Analysis of DNA from Ethnoarchaeological Stone Scrapers
Birgitta Kimura, Bruce L Hardy, Steven A Brandt and William W Hauswirth
ABSTRACT
Analysis of DNA residues on stone tools provides a direct method of determining what the tools were used on. However, little is known about the taphonomy of DNA on tools. The discovery of present-day stone-tool using hide workers in Ethiopia therefore provides a unique opportunity to study the survival and authenticity of DNA residues on stone tools. We collected stone scrapers from contexts ranging from excavated to confirmed use, as well as unused scrapers to test for the presence of DNA residues. We amplified segments of the mitochondrial genome with PCR to determine which animal species a tool was used on. We were able to recover authentic DNA from scrapers of known use, and from a subset of excavated scrapers. We did, however, also occasionally obtain DNA unrelated to the use of the tools. Thus, caution must be used when interpreting results of DNA analysis of stone tools.
Keywords
Ethnoarchaeology / DNA / Stone tools / Biomolecular archaeology / Taphonomy
INTRODUCTION
Stone tools are among the most common, indeed often the only, artifacts found at archaeological sites. Knowledge of what they were used for is important for understanding the behavior of prehistoric groups. Although use-wear studies can determine what ranges of tasks tools were used for, it can only indicate the broad category of use-material. Biological residues, on the other hand, have the potential of revealing which species and subspecies of animal or plant the tool was used onto process. Previous studies have used immunological techniques to identify proteins on stone tools (Tuross et al, 1996, Kooyman et al, 1992, Loy and Hardy, 1992), but there are problems with the technique.
Eisele (1994) was unable to detect protein on blood coated tools that had been buried in damp earth for 1 month, and found only weak reactivity on tools buried for 10 months in dry soil. Fiedel (1996) points out that blind tests have shown inconsistencies when tools have been analyzed with different immunological methods, and that species identification frequently does not fit with the faunal records. Tuross et al (1996) showed that experimental tools only contained low amounts of protein and that UV-irradiated tools had no immunological reactivity. Protein residues on pot sherds are also degraded (Evershed and Tuross 1996). Proteins are quickly degraded, antibodies used for detection frequently react with more then one species, and samples may be contaminated with substances that interfere with detection. In addition, the immunoreactive sites on proteins are often conserved between species, and in particular between subspecies.
DNA, although it also degrades and may be contaminated with substances that interfere with detection, is generally more stable than protein and differs more between species, thus offering better resolution for identifying species and subspecies. Since the development of the polymerase chain reaction (PCR), which enables the detection of minute amounts of DNA, there has been an explosion in studies of ancient DNA. Most of them have focused on analyzing bone, for example the study of rabbit mitochondrial DNA in medieval France by Hardy et al (1995), and the determination of Neanderthal DNA sequences by Krings et al (1997). However, other tissues such as brain (Hauswirth et al, 1994), and plant material (Goulubinoff et al, 1993) have also been analyzed. There are also a few published studies of DNA from stone tools. Loy (1993) and Loy and Dixon (1998) used DNA studies in conjunction with protein analysis and hemoglobin crystallization to determine which species certain collections of Alaskan stone tools had been used on, and Hardy et al (1997) analyzed Middle Paleolithic stone tools from La Quina. Loy (1993), however, used a nested PCR approach, which onlyto characterizes the DNA obtained to the sub-Family Bovinae, and.He did not sequence the product, but identified the species as bison by hemoglobin crystallization. Fiedel (1996) points out that this identification is surprising, as there is no archaeological evidence for bison hunting in the area at the time the tools were used. In the study by Loy and Dixon (1998), the DNA analysis was limited to one tool, and the molecular size pattern obtained by nested PCR was compared to control samples. No DNA sequencing was reported. Hardy et al (1997) obtained animal DNA sequences from 5 of 8 tools. However, only one sequence was confined to a tool, and was not present in sediment samples. Consequently only 1 of 8 tools yielded a sequence consistent with the probable use of the tool.
Although there have been a few experimental studies on taphonomy of DNA residues ( Hardy et al, 1997, Gaensslen et al, 1994, Tuross, 1994), little is known about how fast DNA degrades in an archaeological environment, what conditions are best for preservation of DNA or even whether it is practical to obtain DNA from stone tools. In addition, there is the question of whether DNA obtained from a tool is related to it's use, or whether it may reflect in situ contamination (Brown and Brown, 1992). Thus there is a need to study the reliability of DNA analysis from stone tools.
