Combining aDNA and radiocarbon dating

Humans only live for so long. Thus, if two individuals are identified as having had a close genetic relationship there can only be so many years between their dates of death

The benefits of ancient DNA and radiocarbon dating to archaeological research are undeniable. Ancient DNA (aDNA) helps provide information about genetic relatedness, population movement, and even health, whereas radiocarbon dating is a foundational tool for dating organic remains. Previously, these have been seen as two largely separate techniques, used for different purposes. Now, however, new research led by Jakob Sedig, a member of David Reich’s team at the Harvard Medical School, has examined how aDNA might be useful in helping to refine radiocarbon dates. In this month’s ‘Science Notes’, we explore the results of this study, which were recently published in the Journal of Archaeological Science.

Image: Sedig et al. (2021) Journal of Archaeological Science.

The premise behind combining these two techniques is that humans only live for so long. Thus, if two individuals are identified as having had a close genetic relationship (that is, they are within three degrees of relatedness), there can only be so many years between their dates of death. Utilising this potential maximum difference in ages could, in theory, be used to refine their radiocarbon dates.

To test this hypothesis, the team from Harvard first had to determine ‘date of death’ (DOD) estimates for different degrees of relatives. They began by developing estimates based on biological possibilities. For example, if a 15-year-old couple have a child and the child dies at birth, but both parents live to be 100, then the DOD separation between them would be 85 years. If the opposite were true – the mother died in labour, but the child lived to be 100 – then the separation would be 100 years. For siblings, this maximum could potentially be as high as 130 years, while going as far as third-degree relatives the calculations show that DOD separations could even surpass 195 years.

But these estimates are based on biological possibilities and not real-world probabilities. To determine more realistic DOD separation estimates, the team examined genealogical data and found that the mean value of DOD separation between parent and child was 28.84 years, while for second- to third-degree relatives it was 34.94 years. These estimates are slightly skewed, as most genealogical data comes from individuals of European ancestry in industrialised societies where life expectancy is likely to be higher. Going forward, these estimates have the potential to be refined by social organisation (namely, hunter-gatherer versus pastoralist/agriculturalist), as well as by determinations of biological age of the skeleton using osteological analysis.

Using both the biological maximums and the more likely of the DOD estimates, the team then applied them to genetically related individuals that have previously been identified in the Reich Laboratory database of almost 5,700 published ancient genomes, which are derived from individuals all across the globe and spanning more than 30,000 years. Within the database, 837 individuals have been found as having at least one identified relative, and 203 of these related pairs have previously been radiocarbon dated.

The team first sought out the biggest outliers (that is, those of the related pairs who did not have any overlap in their dates). One example they found was from two Bell Beaker sites in Britain: Amesbury Down and Porton Down. Two individuals, one from each site, were found to share 50% of their autosomal genomes but had different mitochondrial DNA haplotypes, suggesting that they were father and daughter. Their previously recorded radiocarbon dates, however, did not overlap (2480-2280 calBC and 2140-1940 calBC, respectively) – a biological impossibility considering that they were parent and child. Both were then redated, which led to a new date of 2200-2031 calBC for the father, which fits within the expected DOD range.

DNA can also potentially be used to narrow radiocarbon dating ranges. For instance, in the case of the father–daughter pair, using the estimated mean DOD separation of 29 years for parent and child, the radiocarbon dates for these individuals could be refined to 2200-2037 calBC and 2137-1983 calBC, a reduction of nine and 47 years respectively (above). While there is not a huge degree of difference in this particular example, in other instances where date distribution estimates are larger – and particularly for dates where the radiocarbon curve plateaus – this could help narrow dates by hundreds of years.

While, on the whole, radiocarbon dating is incredibly accurate, there are many reasons why results may not be completely precise. For example, if a sample was dated decades ago when the technique was less refined, or there is an unidentified marine reservoir effect skewing the results (see CA 366), or the dates lie within parts of the radiocarbon curve that are not particularly sensitive. In such instances aDNA could be especially helpful. The team developed a statistical programme that narrows ranges by combining DOD estimates and radiocarbon-date distributions, and found that most individuals in a relative pair could have their date ranges reduced by an average of about 50 years. As larger and more refined aDNA databases are collected, the integration of the two techniques could lead to some interesting and unexpected outputs.