INTRODUCTION
Stable carbon isotope measurements on human bone collagen are often used to estimate the amount of dietary protein deriving from marine foods and so can help correct radiocarbon (14C) dates subject to marine reservoir effects (Barrett et al. Reference Barrett, Beukens and Brothwell2000; Yoneda et al. Reference Yoneda, Tanaka, Shibata, Morita, Uzawa, Hirota and Uchida2002; Dewar and Pfeiffer Reference Dewar and Pfeiffer2010; Ascough et al. Reference Ascough, Church, Cook, Dunbar, Gestsdóttir, McGovern, Dugmore, Friðriksson and Edwards2012; Craig et al. Reference Craig, Bondioli, Fattore, Higham and Robert2013). In freshwater aquatic systems, stable nitrogen isotopes have generally been found to be of greater utility (Cook et al. Reference Cook, Bonsall, Hedges, McSweeney, Boronean and Pettitt2001; Shishlina et al. Reference Shishlina, Zazovskaya, van der Plicht, Hedges, Sevastyanov and Chichagova2009; Wood et al. Reference Wood, Higham, Buzilhova, Surorov, Heinemeier and Olsen2013; Bronk Ramsey et al. Reference Bronk Ramsey, Schulting, Goriunova, Bazaliiskii and Weber2014; Fernandez et al. Reference Fernandes, Grootes, Nadeau and Nehlich2015; Schulting et al. 2014; Reference Schulting, Bronk Ramsey, Bazaliiskii and Weber2015; Svyatko et al. 2015, 2017a, Reference Svyatko, Schulting, Poliakov, Ogle and Reimer2017b), sometimes in combination with stable carbon isotope ratios, in situations where these differ isotopically from terrestrial ecosystems (Katzenberg and Weber Reference Katzenberg and Weber1999; Yoshii Reference Yoshii1999; Yoshii et al. Reference Yoshii, Melnik, Timoshkin, Bondarenko, Anoshko, Yoshioka and Wada1999). The degree to which isotopic inferences concerning past diets are effective in correcting for marine and freshwater reservoir offsets in 14C can be assessed through paired dating programs of human and terrestrial mammal bone—the latter usually not subject to significant reservoir effects—from the same graves. In many cases, however, the use of carbon and/or nitrogen isotope ratios still leaves much unexplained variation in observed offsets in 14C years. Here, we present the results of a pilot study aimed at exploring the utility of stable hydrogen isotope ratios as an independent proxy for trophic position, in order to address freshwater reservoir offsets at the Early Neolithic and Early Bronze Age cemeteries at Shamanka II, Lake Baikal, southern Siberia (Figure 1). Any additional information that can be gained will contribute to improving the accuracy of radiocarbon determinations on human remains, which in turn provides the framework for an increasing number of bioarchaeological studies in the region (Waters-Rist et al. Reference Waters-Rist, Bazaliiskii, Goriunova and Katzenberg2015; Lieverse et al. Reference Lieverse, Mack, Bazaliiski and Weber2016; Weber et al. 2016a, Reference Weber, Schulting, Bronk Ramsey, Goriunova and Bazaliiskii2016b). This is important since not all graves contain alternative materials suitable or available for dating (i.e., terrestrial mammalian bone/tooth), and so there is still a strong reliance on directly dating human skeletons. There is the potential for wider application in many other contexts where investigating trophic levels is of interest, whether or not these include the need for 14C reservoir corrections.
Figure 1 Map of the Lake Baikal region showing the location of Shamanka II.
Stable Hydrogen Isotopes
The use of stable carbon and nitrogen isotopes in archaeology is sufficiently common to require no detailed introduction, and many overviews are available (e.g., Lee-Thorp Reference Lee-Thorp2008). In the specific context of this paper, they have seen extensive use in the Lake Baikal region, both for palaeodietary reconstruction and for the investigation of freshwater reservoir effects (FRE) (Katzenberg and Weber Reference Katzenberg and Weber1999; Weber and Bettinger Reference Weber and Bettinger2010; Weber et al. 2011, 2016a, Reference Weber, Schulting, Bronk Ramsey, Goriunova and Bazaliiskii2016b; Katzenberg et al. Reference Katzenberg, McKenzie, Losey, Goriunova and Weber2012; Bronk Ramsey et al. Reference Bronk Ramsey, Schulting, Goriunova, Bazaliiskii and Weber2014; Schulting et al. 2014; Reference Schulting, Bronk Ramsey, Bazaliiskii and Weber2015). Important points to bear in mind are that (1) the subsistence economy of the Early Neolithic to Early Bronze Age cultures of the region was based entirely on fishing, hunting and gathering, with a variable but generally strong contribution from the aquatic resources of the lake itself and its surrounding rivers, and (2) Lake Baikal exhibits an unusually variable stable carbon isotope ecology, with fish bone collagen values from different zones ranging from –10‰ to –25‰ (Katzenberg and Weber Reference Katzenberg and Weber1999; Yoshii Reference Yoshii1999; Yoshii et al. Reference Yoshii, Melnik, Timoshkin, Bondarenko, Anoshko, Yoshioka and Wada1999; Weber et al. 2011).
