Introduction

The continuous rise in greenhouse gas (GHG) emissions has triggered rapid changes in global climate patterns, resulting in the frequent occurrence of extreme weather events such as droughts, floods, and heatwaves. These changes profoundly impact the global ecosystem and human livelihoods1. Methane (CH4), which ranks second only to carbon dioxide in terms of greenhouse gases on Earth, has a global warming potential approximately 82.5 times greater than that of carbon dioxide over a 20-year period. Hence, it is recognized as a major contributor to the greenhouse effect2,3. Ruminants contribute significantly to methane emissions, making them a focal point of research. CH4 emissions from ruminants constitute 16% of global greenhouse gas emissions and 30%-32% of global anthropogenic CH4 emissions4,5. In addition, CH4 emissions represent a substantial energy loss to animals, ranging from 2 to 12% of gross energy intake6. Consequently, mitigating CH4 emissions from ruminants is crucial in the context of sustainable agricultural practices and global climate change mitigation efforts7. CH4 emitted from ruminants is produced primarily by archaea in the gastrointestinal tract (GIT), especially in the rumen8. Therefore, a systematic analysis of the composition and function of archaea in the GIT of ruminant animals is crucial for providing background information.

Archaea play a key role in syntrophic metabolism by consuming the end products of biomass fermentation from other microbiota, thus maintaining the hydrogen balance of the GIT9. Archaea can utilize compounds such as CO2, CO, ethanol, formate, acetate or methyl compounds to form CH4 through three major pathways: hydrogenotrophic, methylotrophic, and aceticlastic. Each trophic type necessitates distinct functions in groups of archaea. Researchers have emphasized that the hydrogenotrophic pathway is responsible for more than 80% of rumen CH4 production10,11,12. Methanobrevibacter spp., which are hydrogenotrophic methanogens, have emerged as the predominant genus involved in rumen methanogenesis12,13. Additionally, Methanosarcinales, Methanosphaera, and Methanomethylophilaceae are also involved in reducing methylamine and methanol to CH414,15,16. Methanosarcinales is notable for its ability to effectively dissolve acetate into carbon dioxide and methane17. Yaks, as native ruminants, have been identified as ‘low methane’ emitters18,19. The variations in CH4 emissions among different ruminants may be attributed to the distinct composition of methanogens in the rumen19,20,21,22. Although numerous researchers have explored the function and composition of the primary methanogens in the GIT of ruminants2,9,11,23, comprehensive and global analyses of the systematic abundance of different ruminant gut archaea are lacking.

In this work, we establish a catalogue of the ruminant GIT archaeome by collecting and assembling 2,270 ruminant gut metagenomic samples, including previously assembled and cultured archaeal genomes. The catalogue comprises a total of 998 dereplicated archaeal genomes (99.9% Average Nucleotide Identity-ANI similarity), and 42,691 protein clusters and 216/556 strains (99% ANI similarity) that were previously undescribed. Furthermore, our study extends to the gut biogeography of the archaeome across various ruminants, revealing host- and gut-segment-dependent characteristics of the archaeal composition. This catalogue expands the taxonomic and functional variation of the ruminant gut archaeome, and also holds the potential to advance our understanding of archaeal ecology in ruminants.

Results

A total of 998 unique archaeal genomes were recovered from ruminant gastrointestinal samples

To explore the diversity of archaea in ruminant GIT samples, we recovered genomes from 2270 global metagenomic samples and combined them with publicly available genomes from previous collections of metagenome-assembled genomes (MAGs) and isolates. The 998 obtained nonchimeric and nonredundant genomes spanned a wide range of taxonomic diversity, including those of Methanobacteriaceae (n = 669; 67.03%), Methanomassiliicoccales (n = 198; 19.84%), Methanocorpusculaceae (n = 90; 9.02%), Methanomicrobiaceae (n = 32; 3.21%), and Methanosarcinaceae (n = 9; 0.90%) (Supplementary Data 1; Fig. 1). Most of the genomes were taxonomically affiliated with the known genus Methanobrevibacter (n = 616; 61.72%), which was consistent with previous studies12,24. However, this proportion was lower than that reported in archaeal genomes recovered from human GIT samples (998/1167; 85.52%)25. Other identified genomes belonged to genera such as UBA71 of the family Methanomethylophilaceae (n = 105; 10.52%), Methanocorpusculum (n = 90; 9.02%), Methanomethylophilus (n = 54; 5.41%), Methanosphaera (n = 51; 5.11%), and Methanomicrobium (n = 32; 3.21%), whereas SIG5 (n = 23; 2.30%) and MX-02 (n = 8; 0.80%) were from Methanomethylophilaceae, Methanimicrococcus (n = 7; 0.70%), and Methanosarcina (n = 2; 0.20%). CADBMS01 of the families Methanobacteriaceae and Methanobacterium were each represented by a single genome. Among the 998 genomes, 8 belonged to Methanomethylophilaceae (0.80%) and could not be assigned to any previously described genus, whereas 140 genomes (14.03%) did not match any known species (Supplementary Data 1). An equally large proportion of the genomes not matching any known species (n = 70; 7.01%) were affiliated with the families Methanobacteriaceae and Methanomethylophilaceae. A phylogenetic tree was constructed for the 279 archaeal genomes at the strain level with a completeness of >80% and contamination of <5% (Fig. 1). Among these genomes, 47 were newly recovered from this study, with 148 genomes derived from the rumen and 105 genomes recovered from mixed assemblies with GIT metagenomic samples (Fig. 1). A phylogenetic tree was also constructed for the 556 archaeal genomes at the strain level with a completeness of >50% and contamination of <5%, of which 216 were previously undescribed (Supplementary Fig. 1).

