Introduction

The ability to obtain sufficient nutrient and energy intake for growth, reproduction, and maintaining a healthy immune system is a fundamental challenge for wild animals1,2. Numerous studies have shown that species have evolved behavioral and physiological strategies, including hibernation, seasonal reproduction, shifts in diet and range use in response to temporal changes in food distribution and availability, and increased mobilization of fat stores to meet changing nutritional and reproductive demands3,4,5,6,7. In addition, abiotic factors such as seasonal variation in temperature, rainfall, and photoperiod can serve as triggers that affect animal physiology and result in seasonal shifts in metabolic requirements and activity patterns8.

In this regard, advances in microbiome research have demonstrated the significant role and plasticity of the host gut microbiome in adjusting to changing social and environmental conditions9,10,11,12,13,14,15. For instance, a study on wild semi-provisioned Tibetan macaques (Macaca thibetana) found that during the spring, their gut microbiome was enriched in bacteria such as Prevotella, which functioned in carbohydrate and energy metabolism. In the winter, their gut microbiome showed an increase in Succinivibrio, which is reported to facilitate more efficient absorption of difficult-to-digest structural carbohydrates including cellulose and hemicellulose16. Similarly, a study of Ethiopian geladas (Theropithecus gelada), identified seasonal differences in the composition of their gut microbiome that were attributed to the combined effects of changes in food availability and temperature17. During the rainy season, when grasses dominated their diet, fiber-degrading and fermenting bacteria such as Prevotellaceae and Bacteroidaceae dominated the gelada gut microbiome. In contrast, during the dry season, starch-degrading bacteria were more prominent. Moreover, during cold and dry periods, the gelada gut microbiome experienced an increase in bacteria related to energy, amino acid, and lipid metabolism (i.e. Victivallales, Christensenellaceae and Methanobrevibacter). These microbes communities appeared to play an important role in regulating host body temperature and offsetting the elevated energetic demands of remaining thermoneutral17. In this regard, a controlled study of laboratory mice exposed to cold stress over a 10-day period (temperatures of 6 °C) found that their microbiome became enriched in bacteria of the genus Firmicutes. When this enriched microbiome was transplanted into germ-free mice, there was evidence of remodeling of fat stores and intestinal tissues associated with an increase in the ability of the mice to withstand periods of high energy demand associated with cold temperatures18.

In the present study, we examined seasonal differences in the gut microbiome of the golden snub-nosed monkey (Rhinopithecus roxellana, subfamily Colobinae), including individuals from a wild population inhabiting Shennongjia National Park and a captive population housed at the Beijing Wildlife Park. Golden snub-nosed monkeys are an Endangered species of foregut fermenting primates that inhabit high-altitude mountainous temperate forests (from 1500 to 3400 m) across central China. This region is characterized by seasonal extremes in temperature19,20, with nighttime winter temperatures averaging 0 °C and dropping to as low as −14 °C. During the winter, golden snub-nosed monkeys can encounter severe snowstorms, and snow cover can last up to 4 months21,22. These conditions make foraging difficult and increase the challenge of maintaining an energy balance and remaining thermoneutral.

To improve habituation and ensure individuals have sufficient year-round food, some wild populations are semi-provisioned, receiving limited food supplementation from local management authorities. In the Shennongjia population, golden snub-nosed monkeys primarily feed on young leaves and buds during the summer, which account for more than 60% of their feeding time. In winter, they consume lichen and tree bark, which are typically considered low-nutrient foods, accounting for approximately 40% of feeding time23,24 (Supplementary Table 1). At a site in the Qinling Mountains, Guo et al.25 found that wild golden snub-nosed monkeys increased their energy intake by a factor of 1.8 in winter compared to spring by exploiting a high lipid and high carbohydrate diet (e.g., bark from Morus australis, Litsea pungens as lipid sources, and bark from Malus baccatan and Cornus hemsleyi as carbohydrate sources)25, which can offset the increased winter energy requirements of remaining thermoneutral. Similarly, Hou et al. reported that these monkeys lost an average of 14% of their body weight during winter, a loss that was compensated for by increased daily energy intake during summer and fall, with much of this energy being stored as fat26.

Traditionally, studies investigating the primate gut microbiome have been primarily based on 16S rRNA data. This approach, while generating data on microbiota species composition and diversity, offers limited insight into the potential functionalities of the gut microbiota community. By contrast, metagenomic and metatranscriptomic sequencing offer a more detailed profile of the gut microbiome by including more accurate microbiota identification, classification, and comprehensive genomic information27,28. Using metagenomic and metatranscriptomic sequencing enabled us to gain greater insight into the metabolic potential, functional features, and contributions of the microbiome to the host’s ability to respond to seasonal shifts in diet and nutrient intake. Additionally, the metagenome-assembled genomes (MAGs) technique has gained popularity for reconstructing bacterial draft genomes through metagenomic binning29. Compared to traditional methods that rely on database annotations, MAGs enable the reconstruction of gut microbiome genomes from rare community members, offering significant advantages in discovering new microbes and expanding our understanding of unknown microbial characteristics and functions30.

Thus in the current study, we sought to reconstruct MAGs based on fecal metagenomic reads and to investigate gene expression profiles through metatranscriptomics, aiming to explore the gut microbiome at the genome level in both captice and wild golden snub-nosed monkeys. We collected 29 winter and 17 summer fecal samples from a wild population of golden snub-nosed monkeys. For 12 individuals, we collected both winter and summer samples, these are referred to as the same individual group (n = 12). We also collected winter (n = 10) and summer (n = 7) fecal samples from individuals in a captive population of golden snub-nosed monkeys that consume a more consistent diet throughout the year. In total, we reconstructed 578 non-redundant MAGs from the fecal metagenomes of golden snub-nosed monkeys, 76.5% of which did not have exact matches in reference databases. Finally, we annotated 4959 KEGG Orthologs (KOs) from the metatranscriptome and performed paired analyses with the metagenomic data. Our results provide new insight into how changes in the gut microbiome respond to seasonal variation in diet and energy demands for golden snub-nosed monkeys and other wild animals.