Stable hydrogen isotopes (δ2H, or δD), on the other hand, have seen relatively limited use in archaeology and so require additional discussion. They have been mainly used in forensic applications, food sourcing and ecological studies, particularly to trace migration patterns in animals. They often covary with δ18O ratios and so are used to track mobility and seasonality (Hobson et al. Reference Hobson, Bowen, Wassenaar, Ferrand and Lormee2004; Kirsanow et al. Reference Kirsanow, Makarewicz and Tuross2008; Bowen et al. Reference Bowen, Ehleringer, Chesson, Thompson, Podlesak and Cerling2009). But many studies have also demonstrated that δ2H is subject to a marked trophic level enrichment, such that it combines climate (drinking water) and dietary signals (Birchall et al. Reference Birchall, O’Connell, Heaton and Hedges2005; Reynard and Hedges Reference Reynard and Hedges2008; Soto et al. Reference Soto, Wassenaar, Hobson and Catalana2011; Peters et al. Reference Peters, Wolf, Stricker, Collier and Martinez del Rio2012; Topalov et al. Reference Topalov, Schimmelmann and Polly2013). Perhaps because it combines these two signals, δ2H has been suggested to be particularly useful in distinguishing terrestrial and aquatic systems (Finlay et al. Reference Finlay, Doucett and McNeely2010). While a positive correlation with δ15N ratios would be expected—as these are also a useful proxy for trophic level (Hedges and Reynard Reference Hedges and Reynard2007)—there does seem to be the potential for additional information with δ2H (e.g., Birchall et al. Reference Birchall, O’Connell, Heaton and Hedges2005, Figure 1). This may particularly be the case in dealing with situations where δ15N ratios are affected by factors other than trophic level enrichment, such as aridity or manuring (Amundson et al. Reference Amundson, Austin, Schuur, Yoo, Matzek, Kendall, Uebersax, Brenner and Baisden2003; Bogaard et al. Reference Bogaard, Heaton, Poulton and Merbach2007), though we do not expect either to be relevant in the case of Lake Baikal hunter-gatherers.
MATERIALS AND METHODS
Paired dates were obtained on human bone and terrestrial faunal (marmot: Marmota sibirica; and red deer: Cervus elaphus) dentine collagen from 10 graves at the Early Neolithic (EN) and Early Bronze Age (EBA) cemeteries of Shamanka II, on the southwest shore of Lake Baikal. Of these, three had been previously dated and were included in a study of the region’s FRE (Bronk Ramsey et al. Reference Bronk Ramsey, Schulting, Goriunova, Bazaliiskii and Weber2014; Schulting et al. Reference Schulting, Bronk Ramsey, Goriunova, Bazaliiskii and Weber2014). All the selected humans are of post-weaning age, with the youngest being aged 5–6 yr (see Waters-Rist et al. Reference Waters-Rist, Bazaliiski, Weber and Katzenberg2011). Eight of the graves were known to date to a large Early Neolithic (ca. 7500–6700 cal BP) cemetery based on their mortuary protocols, while two graves were selected from a smaller Early Bronze Age (ca. 4600–3700 cal BP) cluster of burials (Weber et al. 2016a, Reference Weber, Schulting, Bronk Ramsey, Goriunova and Bazaliiskii2016b) in order to look at possible temporal variation in diets and 14C offsets. A number of samples were dated multiple times. These are combined using the R_Combine function in OxCal 4.2 (Bronk Ramsey Reference Bronk Ramsey2013).