Fig. 1: Archaeal genomes (279) at the strain level from the ruminant gastrointestinal tract reveal taxonomic expansion of the archaeome.
figure 1

The phylogenetic tree depicts archaeal genomes with a completeness of >80% and contamination of <5%, clustered at 99% similarity (strain level). The characteristics displayed from the center to the outside include the branch of the phylogenetic tree with ultrafast bootstrap values. Newly recovered genomes and isolates are indicated by red solid stars (New) and red stars (Isolate), respectively. The taxonomic affiliations of the MAGs are shown at the order and family levels. The genome size of each MAG is represented by brown bars in megabases (Mb). The different colors of the branches correspond to the colors of the classified families.

Full size image

The archaeal protein profile correlated with different taxonomies

In total, 1,604,531 proteins were identified from the 998 genomes. A protein catalogue of all 998 archaeal genomes was generated by clustering the predicted genes across all the genomes and excluding singleton clustering, resulting in 42,691 cluster representatives ( > 50% amino acid identity and >80% coverage). A total of 5,456 proteins were shared among >50 genomes in our dataset, revealing the taxonomic distance of the three most abundant families: Methanobacteriaceae, Methanomethylophilaceae, and Methanocorpusculaceae (Fig. 2a). The NMDS plot also demonstrated significant separation among five archaeal families: Methanobacteriaceae, Methanomethylophilaceae, Methanocorpusculaceae, Methanomicrobiaceae, and Methanosarcinaceae (Fig. 2b). To investigate differences at the genus level in the main family, Methanobacteriaceae, in the ruminant gut, we created NMDS plots (Fig. 2c). Interestingly, the plots revealed that, according to the GTDB classification, Methanobrevibacter, Methanobrevibacter_A, and Methanobrevibacter_B exhibited significant separation. The results were consistent with the classification rules of the GTDB, where different tails (_A/_B) signify different genera, although they have not yet been named (Fig. 2c). Another significant separation was observed for the genus Methanosphaera (Fig. 2c).

Fig. 2: Archaeal genomes from the ruminant gastrointestinal tract distribution and the corresponding unified protein catalogue.
figure 2

a Unified ruminant gut archaeal protein catalogue based on protein clustering at 50% sequence identity and 80% coverage via MMseqs2 of all 998 archaeal genomes. The heatmap depicts the presence of 5456 proteins (columns) across the 998 archaeal genomes (rows). Heatmap visualization was performed via the pheatmap package in R. White indicates that no data were available. b Nonmetric multidimensional scaling (NMDS) plot of the proteins selected for the heatmap. The NMDS plot shows five distinct clusters corresponding to the archaeal families Methanobacteriaceae, Methanomethylophilaceae, Methanocorpusculaceae, Methanomicrobiaceae, and Methanosarcinaceae, with the central dots representing the mean NMDS scores for each genome. c The NMDS plot shows six distinct clusters corresponding to the archaeal genera Methanobrevibacter, Methanobrevibacter_A, Methanobrevibacter_B, Methanosphaera, Methanobacterium, and CADBMS01 of the Methanobacteriaceae family with the central dot representing the mean NMDS scores for each genome. The stress value was calculated via metaMDS functions from the vegan package with Bray-Curtis distances (b, c). The P value was calculated by permutational multivariate analysis of variance (PERMANOVA) analysis (b, c). Source data are provided as a Source Data file.

Full size image

Composition of archaeal genomes shaped by breed and gut biogeography

Read-based community profiling revealed variations across breeds and gut biogeography (Supplementary Data 2, Fig. 3, and Fig. 4). For the three major families of archaeal genomes, Methanobacteriaceae and Methanomethylophilaceae, relative abundance and significance analyses were conducted between breeds in the gastrointestinal and rumen microbiomes (Fig. 3a–d). The relative abundance of Methanobacteriaceae was significantly greater in the gastrointestinal and rumen microbiomes of beef cattle and cows than in those of yak and buffalo (P < 0.0001; Supplementary Data 3, Fig. 3a, b). However, there was no difference between the buffaloes and yaks (Fig. 3a, b). In contrast, the relative abundance of Methanomethylophilaceae was greater in the entire gastrointestinal and rumen microbiomes of yaks than in those of beef cattle, cows, and buffaloes (P < 0.0001; Supplementary Data 3, Fig. 3c, d). A detailed investigation of the relative abundance and composition of families in the rumen was conducted (Supplementary Data 2 and Fig. 3e). The overall relative abundances of the archaeal genomes in the rumens of beef, goat, bison, cow, sheep, zebu, deer, buffalo, yak, and camel were 1.76%, 0.97%, 0.95%, 0.67%, 0.72%, 0.47%, 0.30%, 0.40%, 0.26%, and 0.14%, respectively (Supplementary Data 2). The family Methanocorpusculaceae, which includes a special host-associated genus, Methanocorpusculum, which is related to ‘low methane’ emitters, was only read-assigned to the rumen microbiome of sheep, deer, and buffalo and had a very low relative abundance (Fig. 3e)21. However, a high relative abundance of the family Methanocorpusculaceae was detected in the hindgut, including the cecum, colon, rectum, and feces of buffaloes, cows, and yaks (Fig. 4c). For the families Methanomethylophilaceae and Methanomicrobiaceae, the relative abundances of the total archaeome were 83.30% and 85.75%, respectively, in the rumens of camels and yaks (Fig. 3e). In summary, the composition of archaeal genomes in the rumen was influenced by breed (Figs. 3e and 4a)26.