Results

Construction of MAGs and annotation of fungi

In total, we obtained 2,213,582,166 raw reads from 63 fecal samples, generating 960 Gb of raw data. After quality filtering and the removal of host sequences, 951 Gb (mean ± SD: 15.09 ± 0.89) of short reads were used for the overall metagenome assemblies. For the metatranscriptome, we obtained 404,818,270 raw reads from 22 samples, resulting in 119.5 Gb (mean ± SD: 5.43 ± 0.42) of data after quality filtering and removal of host sequences (Fig. 1A).

Fig. 1: Analysis pipeline for this study and the phylogenetic tree of nonredundant MAGs.
figure 1

A Pipeline for constructing metagenome-assembled genomes (MAGs) and metatranscriptomics gene expression profiles of golden snub-nosed monkeys. B Geographic distribution of the 63 fecal samples from the golden snub-nosed monkeys. Metagenomic sequencing data were obtained from 29 fecal samples in the wild winter group and 17 fecal samples in the wild summer group. For 12 wild golden snub-nosed monkeys, samples were collected in both seasons, designating them as the ‘same individual group’. The captive winter group included ten samples, and the captive summer group included seven samples, all collected at Beijing Wildlife Park. C Phylogenetic tree of the 578 nonredundant MAGs. Clades are colored according to taxonomic phylum. From the inner to the outer layers: rectangular symbols indicate MAG novelty status: green for novel MAGs and yellow for known MAGs; Stacked bars represent genome completeness (orange); Heatmap shows the N50 values of contigs, with colors ranging from light yellow to dark yellow indicating increasing N50 length. Bar plots show the mean relative abundance of each MAG in the four groups, captive summer group (light blue), wild summer group (red), captive winter group (purple), and wild winter group (dark blue).

Full size image

Microbiome genomes were constructed from the metagenomic sequencing data obtained from the 63 samples described above. Metagenomic sequencing generated 3602 MAGs with a threshold of >50% completeness and contamination of ≤20% (Supplementary Table 4). These reconstructed gut microbiome genomes were then compiled and dereplicated at 99% of the average nucleotide identity (ANI), which resulted in a final set of 578 non-redundant MAGs (NR-MAGs) (Fig. 1C). All of the NR-MAGs met the medium-quality criteria (more than 50% completeness and <5% contamination), with 398 classified as high-quality (more than 90% completeness and <5% contamination)31. The 578 NR-MAGs were subsequently classified into taxa using the Genome Taxonomy Database Toolkit (GTDB-Tk), with 574 and 4 assigned as bacteria and archaea, respectively. Of all NR-MAGs, only 136 (23.5%) could be assigned to known species (mean ± SD: 10.44% ± 11.92%), and 442 (76.5%) could not be assigned to the species level (mean ± SD: 56.78% ± 15.47%). Additionally, 14 MAGs (2.4%) could not be assigned to the genus level (mean ± SD: 0.8% ± 0.83%), and 1 MAG could not be assigned to the family level (Fig. 1C, Supplementary Table 5). Over 70% of the genomes could not be assigned to an existing species using the GTDB-tk database, confirming that the majority of the golden snub-nosed monkey gut microbiota lack representation in current reference databases.

The all NR-MAGs were distributed across 17 phyla, 22 classes, 44 orders, 78 families, and 284 genera. The most abundant MAGs in the gut microbiome of the golden snub-nosed monkeys were primarily from the phylum Bacteroidota. Among the top ten most abundant MAGs, five belonged to the phylum Bacteroidota (mean ± SD: 13.43% ± 7.83%), and nine were classified as novel MAGs (mean ± SD: 15.81% ± 9.97%). This reflected the novelty and potential for further exploration of the gut microbiome of golden snub-nosed monkeys.

A total of 243 fungal species were identified in the gut of golden snub-nosed monkeys. These spanned five phyla, 23 classes, 57 orders, and 104 families, the majority (66 families, 63.5%) of which belonged to the phylum Ascomycota (Supplementary Table 6).

Gut microbiome community differences between captive and wild groups

To identify clustering patterns and diversity in the gut microbiome of wild golden snub-nosed monkeys, we performed a PCoA analysis based on Bray-Curtis distances. We found the gut microbiome of wild populations showed distinct separation between different seasons, while the captive population exhibited a more convergent trend across seasons Fig. 2A. Compared to PCoA, however, enterotype analysis, which does not rely on grouping information, can better reflect the actual existing clustering effect of the gut Compared to PCoA, however, enterotype analysis, which does not rely on grouping information, can better reflect the actual existing clustering effect of the gut microbiome in high-dimensional data space in high-dimensional data space32. Our results showed that all individuals in the captive group, regardless of season, belonged to Enterotype 1 (Fig. 2B). In contrast, the wild summer group was predominantly classified as Enterotype 2 (76.5%), while the wild winter group was predominantly classified as Enterotype 3 (70%) (Fig. 2B). These results indicated that enterotypes are associated with environmental conditions and seasonal fluctuation in diet and energy requirements, which is important for the management of captive animals and their reintroduction of individuals into the wild.

Fig. 2: Seasonal influences on the gut microbial community characteristics and assembly mechanisms of golden snub-nosed monkeys.
figure 2

A Principal coordinates analysis (PCoA) revealed four distinct clusters of microbial communities belonging to the four groups. B Principal coordinate analysis plots based on the Jensen-Shannon distance between samples, revealing three enterotypes in the gut microbiota. The lines connected to the center of each ellipse correspond to the group affiliation. C The box plots show the alpha diversity (Simpson index) of the four groups. ∗:p < 0.05. D The Venn diagram shows the number of MAGs shared and uniquely present among the wild summer, wild winter, captive summer, and captive winter groups. E–H Co-occurrence network among core species of microbiota of the four golden snub-nosed monkey groups. E wild summer group, F wild winter group, G captive summer group, H captive winter group. I–L Relative importance of different bacterial community assembly mechanisms of the gut microbiota in four groups. I wild summer group, J wild winter group, K captive summer group, L captive winter group.