Stable Isotope Measurements
Stable carbon and nitrogen isotope ratio measurements were undertaken using the same collagen preparation used for radiocarbon dating (see below). Measurements were made on a Sercon continuous flow IRMS, with a precision of±0.2‰ for both δ13C and δ15N. An alanine standard was used for drift correction on the IRMS, with additional alanine and glutamic acid standards (USGS40: δ13C=–26.4‰, δ15N=–4.5‰; USGS41: δ13C=+37.6‰, δ15N=+47.6‰) used in a three-point calibration of the results (Coplen et al. Reference Coplen, Brand, Gehre, Gröning, Meijer, Toman and Verkouteren2006). The values reported here are the means of measurements in triplicate.
The same prepared collagen was used for stable hydrogen isotope analysis. Measurement of δ2H presents additional challenges, since a minor but not insignificant proportion (ca. 20% for collagen) of the hydrogen in proteinaceous tissues (including collagen and keratin) is prone to exchange with hydrogen in ambient water vapor found in the laboratory undertaking the analysis (Cormie et al. Reference Cormie, Schwarcz and Gray1994; Wassenaar and Hobson 2000, Reference Wassenaar and Hobson2003; Bowen et al. Reference Bowen, Chesson, Nielson, Cerling and Ehleringer2005; Reynard and Hedges Reference Reynard and Hedges2008; Chesson et al. Reference Chesson, Podlesak, Cerling and Ehleringer2009; Meier-Augenstein et al. 2011, Reference Meier-Augenstein, Hobson and Wassenaar2013). While there is a protocol in place for the analysis of keratin making use of international standards (Wassenaar and Hobson Reference Wassenaar and Hobson2003; Bowen et al. Reference Bowen, Chesson, Nielson, Cerling and Ehleringer2005; Meier-Augenstein et al. 2011, Reference Meier-Augenstein, Hobson and Wassenaar2013), there are as yet no such standards for collagen. A further issue has been noted recently affecting both keratin and collagen, involving hydrogen fractionation during the formation of HCN during measurement in the presence of nitrogen (Nair et al. Reference Nair, Geilmann, Coplen, Qi, Gehre, Schimmelmann and Brand2015; Coplen and Qi Reference Coplen and Qi2016; Reynard and Tuross Reference Reynard and Tuross2016). Given these problems, we report both the measured and exchange-corrected values of δ2H, using the latter in the analysis. Since the relationship between them is perfectly linear, the use of either set of values will give the same results in terms of our discussion.
The primary reference material used was IA-R002 (mineral oil, δ2HVSMOW=–111.2‰), traceable to NBS-22 (mineral oil, δ2HVSMOW=–118.5‰), an inter-laboratory comparison standard distributed by the International Atomic Energy Agency (IAEA). In addition, inter-laboratory comparison standard IAEA-CH-7 (polyethylene foil, δ2HV-SMOW=–100.3‰) and FIRMS 221-1 (nylon, δ2HVSMOW=–160.9‰) (http://www.lgcstandards.com) were measured for quality control. Also included in the runs were the keratin standards USGS42 (human hair, non-exchangeable δ2HVSMOW=–72.9±2.2‰), USGS43 (human hair, non-exchangeable δ2HVSMOW=–44.4±2.0‰), and Eurofins 11/2/C (casein, non-exchangeable δ2HVSMOW=–113.4±3.8‰). The δ2HVSMOW ratios for inter-laboratory comparison standards USGS42 and USGS43 have recently been revised (Coplen and Qi Reference Coplen and Qi2016), and we employ the new values here. Eurofins 11/2/C is an inter-laboratory quality control sample provided by Eurofins Scientific. In addition, we included cow and bison collagen standards previously prepared at the Research Laboratory for Archaeology and the History of Art (RLAHA), Oxford (Reynard Reference Reynard2007), which underwent equilibration in our study with two waters with known δ2HVSMOW ratios (depleted Water A=–43.5±0.21‰, and enriched Water B=+110.0±1.42‰, calibrated against in-house standards IA-R053 and IA-R055 at Iso-Analytical), differing by more than 100‰ as recommended by Meier-Augenstein et al. (Reference Meier-Augenstein, Chartrand, Kemp and St-Jean2011). Since the δ2H ratios of the cow and bison (Table 1) do not entirely bracket the range of human values in our study, and therefore are not entirely appropriate for their calibration, we are in the process of preparing a new marine seal bone collagen standard regularly used as an internal standard for stable carbon and nitrogen isotope measurements at Oxford.
Table 1 δ2H results for standards δ2HVSMOW(non-exchangeable)=δ2HVSMOW(measured) – (–36.752/0.892).