Fig. 3: Percentage and relative abundance of archaeal genomes at the family level across different ruminant species.
figure 3

a Percentage of Methanobacteriaceae in the intestine. b Percentage of Methanobacteriaceae in the rumen. c Percentage of Methanomethylophilaceae in the intestine. d Percentage of Methanomethylophilaceae in the rumen. e Heatmap showing the composition of archaeome at the family level in the rumen. From red to purple indicates values from high to low. The number of samples for breeds in the intestine (a, c) is cattle: n = 638, goat: n = 64, bison: n = 8, cow: n = 366, sheep: n = 92, zebu: n = 24, yak: n = 79, deer: n = 105, buffalo: n = 682, and camel: n = 50. The number of samples for breeds in the rumen (b, d) is cattle: n = 606, goat: n = 6, bison: n = 8, cow: n = 111, sheep: n = 47, zebu: n = 24, yak: n = 44, deer: n = 26, buffalo: n = 125, and camel: n = 44. The middle horizontal line represents the mean value (a–d). The statistical significance of the abundance differences was calculated by the two-sided Kruskal-Wallis test corrected for multiple comparisons via Dunn’s test (a–d). Different letters indicate statistically significant differences (P < 0.05), with the specific pairwise comparison P-values as Supplementary Data 3 demonstrates (a–d). Source data are provided as a Source Data file.

Full size image
Fig. 4: Percentages of archaea in the rumen across ruminant species and relative abundances determined through gut biogeography.
figure 4

a Nonmetric multidimensional scaling (NMDS) plot depicting the percentage of archaea in the rumen microbiome across ruminant species, with the central dot representing the mean NMDS score for each sample. b NMDS plot illustrating the percentage of archaea in ruminant gut microbiomes across different gut biogeographies of buffalo, cows, and yaks with the central dot representing the mean NMDS scores for each sample. The stress value was calculated via metaMDS functions from the vegan package with Bray-Curtis distances (a, b). The P value was calculated via permutational multivariate analysis of variance (PERMANOVA) analysis (a, b). The black circle, triangle, and square symbols represent buffalo, dairy cow, and yak, respectively. c The relative abundance of archaea in ruminant gut microbiomes across gut biogeographies of buffalo, cows, and yaks. The black circles, triangles, and squares represent buffalo, dairy cows, and yaks, respectively. Rum Rumen, Ret Reticulum, Oma Omasum, Abo Abomasum, Duo Duodenum, Jej Jejunum, Ile Ileum, Cec Cecum, Col Colon, Rec Rectum, and Fec Feces. The schematic diagram of the ruminant gut was hand-drawn and authorized by Jingbo Xia, a friend of J.D. M. (c). Source data are provided as a Source Data file.

Full size image

To support previous reports24,27,28, highlighting that gut biogeography is a crucial factor shaping the diverse composition and functions within animal gut microbiomes, we investigated whether the archaeal genomes of ruminants are determined by gut biogeography. Buffalo, cow, and yak were selected for nonmetric multidimensional scaling (NMDS) analysis, considering the adequate number of representative samples from different gut locations (Fig. 4b). The results demonstrated that the composition of archaeal genomes in buffaloes, cows, and yaks could be differentiated on the basis of gut location (Fig. 4b). Notably, for buffalo and yak, the archaeal genomes could be separated into three distinct gut compartment groups (stomach: rumen, reticulum, omasum, and abomasum; small intestine: duodenum, jejunum, and ileum; and large intestine: cecum, colon, rectum, and feces) at the taxonomic level (Fig. 4b). The analysis also suggested substantial changes in archaeal taxa across the 10 ruminant GITs and feces (Fig. 4c). For example, Methanobacteriaceae and Methanomethylophilaceae dominated in the stomach region, whereas Methanobacteriaceae were more abundant in the small intestine than in the stomach and large intestine, which agreed with the findings of a previous study29. Methanobacteriaceae and Methanocorpusculaceae were more prevalent in the large intestine and feces. These results underscore that gut biogeography exerts great selective pressures on the ruminant archaeal genome communities.