Full size image

The alpha diversity of the gut microbiome community composition (i.e., taxonomic composition) in wild golden snub-nosed monkeys was significantly higher in summer than in winter (Simpson index, p < 0.05). No such seasonal differences were detected in the captive study group (Fig. 2C). These results indicated that the effects of seasonal changes on the gut microbial of golden snub-nosed monkeys were more pronounced in wild populations, whereas captive groups were less affected by seasonal variations. A Venn diagram revealed that all four groups (wild winter and summer and captive winter and summer) shared 188 MAGs (32.53%) (Fig. 2D). The captive group harbored 23 unique MAGs, 12 of which could be assigned to the known species. The wild group contained 24 unique MAGs, all of which lacked species-level annotation. Notably, both winter wild and captive groups exhibited more group-unqiue MAGs than did summer groups (wild winter: 21, captive winter: 13 vs. wild summer: 3, captive summer: 10) (Fig. 2D). Among the MAGs unique to the wild winter group, most were taxonomically assigned to the orders Oscillospirales and Lachnospirales.

Compared to the standard alpha/beta diversity metrics widely used in microbial ecology, co-occurrence networks can provide a powerful approach to exploring microbial ecosystem stability33. Co-occurrence network analyses revealed significant seasonal differences in the complexity of the gut microbiome network structure in the wild golden snub-nosed monkeys. The wild summer group exhibited greater robustness (average clustering coefficient: 0.101, nodes: 375, edges: 680), whereas the wild winter group showed lower community stability compared to the summer group (average clustering coefficient: 0.039, nodes: 419, edges: 242) (Fig. 2E, F). Additionally, under different environmental conditions, regardless of season, the co-occurrence networks of the wild monkeys were significantly sparser, with fewer edges and nodes, compared to those of the captive monkeys (winter/summer, average clustering coefficient: 0.212/0.308, nodes: 400/375, edges: 4250/5955) (Fig. 2E–H).

Unraveling the drivers controlling gut microbial community assembly is a central issue in microbial ecology34. Specifically, we partitioned 578 non-redundant MAGs into a series of phylogenetic bins (≤12 MAGs per bin) to evaluate the role of different bacterial lineages in community assembly processes. Our results revealed that dispersal limitation (DL) was the predominant assembly process shaping the gut microbial of golden snub-nosed monkeys. DL accounted for the majority of community assembly dynamics across all groups (wild summer: 92.5%, wild winter: 92.6%, captive summer: 79%, captive winter: 82%). In contrast, homogeneous selection (HoS) and heterogeneous selection played relatively minor roles, contributing approximately 2.6–15% and 2–4.9%, respectively (Fig. 2I–L). In the case of the wild snub-nosed monkeys, although the overall contribution of deterministic processes to gut microbial community assembly remained unchanged from summer to winter, the proportion of heterogeneous selection (HeS summer: 4.9%; winter: 2.7%) decreased, while the proportion of HoS (HoS summer: 2.6%; winter: 4.7%) increased (Fig. 2I, J). Compared to the captive snub-nosed monkeys, however, the proportion of DL (DL wild: 92.5–92.6%; captive: 79–82%) increased in the wild, while the proportion of HoS (HoS wild: 2.6–4.7%; captive: 15–19%) decreased. The prevailing view is that captive animals live in smaller and more homogeneous environments, which may increase the influence of HoS on their gut microbiota35. In contrast, wild populations experience greater ecological freedom and environmental variability, which could lead to more pronounced DL36, as reflected by the more scattered clustering patterns in the PCoA plots (Fig. 2A).

Seasonal fluctuations of the gut microbiome composition

For captive group, our results did not detect any microbiota taxa with significant seasonal differences (Supplementary Table 7), consistent with our findings on enterotype and community structure, which indicated that the gut microbiota of captive golden snub-nosed monkeys exhibited minimal fluctuations across seasons.

In the wild monkeys, 24 MAGs exhibited significantly higher abundance in the summer compared to the winter (pFDR < 0.05) (Supplementary Table 8) (see Methods 2.8) (Fig. 3A, Supplementary Table 9). This suggested that these 24 MAGs may better reflect the effects of seasonal fluctuations on the host’s gut microbiota, which primarily included Oscillospiraceae (12 MAGs, mean completeness 94.16, mean contamination 1.58), Lachnospiraceae (4 MAGs, mean completeness 91.41, mean contamination 1.58), and Eggerthellaceae (2 MAGs, mean completeness 99.0, mean contamination 0.85) (Fig. 3A, Supplementary Fig. 2). Similarly, in the wild winter group, we identified 21 MAGs with significantly higher abundance compared to the wild summer group (pFDR < 0.05) (Supplementary Table 8) (see Methods 2.8). Most of these belonged to the taxonomic groups Lachnospiraceae (9 MAGs, mean completeness 93.99, mean contamination 2.62) and Oscillospirales (3 MAGs, mean completeness 96.49, mean contamination 0) (Fig. 3A, Supplementary Figure 2).

Fig. 3: Seasonal differences in MAGs of golden snub-nosed monkeys.
figure 3

A The circos barplot reveals enriched MAGs between wild summer (red) and wild winter (blue) groups. The bars indicate abundance. The barplot backgrounds are colored according to the taxonomic phylum. B The Sankey diagram reveals that 9 MAGs belong to the family Lachnospiraceae and share several unique genes and pathways. C The circos barplot reveals the enriched fungi between wild summer (red) and wild winter (blue) group.

