Insights into demographic and cultural influences on the oral microbiome from historical Japanese dental calculus Authors: Yuki Kuriyama, Fuzuki Mizuno, Takashi Yamada, Masahiko Kumagai, Masaki Tanaka, Izumi Naka, Kazuaki Hirata, Soichiro Mizushima, Shiori Yonemoto, Kyoko Funahashi, Noriko Seguchi, Kazuhiro Sakaue, Eisaku Kanazawa, Masami Matsushita, Takayuki Matsushita, Aiko Saso, Takashi Nara, Ryohei Sawaura, Chiaki Katagiri, Tomoki Maekawa, Kunihiko Kurosaki, Hisako Saitoh, Shuuji Namiki, Shintaroh Ueda, Jun Ohashi Published: 2026 ABSTRACT Recent advances in genomic technologies have enabled detailed analyses of ancient microbiomes using dental calculus. While most studies have focused on European and North American populations, ancient Japanese oral microbiomes remain largely unexplored. Here, we analyzed dental calculus primarily from Edo period individuals (17th–19th centuries) to investigate the compositional, functional, and phylogenetic diversity of ancient oral microbiomes. Our results revealed clear differences between ancient and modern Japanese microbiomes, as well as phylogenetic divergence between the Final Jomon (ca. 1000 BCE) and Edo periods. We also identified regional variation in ancient oral microbiomes and clade-level diversity within the periodontitis-associated archaeon Methanobrevibacter oralis. Interestingly, individuals with traces of tooth blackening (ohaguro), a custom practiced by Edo period women, were all assigned to the same clade, suggesting cultural influences on the oral microbiome. These findings highlight the important role of human culture and demography in shaping the evolutionary dynamics of microbiomes.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-55286-2. Introduction Distinct bacterial communities are formed in various parts of the human body, including the skin, oral cavity, and intestine. These microbiomes play a crucial role in host physiology and have been shown to influence human health and development. In particular, the intestinal microbiome interacts with the host’s immune, endocrine, and nervous systems, and has been implicated in the development of various diseases, such as inflammatory bowel disease, cancer, and autism1,2. Similarly, dysbiosis in the oral microbiome has been associated with oral diseases, including dental caries and periodontitis3. Although individual microbiomes are generally stable over time, they can undergo plastic changes in response to lifestyle factors such as diet4 and antibiotic use5. The impact of modernization on the microbiome is particularly significant, raising the possibility that ancient humans harbored microbiomes with different diversities compared to modern humans6. Furthermore, it has been suggested that some lifestyle diseases commonly observed in developed countries, such as diabetes and obesity, may result from deviations from the microbiome traditionally shaped under a hunter-gatherer lifestyle7,8. Recent studies have demonstrated that ancient human microbiomes can be reconstructed and analyzed using coprolites (fossilized feces) and dental calculus (calcified plaque)9,10. By comparing these ancient microbiomes with those of modern humans, researchers can investigate temporal changes in microbiome diversity and assess their biological significance in relation to changes in human lifestyles11–14. In addition to overall microbiome diversity, the phylogenetic diversity of specific bacteria, including those associated with disease, can also be examined across time. This approach enables the identification of ancient-specific bacterial lineages, the reconstruction of bacterial virulence evolution15–17, and the inference of human population history based on phylogenetic similarities among oral bacteria11,18. This methodology is particularly valuable in regions such as East Asia and Oceania, where high temperatures and humidity hinder the preservation of ancient human genomes. In these areas, the well-preserved bacterial DNA in dental calculus provides an alternative means of investigating human population history19. In Japan, research on ancient microbiomes using dental calculus has also progressed. It has been reported that clusters of bacteria associated with periodontitis underwent historical changes20. Additionally, Methanobrevibacter oralis, an archaeon linked to periodontitis21, was more frequently detected in the dental calculus of women from the Edo period (17th–19th centuries) than in that of men, possibly due to the practice of ohaguro (Fig. S2, a custom in which married women stained their teeth black). Furthermore, phylogenetic analysis of Anaerolineaceae sp. oral taxon 439 has revealed that lineages of the Final Jomon period (ca. 1,000 BCE) were distinct from those of the Edo period, suggesting the existence of era-specific phylogenetic diversity22. However, compared to Western countries, studies utilizing ancient Japanese dental calculus remain scarce, and the diversity of ancient Japanese oral microbiomes has yet to be fully elucidated. Samples from the Edo period offer several advantages: Japan’s national isolation (sakoku) minimized foreign influence; restricted travel between domains allowed for the detection of regional variation; and the relatively recent date ensures better preservation of microbial DNA. These factors make Edo period dental calculus particularly suitable for investigating microbiome diversity in ancient Japan and for identifying commonalities and differences with modern microbiomes. In this study, we conducted shotgun sequencing primarily on ancient dental calculus from the Edo period. We analyzed the metagenomic data to investigate the taxonomic composition, functional characteristics, and phylogenetic diversity of ancient Japanese oral microbiomes. Additionally, we examined how regional and social factors, including lifestyle practices such as ohaguro, influenced microbiome diversity. Comparison of ancient and modern Japanese oral microbiomes may provide insights into historical changes in microbial diversity and their biological significance. Methods Sample collection Dental calculus samples were collected from 9 modern Japanese individuals (Niigata: n = 9) and 118 ancient Japanese individuals, mainly from the Edo period (Tokyo: n = 52, Saitama: n = 4, Yamanashi: n = 15, Fukuoka: n = 8, Okinawa: n = 39) (Table 