Technical Comments

Response to Comment on “Sedimentary DNA from a submerged site reveals wheat in the British Isles 8000 years ago”

See allHide authors and affiliations

Science  17 Jul 2015:
Vol. 349, Issue 6245, pp. 247
DOI: 10.1126/science.aab2062


Bennett questions the rigor of the dating of our sample from which sedimentary ancient DNA was obtained and the reliability of the taxonomic identification of wheat. We present a further radiocarbon date from S308 that confirms the lateral consistency of the palaeosol age. The suggestion of taxonomic false positives in our data illustrates a misinterpretation of the phylogenetic intersection analysis.

Bennett (1) raises the point that possible variances in the depositional environment make it inappropriate to apply radiocarbon dates from nearby monoliths MS-05 and MS-08 in our study (2), raising the possibility that the wheat DNA we identified could possibly be younger than the 8000 years we claim. The evidence from the radiocarbon dates at the site and those we presented is that the palaeosol is of a consistent age over a 420-m area reaching between the sites of Bouldnor Cliff II, on which our study was based, and Bouldnor Cliff V (3). We therefore felt that the site was dated securely enough in this study. However, we accept that Bennett’s argument is most robustly refuted by obtaining a radiocarbon date from the sample S308, which is a box sample taken from the cliff as outlined in the methods and not, as Bennett suggests, a monolith. We attempted to date both a twig from the sample and the humic acid fraction of the sediment itself. Sediment dating is problematic because of the risk of inclusion of older carbon sources and can lead to overestimation of age. In this case, the humic acid content of the sandy clay was too low to provide a direct sediment date, but the twig returned an age of 7935 to 7790 calendar years before the present (Beta-406961), confirming the age of the sample from which the sedimentary ancient DNA (sedaDNA) was obtained.

Bennett asserts that the phylogenetic intersection analysis (PIA) fails to filter out exotic species such as Ursus maritimus (polar bear), Cervus nippon (Sika deer), and tropical panicoids, and as such casts doubt on the validity of the identity of the wheat DNA. This is incorrect and shows some misunderstanding of the DNA analysis. The analysis is shown to be highly robust and does not in any way falsely identify polar bears, Sika deer, or tropical grasses as being present at the site. These are instances of closest match between sedaDNA reads and the database; it would be naive to interpret these as a species identification, for reasons we explained in some depth in the supplementary materials of (2). These are not instances of reads that have been filtered out by the PIA, but rather meet the criteria of the analysis. The PIA is predicated on the fact that, due to variable database representation, the species of origin may not be present in the database; indeed, often, large taxonomic orders are represented by only a very few species for many genomic regions. The robustness of the analysis comes from examining the phylogenetic range of similar DNAs within a database. In fact, Bennett is alluding to sedaDNA that is attributed robustly to Carnivora, Caniformia, or Ursidae, in the case of the “polar bear,” meaning that the true species of origin lies somewhere in the phylogenetic range encompassed by this order, suborder, or family, respectively. Similarly, cervinae is identified in the case of the “Sika deer,” and various uncontroversial higher taxonomic orders of grasses are identified in the analysis in which the closest database match is from a tropical grass such as rice.

Factors such as genome size and how well genomes are represented in databases have a great influence on the resulting frequency in the DNA profile. Wheat has a large and well-characterized genome, and, furthermore, many wheat species and the sister species Aegilops have been characterized. This means that we have a great power to detect wheat relative to other organisms from metagenomic profiles using the PIA, which is reflected in the relatively large number of wheat sequences we identify. Given the high level of accuracy (81%) of the analysis, the evidence for the presence of wheat at Bouldnor is overwhelming.

Finally, Bennett asserts that sedaDNA requires the existence of macrofossil sources of DNA. The sandy clay from which we took the sedaDNA sample was largely devoid of macrofossils save for a few twigs, with a low organic content, despite the large number of taxonomic orders that were identified. Bouldnor Cliff has not been extensively excavated yet, and it is possible that the quantities of grain involved may be very small and that such wheat macrofossils remain to be discovered. However, the presence of sedaDNA in the absence of macrofossils is precedented (4). We can make no conclusion about the source of the DNA without further evidence. However, if rapid transport occurred—such as might be expected, for instance, using boats associated with pioneer groups on the western coast of France—then the source may even have been flour rather than grain.

References and Notes

  1. Acknowledgments: Funding support for O.S. was by the Natural Environment Research Council (NE/L006847/1).
View Abstract

Navigate This Article