Sediment constitutes a ubiquitous feature of archaeological sites. Previous studies have shown that mitochondrial (mt) or chloroplast DNA can be recovered from ancient sediment by PCR [1, 2]. Here we describe the retrieval of mammalian mtDNA from sediments by hybridization capture [3].
We tested the ability of six DNA extraction methods to retrieve DNA of different sizes bound to 150 mg of clay or lime. We then extracted DNA from 83 sediment samples from six sites: Caune de l’Arago (France), Chagyrskaya Cave (Russia), Denisova Cave (Russia), Les Cottés (France), Trou Al’Wesse (Belgium) and Vindija Cave (Croatia). Aliquots of each DNA extract were converted into DNA libraries and subjected to hybridization capture using probes spanning the complete mitochondrial genomes of 242 mammals, and the isolated DNA fragments were sequenced.
We used simulated datasets composed of different proportions of mammalian and bacterial DNA sequences of different lengths, where varying levels of nucleotide substitutions typically present in ancient DNA [4] had been introduced, to test our ability to identify taxa. By comparison to a database of mammalian mtDNA sequences using BLAST [5] and assignment to taxa using MEGAN [6], taxon compositions at the family level were accurately reconstructed even in the presence of high levels of ancient DNA-like nucleotide substitutions. Biases potentially introduced by the capture procedure and the effect of sequencing depth on taxon composition were also investigated.
DNA sequences were retrieved from all archaeological sediments tested. Cytosine to thymine substitutions at terminal positions, an indication that DNA fragments are of ancient origin [4], were seen at five of the sites. A variety of taxa from twelve families were identified, including mammoths and woolly rhinoceroses. We conclude that molecular analysis of sediment by hybridization capture is a highly parallelizable and non-destructive approach to identify the past presence of animals at archaeological sites.