1, Fig. S3). For the modern Japanese individuals, oral health status was clinically evaluated by a licensed dentist. The mean probing pocket depth (PPD) across the nine participants was 5.2 ± 2.2 mm, indicating a generalized state of periodontitis among the cohort. For the ancient samples, dental calculus was primarily collected from teeth remains where the relationship between the calculus deposit and the underlying alveolar bone could be observed. Visual inspection of the mandibles and maxillae revealed varying degrees of alveolar bone resorption at the sites of calculus attachment, consistent with chronic periodontal attachment loss in these historical individuals. Additionally, two soil samples were collected in Japan. Dental calculus samples from modern individuals were obtained and analyzed after obtaining informed consent from all participants. All procedures involving human participants were conducted in accordance with the Declaration of Helsinki. The studies had been approved by the Ethics Committee of Toho University School of Medicine (A21001 and A23105). Table 1Burial sites, periods, and number of ancient Japanese dental calculus samples obtained in this study.Burial sitePrefecturePeriodNumber of samplesOoka family cemeteryTokyoEdo period3Nagai family cemeteryTokyoEdo period13Makino family cemeteryTokyoEdo period8Ikenohata Shichiken-choTokyoEdo period2Enouji temple siteTokyoEdo period4Sogenji templeTokyoEdo period4Hitotsubashi High SchoolTokyoEdo period18Tozuka TateyamaSaitamaEdo period4ShiokawaYamanashiEdo period15Ogasawara family cemetery (Sogenji temple site)FukuokaEdo period2KyomachiFukuokaEdo period2Haruda siteFukuokaEdo period4Kanjinbaru ancient tombsOkinawaEarly modern period7Yacchi-no-GamaOkinawaEarly modern period15Isobe gas stationOkinawa17th–19th century1NosokoOkinawa16th–19th century4TonoshiroOkinawa15th–19th century2KuramotoOkinawa15th–16th century3ArakawaOkinawa15th–16th century1Kanzabaru ancient tombsOkinawaGusuku period6 Table 1Burial sites, periods, and number of ancient Japanese dental calculus samples obtained in this study.Burial sitePrefecturePeriodNumber of samplesOoka family cemeteryTokyoEdo period3Nagai family cemeteryTokyoEdo period13Makino family cemeteryTokyoEdo period8Ikenohata Shichiken-choTokyoEdo period2Enouji temple siteTokyoEdo period4Sogenji templeTokyoEdo period4Hitotsubashi High SchoolTokyoEdo period18Tozuka TateyamaSaitamaEdo period4ShiokawaYamanashiEdo period15Ogasawara family cemetery (Sogenji temple site)FukuokaEdo period2KyomachiFukuokaEdo period2Haruda siteFukuokaEdo period4Kanjinbaru ancient tombsOkinawaEarly modern period7Yacchi-no-GamaOkinawaEarly modern period15Isobe gas stationOkinawa17th–19th century1NosokoOkinawa16th–19th century4TonoshiroOkinawa15th–19th century2KuramotoOkinawa15th–16th century3ArakawaOkinawa15th–16th century1Kanzabaru ancient tombsOkinawaGusuku period6 Burial sites, periods, and number of ancient Japanese dental calculus samples obtained in this study. Publicly available datasets Publicly available dental calculus and soil datasets were obtained from NCBI (National Center for Biotechnology Information) and ENA (European Nucleotide Archive), including data from eight modern Spanish individuals11, ten Edo period Japanese individuals, five Jomon period individuals22(Table 2), eleven ancient Italian individuals15, four Neanderthals13, and ten soil samples23. Table 2Burial sites, periods, and number of ancient Japanese dental calculus samples obtained from previous study (Eisenhofer et al., 2020).Burial sitePrefecturePeriodNumber of samplesIkenohata Shichiken-choTokyoEdo period10Ebishima shell middenIwateJomon period1Miyano shell middenIwateJomon period3Ikawazu shell middenAichiJomon period1 Table 2Burial sites, periods, and number of ancient Japanese dental calculus samples obtained from previous study (Eisenhofer et al., 2020).Burial sitePrefecturePeriodNumber of samplesIkenohata Shichiken-choTokyoEdo period10Ebishima shell middenIwateJomon period1Miyano shell middenIwateJomon period3Ikawazu shell middenAichiJomon period1 Burial sites, periods, and number of ancient Japanese dental calculus samples obtained from previous study (Eisenhofer et al., 2020). Modern dental calculus DNA extraction, library preparation and sequencing Modern dental calculus samples were processed in a standard molecular biology laboratory dedicated to modern DNA work. Surface decontamination was performed by ultraviolet (UV) irradiation for 1 min on a single side of each sample. Samples were subsequently washed twice with molecular-grade water and mechanically crushed using sterile plastic pestles. DNA extraction from modern dental calculus was performed using the DNeasy PowerBiofilm Kit (QIAGEN) following the manufacturer’s protocol. DNA was eluted three times with 20 µl, 20 µl, and 11 µl of EB buffer, yielding a total volume of 51 µl. DNA concentrations were quantified using the Qubit High Sensitivity dsDNA Assay Kit (Thermo Fisher Scientific). Double-stranded DNA libraries were constructed using the NEBNext Ultra II FS DNA Library Prep Kit for Illumina (New England Biolabs) with dual-index primers, according to the manufacturer’s instructions with minor modifications. This kit includes an enzymatic fragmentation step prior to adapter ligation, which was applied to the modern dental calculus extracts. Library quality and concentration were assessed using an Agilent Bioanalyzer and the Qubit dsDNA HS Assay Kit. Libraries were then pooled at equimolar concentrations and sequenced on the Illumina MiSeq platform using the 150-cycle v3 kit. Ancient dental calculus DNA extraction, library preparation and sequencing Dental calculus samples were processed in dedicated cleanspace facilities designed for ancient DNA (aDNA) work to minimize contamination. Surface decontamination was performed by ultraviolet (UV) irradiation for 1 min on a single side of each sample. Samples were then washed twice with molecular-grade water and mechanically crushed using sterile plastic pestles. DNA extraction was performed using a silica-based protocol optimized for degraded DNA. Digestion was carried out in 400 µl of 0.5 M EDTA containing 10% Proteinase K and 2% SDS, with incubation for 24 h at 56 °C, repeated twice under rotation. Following digestion, samples were centrifuged at maximum speed for 1 min to