USGS42, USGS43 and Eurofins 11/2/C were weighed into open capsules and simultaneously equilibrated alongside the archaeological samples and collagen standards with ambient water vapor at Iso-Analytical (Crewe, Cheshire, UK) for 12 days prior to analysis (cf. Wassenaar and Hobson Reference Wassenaar and Hobson2003). These were only sealed and added to the sample carousel once batch analysis had begun. They should therefore be subject to the same exchange between the different runs, adhering to the “Principle of Identical Treatment” (Werner and Brand Reference Werner and Brand2001). Samples were kept in a sealed glass container with dessicants until ready for analysis. Measurements were undertaken at Iso-Analytical by Elemental Analyser–Isotope Ratio Mass Spectrometry (EA-IRMS). Following equilibration, samples and references were placed in silver capsules and loaded into a zero-blank auto-sampler, flushed with 99.9992% helium at a flow rate of approximately 50 mL/min before and during the entirety of batch runs. The samples were then dropped into a furnace at 1080°C and thermally decomposed to H2 and CO over glassy carbon. Any traces of water produced were removed by magnesium perchlorate, and any traces of CO2 were removed via a Carbosorb™ trap. H2 was resolved by a packed column gas chromatograph held at 35°C. The resultant chromatographic peak entered the ion source of the IRMS where it was ionized and accelerated.
The measured δ2H ratios for USGS42, USGS43 and Eurofins 11/2/C were used to obtain a 3-point linear calibration (y=0.892x−36.752, r 2=0.987, n=36), which was used for the exchangeable hydrogen correction. We assume that keratin and collagen powders will behave similarly in terms of exchangeable H, though this may not be valid because of differences in their amino acid composition (Reynard and Tuross Reference Reynard and Tuross2016; Soto et al. Reference Soto, Koehler, Wassenaar and Hobson2017). We have not applied a separate calibration using the collagen standards, pending the dual water calibration of the seal collagen. While the reported values may thus be subject to revision, any changes will apply equally to all the archaeological samples and so will not affect our discussion here. It is worth noting that the sub-fossil bison collagen we employed as one of our standards is the same material used in a recent study by Reynard and Tuross comparing different protocols, which yielded a nearly identical measured δ2HVSMOW-SLAP ratio of –151.4±1.7‰ (n=5) compared to that of –151.9±4.2‰ (n=12) obtained in our study (Table 1), despite being analyzed in—and equilibrated to the atmosphere of—a laboratory on the eastern seaboard of the United States (Reynard and Tuross Reference Reynard and Tuross2016: tab. 1). While on opposite sides of the Atlantic, the two laboratories are in zones sharing similar δ2H and δ18O precipitation values (Bowen Reference Bowen2010), and so their equivalence is not surprising but does provide additional confidence in the results. Measurement precision during the runs for the non-organic standards (mineral oil, polyethelene and nylon) was 1.9‰ (n=67), and that for the organic standards (keratin and casein) was 2.9‰ (n=45) (Table 1).
Radiocarbon Dating
Radiocarbon measurements were undertaken following the standard protocols in place at the Oxford Radiocarbon Accelerator Unit (ORAU) (Brock et al. 2007, Reference Brock, Higham, Ditchfield and Bronk Ramsey2010). Briefly, bone surfaces are cleaned by shot-blasting after which they are crushed and then demineralized using an acid-base-acid treatment (0.5M hydrochloric acid – 0.1M sodium hydroxide – 0.5M HCl). The resulting “collagen” is then gelatinized (after Longin Reference Longin1971) and filtered with an Ezee filter. The filtrate then undergoes a 30 kD ultrafiltration step, in which the >30 kD gelatin fraction is retained, washed with milliQ ultrapure water, and freeze-dried pending conversion to CO2 for 14C AMS measurement (Brock et al. Reference Brock, Higham, Ditchfield and Bronk Ramsey2010).
RESULTS AND DISCUSSION
Results of the paired dating program are provided in Table 2. All C:N ratios fall between 3.1 and 3.4, indicating well-preserved collagen (DeNiro Reference DeNiro1985; Ambrose Reference Ambrose1990). The human-faunal offsets range between 0 (i.e., no offset) and 679 14C yr. The paired dating in Grave 104 showing no offset is clearly an outlier, the removal of which results in an average offset of 515±175 14C yr. The individual in Grave 104 may have been an outsider who died and was buried at Shamanka II not long after arriving there. The fact that their δ13C, δ15N and δ2H values all suggest a significant contribution of aquatic resources from Lake Baikal itself (or a connecting river) is puzzling, since a 14C reservoir offset would then be expected. It may be that other as yet unidentified river systems in the Cis-Baikal region are also 13C-enriched, but lack a significant reservoir age.