Full size image

In response to winter low temperatures and seasonal variation in food distribution and availability, golden snub-nosed monkeys appear to rely on fat metabolism to maintain energy balance26. Our results suggested that Lachnospiraceae may play a significant role in the energy metabolism of the golden snub-nosed monkeys in the winter. The core genes of Lachnospiraceae (shared among the nine enriched MAGs from the winter) and Oscillospirales (shared among the 12 enrichweds MAGs from the summer) revealed that the enriched MAGs from Lachnospiraceae in winter shared several unique genes, including K00074, K00817, and K11358 (ko00360; phenylalanine metabolism); K01610, K01647, and K01681 (ko00020; citrate cycle); K00937 (oxidative phosphorylation); K00248 (ko00071; fatty acid degradation); and K00626 (involved in ko04975; fat digestion and absorption, ko00071; fatty acid degradation, and ko00380; tryptophan metabolism) (Fig. 3B). Our results show that these genes were predominantly associated with pathways, such as map00020 (Citrate cycle (TCA cycle)), map00360 (Phenylalanine metabolism), map00071 (Fatty acid degradation), map00190 (Oxidative phosphorylation), map00380 (Tryptophan metabolism), and map04975 (Fat digestion and absorption) (Fig. 3B), suggesting that these genes may aid the golden snub-nosed monkeys to more efficiently utilize fat for energy and heat production during the winter.

We also analyzed the fungi in the gut microbiota of golden snub-nosed monkeys using the same methods. In accordance with results of the MAGs, no significantly different fungal taxa at the family level were observed in the captive group (Supplementary Table 10). In the wild group, we found that Synchytriaceae was significantly more abundant in the summer, while Clavicipitaceae, Parmeliaceae, Enterocytozoonidae, Dacrymycetaceae, and Ramalinaceae were significantly more abundant in the winter (Fig. 3C, Supplementary Table 11, 12). Studies have found that Parmeliaceae37 and Ramalinaceae38 belong to lichen-forming fungi, thus our results indicated that fungi in the gut microbiota may play an important role in facilitating dietary adaptation as the golden snub-nosed monkeys shift their diet from leaves, buds, and fruits in the summer to lichen, tree bark, and pine seeds in the winter24.

The effect of seasonal fluctuation on functional pathways and CAZymes

Based on our analysis of the KEGG database functional annotation, we found significant enrichment in 29 pathways during the winter and in 69 pathways during the summer (Kruskal–Wallis test, p < 0.05) (Supplementary Table 13). During the summer, the gut microbiome of wild golden snub-nosed monkeys was enriched in functional pathways related to the digestion and metabolism of potential plant secondary compounds (Fig. 4A). These pathways include “folate biosynthesis”, “porphyrin and chlorophyll metabolism”, “synthesis and degradation of ketone bodies”, and “terpenoid backbone biosynthesis”. In winter, we found that the “biosynthesis of unsaturated fatty acids”, “longevity regulating pathways”, and “taurine and hypotaurine metabolism” were significantly enriched (Fig. 4A). These pathways are essential for host immune health, energy storage, and physical development.

Fig. 4: Seasonal differences in gut microbiome functional pathways and gene expression activity of golden snub-nosed monkeys.
figure 4

A Clustering heatmap of the abundance of level 3 pathways based on the KEGG database. B Up- and down-regulated pathways in the gut metatranscriptome of golden snub-nosed monkeys, and the significantly lower pathway in the winter metagenomic group. C Clustering heatmap (left) and histograms (right) of the abundance of CAZy gene families. Yellow asterisk corresponds to summer and is significantly higher than winter; blue asterisk corresponds to winter and is significantly higher than summer ∗:p < 0.05. D Up- and down-regulated Metabolic pathways in the gut metatranscriptome of golden snub-nosed monkeys. Each scatterplot illustrates the average gene (DNA) and transcript (RNA) relative abundance for 2883 KOs from all samples. Blue circles correspond to KOs where RNA > DNA; red circles correspond to KOs where DNA > RNA. Marks on the x or y axis margins represent KOs with zero measured abundance in one dataset but non-zero abundances in the other.

Full size image

In the wild snub-nosed monkeys, we identified significant seasonal variation in enzyme levels. In the summer, the levels of GH11, GH77, GT1, GT2, and GT66 were significantly higher than in the winter (p < 0.05). Alternatively, the levels of lichen-degrading enzymes GH16, GH26, GH5, and GH76 in the gut microbiome of wild golden snub-nosed monkeys were significantly higher in winter compared to summer (p < 0.05). This supports field observations that lichen is an important food source for golden snub-nosed monkeys in winter (Fig. 4C).

We also examined the expression of functional genes in the golden snub-nosed monkey gut microbiome. In the winter, metatranscriptomic analysis revealed that gene expression in the gut microbiome of wild golden snub-nosed monkeys was downregulated in metabolism-related genes (Fig. 4D), including those associated with Lysine biosynthesis, Pyrimidine metabolism, Peptidoglycan biosynthesis, Terpenoid backbone biosynthesis, Folate biosynthesis, Pentose and glucuronate interconversions, the Pentose phosphate pathway and Thiamine metabolism (Fig. 4B). These pathways were also significantly lower in winter based on our metagenomic analysis, indicating a consistent relationship between gene abundance and gene expression (i.e., significantly downregulated genes in the metagenome also show significantly decreased expression in activity in the metatranscriptome). In addition, K02794 [EC:2.7.1.191] and K02800 [EC:2.7.1.197] were found to be significantly upregulated in the winter metatranscriptome of wild golden snub-nosed monkeys (Fig. 4B). These genes primarily participate in encoding enzymes involved in the degradation and digestion of mannitol and mannose, which are major components of lichens39,40, this is consistent with our findings in the CAZymes analysis.