pellet debris. The supernatant was purified using the Qiagen MinElute Reaction Cleanup Kit according to the manufacturer’s protocol, with minor modifications. DNA was eluted twice with 11 µl of EB buffer, yielding a total volume of 21 µl. DNA concentrations were quantified using the Qubit High Sensitivity dsDNA Assay Kit (Thermo Fisher Scientific). Double-stranded DNA libraries were constructed using the NEBNext Ultra II DNA Library Prep Kit for Illumina (New England Biolabs) with dual-index primers, following the manufacturer’s instructions with minor modifications. Library quality and concentration were evaluated using an Agilent Bioanalyzer and the Qubit dsDNA HS Assay Kit. Libraries were pooled at equimolar concentrations and sequenced on the Illumina MiSeq platform using the 150-cycle v3 kit. Quality control Quality control of shotgun metagenomic data was performed using fastp (ver. 0.22.0) with the parameters -q 20 -f 1 -F 1 -t 1 -T 1 -l 20, followed by adapter removal (--detect_adapter_for_pe) and paired-end read merging (-m --overlap_len_require 11 --include_unmerged)24,25. Human reads were removed using KneadData (ver. 0.10.0) with the hg37 reference genome. Soil contamination was assessed on a per-sample basis using decOM (ver. 1.1)26. Taxonomic classification Sequence similarity searches against the NCBI nr database were performed using DIAMOND (ver. 2.1.6)27, and taxonomic assignment was conducted using MEGAN (ver. 6.25.10) with the Lowest Common Ancestor (LCA) algorithm28–30. LCA parameters were set as follows: LCA Algorithm = naïve, minimum bit score = 50, maximum E-value = 0.01, top percent = 10, and minimum support percent = 0.01. To minimize the effects of soil contamination—particularly relevant to ancient microbiome analysis—only oral taxa were extracted based on the Human Oral Microbiome Database (HOMD ver. 3.1)31, and diversity analysis was conducted at the species level. Microbiome composition analysis Bray–Curtis distances were calculated based on the proportion of reads assigned to each species. Principal Coordinate Analysis (PCoA) was then conducted to visualize microbial community diversity. Covariance between species proportions and sample positions on the PCoA axes was assessed, and strongly correlated taxa were visualized using biplots. Group differences based on region and social status were assessed using PERMANOVA (Bray–Curtis distance, 9,999 permutations). Differences in the presence of specific taxa between groups were evaluated using Welch’s t-test, and the Wilcoxon rank-sum test was additionally employed in selected cases. All statistical analyses were performed in R. Functional profiling of the microbiome Following fastp quality control, adapter trimming, paired-end merging, and human read removal with KneadData, similarity searches were performed using DIAMOND against the NCBI nr database. Functional annotation was carried out using the MEGAN mapping file (megan-map-Feb2022.db), and functions were categorized at SEED level 232,33. As with taxonomic analyses, functional diversity was evaluated using Bray–Curtis distances and PCoA with biplots, and group comparisons were conducted using PERMANOVA and Welch’s t-test in R. Phylogenetic analysis of the oral microbiome Microbial DNA sequences were extracted from the shotgun data for phylogenetic analysis using two approaches: a reference-based method, which extracts species-specific sequences via mapping to reference genomes, and an assembly-based method, which reconstructs metagenome-assembled genomes (MAGs) based on sequence features. Reference-based approach After preprocessing with fastp and KneadData, species-specific marker genes were extracted using MetaPhlAn (ver. 4.0.6). For taxa with marker genes detected in multiple samples, SNP alignments were performed using StrainPhlAn (ver. 4.0.6)34,35. Phylogenetic trees were constructed using the GTR-GAMMA model and the maximum likelihood method in MEGA (ver. 11.0.13)36. To increase sample coverage for specific taxa in which only a limited number of samples passed the default marker detection thresholds, the parameters in StrainPhlAn were relaxed (--marker_in_n_samples 50 --sample_with_n_markers 50). This adjustment was applied selectively to taxa for which phylogenetic reconstruction was not feasible under the default settings. Assembly-based approach Paired-end data were preprocessed with fastp for quality control and deduplication (-D), and then subjected to human read removal using KneadData. The resulting reads were assembled into contigs using MEGAHIT (ver. 1.2.9)37. Binning was performed using MaxBin2 (ver. 2.2.7, -min_contig_length 500)38 and MetaBAT1/2 (ver. 2.15, -m 1500)39,40, based on coverage profiles obtained with Bowtie2 (ver. 2.4.1, --sensitive)41. Resulting bins were integrated into MAGs using DAS Tool (ver. 1.1.6, --search_engine blastp)42, and their quality was assessed using CheckM (ver. 1.2.3)43. Taxonomic assignment of MAGs was performed using PhyloPhlAn (ver. 3.1.68), with the SGB.Jun23 database as the reference for classification44. MAGs with completeness ≥ 50% and contamination ≤ 5%, whose top hit in PhyloPhlAn belonged to the same species, were subjected to SNP extraction and alignment using SKA (ver. 0.3.7, -k 31, --min-freq 0.8) and SNP sites (ver. 2.5.1)45,46. Phylogenetic trees were then constructed using MEGA with the GTR-GAMMA model. To assess coverage and DNA damage, deduplicated and merged reads were mapped to species-level MAGs using BWA (ver. 0.7.18, aln -l 16500 -n 0.01)47, and coverage was calculated using SAMtools (ver. 1.6)48. DNA damage patterns at read termini were evaluated with mapDamage249. Comparative genomic analysis of MAG clades For the obtained MAGs, coding regions were predicted using DFAST (ver. 1.3.1)50 and annotated using eggNOG-mapper (ver. 2.1.12)51. Ortholog detection was performed using SonicParanoid2 (ver. 2.0.7)52 to identify clade-specific genes. MAGs with ≥ 80% completeness and ≤ 5% contamination were used to extract genes present in ≥ 80% of one clade but ≤ 10% in another. To detect clade-specific mutations in shared genes, amino acid and codon-based alignments were performed in MEGA using sequences with consistent annotations from DFAST and eggNOG-mapper. Divergence time estimation After phylogenetic analysis, divergence times were estimated using