samples were collected. On the other hand, the Paleolithic way of life and the natural environment for these regions in the second part of the Late Pleistocene were similar (e.g., Allen et al. 2010; Velichko et al. 2017: 518), and this allowed us to use the available information.

Results and discussion

Siberia

One sample of presumably pre-modern hominins, called Denisovans, was studied by Dobrovolskaya and Tiunov (2011) (see Table 1). It is not clear, however, if it belongs to a Denisovan because this cranial vault fragment (see Dobrovolskaya and Tiunov 2011: 82) is not listed in the catalog of hominin fossils from Denisova Cave (Douka et al. 2019, Supplementary Information, pp. 27–34). It is possible that this is the non-human parietal fragment from Layer 11 (the Denisova 7 find, according to Douka et al. 2019), identified as a cave bear, although its δ15N value (16.0‰) is too high for this mammal. Therefore, we can only tentatively assume that this sample represents a Denisovan individual. Using the δ15N value, it is possible that a certain amount of aquatic food (freshwater fish) supplemented the terrestrial animals as a source of protein (Fig. 2).

Fig. 2
figure 2

δ13C vs. δ15N plot for possible Denisovan, Neanderthal/Denisovan hybrid, and Neanderthals from the Altai Mountains (see Table 1), on the background of animal stable isotope values (see Table 2) (with ± 1 sigma)

One specimen belongs to the Neanderthal/Denisovan hybrid (the Denisova 11 find, according to Douka et al. 2019) discovered in Denisova Cave (Douka et al. 2019) (Table 1). Its δ15N value is very high (16.4‰), and the consumption of some aquatic food can be suggested (Fig. 2). Unfortunately, we do not have any stable isotope data for Late Pleistocene fish from Western Siberia, and only data on the seventeenth century AD (i.e., pre-industrial) fish from the northern West Siberian Plain is available (Kuzmin et al. 2020; see Fig. 2). The upper end of its δ15N interval is 12.8‰, and we suggest that this individual supplemented its terrestrial animal diet with freshwater fish. Abundant fish bones (non-identified to species) were found in Okladnikov Cave (Derevianko and Markin 1992: 207) located not far from Denisova Cave. The role of the arid environment in the creation of a relatively high δ15N level for the Altai mammals should also be taken into account for both Denisova 11 and the possible hominin from Layer 11 of Denisova Cave.

More stable isotope data is now available for Siberian Neanderthals from the Denisova and Okladnikov caves in the Altai Mountains (Figs. 12; Table 1). The sub-adult humerus from Okladnikov Cave with Neanderthal DNA (see Kuzmin and Keates 2020) was originally 14C-dated to ca. 34,190 BP (Krause et al. 2007), but recently, it was found that “… the true age of these bones is beyond the radiocarbon age limit …” (Higham 2019: 1073), and we assume that its age is more than ca. 45,000 BP (Table 1). When we compare the stable isotope values for three Neanderthals and animals from Denisova and Chagyrskaya caves (Tables 12; Fig. 2), it is clear that the hominin diet was based on terrestrial mammals as was recently concluded by Salazar-García et al. (2021) for the Chagyrskaya Cave individual. It is evident that if we assume a 5‰ increase in the δ15N level from herbivores to carnivores (including Neanderthals) (Bocherens and Drucker 2003; Drucker and Bocherens 2004; Hedges and Reynard 2007), it is possible to interpret the δ15N values (12.7–13.8‰) for the Altai Neanderthals in this way (Fig. 2). The variation of δ15N values for carnivores, from 7.0 to 10.6‰ (Fig. 2; Table 2), is noteworthy. For the oldest directly 14C-dated AMH in Asia, Ust’-Ishim (Fu et al. 2014) (Table 1), we used the stable isotope values for pre-LGM mammals in central Western Siberia (Fig. 3; Table 3). Judging from this limited baseline dataset, the elevated δ15N value (14.2‰) for the Ust’-Ishim individual is hard to explain by the consumption of terrestrial protein only (Fig. 3). Also, hunting of the largest representatives of Late Pleistocene megafauna—woolly mammoth and woolly rhinoceros—is less likely. Currently, we have an extremely limited amount of information about the direct procurement of these species by Paleolithic hominins (Zenin 2002; Kuzmin 2008, 2011; see also Stuart 2021: 48, 63). The main prey of Paleolithic people in Siberia, judging from zooarchaeological data, were reindeer, bison, horse, and mountain sheep (Vasil’ev 2003; Stuart 2021: 63). However, when we take into account the overall data on stable isotopes for the pre-LGM mammals from Siberia and Eastern Europe based on published sources (Table 4; Fig. 4), the major contribution of protein for the Ust’-Ishim individual from terrestrial mammals is evident. The consumption of aquatic food for Ust’-Ishim was previously suggested by Fu et al. (2014) without knowledge of the stable isotope composition for the regional fauna (mammals and fish). It is now clear that it is not necessary to take into account freshwater fish as one of the protein sources for this AMH.