Table 2 14C paired dating results on humans and fauna from Shamanka II. M=male; F=female; I=indeterminate. Samples with multiple dates have been combined in OxCal using the R_combine function; in all cases these passed χ2 tests (Ward and Wilson Reference Ward and Wilson1978). Dates from graves marked * have been previously published (Bronk Ramsey et al. Reference Bronk Ramsey, Schulting, Goriunova, Bazaliiskii and Weber2014; Schulting et al. Reference Schulting, Bronk Ramsey, Goriunova, Bazaliiskii and Weber2014).
The reported δ13C, δ15N and δ2H values are averages of measurements made in triplicate (or multiples thereof in the case of δ13C and δ15N where multiple dates were obtained on the same human skeleton). The average standard deviation of the triplicate calibrated δ2HVSMOW measurements on human bone is 3.9±3.2‰. In one case, two different samples from the same individual (Grave 39) were also measured for δ2H, with the resulting six measurements giving very consistent ratios with a mean of –28.9±2.0‰. In two cases, however, the triplicate measurements are more variable (Grave 75: +1.8±7.9‰; and Grave 104: –41.8±11.3‰). The considerable degree of variability in δ2H between individuals at Shamanka II is apparent from the high coefficient of variation (Table 3), and is consistent with previous studies reporting particularly high variance in δ2H values from carnivores (Topalov et al. Reference Topalov, Schimmelmann and Polly2013). This can also be seen in the values of –3.5‰ and +31.7‰ for two archaeological Baikal seals (nerpa, Phoca sibirica) from the site of Sagan Zaba II (Nomokonova et al. Reference Nomokonova, Losey, Goriunova and Weber2013), though these are both still substantially higher than the values of –97.1‰ and –83.3‰ obtained on two unidentified ungulates from the same site (Table 3). The human δ2H average of –29.7±21.9‰ is, as would be expected, intermediate between the seals and the ungulates, though closer to the former. As a point of reference, the waters of Lake Baikal exhibit extremely homogeneous δ18O and δ2H values horizontally, vertically and seasonally, averaging –15.8±0.3‰ and –123.0±1.2‰ (n=32), respectively (Seal and Shanks Reference Seal and Shanks1998).
Table 3 Human δ13C, δ15N and δ2H results from Shamanka II. Both measured and “non-exchangeable” δ2H are reported, with the caveat that the latter have not been corrected with like-for-like standards (i.e., collagen).
Considering the human stable isotope results, there is no significant correlation between either δ13C and δ15N (r 2=0.138, p=0.291) or δ13C and δ2H values (r 2=0.306, p=0.098). There is, however, a strong correlation between δ15N and δ2H values (r 2=0.885, p=0.000) (Figure 2), providing further support for the use of δ2H as a proxy for trophic level. This is not unexpected, since an estimated 60% of the non-exchangeable hydrogen in bone collagen derives from food, with the remainder coming from drinking water (Reynard Reference Reynard2007). But, to our knowledge, such a strong correlation has not previously been observed in humans. The explanation may lie in the coincidence of hydrogen from both food (i.e., fish) and drinking water from Lake Baikal. This relationship is carried through in the positive correlations with the observed human-faunal 14C offsets. While there is no significant correlation in the case of δ13C (r 2=0.000, p=0.991) (see also Schulting et al. Reference Schulting, Bronk Ramsey, Goriunova, Bazaliiskii and Weber2014), both δ15N (r 2=0.428, p=0.040) and δ2H (r 2=0.482, p=0.023) are significant predictors of the 14C offsets, accounting for approximately 45% of the observed variability (Figures 3 and 4).
Figure 2 Scatterplot with linear regression of δ2H and δ15N ratios on human bone collaen, showing strong positive correlation (r 2=0.885, p=0.000). Error bars show average±1SD.
Figure 3 Scatterplot with linear regression model of human δ15N values and human-faunal offsets in 14C yr, showing a moderate positive correlation (r 2=0.428, p=0.040). The outlier Gr 104 is identified.
Figure 4 Scatterplot with linear regression model of human δ2H values and human-faunal offsets in 14C yr, showing a moderate positive correlation (r 2=0.483, p=0.023). Error bars show average±1SD for δ2H.