IQ-TREE (ver. 2.3.6, --date-ci 100)53, with calibration points derived from sample ages: recorded ages at death, Edo period range (1600–1868 CE) for undated Edo period samples, century-specific ranges for other samples, and the year of database registration for reference sequences. Results To facilitate interpretation of the results described below, a conceptual overview of the study design, analytical framework, and principal findings is provided in Fig. S1. Taxonomic composition of the modern dental calculus microbiome To evaluate the diversity of microbial communities in modern dental calculus samples, principal coordinates analysis (PCoA) was performed based on Bray–Curtis distances (Fig. 1), which were calculated from species-level taxonomic profiles of modern Japanese (n = 9) and Spanish (n = 8) individuals11. A small cluster composed exclusively of four Japanese individuals was observed at lower PC1 values (Fig. 1), which may suggest minimal geographic variation in the composition of the modern oral microbiome. However, these findings should be interpreted with caution given the limited sample size. To further examine the taxonomic characteristics of this cluster, the relative abundances of key oral bacterial species were compared between the four samples with lower PC1 values and the remaining samples. Streptococcus sanguinis was significantly more abundant in the higher PC1 group (Welch’s t-test, t(12) = −5.3, p < 0.001), whereas Treponema denticola showed higher abundance in the lower PC1 group (Welch’s t-test, t(7.4) = 2.6, p = 0.035). Notably, additional analyses revealed that all three members of the “Red Complex” which are associated with periodontal disease54, exhibited significant differences between the two groups (Wilcoxon rank-sum test, Porphyromonas gingivalis: p = 0.019, Tannerella forsythia: p = 0.045, T. denticola: p = 0.020). Fig. 1PCoA based on species-level taxonomic composition of dental calculus samples from modern individuals (n = 17; Japan: n = 9, Spain: n = 8). PCoA based on species-level taxonomic composition of dental calculus samples from modern individuals (n = 17; Japan: n = 9, Spain: n = 8). Impact of contamination on ancient dental calculus samples Contamination in newly sequenced ancient dental calculus samples was assessed using decOM26. Several samples contained a substantial proportion of non-oral microbial reads (Fig. S4a). In a PCoA plot based on species-level taxonomic profiles, some ancient samples clustered near soil samples, suggesting possible environmental contamination (Fig. S4b). To minimize such effects, the data were filtered to include only known oral taxa based on the Human Oral Microbiome Database (HOMD)31, and diversity was reanalyzed using this subset. The present method was validated using modern samples (n = 17), in which 257 oral species accounted for 74 ± 6.4% of total reads. PCoA results before and after filtering were highly correlated (Fig. S5, PC1: r = 0.994, PC2: r = 0.997), indicating minimal impact on original microbial diversity. In contrast, when the same filtering was applied to soil samples (n = 12), only 1.1 ± 1.5% of reads (8 species) remained, confirming the effectiveness of the method in removing non-oral contaminants. Comparison between ancient and modern japanese microbiomes A total of 285 oral species were identified in modern (n = 9) and ancient (n = 118) Japanese samples. Among samples with ≥ 1,000 assigned reads (modern: n = 9, ancient: n = 99), PCoA revealed a clear separation between modern and ancient samples along PC1 (Fig. S6, PERMANOVA, p = 0.0092). However, PC1 was correlated with the number of detected species (r = 0.63), suggesting that the observed differences might partly reflect variation in DNA recovery. To account for this, we reanalyzed the data using only ancient samples with ≥ 81 detected species (n = 26), corresponding to the minimum observed in modern samples. Even under this condition (r = 0.14 for PC1 vs. species number), ancient–modern differences remained significant (Fig. 2, PERMANOVA, p = 0.021). In both analyses, M. oralis was more abundant in ancient samples, as confirmed by biplot analysis and Welch’s t-test (Fig. S6: t(98) = 8.5, p < 0.001, Fig. 2: t(25) = 4.4, p < 0.001). Fig. 2PCoA based on the composition of oral microbial species in ancient (n = 26) and modern Japanese (n = 9) dental calculus samples, with ≥ 1,000 reads and ≥ 81 detected species. PCoA based on the composition of oral microbial species in ancient (n = 26) and modern Japanese (n = 9) dental calculus samples, with ≥ 1,000 reads and ≥ 81 detected species. Regional and social differences in ancient japanese microbiomes From 128 ancient samples, 259 known oral species were identified, accounting for 45 ± 25% of total reads. Among samples with ≥ 1,000 reads (n = 109), PCoA revealed regional differences along PC3 between samples from Mainland Japan (including Honshu and Kyushu) and those from Okinawa, a subtropical island chain in the south (Fig. S7a, PERMANOVA, p = 0.021). When the analysis was restricted to samples with ≥ 50 detected species (n = 69), slight regional differences appeared along PC2 (Fig. 3a, PERMANOVA, p = 0.012). Among the top three taxa associated with PC2, significant differences in prevalence between Mainland Japan and Okinawa were observed for M. oralis (p = 0.020), Actinomyces sp. oral taxon 414 (p = 0.0044), and Actinomyces dentalis (p = 0.0052). Fig. 3(a) PCoA based on the composition of oral microbial species in ancient Japanese dental calculus samples (n = 69; Mainland Japan: n = 52, Okinawa: n = 17) with ≥ 1,000 reads and ≥ 50 detected species. (b) PCoA based on the composition of oral microbial species in ancient Japanese dental calculus samples from Mainland Japan (n = 52; Daimyo/Hatamoto: n = 20, others: n = 32) under the same conditions. (a) PCoA based on the composition of oral microbial species in ancient Japanese dental calculus samples (n = 69; Mainland Japan: n = 52, Okinawa: n = 17) with ≥ 1,000 reads and ≥ 50 detected species. (b) PCoA based on the composition of oral microbial species in ancient Japanese dental calculus samples from Mainland Japan (n = 52; Daimyo/Hatamoto: n = 20, others: n = 32) under the same conditions. To examine the impact of social status on the oral microbiome during the Edo period, we compared samples from