Using carbon cycling and metatranscriptomic analyses to compare the methane metabolism of the golden snub-nosed monkey in different seasons

Due to the critical role of carbon in metabolism41, we constructed a carbon cycling process within the gut microbiome of the golden snub-nosed monkeys (Fig. 5A). In the wild summer group, we observed that the abundance of Fermentation (pyruvate -> formate pf1D and formate -> CO2 & H2 fdh) and Methanogenesis (acetate -> methane cdhCDE, mcrABG, and methanol -> methane mtaABC) was significantly higher than in winter (p < 0.05) (Fig. 5B). In winter, the processes of Gluconeogenesis (fbp, pck) and Fermentation to succinate were significantly higher than in summer (p < 0.05) (Fig. 5B). Furthermore, the metatranscriptomic results from the wild winter group demonstrated significant upregulation of the transcriptional activity of genes related to methane metabolism (Fig. 5C), primarily involving pathways M00356 and M00567 (Fig. 5D). It has been suggested that methanogens can drive metabolism forward by removing the gaseous fermentation by-products, carbon dioxide and hydrogen, implying that the gut microbiome contributes to maintaining the metabolic processes of golden snub-nosed monkeys

Fig. 5: Seasonal differences in the gut microbiome carbon cycle pathways and methane metabolism of golden snub-nosed monkeys.
figure 5

A Relative abundances of the pathways involved in the carbon cycle. The pie chart indicates the relative abundance of each pathway in each metagenomic sample. The size of pie charts represents the total relative abundance of each pathway. CBB, Calvin-Benson-Bassham cycle; rTCA, reductive citric acid cycle; wL, Wood-Ljungdahl pathway; 3HB, 3-hydroxypropionate bicycle; DHc, dicarboxylate-hydroxybutyrate cycle. B Bar plot of the pathway involved in the carbon cycle for different seasons. ∗:p < 0.05, ∗∗:p < 0.01. C Up- and down-regulated genes in Methane metabolism, with the scatterplot illustrating the average gene (DNA) and transcript (RNA) relative abundance. D Reconstruction of the metabolic pathway associated with Methanogenesis in the gut microbiome of golden snub-nosed monkeys. All K numbers and EC numbers were obtained from the KEGG database.

Full size image

Discussion

In this study, we constructed a genomic catalog comprising 3602 MAGs of wild and captive golden snub-nosed monkeys. After dereplication, we obtained 578 NR-MAGs, among which 442 MAGs could not be assigned to any known reference genome in the GTDB-Tk database. This finding is consistent with previous large-scale studies reconstructing the gut microbiome genomes of non-human primates and highlights the underexplored microbiome diversity in these hosts42. Collectively, our results revealed substantial “dark matter” in the gut microbiome of non-human primates, which could emphasize the value of primate-focused metagenomic research in significantly broadening our understanding of microbiome diversity and function beyond existing reference databases. Furthermore, the MAG-based approach also enhanced our ability to link functional genes with specific microbiota taxa, moving beyond traditional correlation-based inferences. Although MAG-based approaches have greatly advanced microbiome research, they still have certain limitations43. For example, low-abundance taxa are often underrepresented, and the fragmentation of genomes, especially those derived from wildlife samples, poses additional challenges. Therefore, we conducted rigorous quality assessments to retain only high- and medium-quality MAGs for taxonomic and functional analyses. This quality control step was crucial for improving genome completeness, reducing contamination, and enhancing the reliability of microbiota classification and gene-function mapping.

In their natural habitat, golden snub-nosed monkeys encounter considerable seasonal variation in the distribution and availability of food resources, leading to major shifts in diet44. Our study confirmed that the gut microbiome of wild golden snub-nosed monkeys exhibits plasticity in its response to seasonal fluctuation in diet and ambient temperature in a multidimensional manner, encompassing community structure, microbiota taxa, functional genes, and expression activities. As a highly folivorous foregut fermenting primate inhabiting high-latitude regions, wild golden snub-nosed monkeys face significant seasonal challenges that require them to alter their diet and adjust their energy requirements, especially during periods of extreme cold winter temperatures. During winter, the snub-nosed monkey's diet was dominated by lichen, pine seeds, and bark. During summer, their diet was dominated by leaves, buds, and fruits. In contrast, the diet of captive golden snub-nosed monkeys is similar across seasons (Supplementary Table 3), resulting in a more stable and consistent microbiome. Previous research indicated that wild golden snub-nosed monkeys experienced significant seasonal variation in their diet, with abundant and diverse food available in summer, while their diet became more limited and monotonous in winter44.

Our study explored differences in host-microbe interactions between wild and captive golden snub-nosed monkeys, revealing the response mechanisms of the gut microbiota to seasonal fluctuations in the host’s dietary diversity and richness. We found that the gut diversity of wild golden snub-nosed monkeys was higher in the summer, which is consistent with previous studies of wild white-faced capuchins (Cebus capucinus imitator)45 and wild geladas17. Increased microbial diversity may contribute to more efficient nutrient uptake and promote the health of wild primate populations. A study of the gut microbiota of snails found that as the degree of HoS increases, the diversity of the microbial communities decreases46. These results support a strong interaction mechanism between host diet and host gut microbiome.

With the aid of more precise sequencing methods, we found that gut microbiome composition and function exhibited plasticity in response to seasonal changes in food consumption, at both the community level and in specific functional genes and carbohydrate enzymes. From the perspective of functional genes, we found that the microbiome community of wild golden snub-nosed monkeys during the summer aided in plant compound metabolism (Fig. 4A), enhancing the host’s ability to digest leaves and degrade plant secondary compounds. This plasticity enabled the monkeys to consume large amounts of leaves during the summer, helping them accumulate fat reserves essential for surviving winter conditions26. Similarly, studies on geladas have shown that their gut microbiome can assist in digesting large amounts of carbohydrates from underground plant parts during the dry season17. Moreover, during the summer when female golden snub-nosed monkeys are nursing infants, they require an extremely efficient process for denaturing plant secondary compounds that may be harmless to adults but potentially toxic to their nursing infants47.