Comparison of the Methanomethylophilaceae clade in different environments

In yaks, we observed a greater relative abundance of Methanomethylophilaceae in the rumen, which is the main intestinal site for biomass degradation through interactions between bacteria, archaea, fungi, protozoa, and viruses, than in beef cattle and cows. Therefore, we suggest that host-associated Methanomethylophilaceae exhibit taxonomic and functional distinctions from their environmental counterparts due to the unique characteristics of their host environments. According to the phylogenetic tree analysis, Methanomethylophilaceae in the ruminant gastrointestinal microbiome separated into three distinct clades, most of which differed from those of strains from host-associated sources (human, termite, and chicken) and environmental sources (wastewater, soil, and sediment) (Supplementary Fig. 2a). Only two human-associated Methanomethylophilaceae genomes clustered with ruminant-associated Methanomethylophilaceae genomes in the two clades of the phylogenetic tree (Supplementary Fig. 2a). ANI-based analyses of the family Methanomethylophilaceae revealed an overall separation between the genomes of different origins (Supplementary Fig. 2b). On basis of their representative microbiota, the Methanomethylophilaceae strains in this study could be classified into (1) the exclusively found group of the ruminant gut; (2) host-associated (ruminant, human, and chicken) strains exclusive to the termite gut; and (3) strains with various origins (including ruminant, human, chicken, termite, wastewater, sediment, and soil), highlighting the diversity of Methanomethylophilaceae in host-associated and environmental sources30,31,32.

Functional and metabolic interactions of the archaeome with different ruminants

We analyzed features that could reveal the advanced functional and metabolic distribution of the ruminant-associated gut archaeome involved in methane metabolism (Fig. 5). In agreement with a previous report25, in addition to the major components of methanogenesis, such as methyl-coenzyme M reductase (Mcr) and heterodisulfide reductase/[NiFe] hydrogenase (Hdr/Mvh) complexes, the major ruminant gut methanogens (Methanobacteriaceae and Methanomethylophilaceae) possess very distinct methanogenesis pathways (Fig. 5 and Supplementary Data 4). Thus, Mcr is commonly used as a target to reduce methane production by inhibiting its activity, as demonstrated by inhibitors such as 3-NOP33,34. For example, most of the genetic potential for the H2/CO2 pathway comes from Methanobacteriaceae. In the methylotrophic pathway of methanogenesis, most of the genes belong to Methanomethylophilaceae, with additional genes originating from Methanosarcinaceae. The relative abundance of Methanomethylophilaceae varies in ruminant breeds, suggesting that the capacity for methyl compound production by the ruminant gut microbiota might be influenced by different populations and diet selections32,35,36. However, approximately 10% of the MtaABC genes (474 counts), indicating the genetic potential of methanol, were associated with Methanobacteriaceae. The presence of MtaABC genes in some Methanobacteriaceae species strongly indicates that methanol utilization might confer an adaptive advantage in the ruminant guts with high methanol concentrations25. However, the conditions under which Methanobacteriaceae species use methanol need to be investigated to determine whether they differ between ruminant breeds and/or whether methanol serves as a methanogenic substrate within a broader anabolic pathway for Methanobacteriaceae.

Fig. 5: Methanogenic pathways in the ruminant gut-associated archaeome at the species level.
figure 5

a The proportion of species with a predicted protein or protein complex is indicated by the bar for total archaeal genomes at the species level (n = 203). The number on the right of the bar represents the total number of genes predicted for all archaeal genomes. The blue, light green, and dark green squares on the lines represent genes involved in three different methanogenic pathways: hydrogenotrophic, methylotrophic, and aceticlastic, respectively. b Differential enrichment of methanogenesis-associated genes in the rumen across ruminant species. Source data are provided as a Source Data file.

Full size image

A small proportion of genes involved in the methylotrophic pathway of methanogenesis come from Methanosarcinaceae, including MtaABC, MtsAB, MtmBC, MtbABC, and MttBC, which are responsible for the utilization of methanol, methanethiol, methylamin, dimethylamine, and trimethylamine, respectively. This capacity enables Methanosarcinaceae to reduce the concentrations of methyl compounds produced by the ruminant gut microbiota, contributing to methane emissions37,38. Interestingly, genes involved in the acetoclastic methanogenesis pathway, such as AckA, Pta, and CdhA-E, have also been detected in Methanosarcinaceae (Fig. 5a), indicating that in some cases, these Methanosarcinaceae might be essential for energy conservation during acetoclastic methanogenesis39. The main methylotrophic methanogen Methanomethylophilaceae, which also encodes AcsA, AcdAB, and CdhA-E, is involved in the conversion of acetate to acetyl-CoA and methyl-H4MPT. It has been reported that the acetoclastic methanogenic pathway in Methanomassiliicoccales occurs in the monogastric gut40. However, the interactions and regulatory mechanisms of methane production in both the methylotrophic and acetoclastic metabolic pathways in Methanosarcinaceae and Methanomethylophilaceae remain unclear and require further investigation.

Methanocorpusculaceae, the second most important hydrogenotrophic methanogen in the ruminant gut, is involved in all genes in the pathway from CO2 to CH4, with the greatest proportion of genes being FwdA-E (Fig. 5a). The relative abundance of Methanocorpusculaceae is greater in ‘low methane’ emitters, such as the southern hairy-nosed wombat (Lasiorhinus latifrons), which is a foregut fermenter21. The relative abundance of Methanocorpusculaceae was greater in the large intestine than in the stomach and small intestine (Fig. 4c), indicating that Methanocorpusculaceae may be better adapted to the intestinal fermentation environment of nonruminant animals. Methanocorpusculaceae also encodes the highest proportion of the membrane-bound complex EchA-F, constituting a set of genes involved in electron transfer during methane production. Membrane-bound Fpo complexes were more prevalent in Methanomethylophilaceae (Fig. 5a).