The results are in fact strikingly similar, as might be expected given the strong correlation between the two stable isotopes, although the relationship with δ2H is actually slightly stronger (although a regression model incorporating both δ2H and δ15N fails to attain statistical significance (adjusted r 2=0.334; p=0.100)). This is confirmed in a multiple linear regression model (Equation 1), in which both δ13C (p=0.034) and δ2H (p=0.036) ratios contribute significantly to the predicted 14C offset (adjusted r 2=0.656, p=0.024), while, interestingly, δ15N does not (p=0.199). Re-running the regression model (Equation 2) excluding δ15N reduces the contribution of δ13C to the point where it just fails to attain significance (p=0.066)—though nevertheless providing a significant improvement over the single isotope models (adjusted r 2=0.603, p=0.016)—while retaining the significance of δ2H (p=0.007). For both models all standardized residuals are less than two standard deviations.
$$\eqalignno{ & ^{{14}} {\rm C}\,{\rm offset}{\equals}-\hskip-3pt 1903.2662-340.8886\left( {{\rm \rdelta}^{{13}} {\rm C}} \right)-189.8488\left( {{\rm \rdelta}^{{15}} {\rm N}} \right){\plus}21.5779\left( {{\rm \rdelta}^{2} {\rm H}} \right) \cr & F{\equals}6.72,\,{\rm adjusted}\,{r}^{2} {\equals}0.656,\,p{\equals}{\rm }0.024,\,S{\equals}135.92,\,{\rm n}{\equals}10 $$
$$\eqalignno{ & ^{{ 14}} {\rm C}\,{\rm offset}{\equals}-\hskip-3pt 3338.0028-246.2623\left( {{\rm \rdelta}^{{13}} {\rm C}} \right){\plus}10.5748\left( {{\rm \rdelta}^{2} {\rm H}} \right) \cr & F{\equals}7.83,\,{\rm adjusted}\,{r}^{2} {\equals}0.603,\,p{\equals}0.016,\,S{\equals}146.03,\,{\rm n}{\equals}10 $$
That the regression equations obtained here are less precise than those previously produced for the Cis-Baikal region, including Shamanka II (Bronk Ramsey et al. Reference Bronk Ramsey, Schulting, Goriunova, Bazaliiskii and Weber2014; Schulting et al. Reference Schulting, Bronk Ramsey, Goriunova, Bazaliiskii and Weber2014), is largely the result of the much-reduced range of variability in both the 14C offsets and the dietary stable isotope data at this site. These previously published equations, therefore, are still preferred for the correction of 14C dates on human remains in the Southwest Baikal and Angara micro-regions.
While a comparison of the Early Neolithic and Early Bronze Age results is limited by the inclusion of only two graves from the latter period, it is worth noting that the EBA results fall below the entire range seen in the Early Neolithic individuals in terms of their δ15N and δ2H values as well as their 14C offsets (with the exception of the abovementioned Grave 104). They do not differ, however, in their δ13C values (Table 3).
CONCLUSIONS
The average offset of 515±175 14C yr is consistent with that of 537±80 yr previously reported for Shamanka II (Schulting et al. Reference Schulting, Bronk Ramsey, Goriunova, Bazaliiskii and Weber2014). Linear regression models making use of δ13C, δ15N, and δ2H both separately and in combination suggest that stable hydrogen isotope ratios are at least as useful a predictor of the 14C offsets as stable nitrogen isotopes. At the same time, the greater difficulties involved in the measurement of δ2H compared with that of δ15N, and the thus far limited additional information obtained, are factors that do need to be taken into account. Nevertheless, our results provide further support for the utility of δ2H as a proxy for trophic position, and suggest that this isotope system holds promise for future investigations in the Baikal region and elsewhere. In terms of correcting dates on human remains, the previously published equations are still preferred (Bronk Ramsey et al. Reference Bronk Ramsey, Schulting, Goriunova, Bazaliiskii and Weber2014; Schulting et al. Reference Schulting, Bronk Ramsey, Goriunova, Bazaliiskii and Weber2014). This may be modified in future as new results become available.
ACKNOWLEDGMENTS
This research was undertaken as part of the Baikal-Hokkaido Archaeological Research Project, funded by the Social Sciences and Research Council of Canada (Major Collaborative Research Initiative grants Nos. 410-2000-1000, 412-2005-1004, and 412-2011-1001). Many thanks to Linda Reynard and Wolfram Meier-Augenstein for discussions concerning sample preparation and interpretation of stable hydrogen isotopes, and for the comments provided by two anonymous reviewers, which have improved the clarity of the text.