During the winter, we identified an abundance of lichen-forming fungi such as Parmeliaceae37 and Ramalinaceae38 in the gut microbiota of the wild golden snub-nosed monkeys (Fig. 3C). This coincided with an increase in lichen consumption during the winter. These lichen fungi are rich in complex polysaccharides, such as mannopolysaccharides and lichenin, which require an extended period of fermentation for digestion and are generally considered to be of low nutritional quality39,40,41. We found that the gut microbiome of golden snub-nosed monkeys is rich in lichen-degrading enzymes during winter, with GH16 coding for licheninase enzymes, and GH26, GH5, and GH76 coding for mannanase enzymes (Fig. 4C). In addition, K02794 [EC:2.7.1.191] and K02800 [EC:2.7.1.197], which are involved in the degradation and digestion of mannitol and mannose, also exhibited high expression in winter (Fig. 4B). This suggests that the gut microbiota of golden snub-nosed monkeys can help their host digest complex polysaccharides during the winter through lichen-degrading enzymes. This strategy is similar to that employed by Tibetan macaques that rely on their gut microbiome to degrade glycans derived from cellulose and hemicellulose in winter and geladas that efficiently digest starches present in underground plant parts during the dry season16,17. Moreover, lichens are used in traditional medicine across various cultures worldwide and are believed to possess antiseptic and antibacterial properties48. Thus, it is possible that lichens play a role in the golden snub-nosed monkey’s self-medication. We note that our wild golden snub-monkey population was provisioned with a small amount of food throughout the year. This supplemental feeding, which may have reduced their reliance on natural foods, suggests that wild non-provisioned individuals are likely to exhibit even greater seasonal variation in diet and microbiome composition (see Methods for information on the provisioning protocol of our study animals).

In addition to adapting to seasonal variation in the nutrient composition and digestibility of their diet, maintaining energy balance during the winter is also crucial for the golden snub-nosed monkeys. Studies have demonstrated that golden snub-nosed monkeys lose approximately 14% of their body weight (primarily fat) in response to the energy requirements of remaining thermoneutral during the long and cold winter26. We found that the gut microbiome of wild golden snub-nosed monkeys in the winter was enriched with bacteria of the family Lachnospiraceae, which was positively correlated with leaf consumption in the Skywalker hoolock gibbon (Hoolock tianxing)49. During the rainy season, when grasses dominated their diet, the gelada gut microbiome was enriched in Lachnospiraceae17. In the present study, based on the Lachnospiraceae species genomes we assembled, we found that the Lachnospiraceae was enriched with functional genes enabling efficient utilization of host fat stores for energy production, including the tricarboxylic acid (TCA) cycle50, fat digestion and absorption51, fatty acid degradation52, oxidative phosphorylation50, phenylalanine metabolism53, and tryptophan metabolism54 (Fig. 4B). Similar results have been observed in Brandt’s voles (Lasiopodomys brandtii). Huddling voles exposed to cold stress exhibited upregulation of Lachnospiraceae abundance compared to solitary individuals, which resulted in higher concentrations of short-chain fatty acids (SCFAs) to cope with the host’s cold stress55. In addition, several studies of the mammalian gut microbiome (i.e., brown bears, Ursus arctos11, Tibetan macaques16, and geladas17) indicate that the microbiome can assist hosts in efficiently maintaining energy balance in response to seasonal challenges in food availability and environmental conditions. In our study, wild golden snub-nosed monkeys were found to exhibit marked seasonal shifts in microbiome composition and function, despite receiving a small amount of provisioned foods.

Finally, a recent study of humans found that many taxa of microbiota of the family Lachnospiraceae play important roles in bile acid conversion, short-chain fatty acid production, and antibiotic production56. In this regard, our results serve to deepen our understanding of the possible functions of the Lachnospiraceae in lipid metabolism, and for next-generation probiotics56.

In conclusion, we reconstructed a total of 578 MAGs from wild and captive golden snub-nosed monkeys across seasons and environmental conditions (Fig. 1C). We found significant differences in the gut microbiome of wild golden snub-nosed monkeys that appear to closely correspond with both seasonal shifts in diet and periods of cold stress that occur during the winter in their high elevation habitat. Moreover, a pairwise comparison of metagenomic and metatranscriptomic data revealed that pathways associated with lichen degradation (Fig. 4B) and methane metabolism (Fig. 5C) were upregulated during winter, potentially facilitating more efficient food digestion and energy balance needed to remain thermoneutral.

Overall, this study presents a comprehensive, multi-omics investigation of the gut microbiome of an Endangered species of primate across distinct seasonal and environmental conditions. By integrating metagenomic and metatranscriptomic analyses, we reveal how microbiome community composition and function exhibit ecological plasticity in response to seasonal shifts in diet. These findings not only enhance our understanding of gut microbiome dynamics but also have broader implications for primate ecology and conservation. An expanding body of research posits that the gut microbiome is an essential contributor to host nutrition, detoxification of plant secondary compounds, enhances energy metabolism, and plays a crucial role in its host’s ability to adapt to changing environmental conditions. Our study demonstrated that, through the use of multiple omic approaches, regular monitoring of the fecal gut microbiome can serve as an important tool in the field of conservation biology to assess wild animal population health and to gain a more detailed understanding of host-microbe interactions under changing environmental conditions. Finally, host-microbe co-evolution represents an important line of research in understanding the ability of both threatened and non-threatened species to adjust to marked shifts in food availability and environmental perturbations associated with climate change and extreme weather events that characterize the Anthropocene.

