Exploring the potential of dental calculus to shed light on past human migrations in Oceania

Abstract

The Pacific islands and Island Southeast Asia have experienced multiple waves of human migrations, providing a case study for exploring the potential of ancient microbiomes to study human migration. We perform a metagenomic study of archaeological dental calculus from 102 individuals, originating from 10 Pacific islands and 1 island in Island Southeast Asia spanning ~3000 years. Oral microbiome DNA preservation in calculus is far higher than that of human DNA in archaeological bone, and comparable to that of calculus from temperate regions. Oral microbial community composition is minimally driven by time period and geography in Pacific and Island Southeast Asia calculus, but is found to be distinctive compared to calculus from Europe, Africa, and Asia. Phylogenies of individual bacterial species in Pacific and Island Southeast Asia calculus reflect geography. Archaeological dental calculus shows good preservation in tropical regions and the potential to yield information about past human migrations, complementing studies of the human genome.

Fig. 1. Preservation assessment of dental calculus samples.

A SourceTracker analysis of species tables. Each bar represents a sample, coloured by the proportion of each contributing source (sources are the same as in (B)). B PCA of well-preserved calculus from the Pacific islands (blue) with samples from the same sources used in SourceTracker analysis plus additional ancient calculus. C Map of the islands from which samples were collected for this study, with the island name, number of collected samples, and age of the site. Pie charts indicate the fraction of samples from each site that were considered well-preserved, and are coloured by age of the site. Map was produced using the R packages ggmap, plotly, and sf. D Comparison of endogenous oral bacterial DNA and human host DNA in samples with environmental conditions on the islands that may affect DNA preservation. No associations between preservation of either oral bacterial DNA or human host DNA were found with any of the measured environmental factors by ANOVA; values for groups with < 3 samples should be considered unreliable. Boxes show data median, interquartile range (25th–75th percentile) and whiskers indicate minimum and maximum values. E Comparison between the Pacific/ISEA and northern Europe (England/the Netherlands) of the percentage of endogenous oral microbiome DNA in ancient dental calculus and human DNA in bones/teeth. Scale is logarithmic. The number of samples in each group is indicated next to the bars. Ancient dental calculus contains high levels of endogenous oral bacterial DNA in the Pacific/ISEA similar to that seen in northern Europe, in contrast to the lower levels of preserved human DNA in bones/teeth from the Pacific islands compared to northern Europe. Arch. bone - archaeological bone; Anc. calculus - ancient calculus; Mod. calculus - modern calculus.

Fig. 2. Community species profiles are minimally structured by island.

A PCA of well-preserved calculus samples, coloured by the island from which samples were collected; PCs 1, 2, and 3 are shown, accounting for > 50% of the variation in the dataset. B Canonical correlation (CC) analysis comparing the positions of calculus samples in the PCA shown in (A) to environmental and laboratory metadata. Pearson two-sided correlation tests were used to determine if the correlations were significant. Metadata with a p ≤ 0.01 and CC value ≥ 0.4 are marked with an asterisk (*). Ann. evapotransp. - annual evapotranspiration; No. species - number of species. C FAVA values of variance in microbial composition between samples grouped by island. Dotted line indicates the FAVA value of all samples not separated by island. There is only one sample from Watom, so FAVA could not be calculated for this island, and the dot is left unfilled to indicate that FAVA is NA rather than 0. D Specificity of species in calculus samples to island conditions (blue), island location (green), or sequenced library characteristics (red). Dark colours in the violin plots indicate the proportion of species that are significantly associated with that metadata type. No dark colour indicates that no species were specifically associated with that metadata.

Fig. 3. Situating the Pacific/ISEA calculus samples within known ancient dental calculus microbial diversity.

A PCA of Pacific/ISEA calculus samples with ancient dental calculus from additional geographic regions. The additional samples are the same as those in Fig. 1B. B The distance to the centroid of all samples in the PCA, test by ANOVA *** p = 0.001, with Tukey’s Honest Significant Differences; values for groups with < 3 samples should be considered unreliable. Boxes show data median, interquartile range (25th-75th percentile) and whiskers indicate minimum and maximum values. C The number of species in each sample, ordered by PC1 loading, and coloured by the percentage of total reads in the sample that were assigned taxonomy. The trend line is fit with a generalized linear model. D FAVA values of variance in microbial composition between samples grouped by continent. Dotted line indicates the FAVA value of all samples not separated by continent.

Fig. 4. KEGG ortholog (KO) enrichment is associated with sample species composition.

A Heatmap of clustered samples (clusters A and B) and KOs (clusters 1, 2, and 3), showing CLR-transformed copies per million (CPM) for each ortholog. B Percent of KOs in each KO cluster from (A) in KEGG pathways present in all samples. Two-sided Wilcox test with FDR correction, ** p = 0.00272. C Mean percent contribution by genera of KOs enriched in KO Cluster 1, which are enriched in sample Cluster B. Ottowia is the most prevalent genus contributing to these KOs. D Read counts of Ottowia in all samples, aligned with panel (A), showing a higher percentage of samples in cluster B have higher Ottowia read counts than in cluster A.

Fig. 5. Phylogenetic trees show that bacterial genomes from the same island resemble each other.

A A neighbour-joining tree of Anaerolineaceae bacterium oral taxon 439, including only samples with > 5X genomic coverage of the taxon and using only homozygous SNPs, with midpoint rooting. B A neighbour-joining tree of Tannerella forsythia from samples with > 2X genomic coverage, using only homozygous SNPs with midpoint rooting. For both trees, the age of the sample (in years BP) is shown as coloured circles on tree tips, the circle diameter indicates the number of SNPs (x10,000) in that sample, the island of origin is indicated by a coloured box behind sample IDs, the percentage of heterozygous SNPs is shown as a bar, and the mean coverage of the genome as the colour of the bar. Scale bar indicates the genetic distance.