We also confirmed the enrichment of genes involved in methanogenesis between ruminant breeds (Fig. 5b). These results aligned with our hypothesis that genes involved in methylotrophic methanogenesis, including Mta, Mtb, Mtm, and Mtt, were more enriched in the rumens of yaks and camels than in those of zebu, sheep, goats, deer, bison, cows, beef cattle, and buffalo (Fig. 5b). Conversely, the genes involved in hydrogenotrophic methanogenesis, namely, Fwd, Fdh, Ftr, Mtd, Mch, Mer, and Mtr, were more enriched in the rumens of camels and yaks. The Acs gene involved in acetoclastic methane production was also enriched in the rumens of camels and yaks, following the same trend as that of the genes involved in methylotrophic methanogenesis (Fig. 5b). However, no enrichment was detected in the genes Mcr, Hdr, and Mvh, which are involved in hydrogenotrophic, methylotrophic, and acetoclastic methanogenesis, between the breeds.

Ruminant-associated archaeal genomes potentially carry unique resistance genes and other functions

For antibiotic resistance gene (ARG) detection, the results revealed that 7 Methanobrevibacter sp. genomes contained the greatest number of ARGs, primarily tet genes that encode proteins responsible for resistance to tetracycline antibiotics, including tetM, tetW, tetO, tet32, tetO/W, tet44, and tetW/N/W (Supplementary Fig. 3a). This finding was consistent with a previous report on ARGs in the rumen microbiota, where tetW was identified as the dominant ARG41. However, these findings differed from those of previous studies42,43, suggesting potential influences of diet, environment, and management. Two genomes, Methanobrevibacter B boviskoreani and Methanobrevibacter A millerae B, contained both ARGs and mobile genetic elements (MGEs) (Supplementary Fig. 3a). According to the virulence factor database (VFDB), the two major categories were immune modulation (IM) and stress survival (SS), with a large proportion of the genomes belonging to Methanobrevibacter, Methanobrevibacter_A, and Methanobrevibacter_B (Supplementary Fig. 3b). For quorum sensing (QS) detection, only 4-hydroxyl-2-alkyl-quinolines (HAQs)44 (n = 154) and cholera autoinducer-1 (CAI-1)45 (n = 99) were identified as the predominant languages in the ruminant archaeal genomes (Supplementary Fig. 3c). This differed from a previous report, as AI-2 was suggested as a potentially “universal” signal mediating social behavior in an analysis of 981 rumen bacterial and archaeal genomes46. These two dominant QS languages were exclusively found in Methanobrevibacter sp. (Supplementary Fig. 3c). We also detected antibacterial biocide and metal resistance genes in ruminant archaeal genomes. The results revealed that the largest category of resistance genes (n = 13,080) associated with Cu/Ag resistance (Supplementary Fig. 3d) primarily originated from Methanobrevibacter species, including Methanobrevibacter_A, Methanobrevibacter, and Methanobrevibacter_B.

Ruminant lysogenic archaea and proviruses

In total, 164 archaeal genomes in the ruminant archaeome contained at least one viral sequence (mean = 1.99), revealing a lysogeny ratio (LyR) of 16.4% (164/998) among ruminant gut archaea (Supplementary Data 5). This LyR was lower than 28.5% for archaea in marine systems47. We identified 327 viral populations in these archaeome datasets, with the virome distributed among 8 complete, 128 high-quality (>90% completeness), and 191 medium-quality (50–90% completeness) categories (Supplementary Fig. 4a and Supplementary Data 6). Among the identified proviruses, 299 viral species were specific to Methanobrevibacter (including 148 for Methanobrevibacter and 129 and 22 for Methanobrevibacter A and B, respectively) (Supplementary Fig. 4b). This diversity was greater than that reported for the human gut archaeome25. In addition, viromes in extreme environments and the human gut have been extensively explored through metagenomic analysis25,48,49,50,51,52,53,54 because of the importance of the virome in modulating the balance and functions of the ecosystem. The rumen virome database (RVD) was established by mining 975 globally published rumen metagenomes55. However, to our knowledge, this is a systematic description of ruminant gut archaeal lysogens and the virome. On the basis of the taxonomic classification released by ICTV 202256, 301 vOTUs were assigned to the Caudoviricetes class, including the major families Myoviridae, Siphoviridae and Podoviridae, which were removed from the ICTV update (Supplementary Fig. 4c). A total of 1074 AMGs were found in ruminant archaeal viruses. Functional analysis revealed that virome AMGs were involved in miscellaneous (MISC) (n = 332), organic nitrogen (n = 221), carbon utilization (n = 221), energy (n = 174), transporters (n = 119), and carbon utilization (Woodcroft) (n = 7) (Supplementary Fig. 4d). For carbon utilization, the main function was that of glycosyl transferases (Supplementary Fig. 4e). For the energy category, AMG was involved mainly in methanogenesis, including the conversion of CO2, acetate, methanol, methylamine, dimethylamine, and trimethylamine to form methane (Supplementary Fig. 4e). In summary, ruminant archaeal viruses are potentially ecologically important and may function in the ruminant gut microbiome, particularly in strategies to potentially reduce methane production.