Methods

Sample collection

During 2021 and 2022, we collected fresh fecal samples from the wild population of golden snub-nosed monkeys inhabiting Shennongjia National Park (SNJ), Hubei Province, China. The natural diet of the golden snub-nosed monkeys at this site is supplemented twice per day (totaling ∼40 kg or 0.5 kg per monkey), primarily with sweet potatoes, apples, and peaches. The amount of daily provisioned food was 20 kg of sweet potatoes, 10 kg of apples and 10 kg of peaches in summer, and the food was supplemented with 30 kg of sweet potatoes, 5 kg of apples, and 5 kg peaches in the winter, which can provide approximately 10–20% of their daily energy intake (Supplementary Table 2).

We collected 29 fecal samples from the wild population in winter Dec 2020–Jan 2021.1) and 17 fecal samples in summer (June 2021–July 2021). Based on individual identification, we were able to obtain both winter and summer fecal samples for 12 golden snub-nosed monkeys. These samples were designated as the “same individual group” (Fig. 1B). Also, information on sampling location, time and individual age (adult vs. juvenile) was recorded during fecal collection.

In addition, we collected fecal samples (ten samples in winter and seven samples in summer) from captive golden snub-nosed monkeys in Beijing Wildlife Park. For these individuals we integrated dietary records from the wildlife park with nutritional information from the Center for Food Safety (https://www.cfs.gov.hk/) to calculate the nutritional composition of their diet (Supplementary Table 3). Given changes in the number of golden snub-nosed monkeys available to zoo visitors, we were not able to collect winter and summer fecal samples from the same captive individual. All samples were temporarily preserved and transported to the laboratory in dry ice, then subsequently stored in a −80 °C freezer before nucleic acid extraction. The procedures and protocols of all experiments in this study have been approved by the Institute of Zoology, Chinese Academy of Sciences

Based on the sampling season and habitat conditions, the samples were divided into four groups: Wild Summer Group, Wild Winter Group, Captive Summer Group, and Captive Winter Group.

DNA and RNA extraction, and sequencing

Following the manufacturer’s instructions, microbiome genomic DNA was extracted using a cetyltrimethylammonium bromide (CTAB) reagent kit (Macklin, China). To remove RNA contamination, RNase A (omega BIO-TEK, USA) was added and incubated at 37 °C for 15 min. The concentration, purity, and integrity of the extracted DNA were assessed using an Agilent 5400 Fragment Analyzer (Agilent Technologies, USA), and a total of 63 qualified samples were selected for metagenomic sequencing. The NEB Next Ultra DNA Library Prep Kit (NEB, USA) was used to construct a single library, and DNA sequencing was carried out using the 2 × 150 bp paired-end read protocol on the Illumina NovaSeq 6000 platform (Illumina, USA). The amount of raw data obtained from each sample was at least 6 Gb.

Compared to metagenomics, metatranscriptomics provides a more accurate interpretation of the functional activity of the gut microbiome. Due to RNA’s susceptibility to degradation, only 22 of the 29 winter fecal samples from the wild monkeys met the quality standards for metatranscriptomic sequencing and were included in this analysis. We used the BioTeke RP8001 Fecal Total RNA Rapid Extraction Kit (BioTeke, China) to extract total RNA from fecal samples. Agarose gel electrophoresis was used to evaluate RNA integrity and to detect potential genomic DNA contamination; a NanoPhotometer spectrophotometer (Implen, Germany) was used to assess RNA purity based on OD260/280 and OD260/230 ratios; a Qubit 2.0 Fluorometer (Life Technologies, USA) was employed for accurate quantification of RNA concentration; and an Agilent 2100 Bioanalyzer (Agilent Technologies, USA) was used to precisely determine RNA integrity. Samples that passed quality inspection were treated with the TIANSeq rRNA Depletion Kit (TIANGEN BIOTECH, China) to remove rRNA and enrich for mRNA. Next, a standard method was used to prepare the RNA library. Specifically, fragmented mRNA was used as the template, and first-strand cDNA synthesis was performed using M-MuLV reverse transcriptase with random hexamer primers. Subsequently, the RNA strand was degraded using RNase H, and the second-strand cDNA was synthesized using DNA polymerase I and dNTPs. The resulting double-stranded cDNA was purified, followed by end repair, A-tailing, and adapter ligation. Approximately 200 bp fragments were selected using AMPure XP beads. The libraries were then amplified by PCR and further purified using AMPure XP beads. Sequencing was performed on the Illumina NovaSeq 6000 platform (Illumina, USA) using a paired-end 150 nt (PE150) high-throughput sequencing strategy, after the constructed libraries had passed the quality control process. Each sample generated at least 5 Gb of raw data.

Metagenome assembly, genome binning, and taxonomy annotation

To ensure the accuracy and reliability of the subsequent analyses, we pre-processed the raw sequencing data. The original data were filtered using Trimmomatic (version 0.39, parameter: ILLUMINACLIP: adapters_path:2:30:10) to remove adapter sequences57. Trimmed sequences with a quality score below 20 (parameter: SLIDINGWINDOW: 4:20) and sequences shorter than 50 bp (parameter: MINLEN:50) were eliminated. This process results in higher-quality and cleaner reads. To remove host contaminants, we used Bowtie 2 (version 2.3.5.1, parameter: --very-sensitive)58 to map the reads to the Rhinopithecus roxellana genome (Genome assembly: Rrox_v1 GCA_000769185.1).

The host-removed clean reads of each sample were individually assembled into contigs using MEGAHIT (version1.1.3)59. Metagenome binning into MAGs was then conducted using MetaBAT2 (version2.12.1) (parameters: minimum contig length to bin -l 1500)60. The dRep (version 3.5.0) was used to dereplicate bins generated from different samples and ensure representatives and diversity of all MAGs61. Completeness and contamination of the final bins were determined using CheckM (parameters: dRep dereplicate -comp 80 -conn 10)62. MAGs annotation was carried out using GTDB-tk (version 2.3.2) (parameters: gtdbtk classify_wf, --extension fa --skip_ani_screen) and used to construct a species phylogenetic tree63. The phylogenetic tree of the bins was then visualized using tvBOT64. Using coverM (version 0.7.0) ((https://github.com/wwood/CoverM), the non-redundant MAGs were mapped back to clean reads to calculate abundance, with statistical units in count and Reads Per Kilobase Million (RPKM) (parameters: coverm genome -m count, -m rpkm). Gene prediction was performed on the obtained bins using Prokka (version 1.11)65. KEGG (Kyoto Encyclopedia of Genes and Genomes) database annotation was conducted using DIAMOND BLASTP (version 2.1.8)66 to obtain functional information at the bin level, with a primary focus on the families Oscillospiraceae and Lachnospiraceae.