The discovery of contemporary stone tool using hide workers in Ethiopia (Gallagher, 1977, Clark and Kurashina, 1981, Brandt, 1996, Haaland, 1987, Brandt et al, 1996, Brandt and Weedman,1997) provides a unique opportunity to study the survival and authenticity of DNA on stone tools in a natural setting. Using stone tools from contemporary knappers to study taphonomy has several advantages over experimental studies. First, the bias of the investigator is reduced. The hide workers make and use the stone tools as part of their everyday activities, and thus the survival of DNA is more likely to reflect the situation with prehistoric tools. Even in careful experimental studies, it is unlikely that the investigator will recognize and control for all variables which have bearing on the taphonomy of DNA. Although one cannot make the analogy that all stone tools in prehistory were treated as those of the Ethiopian hide workers, analysis of these samples should give an indication of additional variables to examine. Second, tool use, storage patterns and discard practices are known. Thus, the determination of species based on the DNA can be checked against the known use and possible contaminants. This is not possible with archaeological tools. Third, observations of the hide workers will, for example, aid in setting up models about how much use is needed to recover sufficient DNA to analyze, whether general household waste will contaminate DNA on stone tools, and what type of storage and discard practices maximize chances for survival of DNA.
STUDY AREA AND SAMPLE PROCURMENT
In June and July 1995 one of us (SAB) directed a pilot ethnoarchaeological survey of the hide workers of southern Ethiopia (Brandt 1996, Brandt and Weedman 1997, Brandt et al 1997). Focusing on the subsistence farming communities of the Sidama, Konso, Gamo, Wolayta, and Gurage ethnic groups (Figure 1), the main goal of the survey was to determine how many hide workers were still using stone rather then metal to process animal skins into useful products. The survey identified many male and female hide workers who still manufacture and use flaked stone scrapers, hafted into wooden handles, to scrape cow, goat and wild animal hides into skins that are made into bedding, bags and clothing.
The scrapers in this study, examples of which are presented in Figure 2, were obtained from a sample of hide workers from each ethnic group. The hide workers were first interviewed and videotaped for information on the life cycle of stone scrapers from production to discard, followed by the collection of debitage as well as scrapers at various stages of use. A sample ofSome hide workers' trash pits were also excavated for purposes of recovering a sample of buried lithic material.
The Sidama, Wolayta and Gurage manufacture scrapers exclusively from obsidian obtained from various sources, while the Gamo use chert and obsidian and the Konso use chert and quartz. In all cases primary and secondary flaking is done by direct percussion using an iron bar or hoe as a percussor. While all the scrapers are used as end scrapers, typologically some can be described as side and end scrapers. All scrapers are hafted into wooden handles with single or double sockets filled with mastic. The only exception are certain groups of Gamo hide workers, who insert a scraper into the split end of a straight piece of wood, securing the scraper in the open haft by tightly wrapped cord.
The composition of mastic differs between areas, but is composed largely of hardwood resins. However, some hide workers include sheep hair or wool in the mixture. Scrapers are used to scrape off the fatty tissue on the inside of hides that have been dried in the sun for several days. The hides are moistened with water before they are scraped, usually by spraying with the mouth, but in Konso one hide worker applied it with a brush. The Sidama and Gurage only scrape cattle hides, whereas the Konso and Wolayta scrape cattle, goat and occasionally wild animal hides. Most groups keep their cores and unused and used scrapers in baskets, while debris from scraper manufacture and retouch as well as discarded scrapers are dumped away from the living areas into pits, tree hollows or gardens and is rarely intentionally buried. Debris is often collected into a container and periodically emptied into pits.
MATERIALS AND METHODS
Samples for DNA analysis were collected using disposable gloves and put in new plastic bags. Stone scrapers were divided into the following categories: 1) Manufactured, defined as collected directly after observed knapping; 2) "Unused", defined according to hide worker's reports and often stored with other supplies; 3) Known use, defined as scrapers taken out of the handles after observed use by us; 4) Used or usable, defined according to hide worker's reports, and stored among the hide worker's supplies or provided by the hide worker; 5) Excavated, defined as collected from the hide workers disposal areas; 6) Surface collected, and;7) Unknown, defined as scrapers obtained from the hide workers without any information. Three disposal pits were excavated in levels. One in Sidama contained only lithics and had been in use for 20 years according to the hide worker. One in Gamo contained both lithics and other household waste, such as animal bones, pottery fragments and glass, and had been in use for 3 years. The third was a Gurage pit which had been used solely for lithics, but was situated close to where cattle were kept. Other pits containing lithics and mixed household waste in Sidama and Wolayta were sampled by shovel tests. The majority of the samples were analyzed by B.K. at the University of Florida, and a subset by B.L.H. at Indiana University. Some scrapers were examined by B.K. for presence of biological material under low magnification.