individuals of higher social rank (Daimyo and Hatamoto) with those from other social classes within Mainland Japan. Although no clear clustering was observed by PCoA among samples with ≥ 1,000 reads (n = 80), PERMANOVA revealed a significant difference (Fig. S7b, p = 0.039). In contrast, when limited to samples with ≥ 50 detected species (n = 52), neither clustering nor statistical significance was observed (Fig. 3b, PERMANOVA, p = 0.089). Lastly, regional differences were analyzed between two sites within Honshu: urban Tokyo (Hitotsubashi High School) and rural Yamanashi (Shiokawa). Among samples with ≥ 1,000 reads (n = 28), separation along PC3 was observed (Fig. S8a, PERMANOVA, p = 0.027). However, when restricted to samples with ≥ 50 detected species, the difference was no longer statistically significant (n = 11) (Fig. S8b, PERMANOVA, p = 0.18). Functional profiling of the microbiome Functional profiles at SEED level 2 were predicted from modern (n = 9) and ancient (n = 118) Japanese dental calculus samples (Fig. 4). PCoA indicated that, with the exception of a single ancient sample, modern and ancient microbiomes formed two separate clusters along PC2, suggesting potential temporal differences in predicted functional profiles, although this was not statistically supported by PERMANOVA (p = 0.072). Fig. 4PCoA based on predicted functional profiles (relative abundances of each function) classified at SEED level 2 in ancient (n = 118) and modern Japanese (n = 9) dental calculus samples. PCoA based on predicted functional profiles (relative abundances of each function) classified at SEED level 2 in ancient (n = 118) and modern Japanese (n = 9) dental calculus samples. Phylogenetic separation by period among microbial species To elucidate the phylogenetic relationships of Anaerolineaceae sp. oral taxon 439 between the Jomon and Edo periods, we conducted a reference-based phylogenetic analysis of newly sequenced ancient Japanese dental calculus samples, using species-specific marker genes identified by MetaPhlAn and StrainPhlAn. The analysis supported the presence of a Jomon-specific clade, consistent with previous findings22(Fig. 5a). In addition, we identified phylogenetic divergence between the Edo and Jomon periods in several species not examined in the prior study, including Streptococcus sinensis, Eubacterium minutum and Candidatus Nanosynsacchari sp. TM7_ANC_38.39_G1_1 (Fig. 5b, Fig. S9). Fig. 5(a) Phylogenetic tree of Anaerolineaceae sp. oral taxon 439, constructed using a reference-based approach, with the W11661 strain from NCBI used as the reference. (b) Phylogenetic tree of Streptococcus sinensis, constructed using a reference-based approach, with the HKU4 strain from NCBI used as the reference. Node color indicates the time period of the sample; node shape indicates the site of recovery; italicized numbers denote bootstrap support values based on 100 replicates. SNPs were extracted from marker genes present in ≥ 80% of samples, including only those samples in which ≥ 80% of marker genes were detected. Jomon samples were designated as the outgroup for tree construction. (a) Phylogenetic tree of Anaerolineaceae sp. oral taxon 439, constructed using a reference-based approach, with the W11661 strain from NCBI used as the reference. (b) Phylogenetic tree of Streptococcus sinensis, constructed using a reference-based approach, with the HKU4 strain from NCBI used as the reference. Node color indicates the time period of the sample; node shape indicates the site of recovery; italicized numbers denote bootstrap support values based on 100 replicates. SNPs were extracted from marker genes present in ≥ 80% of samples, including only those samples in which ≥ 80% of marker genes were detected. Jomon samples were designated as the outgroup for tree construction. Phylogenetic diversity of M. oralis and its association with host sex Among all detected species, M. oralis yielded the most comprehensive phylogeny based on the reference-based approach, encompassing a total of 49 samples. Phylogenetic analysis revealed two major clades of M. oralis in ancient Japanese samples. A statistically significant association between clade affiliation and host sex was observed (Fig. S10; Fisher’s exact test, p = 0.042). In a father–daughter pair, M. oralis from the father clustered within the male-associated clade, whereas that from the daughter belonged to the female-associated clade. Furthermore, all previously reported female samples exhibiting traces of tooth blackening (ohaguro)22 were assigned to the same clade. Using the same species, we also performed a phylogenetic analysis based on an assembly-based approach, as M. oralis yielded the largest number of MAGs that satisfied quality thresholds (completeness ≥ 50%, contamination ≤ 5%, n = 34). The analysis again revealed a bifurcation into two clades, which appeared to be associated with the sex of the host, consistent with the findings from the reference-based analysis, although the statistical significance was marginal (Fig. 6a, Fisher’s exact test, p = 0.057). To rule out the possibility that the observed clade structure resulted from contamination with modern human saliva, we assessed terminal DNA damage patterns. Ancient DNA damage signatures were detected in MAGs from both clades, supporting the authenticity of the sequences (Fig. 6b). The two clades were designated as clade A (male-associated) and clade B (female-associated), and were subjected to further comparative genomic analysis to identify clade-specific genes and mutations. Fig. 6(a) Phylogenetic tree of Methanobrevibacter oralis constructed using an assembly-based approach. Node color indicates the sex of the host individual; node shape indicates the site of recovery; italicized numbers denote bootstrap support values based on 100 replicates. The tree was midpoint-rooted and constructed from SNPs detected in ≥ 80% of samples, based on MAGs with ≥ 50% completeness and ≤ 5% contamination. Reference genomes (DSM 7256, JMR01, M2 CSUR P5920, MGYG-HGUT-02162, and YH) were retrieved from NCBI. Asterisks (*) indicate a father–daughter pair; daggers (†) indicate samples reported to exhibit signs of traditional tooth blackening (ohaguro) in a previous study (Eisenhofer et al., 2020). (b) DNA damage plot at the 3’ ends of sequence reads. Red and blue