Fig. 3
figure 3

δ13C vs. δ15N plot for the Ust’-Ishim AMH (see Table 1) on the background of major food sources from Western Siberia (see Table 3) (with ± 1 sigma)

Fig. 4
figure 4

δ13C vs. δ15N plot for the AMHs from Siberia and Eastern Europe (see Table 1) on the background of major food sources from Siberia and Eastern Europe/Urals (see Table 4) (with ± 1 sigma). Abbreviations: BK – Buran-Kaya III; K-1 – Kostenki 1; U-I – Ust’-Ishim; K-14 – Kostenki 14; S-5 – Sungir, S-5; M – Malta; K-8 – Kostenki 8; S-1 – Sungir, S-1; S-2 – Sungir, S-2; S-3 – Sungir, S-3; ML – Maly Log 2

For two other AMHs from Siberia, Maly Log 2 and Malta (Table 1), a diet based on terrestrial herbivores can be estimated. The δ15N values for these individuals are similar to those for carnivores and omnivores—gray wolf, brown bear, and cave lion (Fig. 4; Table 4). The δ15N value for Malta (12.2‰), which was previously used by Richards et al. (2001) to assume the consumption of aquatic foods, can now be explained by the contribution of protein from terrestrial sources only.

Eastern Europe

The stable isotope values for the early AMHs from Eastern Europe are quite variable (Table 1; Fig. 4). Some of them—Sungir 1, 2, and 3; and Kostenki 8—are similar to Siberian AMHs from Malta and Maly Log 2; for them, a diet based on terrestrial herbivores can be established (see also Drucker et al. 2021: 9–10). Unfortunately, very little information exists about the stable isotope composition for the pre-LGM mammals of Eastern Europe (Iacumin et al. 2000; Drucker et al. 2017). For the Sungir site, values on reindeer (δ13C =  − 19.1 ± 0.3‰; δ15N =  5.7 ± 0.1‰) (Trinkaus et al. 2014) are within one trophic level of δ15N enrichment for hominins (ca. 3–5‰) who consumed reindeer protein. The values on gray wolf from Sungir are δ13C =  − 19.8‰; δ15N =  9.8‰ (Trinkaus et al. 2014). Comparison with stable isotope data for Siberian mammals demonstrates that the Sungir 1, 2, and 3 individuals are similar to carnivores like cave lion and gray wolf, and to an omnivore like the brown bear (Fig. 4). Zooarchaeological data show that most of the animal bones from Sungir belong to reindeer, Arctic fox, and horse; woolly mammoth and gray wolf are also frequently present (Alekseeva 1998).

The Sungir 5 individual has a higher δ15N value (12.9‰). Available parameters for amino acids used for measurements of stable isotopes and 14C age are within a reasonable range; for example, C:Natom = 3.4 (Sikora et al. 2017, Supplementary Materials, Table S10). Because the δ13C and δ15N values for Sungir 5 were not measured on proper collagen but on amino acids (Sikora et al. 2017, Supplementary Materials, sections S3.1.2 and S3.2.2), data for this individual should be treated as preliminary.

Four AMHs from Eastern Europe—Buran-Kaya III (three individuals) and Kostenki 1—have the highest δ15N levels (15.3–16.8‰) known for the entire inland of Eurasia. The compound-specific (phenylalanine and glutamic acid) δ13C and δ15N values for the Buran-Kaya III humans (Drucker et al. 2017) indicate the consumption of mainly mammoth (40–70% of total diet) and saiga (10–25%) proteins (see Wißing et al. 2019). Therefore, the AMHs from Buran-Kaya III exhibit a purely terrestrial diet without any contribution of aquatic resources. The elevated δ15N levels for mammals and plants from Buran-Kaya III can be explained by the high aridity of the Crimean landscape, compared with northwest Europe (Drucker et al. 2017), during human occupation at ca. 36,300–36,800 cal BP.