A brief description of the DNA analysis is given in this section, while details of the protocols for extraction and analysis are presented in the appendix. To prevent contamination with modern DNA and cross contamination between samples, we used standard precautions suggested for work with ancient DNA (Handt et al, 1994, Stoneking, 1995). Two different approaches were used for the determination of which species the tools had been used on. Both involve comparisons of obtained species-specific DNA sequences with published sequences of the mitochondrial cytochrome b gene (Irwin et al, 1991). This is the gene of choice, because it differs substantially between different species, and has been sequenced for many mammalian species. In one approach, universal primers, capable of amplifying all mammalian DNA were used to amplify a 116 basepair long fragment (Hardy et al 1997). In the second, we used specific primers for a different 121 base pair long region of the gene, that preferentially amplify bovine and caprine DNA. The latter will not amplify human DNA, which is likely to be present on the tools due to handling by the hide workers. The DNA segment used to differentiate between species differ between goat and cow at 9 positions when the universal primers are used, and at 12 positions for the species specific primers. The human sequence differs from the cow sequence at 16 and 21 positions respectively. Generally only one primer was used for sequencing, and species were determined based on a minimum of 8 diagnostic nucleotides. To ascertain that the sequences made phylogenetic sense (Handt et al, 1994, Stoneking, 1995) a 138 base pair long segment of the mitochondrial control region, which is the most variable part of the mitochondrial genome, was amplified for some samples. In this region the African cattle consensus sequence differs from the Indian cattle consensus sequence at 16 positions, and from the European cattle consensus sequence at 2 positions. DNA was extracted from tools at the University of Florida following the silica method developed by Höss and Pääbo (1991), and by guanidine hydrochloride extraction, followed by dialysis at Indiana University (Hardy et al, 1997). Some samples contained a potent inhibitor of PCR, and those were further purified on G-50 Sephadex (Pharmacia) spin columns. All extractions were done in a laboratory separate from that used for subsequent analyses. PCR reactions were set up in a dedicated biological hood which was UV-radiated before use. PCR was performed for 30-45 cycles, the amplified products separated by gel electrophoresis and visualized by ethidium bromide staining. Controls included an extraction blank for each set of samples and a no template PCR control. Samples that yielded a DNA band of the expected size in the PCR were sequenced for subsequent species identification.
RESULTS
We attempted to extract DNA from 42 scrapers with the silica method, and extracts were subjected to PCR using the species specific primers. Universal primers were used for 20 scrapers. The provinience, the presence of biological material under low magnification, and the detection of mastic of the samples are given in Table 1. A summary of the PCR and sequencing results are given in Figure 3 and Table 2.
Species specific primers
All the scrapers collected after observed use yielded PCR products corresponding to the species they were used on, namely cattle. Figure 2B illustrates one of them. It has traces of mastic and yielded a cattle sequence. One of the others differed at one nucleotide position from the published cattle sequence. The difference is in the third position of a codon for asparagine, and would not change the amino-acid incorporated into the cytochrome b protein. It is therefore likely that it is a silent mutation and a natural polymorphism.
Three out of four scrapers obtained as "unused" also yielded cattle sequences. All 3 had, however, shown presence of biological material when examined under low magnification. One of them also had traces of mastic, indicating that it may have been hafted. Two of the four scrapers obtained as used from the hide workers gave a PCR product. One was sequenced as cow (Figure 2A), while the other showed a mixed sequence of cow and ovicaprid. The latter came from Konso, where a single scraper may be used to scrape both cattle and goat hides.
Only 5 of the 20 excavated scrapers gave a PCR product. This product was of the correct length, but only 2 of them yielded enough DNA of sufficient quality to be sequenced (Figure 2C and D). One of the sequences was from cow, the other was a mixed cow/ovicaprid sequence. The latter came from Gurage, where although only cattle skins are scraped, one of the ingredients used in the mastic is sheep hair. One of the three scrapers of unknown provinience yielded a cattle sequence. None of the surface collected scrapers showed any evidence for the presence of DNA. One of the five scrapers collected directly after manufacture gave a product of the correct length after PCR. However, when that sample was sequenced, it showed no resemblance to mammalian sequences, and part of the sequence obtained consisted of repeated sequences of the primer that was used. It is therefore likely to be a PCR artifact. Amplification of a segment of the mitochondrial control region for 6 samples positive for cattle showed 3 separate sequences that matched African cattle sequences (Bailey et al 1996). Sequences obtained are shown in Table 3.
Universal primers
When universal primers were used to amplify DNA from the scrapers, no cattle or goat sequences were obtained. However, 3 human and 3 dog sequences were found on 4 scrapers. All of the scrapers had been used, and two had both dog and human DNA on them. One of the dog sequences was found on a scraper from Gurage, where the hide worker had several puppies in the hide working area, while the others came from Gamo and Konso.
DISCUSSION
DNA analysis of stone tools is advantageous because it has the potential to give more specific information on stone tool use than immunological methods used to detect protein, and it can add to information obtained from use wear analysis. DNA analysis is also non-destructive, which preserves the tool for other future morphological studies. In addition, the initial extraction is simple enough to do in the field. This analysis shows that it is possible to obtain DNA of sufficient quality for species determination from stone tools. Although the scrapers used in this study had only been discarded for a short time, 0-20 years, most degradation seems to take place shortly after cell death (Brown and Brown, 1992), and it shouldmay therefore be possible to isolate DNA from archaeological tools as well.