lines indicate C-to-T and G-to-A substitutions, respectively. Results are shown for representative samples from each clade that exhibited the most pronounced ancient DNA damage signatures. (a) Phylogenetic tree of Methanobrevibacter oralis constructed using an assembly-based approach. Node color indicates the sex of the host individual; node shape indicates the site of recovery; italicized numbers denote bootstrap support values based on 100 replicates. The tree was midpoint-rooted and constructed from SNPs detected in ≥ 80% of samples, based on MAGs with ≥ 50% completeness and ≤ 5% contamination. Reference genomes (DSM 7256, JMR01, M2 CSUR P5920, MGYG-HGUT-02162, and YH) were retrieved from NCBI. Asterisks (*) indicate a father–daughter pair; daggers (†) indicate samples reported to exhibit signs of traditional tooth blackening (ohaguro) in a previous study (Eisenhofer et al., 2020). (b) DNA damage plot at the 3’ ends of sequence reads. Red and blue lines indicate C-to-T and G-to-A substitutions, respectively. Results are shown for representative samples from each clade that exhibited the most pronounced ancient DNA damage signatures. Comparative genomic analysis between M. oralis clades For comparative genomic analysis, we selected only high-quality MAGs that met stringent thresholods (Table S1, completeness ≥ 80%, contamination ≤ 5%), comprising 15 from clade A and 16 from clade B. Clade-specific genes were defined as those present in ≥ 80% of MAGs within one clade and ≤ 10% in the other. Using this criterion, we identified 18 clade-specific genes in clade A and 7 in clade B. Among these, 7 genes from clade A and 2 from clade B were successfully annotated with predicted functions using eggNOG-mapper (Table 3). Notably, one of the clade A-specific genes was annotated as containing a SpoVT/AbrB-like domain. Table 3Functionally annotated genes specific to each clade.Clade AClade BAminotransferase class I and IIATPase domain predominantly from ArchaeaDomain of unknown function (DUF1848)Sugar (and other) transporterEpimerase dehydratase-Predicted membrane protein (DUF2085)-Serine dehydrogenase proteinase-SpoVT/AbrB like domain-zinc-ribbon domain-*Only genes for which functional annotations were available (via eggNOG-mapper) are shown. Table 3Functionally annotated genes specific to each clade.Clade AClade BAminotransferase class I and IIATPase domain predominantly from ArchaeaDomain of unknown function (DUF1848)Sugar (and other) transporterEpimerase dehydratase-Predicted membrane protein (DUF2085)-Serine dehydrogenase proteinase-SpoVT/AbrB like domain-zinc-ribbon domain-*Only genes for which functional annotations were available (via eggNOG-mapper) are shown. Functionally annotated genes specific to each clade. *Only genes for which functional annotations were available (via eggNOG-mapper) are shown. The traditional Japanese practice of tooth blackening (ohaguro), which was particularly common among adult women during the Edo period55, may have contributed to the observed sex-associated clade divergence of M. oralis. Ohaguro involved the application of a mixture primarily composed of ferrous acetate (derived from ferrous acetate solution) and tannins (from gallnut powder), suggesting a potential link to iron metabolism. To explore this possibility, we investigated clade-specific amino acid substitutions in genes implicated in iron-associated enzymatic functions. Within the methyl viologen hydrogenase (Mvh) gene family, which encodes a class of ferredoxins involved in methanogenesis via redox reactions of hydrogen molecules56, we identified two clade-specific substitutions in mvhA and three in mvhB. Comparative analysis with reference sequences from related taxa showed that all clade A variants were identical to the M. oralis reference, whereas two substitutions in mvhB (positions 173 and 372) were unique to clade B (Table 4). Table 4Comparison of clade-specific substitutions in mvhA and mvhB with reference sequences from related Methanobrevibacter species.siteClade A/BM.oralisM.smithiiM. arboriphilusM. ruminantiumM. woliniimvhA 306Glu/Ala(GAA/GCA)Glu(GAA)Ala(GCA)Ala(GCT)Ala(GCA)Glu(GAA)mvhA 423Asp/Gly(GAT/GGT)Asp(GAT)Gly(GGT)Gly(GGT)Gly(GGT)Gly(GGT)mvhB 173Asp/Gly(GAC/GGC)Asp(GAC)Asp(GAT)Pro(CCT)-Glu(GAA)Glu(GAA)mvhB 236Asp/Asn(GAC/AAC)Asp(GAC)Asn(AAC/AAT)-Asp(GAC)Ser(TCT)mvhB 372Ile/Met(ATT/ATG)Ile(ATT)Ile(ATC)Val(GTA)-Ile(ATT)Ile(ATT)*Alignments were performed at the amino acid and codon levels. In the first column, the sequences shown on the left represent those from Clade A, and those on the right represent sequences from Clade B. Positions highlighted in bold (mvhB 173 and mvhB 372) indicate sites where Clade B–specific substitutions were detected. For mvhA position 423, one sample in Clade B showed a Val (GTT) substitution. Reference sequences were obtained from the following strains: M. oralis (strains YH and JMR01), M. smithii (TS95C and TS146C), M. arboriphilus (JCM 9315), M. ruminantium (M1), and M. wolinii (SH). Table 4Comparison of clade-specific substitutions in mvhA and mvhB with reference sequences from related Methanobrevibacter species.siteClade A/BM.oralisM.smithiiM. arboriphilusM. ruminantiumM. woliniimvhA 306Glu/Ala(GAA/GCA)Glu(GAA)Ala(GCA)Ala(GCT)Ala(GCA)Glu(GAA)mvhA 423Asp/Gly(GAT/GGT)Asp(GAT)Gly(GGT)Gly(GGT)Gly(GGT)Gly(GGT)mvhB 173Asp/Gly(GAC/GGC)Asp(GAC)Asp(GAT)Pro(CCT)-Glu(GAA)Glu(GAA)mvhB 236Asp/Asn(GAC/AAC)Asp(GAC)Asn(AAC/AAT)-Asp(GAC)Ser(TCT)mvhB 372Ile/Met(ATT/ATG)Ile(ATT)Ile(ATC)Val(GTA)-Ile(ATT)Ile(ATT)*Alignments were performed at the amino acid and codon levels. In the first column, the sequences shown on the left represent those from Clade A, and those on the right represent sequences from Clade B. Positions highlighted in bold (mvhB 173 and mvhB 372) indicate sites where Clade B–specific substitutions were detected. For mvhA position 423, one sample in Clade B showed a Val (GTT) substitution. Reference sequences were obtained from the following strains: M. oralis (strains YH and JMR01), M. smithii (TS95C and TS146C), M. arboriphilus (JCM 9315), M. ruminantium (M1), and M. wolinii (SH). Comparison of clade-specific substitutions in mvhA and mvhB with reference sequences from related Methanobrevibacter species. mvhA 306 Glu/Ala (GAA/GCA) mvhA 423 Asp/Gly (GAT/GGT) mvhB 173 Asp/Gly (GAC/GGC) Asp(GAT) Pro(CCT) mvhB 