Given that reconstructing the eukaryotic genome (particularly fungi) from host fecal samples is highly challenging67,68 and considering the crucial role fungi play in the environmental adaptation of golden snub-nosed monkeys, taxonomic annotation methods based on reads offer valuable insights into the classification and abundance of fungi within these samples. Therefore, we utilized Kraken 2 (version 2.0.7-beta) to generate classification tags for the metagenomic DNA sequences of the clean reads69. After classification, we used Bracken (Bayesian Re-estimation of Abundance after Classification with Kraken) (version 2.0) to estimate the abundance of each sample at various phylogenetic levels, including phylum, class, order, family, genus, and species70, and only retained the result related to fungi (k_Fungi). The workflow of data processing is shown in Fig. 1A.

Metagenomic and metatranscriptomic function annotation

The obtained clean reads were compared to the UniRef90 database using DIAMOND (version 0.8.22)66. We then employed the HUMAnN3 (version 3.6)71 hierarchical search strategy to quantify the abundance of KEGG Orthology (Kyoto Encyclopedia of Genes and Genomes, KO) and CAZy families annotated from the KEGG and CAZy (Carbohydrate-Active enZYmes) databases, based on the relationship between UniRef90 ID and these databases72. The functional gene abundance was then converted into relative abundance in units of copies per million (CPM) using the ‘humann_renorm_table’ script provided by HUMAnN3.

For metatranscriptomic data, RNA clean reads were obtained using the same quality control filter method as for the metagenome. These reads were then indexed with Bowtie258, based on the ChocoPhlan database (v 0.1.1). Using HUMAnN371 software we calculated gene families, path abundance, and path coverage. We then mapped the resulting data onto the UniRef90 and KEGG databases72. We derived the abundance of functional gene expression in the golden snub-nosed monkeys’ gut microbiome using the same methodology for the metagenomic data. The workflow of data processing is shown in Fig. 1A.

Co-occurrence network analysis

Co-occurrence networks for the gut microbiome of golden snub-nosed monkeys were constructed based on MAGs abundance. We used R (version 3.8) to conduct Spearman’s correlation calculations on the MAGs. The statistically robust correlations were used in the analysis of co-occurrence networks under the P value > 0.05 and |correlation coefficient | > 0.7. Gephi (version 0.10.1) software was used to calculate the network diagram parameters and visualize the network73.

Microbial Assembly Mechanism Analysis

According to Ning et al. (2020), “infer community assembly mechanisms using the Phylogenetic-bin-based null model analysis” (iCAMP)34 were applied to test the gut microbial community data of the golden snub-nosed monkeys. We aimed to determine the relative importance of stochastic and deterministic processes in microbial community assembly, and to further classify them into different subprocesses. Specifically, a pairwise βNRI value < −1.96 or > 1.96 indicates HoS or heterogeneous selection (HeS), respectively. The fraction of pairwise comparisons with |βNRI | ≤ 1.96 and RC > 0.95 indicates DL, while |βNRI | ≤ 1.96 and RC < −0.95 represents homogenizing dispersal (HD). The remaining |βNRI | ≤ 1.96 and |RC | ≤ 0.95 are used to identify the influence of drift (DR) or undominant processes.

Enterotype classification

Enterotype clustering was conducted at the bin level. Based on the Jensen-Shannon distance between samples, clustering was performed using the Partitioning Around Medoids (PAM) algorithm (R package). The optimal number of clusters was determined using the Calinski-Harabasz (CH) index74.

Statistical Analysis

In this study, we used vegan (R package) to compute and assess alpha diversity richness and evenness of microbial communities in each sample. We carried out differential analyses using the K-W dunn.test (R package) to compare across groups. Beta diversity was assessed using principal coordinates analysis (PCoA) based on Bray-Curtis distances to compare the differences in gut microbiota structure between seasons and environmental conditions. Differences in the taxonomic abundance of microbiota, functional genes, functional pathways, and carbohydrate enzyme families across environmental conditions and seasons were analyzed using the Kruskal-Wallis test.

To eliminate different seasonal microbiome variations caused by differences in host genetic backgrounds, we employed the following statistical methods: We conducted a Kruskal-Wallis test to identify differentially abundant taxa between all winter samples (N = 29) and summer samples (N = 17), with FDR correction applied to P-values. Effect sizes (η² and Cohen’s f²) and statistical power were also calculated to assess the robustness of the results. Additionally, we performed the same analysis of the samples from the same individuals group (N = 12) across different seasons. We compared the results obtained from the two analyses to identify the shared differential microbiota taxa (Supplementary Fig. 1). This provided a level of control not commonly found in wild field studies, allowing us to minimize the variability caused by different individuals represented in the two seasons.

For the expression of functional genes and to better approximate normality, we performed an arcsine square root-transformation on the relative abundance values of functional genes obtained from the annotation of metagenomes and metatranscriptomes to variance-stabilize data. Then, we used the t test and fold change to compare differences in gene expression. The resulting data were visualized as scatter plots using R (ggpubr)75. For all analyses, P value was set at 0.05.

The analysis of key genes involved in carbon metabolism is based on the DiTing software76. Statistical tests of pathways involved in carbon metabolism were performed using Kruskal-Wallis test.