New Paper Exposing The Challenges and Limitations Inherent to Ancient Dietary Reconstructions Using Dental Calculus Metagenomic Data

New research highlights and tests the limitations of dental calculus (i.e., calcified dental plaque), a microbiome substrate regularly used to reconstruct ancient foodways. Under the direction of Allison E. Mann from Clemson University, a collaborative team analyzed both synthetic and ancient dental calculus datasets to demonstrate the intrinsic challenges related to identifying, authenticating, and interpreting ancient microbiome dietary information.

Essential to human culture, biology, and evolution, ancient diet reconstructions help us to understand a fundamental component of the human experience. Dental calculus contains host, microbial, and to a lesser extent, dietary biomolecules, making it an attractive and commonly used substrate for diet reconstruction using metagenomic techniques. However, several factors hinder the accurate detection of dietary biomolecules, including: (1) the fragmented and damaged nature of the DNA molecules, making sequence identification and authentication particularly challenging, (2) the relatively low number of dietary biomolecules as compared to those derived from the endogenous and environmental microbial community, (3) differences in genome size and similarity across different species, and (4) reference database errors or lack of representation.

To assess the challenges intrinsic to genetic analysis of diet, various metagenomic communities were generated. Results emphasized the inherent challenges to current ancient diet reconstructions, calling for more stringent and consistently applied guidelines. Authentic damage patterns require at least ~500 reads for eukaryotes, with microbes needing thousands of reads (Figure 1 and 2). All of the dietary and host organismal reads incorporated into the synthetic datasets were not able to be assigned to any taxonomic level, with many regions across eukaryotic genomes mapping equally well to unrelated sources (Figure 3). Mapping mismatches direct dietary interpretations to organisms that ‘make sense’, relative to archaeological and temporal contexts. However, relying upon reads identified to organisms  expected/suspected to be contextually relevant obstructs the greater complexity of eukaryotic results from metagenomic data and directs interpretations post hoc.

Importantly, expansion of eukaryotic genomic reference databases can help alleviate the challenges identified in the paper. With more, well-covered eukaryotic genomes, uninformative reads mapping to conserved genome regions can be pushed up to broader taxonomic levels through least common ancestor identification methods, as well as providing diverse comparative references. The authors list key evaluation criteria for assessing studies conducting dietary analyses from ancient dental calculus. In following the evaluation points, more study transparency can be gained, and the validity of study claims can be assessed.

Read the full results in: Mann, Allison E., James A. Fellows Yates, Zandra Fagernäs, Rita M. Austin, Elizabeth A. Nelson, and Courtney A. Hofman. “Do I have something in my teeth? The trouble with genetic analyses of diet from archaeological dental calculus.” Quaternary International (2020).

Figure 1. Typical damage patterns indicative of ancient and modern samples and distribution of genomes in NCBI. (a) Authentic ancient samples are characterized by an increase in cytosine to thymine changes at the end of reads due to chemical damage on single stranded overhangs typical of short fragmented ancient DNA reads (ERR2900752). (b) Modern DNA exhibits no pattern of cytosine to thymine changes (ERR3307054). Plots generated using map- Damage2 (Jo ́nsson et al., 2013). (c) Genome sizes per species (with horizontal line showing median). (d) Total number of sequenced genomes, with number on bars showing percentage of genomes from species generally considered to be model organisms. (Taken from Mann et al. 2020)
Figure 2. Effects of low read counts on damage profiles. Four different ancient DNA sample sources were mapped against their respective reference genomes. Each mapped file was then downsampled to different levels of mapped reads, representing different levels of typical dietary hits. (a) represents damage profiles of 100 downsamplings to each set of number of reads. (b) represents the original damage profile of each reference genome with all mapped reads. The number of reads informing each species in (b) are: H. sapiens – 7943448, G. morhua – 5963543, G. vaginalis – 36721, T. forsythia – 134372. Note that for both G. vaginalis and T. forsythia, 100 downsamplings were not always reached for 25-100 read levels, due to insufficient coverage. (Taken from Mann et al. 2020)
Figure 3. Most synthetic ancient reads generated from dietary genomes cannot be classified to the species level. Lollipop plots indicate the proportion of reads at each spike-in level (5000, 500, 50) that were correctly identified at the species level, only to the genus level or the correct genus but wrong species, the incorrect genus or higher taxonomic group (“other”), or could not be assigned to any taxonomic level. Species images downloaded from (Taken from Mann et al. 2020)


Leave a Reply

Your email address will not be published. Required fields are marked *

Please reload

Please Wait