236 Asp/Asn (GAC/AAC) Asn (AAC/AAT) mvhB 372 Ile/Met (ATT/ATG) Ile(ATC) Val(GTA) *Alignments were performed at the amino acid and codon levels. In the first column, the sequences shown on the left represent those from Clade A, and those on the right represent sequences from Clade B. Positions highlighted in bold (mvhB 173 and mvhB 372) indicate sites where Clade B–specific substitutions were detected. For mvhA position 423, one sample in Clade B showed a Val (GTT) substitution. Reference sequences were obtained from the following strains: M. oralis (strains YH and JMR01), M. smithii (TS95C and TS146C), M. arboriphilus (JCM 9315), M. ruminantium (M1), and M. wolinii (SH). We estimated the divergence time between the two M. oralis clades using IQ-TREE, incorporating sample ages as calibration points. The reference-based analysis dated the divergence to 366 CE (95% confidence interval: 209–1,096 CE), and the assembly-based analysis yielded an estimate of 289 BCE (CI: 831 BCE–469 CE). These results suggest that the clade divergence likely occurred within several centuries surrounding the beginning of the Common Era. Discussion In our analysis of modern dental calculus, a cluster consisting of four Japanese samples was observed at lower PC1 values (Fig. 1). Although this pattern might initially appear to reflect geographic variation, further examination of the taxonomic composition suggests an alternative explanation. The cluster was characterized by higher abundances of members of the “Red Complex,” including T. denticola, P. gingivalis, and T. forsythia, which are well known to be associated with periodontal disease54. In contrast, S. sanguinis, a species commonly associated with oral health57, was more abundant in samples outside this cluster. Clinical assessments revealed that the mean probing pocket depth (PPD) of the modern Japanese individuals was 5.2 ± 2.2 mm, indicating that all participants suffered from moderate to severe periodontitis. This clinical status is closely associated with the taxonomic clustering observed in Fig. 1. Specifically, the high abundance of “Red Complex” taxa, such as T. denticola, in these samples aligns with the high PPD values. In the absence of direct clinical assessment for the ancient individuals, dental calculus was collected from specimens in which the relationship between calculus deposits and the underlying alveolar bone could be evaluated. Visual inspection revealed varying degrees of alveolar bone resorption consistent with chronic periodontal attachment loss. Importantly, these observations are consistent with the distribution of ancient samples in Fig. 2, where some of them are positioned between the two clusters observed in modern individuals along PC2, and therefore do not contradict the interpretation that these clusters reflect differences in oral health status. Taken together, these observations suggest that the clustering pattern may reflect differences in oral hygiene or periodontal status rather than geographic variation. More broadly, these findings indicate that, particularly in modern populations from industrialized countries, widespread lifestyle westernization may have contributed to reduced inter-individual diversity within the oral microbiome. In contrast, results from ancient dental calculus studies have reported greater microbial diversity across different regions of Europe and Asia18. Consistent with a previous 16 S rRNA-based study20, our shotgun metagenomic analysis revealed clear compositional differences between modern and ancient Japanese oral microbiomes (Fig. 2). Notably, M. oralis was frequently detected in ancient samples, consistent with findings from the UK58, and its relative abundance varied substantially across individuals [Fig. S11], in agreement with a previous report59. In contrast, M. oralis was not detected in modern Japanese samples in this study, possibly due to the limited sample size. These observations suggest that Methanobrevibacter species exhibit distinct prevalence patterns in ancient dental calculus. Beyond taxonomic composition, functional differences were also evident between modern and ancient samples (Fig. 4). Previous studies of paleofeces have highlighted genes related to the starch utilization system (SUS) as distinguishing features between ancient and modern populations14. Motivated by these observations, we considered whether similar differences might be detectable in our dental calculus data and found that the relative abundance of SUS genes was significantly different between ancient and modern samples (Welch’s t-test, t(16) = −5.8, p < 0.001). This result suggests that the transition to starch-rich diets in modern populations may have influenced the structure of the oral microbiome. Furthermore, analysis of ancient individuals revealed subtle regional differences in microbial composition between Mainland Japan and Okinawa (Fig. 3a), potentially reflecting lifestyle differences between the two regions. Notably, M. oralis and S. sanguinis were identified in the biplot of Fig. 3a, consistent with previous findings from ancient dental calculus in the Netherlands60, suggesting that these taxa may represent common features of ancient oral microbiomes across different geographic regions. By contrast, no clear differences were observed by social status within Mainland Japan (Fig. 3b) or between urban and rural areas (Fig. S8). Our results also showed that altering thresholds for read count and species richness affected both PCoA and PERMANOVA results, highlighting how environmental contamination, DNA fragmentation, and deamination complicate accurate reconstruction of the ancient oral microbiome. Notably, the use of the HOMD—a reference database constructed primarily from Western populations—may have led to underestimation of microbial diversity in ancient Japanese samples. Future studies incorporating more comprehensive oral microbiome data from East Asian populations may improve the accuracy of ancient oral microbiome reconstruction by enabling the identification of truly ancient and oral-specific taxa, based on DNA damage signatures and their presence in modern Japanese datasets. The Japanese archipelago was originally inhabited by hunter-gatherer Jomon people, who were later admixed with agricultural migrants from the Asian continent around