Richards et al. (2001) previously suggested a large contribution of aquatic resources (50% and more of total protein) to the diet of the Kostenki 1 individual. However, the lack of data on the stable isotopes for animals from Kostenki in particular, and Eastern Europe in general, to some extent hampers this conclusion. Some information about the landscape around the Kostenki 1 site at ca. 36,500 cal BP (direct 14C date of the Kostenki 1 individual; see Table 1) is available. According to the summary publications (Spiridonova 1991; Hoffecker 2002: 31–32; Hoffecker et al. 2008), the climate at that time was oscillating between colder and warmer conditions, reflected in the amounts of pollen of arboreal (trees and shrubs) and grass species. The Kostenki 1 individual correlates with the cold GS-8 stadial (see Kuzmin 2019); pollen spectra of cultural layer 3 with the human burial show a gradual change of the vegetation from grass-dominated semi-open landscapes to dense spruce forests of the taiga type (Spiridonova 1991: 48–49). The AMH from Kostenki 1 most probably corresponds to the early part of layer 3, and the relatively dry and cold climate can to some extent account for the elevated δ15N value of this individual. This, however, does not completely exclude the consumption of aquatic food (Richards 2009). More work is still needed to determine the diet of the Kostenki 1 individual using compound-specific δ13C and δ15N values for its bone collagen (Naito et al. 2016).

Comparison with Central Europe and neighboring regions, and other issues

The stable isotope values for three Siberian Neanderthals are similar to those for Western Europe (e.g., Wißing et al. 2016, 2019; Salazar-García et al. 2021). The only difference is the slightly elevated δ15N level for Siberian individuals, but this can be explained by the arid environment during their existence.

The stable isotope values for early AMHs in Siberia and Eastern Europe are quite variable compared to Central Europe (Figs. 45), consistent with previous studies (Richards and Trinkaus 2009). For some Central European individuals, like Peştera cu Oase 1 (Fig. 5), the consumption of freshwater food was suggested, although with some reservations related to the small amount of faunal stable isotope values available (see Bocherens, 2019). Similar conclusions were made for Kostenki 1 (Richards 2009) and Ust’-Ishim (Fu et al. 2014). It is now clear that the Ust’-Ishim individual consumed mainly terrestrial animals. The wide variations of the δ15N value for AMHs from Eastern and Central Europe without considering possible variations of the faunal isotopic baseline led to the suggestion of an increase of the food spectrum compared to Neanderthals (Richards and Trinkaus 2009; Trinkaus et al. 2009), although for Western Europe, it was found that the diets of Neanderthals and AMHs were similar (Wißing et al. 2019).

Fig. 5
figure 5

Comparison of δ13C and δ15N values from AMHs in Siberia, Eastern Europe, and Central Europe (see Table 1). Abbreviations: PO – Peştera cu Oase 1; PCU – Peştera Cioclovina Uscată 1; P – Předmostí 1; PM-1 – Peştera Muierii 1; PM-2 – Peştera Muierii 2; BF-2 – Brno-Francouzská 2; DV-35 – Dolní Věstonice 35; Pav – Pavlov; DVII13–43 – Dolní Věstonice II (different individuals); for other abbreviations, see Fig. 4

As for the question of woolly mammoth protein as one of the possible main food sources for AMHs in Siberia and Eastern Europe, it can be suggested that these people could have consumed mammoth meat, although evidence for it is non-existent. At some sites, like Malta and Sungir, reindeer is the most abundant species (Alekseeva 1998; Kuzmin 2008). At Kostenki 1 (layer 3), the dominant species are Arctic fox, horse, reindeer, and wolf (Praslov and Rogachev 1982). Because direct hunting of woolly mammoths was uncommon (see above), in the presence of plentiful prey like horse, bison, and reindeer, it is less likely that Upper Paleolithic AMHs in Siberia and Eastern Europe (first of all, Ust’-Ishim and Kostenki 14 individuals) regularly practiced scavenging of meat from recently dead woolly mammoths. It is clear that more research is needed for Eastern Europe, especially for the Kostenki site cluster, to collect data on the stable isotope composition for animal bone collagen as regional and local baselines.

Information about aquatic food consumption could have implications with respect to estimating the freshwater reservoir effect (FRE) when hominin bones are 14C-dated. According to currently available data about pre-industrial fish in Western Siberia, the FRE is up to ca. 2000 14C years (Kuzmin et al. 2020).

Conclusions

The analysis of stable isotope values for the majority of pre-LGM hominins from Siberia and Eastern Europe (Neanderthals and AMHs) allows us to suggest that terrestrial food resources (especially, herbivores like reindeer, horse, and bison) were the main suppliers of protein. Some of the oldest AMH individuals like Kostenki 1 with a very high δ15N value could have possibly consumed a certain amount of aquatic food, most probably freshwater fish from nearby rivers. The very high level of δ15N for the Neanderthal/Denisovan hybrid from Denisova Cave in Siberia may testify to the consumption of aquatic food.

Overall, the δ13C and δ15N values for pre-LGM hominins from Siberia and Eastern Europe vary greatly, and this reflects the wide range of protein sources used by Paleolithic hunter-fisher-gatherers. More data on the stable isotope composition from contemporaneous fauna is still needed for these regions of northern Eurasia for a better understanding of the patterns of hominin diet.