In order to use stable carbon and nitrogen isotope analysis of bone collagen to recreate human diet, the sample collagen must be isolated from the bone. The method for extraction is the same for the pretreatment of radiocarbon dated bone, but there is variation between laboratories as to the protocol followed. Most laboratories follow a version of the Longin (1971) method which involves the removal of the bone mineral portion using acid and then
27 gelatinization of the ‘collagen’ in heated water. The variable steps include the addition of ultra-filtering after gelatinization, or an alkali rinse during collagen isolation. Pretreatment using an acid-base-acid protocol developed by Berger and Libby (1966) is still used at some laboratories. Samples are rinsed in NaOH in-between HCl rinsing steps as a means of removing humic contaminants and lipids, but the effectiveness of the alkali step has been questioned and is subsequently not included in SUERC’s protocol (Arslanov and Svezhentsev, 1993, Liden et al., 1995, Jørkov et al., 2007, Dunbar et al., 2016).
The pretreatment method at SUERC follows a revised Longin (1971) method and, where requested, will add the extra ultrafiltration step proposed by Brown et al. (1988).
Ultrafiltration is sometimes employed as a method of separating degraded collagen, lipids and other contaminants from the sample. Research that compares the two methods (with and without ultrafiltration) has found that there is no difference larger than the accepted measurement variation (1‰) for δ13C measurements and no significant difference in δ15N measurements (Jørkov et al., 2007). Protocols should therefore be chosen based on the likely preservation and contamination of the sample. Of course, this is very difficult to determine macroscopically in bone samples, therefore suitable quality control indicators must be employed for every sample. Those with C/N ratios outside of the range 2.9-3.6 (the range of C/N of fresh bone) are indicative of well-preserved collagen (DeNiro, 1985). This range has been refined by Van Klinken (1999) who stated that the accepted range employed at Oxford radiocarbon laboratory is 3.1-3.5. In addition, Van Klinken (1999) recommended that collagen yield be measured to ensure that it is >0.5% of the total sample weight which indicates the sample is unlikely to be contaminated. Adoption of quality indicators such as these mitigates the potential effect of there being several different pretreatment protocols to consider. Ultimately, the choice of protocol among those described above is inconsequential as long as the prepared sample passes these tests.
When measuring stable isotope values of modern bone collagen and modern flesh, the lipids must be removed from the sample. This is because lipids are enriched in 12C compared to bone collagen and flesh. In this research, we are only concerned with the measurement of protein values from flesh and bone; this is following the assumption that dietary protein is routed to tissue protein. The application of ultrafilters or NaOH treatment have proved ineffective in removing lipids, so to achieve an accurate measurement of collagen and flesh values, solvent extraction is commonly applied (Guiry et al., 2016, Liden et al., 1995). Like the practice of collagen extraction, there are a variety of pretreatment protocols used to
28 remove lipids; however, there are many different variations and no one is favoured over another.
Many lipid extraction methods employ various volumes and ratios of chloroform (CHCl3) and methanol (MeOH), under the assumption that a polar and a non-polar solvent will produce the best total lipid extraction effect (Folch et al., 1957, Bligh and Dyer, 1959, Lee et al., 1996). This is despite research in food chemistry which found that dichloromethane (CH2CI2) was more effective than chloroform in removing lipids from fish muscle (Cequier-Sánchez et al., 2008). There is very little consensus over which method is most effective, or indeed, whether precise ratios are required. Furthermore, there is disparity over whether solvent lipid extraction has an unpredictable effect on δ15N values, necessitating the need to measure two sample aliquots – one extracted and one non-extracted (Sotiropoulos et al., 2004, Guiry et al., 2016). Ferraz et al. (2004) argued that alternatives to solvent extraction should be sought, since solvents were found to extract some protein along with the lipids, although this was demonstrated in tests on human serum, so the effect on bone is unknown.
The lack of a standard lipid extraction method in the ecological literature means that it is difficult to judge which would be the best to apply to archaeological research.
Most methods of lipid extraction are rooted in ecological studies. These methods concentrate mostly on animal soft tissue, whereas the research presented here required a method which would be effective for animal soft tissue and bone. There is a paucity of research in archaeology which includes the lipid extraction of bone, with some exceptions. Evershed et al. (1995) used a 2:1 v/v chloroform:methanol solvent system with ultrasonication to remove lipids in modern and ancient bone samples. However, the aim of the study was to investigate the viability of measuring ancient lipids using GC/MS, so the methods cannot be confidently applied to this research. Howland et al. (2003) used an alternative solvent system (10:5:4 v/v methanol/chloroform/water) with ultrasonication on modern pig bone collagen to investigate dietary isotope signals. Unfortunately, the focus of the research was on the viability of fatty and amino acids as opposed to collagen, so the methods are not transferable to this research.
The lipid extraction of modern bone collagen samples in archaeological dietary isotope research is rare. Where lipid extraction has been employed it is not always readily apparent in the publications, such as in the analysis of modern deer bone collagen isotope values following 2:1 v/v chloroform:methanol extraction to investigate the canopy effect (Stevens et al., 2006). While the quality of this research is not questioned, the pretreatment methods
29 are only very briefly described as following the method after O’Connell et al. (2001), with no reference to whether lipid extraction was warranted or used (Stevens et al., 2006). Since lipid extraction of bone collagen is not a standard practice in archaeology, it would have been helpful if the method in this case had been more thoroughly addressed. It can also be argued that, given the lack of a standard lipid extraction method of bone collagen in either ecology or archaeology, that more work is required to establish a consensus protocol.
3.2.3 The strengths and limitations of stable isotope analysis for dietary