the 10th century BCE, marking the beginning of the Yayoi period and introducing wet-rice farming to Japan61. This admixture and subsequent cultural and demographic continuity ultimately gave rise to the modern Japanese population through later historical periods, including the Edo period. With respect to temporal variation, our analysis confirmed the previously reported phylogenetic divergence of Anaerolineaceae sp. oral taxon 439 between the Edo and Jomon periods22, using newly sequenced samples from multiple archaeological sites (Fig. 5a). In addition, several other species—including S. sinensis and E. minutum—formed Jomon-specific clades (Fig. 5b, Fig. S9), suggesting phylogenetic turnover associated with the migration, admixture, and cultural interactions that began in the Yayoi period and continued thereafter, leading to the transition from hunting and gathering to agriculture and the resulting transformation of dietary habits. Further analysis of dental calculus from other periods (e.g., the Yayoi period) may provide additional insights into the relationship between bacterial evolution and human demographic history in ancient Japan. In particular, the successful phylogenetic analysis of S. sinensis, a species reported to be abundant in ancient dental calculus but rare in modern industrialized populations59, suggests that ancient dental calculus may preserve evolutionary diversity of oral bacterial taxa that are now diminishing in modern human populations. We also focused on M. oralis, a periodontal-associated species62 frequently detected in ancient dental calculus, and performed phylogenetic analyses using both reference-based and assembly-based approaches. Both methods consistently revealed two major clades (Fig. 6a, Fig. S10), one of which diverged from the M. oralis reference sequence. This finding suggests that diversification occurred not only at the species level within Methanobrevibacter15 but also within M. oralis itself in ancient Japan [Fig. S12]. Estimated divergence times for the two clades indicate that their split occurred several centuries before or after the beginning of the Common Era, raising the possibility that this diversification emerged domestically. Notably, all 11 samples from Hitotsubashi High School (Tokyo) belonged to a single clade, suggesting potential geographic structuring in clade distribution. We also observed a statistically significant association between clade membership and host sex; however, this pattern should be interpreted with caution given the substantial proportion of individuals with unknown sex. In one father–daughter pair, the father’s M. oralis strain belonged to one clade, whereas the daughter’s strain belonged to the other. This observation may reflect sex-associated patterns in clade distribution and could, for example, be consistent with processes such as sex-specific adaptation or transmission dynamics. Comparative genomic analysis of the two clades revealed that clade A, which includes the reference strain, possessed a greater number of clade-specific genes (Table 3). Among these, the presence of the SpoVT/AbrB-like domain, known to regulate the transcription of genes involved in antibiotic resistance and biofilm formation63,64, suggests possible adaptation to antibiotic-influenced oral environments in modern individuals However, due to the limited detection of M. oralis sequences in modern Japanese samples, it remains unclear whether this phylogenetic diversity persists in present-day populations. Broader analyses of the modern Japanese oral microbiome will be necessary to clarify the origin, functional roles, and health implications of M. oralis clades. Previous studies have suggested that M. oralis interacts with diverse oral bacteria through metabolic processes such as hydrogen consumption during methanogenesis, and has also been implicated in extra-oral, systemic conditions21. If M. oralis can be more consistently detected in modern Japanese populations, it may be possible to investigate whether such microbial interactions and potential health associations have changed over time. All samples previously reported to exhibit traces of tooth blackening (ohaguro)22 were assigned to clade B, suggesting a potential link between this Edo period custom—predominantly practiced by adult women—and the observed sex-specific clade distribution. Since ohaguro involved the application of iron-rich substances55, we investigated clade-specific variants in iron-associated metabolic genes. Notably, substitutions unique to clade B were identified in mvhA and mvhB, which encode methyl viologen hydrogenase—a ferredoxin-type enzyme involved in methanogenesis that requires iron at its active site56. Comparison with reference sequences from related taxa indicated that two of the mvhB substitutions were exclusive to clade B (Table 4). One possible hypothesis is that these variants may have conferred an energetic advantage under iron-rich oral conditions associated with ohaguro. Alternatively, if the substitutions were metabolically disadvantageous, the iron-rich environment may have enabled the persistence of otherwise suboptimal strains. As our study did not systematically record ohaguro traces on a per-sample basis, future research incorporating standardized and objective assessment of tooth blackening will be essential to further investigate the relationship between this cultural practice and the oral microbiome. In addition, detailed chemical analyses of residual compounds may help to evaluate whether antimicrobial substances, such as flavonoids65, which may be present in plant-derived materials used in ohaguro, influenced oral hygiene and microbial community composition. Consistent with previous studies20,22, our results confirmed distinct differences in oral microbiome composition between ancient and modern Japanese individuals, as well as phylogenetic divergence between the Jomon and Edo periods. We also identified regional variation in ancient microbiome composition and clade-level diversity within the periodontal-associated species M. oralis. Continued progress in oral microbiome research will likely yield valuable insights into the dynamic relationships between microbial evolution, cultural practices, and human demographic history in Japan. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2