Discussion

Although extensive research has been conducted on the rumen and intestinal microbial communities of ruminants, with a primary focus on bacterial populations29,57,58,59,60, systematic analyses of the ruminant gut archaeome remain insufficient. This study provides insights into the biology of the ruminant gastrointestinal tract (GIT) archaeome by assembling and cataloging 998 nonredundant archaeal genomes. Initial associations between the diversity and function of ruminant gut-associated archaea of different breeds and enterotypes were established. However, some intestinal segments excluding the rumen from many geographic locations, such as regions in America, Africa, Oceania, and Russia, are inadequately sampled. Further efforts should aim to increase the analysis of the gut microbiota in these areas or other environments.

Since a small subset of genomes (21/998 = 2%) has been obtained from cultured archaeal representatives29,59, it is crucial to obtain more cultured isolates from the ruminant gut to better understand the ecology, evolution, and function of archaeal genomes. Advancements in high-throughput cultivation methods, machine learning, and Raman microspectroscopy technologies can accelerate the isolation of many isolates on demand, overcoming the limitations of traditional labor-intensive methods61,62,63. Moreover, the dataset of archaeal genomes can serve as a reference and a starting point for targeted cultivation of new members of the ruminant gut archaeome. Additionally, the use of long-read metagenomics sequencing techniques, such as Oxford Nanopore Technology (ONT) and PacBio, has the advantage of longer read lengths (capable of obtaining 10 kb to 1 Mb fragments), which can aid in assembling and recovering more complete MAGs64. The use of specialized methods to reduce host contamination, increase cell lysis, and improve DNA extraction might also reduce the amount of information available on archaea from different gut biogeographies65,66. Devices used to sample regions of the human intestinal tract might also be adapted to easily sample wild ruminant animals that cannot be slaughtered67. In summary, the use of adapted technologies might improve and enable profound capture of more diverse ruminant gut archaeomes.

The observed percentage of archaea in the ruminant gut microbiome varied by breed and gut biogeography, ranging from 0.14% to 1.76% in the rumens of different ruminants (Fig. 3g). This finding was similar to the average percentages reported for human gut archaea (~1.2%)25. However, similar to that in humans, the abundance of methanogenic archaea in ruminants is highly variable and positively correlated with methane exhalation25. Interestingly, low-methane emitters, such as yaks, present different archaeal compositions in the rumen. The Methanomethylophilaceae enterotype dominates the landscape of the archaeal community composition according to the definition of enterotypes26,32. This finding agrees with studies on methane emissions from yak18,36 and the comparison of methanogen diversity between yak (Bos grunniens) and cattle (Bos taurus)68. The rumen archaea of camels also contain the Methanomethylophilaceae enterotype, which was also observed in camels that produce less methane than in those of ruminants such as cattle, sheep and goats69. However, the mechanism underlying the archaeal enterotypes and activity of high- and low-methane emitters remains largely unclear, revealing a potential avenue for reducing methane emissions from ruminants. Further research is essential to elucidate the intricacy of biochemical processes, such as the transfer and utilization of major methyl substrates and dissolved hydrogen, as well as microbial interactions involving bacteria, archaea, viruses, fungi, and protozoa. Advanced techniques such as metagenomics, metatranscriptomics, and stable isotope probing can offer valuable insights into the metabolic activities and functional roles of methylotrophic methanogens in the context of ruminant digestion. Moreover, the rumen, as one of the most efficient anaerobic fiber fermentation systems involving archaea, holds tremendous potential for the development of biomass energy70.

The presented establishment of ruminant gut-associated archaeal genomes, along with a catalogue of 1.6 million predicted proteins, provides foundational information and serves as a specialized data source for addressing major questions in future research. Understanding the shaping of the diversity and function of the archaeome by ruminant host and archaeome-host interactions, potentially through microRNA secretion by host cells, is crucial for breeding ruminants with low CH4 emissions and guides microbiome-informed breeding strategies11. The ruminant gut microbiome, particularly in the rumen, functions through trophic-like levels to ferment the diet and generate nutrients to meet the host’s requirements11. Further research is warranted to investigate how the archaeome, as the third sublevel, interacts with various microorganisms (bacteria, viruses, fungi, and protozoa) at other trophic-like levels, ultimately influencing the host’s digestion and utilization efficiency of feed. The ruminant gut archaeome potentially carries virulence and resistance genes, with a high prevalence of tet-type category ARGs that confer resistance to tetracycline. Considering archaea, conducting global-scale studies is imperative to evaluate the risks associated with gene transfer, environmental impact, food safety, and human health implications41,42. The archaeal virome also encodes auxiliary metabolic genes (AMGs), particularly those related to methane metabolism (Supplementary Fig. 4). However, exploring the process and extent to which the archaeal virome influences methane emissions in ruminants and the potential for targeted reduction of methane emissions by focusing on specific archaea requires further research1,55.

In summary, we successfully delineated the landscape of the archaeome in the GIT of ruminant animals by assembling 2270 ruminant gastrointestinal metagenomes and incorporating previously assembled and isolated archaeal genomes. We confirmed two distinct enterotypes in ruminant GIT archaea, namely, the Methanobacteriaceae enterotype and the Methanomethylophilaceae enterotype, with their composition and dominant methanogenesis pathways showing variations on the basis of breed and gut biogeography. The viral community infecting archaea in the GIT of ruminants carries many methanogenesis genes, offering the potential to regulate archaea to reduce methane emissions. These findings provide comprehensive insights into the ecology and functionality of the gut archaeome in ruminant animals, laying a solid foundation for further exploration in mitigating greenhouse gas emissions through precise regulation of archaea. However, the limitations of this study include the lack of additional culture and in-depth mechanistic validation of relevant methanogens8 and the lack of incorporation of third-generation sequencing data to improve the quality and recovery rate of the assembly. In further studies combining culturomics and the long sequence lengths of third-generation sequencing, complete genomes and MAGs with higher quality could be obtained. This would contribute to gaining deeper insights into higher-resolution strains, allowing for a more precise understanding of their genetic diversity, functional capabilities, and ecological roles in the gut of ruminants.

Methods

Dataset description

A comprehensive catalogue encompassing 998 genomes from the ruminant gut archaeome has been compiled on a global scale. This extensive collection aims to explore the diversity of archaeal genomes and investigate variations in the formation and metabolism of CH4 across ruminant breeds and gut locations. We gathered 2,270 metagenomic samples from publicly available databases (Supplementary Data 7 and 8; Supplementary Fig. 5)9,19,23,29,57,58,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93. First, we cleaned the metagenomic data using fastp v.0.23.248 (with parameters –detect_adapter_for_pe and –dont_eval_duplication -w 16), followed by the removal of host contamination specific to ruminant breeds using BWA v.0.7.17 (with parameters -k 31 -p -S -K 200000000)94. Second, we assembled the cleaned metagenomic data using Megahit v.1.2.9 with a single-sample assembly model (with a -k-list of 39, 59, and 79)95. Contigs less than 2 kb in length in the assembly sequences were then removed with seqkit v2.3.096. Next, a sorted bam file for each sample was generated using minimap2 v2.17-r941 (-ax sr)97 and samtools v1.698 with default parameters. The depth file was calculated with the jgi_summarize_bam_contig_depths script in metabat2 v2.1599. Metagenome-assembled genomes (MAGs) were subsequently assembled using metabat2 v2.15 with default parameters. Finally, we employed checkM v1.2.2100 to assess the quality of the MAGs and selected those identified as archaea with an estimated completeness of >50% and contamination of <5%. In total, 815 MAGs were obtained through these assembly procedures. The quality of the collected MAGs (n = 517) (details are shown in the Data availability section) from previous studies was assessed, and only those with a completeness of >50% and contamination of <5% (n = 428) were retained. Therefore, the total number of archaeal genomes reached 1243 (n = 815 + 428).

Genome quality and taxonomic classification

The complete dataset comprising 1243 archaeal genomes underwent dereplication at 99.9% for nonredundant archaeal genomes (n = 998), 99% for individual strains (n = 556), and 95% for species (n = 203) ANI values using DRep v3.4.0101 (Supplementary Fig. 6). A subset of archaeal genomes at the strain level, totaling 279 and meeting completeness criteria (>80%) and contamination thresholds (<5%), was utilized for constructing the phylogenetic tree. Tree matrices were generated for 279 archaeal genomes with the criteria of >80% completeness and <5% contamination at the strain level (99% ANI similarity) using PhyloPhlAn v3.0.67102. Dendrograms, constructed on the basis of the ANI tree matrix, were annotated using the iTOL tool (Interactive Tree of Life)103. For cultured archaeal genomes, 19 and 16 out of the total 21 were retained at the individual strain (99%) and species (95%) levels, respectively. Taxonomic annotation of all archaeal genomes was performed using the GTDB Toolkit v2.1.1 (database release207_v2)104 with default parameters, employing a set of 122 marker genes to identify archaeal MAGs. The taxonomy of the archaeal genomes was summarized via the gtdb_to_ncbi_majority_vote.py script within the GTDB Toolkit.

Genome annotation and protein catalogue

Protein genes were predicted via Prodigal v2.6.3 with default parameters employing a single model105. The protein-encoding genes were subsequently annotated via eggNOG-mapper v2.1.9 and eggNOG database v5.0 with default parameters106,107. The protein catalogue was generated by combining all the predicted CDSs (totaling 1,604,531) derived from the 998 nonredundant archaeal genomes. MMseqs2 v14.7e284 linclust was utilized to cluster the concatenated protein dataset with the options ‘–cov-mode 1 -c 0.8 –kmer-per-seq 80 –min-seq-id 0.5’108. Proteins were clustered at various percentage identities, and the resulting number of unique proteins per cluster for total taxonomic families was computed and visualized. To minimize the risk of contaminants, nonclustered proteins were filtered out. For clear protein visualization, proteins that clustered only above 2 genes per cluster and were present in more than 50 genomes were displayed via pheatmap and R with NMDS109.

Relative abundance of archaea in ruminant metagenomes

To assess the relative abundance of archaeal genomes at the species level in the collected metagenomic datasets, we employed CoverM v0.6.1 (https://github.com/wwood/CoverM) and utilized the relative abundance calculation method with genome patterns.

Methane metabolism pathway establishment

A total of 203 genomes at the species level, with completeness exceeding 50% and contamination below 5% in protein sequences, were annotated against the Kyoto Encyclopedia of Genes and Genomes (KEGG) orthologs database using eggNOG-mapper v2.1.9 with default settings106. The assigned hits were consolidated for all genomes using KO numbers. The annotated results were further refined, and KOs associated with methane metabolism were retained on the basis of previous studies13,15,16,25,110,111,112,113,114. These included genes involved in hydrogenotrophic, methylotrophic, and acetoclastic pathways contributing to methane formation, as well as electronic receptors and transporters on the membrane. The copy numbers of these genes in the archaeal genomes were computed, and the proportion of each gene at the family level was determined on the basis of the summary copy numbers of the total archaeal genomes. The methane metabolism pathways and the proportion of each gene at the family level were established and visualized for ruminant archaeal genomes. The copy numbers of genes involved in methane metabolism were combined with the relative abundance of archaeal genomes to compare the differences in these genes between different breeds and visualized using pheatmap109.

Detection of virulence and resistance genes and quorum sensing

To predict potential virulence genes in all 998 archaeal genomes, Diamond v2.0.15115 was employed to BLASTp against the following databases: SARGs v3.2.3116, VFDB117, BacMet118, MGEs119, and Quorum sensing120, with parameters set at an e value of 1e−7, a query coverage of 85%, and an identity of 60%. Among the 998 genomes, ARGs were identified in only 74, MGEs in 20, QS in 147, VFDB in 779, and BacMet in 635.

Viral identification, quality estimation and AMG detection

To evaluate the presence of viruses, VirSorter2 v.2.2.320 was utilized to search all 998 unique archaeal genomes. To mitigate potential contamination issues arising from the binning process, we focused on proviruses flanked within archaeal contigs for this analysis. The identification of viral sequences followed the viral sequence identification SOP as outlined in a previous report (dx.doi.org/10.17504/protocols.io.bwm5pc86). In summary, CheckV v1.0.1 (database v.1.4)121 was used to assess the quality of VirSorter2-predicted viral sequences, and a quality control check was performed to eliminate false positives. The criteria for virus sequence inclusion were as follows: keep 1 (viral_gene > 0) and keep 2 (viral_gene = 0 and host_gene = 0 or score >= 0.95 or hallmark > 2). However, only viral sequences of medium quality (50-90% completeness), high quality (>90% completeness), and complete quality were selected for further analysis. A total of 327 viral sequences were identified in our collected archaeal genome datasets. This result was obtained by clustering 8 complete, 128 high-quality (>90% completeness), and 191 medium-quality (50–90% completeness) viral sequences at 95% ANI and 85% AF via the supporting code of CheckV. Taxonomy was determined via geNomad v.1.2.0122, and the auxiliary metabolic genes (AMGs) of the viruses were annotated via DRAM-v v1.4.6123.

Comparison of different environmental Methanomethylophilaceae

To assess the distinction of Methanomethylophilaceae between ruminants and other environments, we assembled a dataset comprising 141 archaeal MAGs identified from diverse environmental samples available in the NCBI database. This included but was not limited to the human gut, termite gut, soil, sediment, wastewater, and chicken gut. This dataset served as a reference for comparison with the Methanomethylophilaceae set collected and assembled from the ruminant gut microbiome14,15,16,111,112,124. All the genomes from the different environments and ruminants used in the analysis exhibited completeness exceeding 80% and contamination below 5% and were deduplicated at the strain level (99% similarity) using DRep v3.4.0101. A total of 124 Methanomethylophilaceae genomes, including 64 from ruminant samples, were retained for further analysis (Supplementary Data 9). We constructed a phylogenetic tree incorporating all the Methanomethylophilaceae genomes mentioned using PhyloPhlAn v3.0.67102 with the parameters –diversity low –fast –min_num_markers 1. The tree was annotated using the iTOL tool103. To estimate the pairwise ANI distance between the Methanomethylophilaceae genomes from the ruminant gut microbiome and other environmental Methanomethylophilaceae genome datasets, we employed fastANI v1.33125. ANI values ranging from 75% to 100% were selected for plotting using the pheatmap package109. The phylogenetic tree was generated using PhyloPhlAn v3.0.67 and visualized via the iTOL website103.

Statistical analysis

The ‘vegan’ package v2.6.4 of R v4.2.2 was used to examine the differences in functional and compositional structure, and the ‘adonis’ function was used to perform permutational multivariate analysis of variance (PERMANOVA) via the Bray‒Curtis method with 1000 permutations. Nonmetric multidimensional scaling (NMDS) analyses and stress values were performed via metaMDS functions from the vegan package with Bray‒Curtis distances. The rules of thumb for stress values are as follows: <0.05 is excellent, 0.05-0.10 is good, 0.10-0.20 is fair, and >0.20 is poor. NMDS plots were made with the ggplot2 package v3.4.1. The differences in the relative abundance of archaeal composition across different ruminants were analyzed via the Kruskal‒Wallis test with Dunn’s multiple comparisons test via Prism 9.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.