Sex and social group altered the gut microbiome and fecal metabolome in the critically endangered Yangtze finless porpoise (Neophocaena asiaeorientalis asiaeorientalis)

Abstract

Factors like sex, diet changes, hormone levels, and stressors disrupt animals' symbiotic bacterial communities. Maintaining healthy bacterial communities is particularly challenging for social species, as group membership, social relationships, microbial transfer, and social stressors influence their microbiotas. This study investigated the influence of sex and social dynamics on the gut microbiome and associated metabolites in the captive Yangtze finless porpoise (YFP), employing 16S rRNA gene sequencing and ultra-high-performance liquid chromatography with tandem mass spectrometry–based metabolomic analyses. The present study reveals that sex and social grouping, that is, male–male (MM), female–female (FF), and male–female (MF) groups, significantly influence the alpha and beta diversity in the captive YFP. The phylum Firmicutes were increased considerably in the FF social group, while Proteobacteria, Cyanobacteria, and Fusobacteriota were significantly increased in the MM group, while Desulfobacterota were risen considerably in the MF group. The genera Macrococcus, Clostridium_sensu_stricto_13, and Cetobacterium were considerably raised in the MM group, Paeniclostridium and Turicibacter were substantially raised in the FF group, while the genus Peptostreptococcaceae were substantially raised in the MF group. The current research also presented significant metabolite variations in the sex and social groups which significantly altered the metabolic pathways such as bile secretion, glycerophospholipid metabolism, protein digestion and absorption, citrate cycle, and carbohydrate digestion in the captive YFPs. Additionally, the research identified a significant correlation between the gut microbiome and fecal metabolome across different sex and social groups. In conclusion, this research highlights the connection between changes in fecal microbiota and host metabolism in captive YFP. It shows how sex and social group dynamics affect both metabolic and bacterial variations, offering valuable insights for improving health and social welfare management in captive YFPs.

Graphical Abstract

Graphical Abstract.

Gut microbiota is essential for managing the host's immunity, intestinal health, nutritional absorption and digestion, and other physiological processes (Fan and Pedersen 2021; Nathan et al. 2021). Various factors, such as diet, living environment social interactions, and physiological conditions, can influence the composition of intestinal microbiota in animals (Palmer et al. 2007; Shah et al. 2025a, 2025b). However, social interactions significantly predict gut microbiota composition in certain host species, outperforming temporal differences in diet (Degnan et al. 2012; Perofsky et al. 2017; Rudolph et al. 2022). Social interactions among group members are crucial for maintaining beneficial host-associated microbes within a specific habit or season (Kuthyar et al. 2019; Sarkar et al. 2020). Group-living species encounter distinct social factors that impact their microbiota, including shifts in group composition, intra-group relationships, and microbial exchange among individuals. In adult yellow baboons, alteration in the gut microbiome has been related to interaction rates, highlighting the importance of direct physical contact among social partners for the transmission of gut microbial species. This relationship persists even when accounting for diet, kinship, and shared environmental factors (Tung et al. 2015). Microbiota communities in semi-feral Welsh ponies were found to differ based on spatial structure and specific social interactions within groups, such as female–female (FF) and mother–offspring relationships (Antwis et al. 2018). Additionally, laboratory studies have established a link between social stressors and microbiota alterations, with prolonged aggression and maternal separation leading to microbiota changes in mice and infant rhesus monkeys (Bailey and Coe 1999; Bailey et al. 2010). Social interactions among individuals can enhance the similarity of their gut microbiome composition (Song et al. 2013; Turroni et al. 2017).

Beyond social factors, sex differences have also been identified as a key variable influencing gut microbiota, although its precise role remains unclear (de la Cuesta-Zuluaga et al. 2019; Ma and Li 2019; Zhao et al. 2019). Research in mice has shown that genetic background has a greater influence on gut microbiota than sex differences (Kovacs et al. 2011). Among 69 factors associated with gut microbiota variation, sex ranked 10th in contribution weight (Falony et al. 2016). Studies exploring sex-specific relations between the microbiome and fat distribution indicate the presence of sex-related alterations in gut microbiota (Min et al. 2019). Similarly, a latest study on wild blue sheep revealed a correlation between sexual dimorphism and intestinal microbiota composition (Zhu et al. 2020). Additionally, young women exhibit higher microbial diversity than men, potentially due to earlier sexual maturity (Harada et al. 2020; Ma et al. 2020). Research on spotted hyenas has demonstrated that the microbial communities within their anal scent glands exhibit significant variations influenced by the reproductive state, host's sex, and social groups (Theis et al. 2013). Current research generally indicates that sex exerts limited influence on the gut microbiota composition of marine mammals (Bates et al. 2023); however, emerging studies have identified potential sex-specific variations in intestinal microbial communities, which may be associated with physiological distinctions between male and female wild harbor seals (Pacheco-Sandoval et al. 2022).

The Yangtze finless porpoise (YFP) (Neophocaena asiaeorientalis asiaeorientalis), with an estimated wild population of only 1,012 individuals, making it a highly threatened freshwater cetacean (Huang et al. 2020). The YFP has been identified as a distinct species, exhibiting reproductive isolation from other 2 known populations of finless porpoises (Zhou et al. 2018). The YFP stands as the sole surviving cetacean species in the Yangtze River ecosystem following the functional extinction of the Baiji (Lipotes vexillifer) (Turvey et al. 2007). Stewart et al. (2017) underscore the importance of sustained environmental conservation initiatives that integrate both in situ and ex situ reserve strategies to protect YFP populations. Effective conservation of YFPs necessitates specialized rescue operations, comprehensive health management, and targeted disease treatment within both in situ and ex situ reserves. Recognizing its critical status, China has designated the YFP as a first-class protected species under the Key State Protected Wild Animals list (Wang et al. 2023). The YFP are grouped into several social groups in captivity because of their role in breeding and caring for pregnant females. Although social grouping may induce behavioral stress in captive individuals (Serres 2019; Serres et al. 2020a, 2020b, 2021; Shah et al. 2024a), its impact on microbiota–metabolome interactions remains uncharacterized. Therefore, the current research employs high-throughput sequencing and ultra-high-performance liquid chromatography with tandem mass spectrometry (UHPLC-MS/MS) methodologies to investigate sex and social group-related alterations in the gut microbiota and metabolic profiles of the captive YFPs. Studying these variations is made possible by the controlled environment and similar diets of captive animals. This study may offer a theoretical basis for assessing intestinal health and enhancing captive YFP management.

Materials and methods

Sample collection and study animals

Fecal samples were taken from both male and female YFPs at the Yangtze Cetacean Breeding and Research Centre in Wuhan, China. The YFP are grouped into male–male (MM), female-female (FF), and male–female (MF) social groups in captivity because of the breeding and care of pregnant females. Animal and sampling details are provided in Supplementary Table S1, which includes the 27 fecal samples collected from the 3 animal groups (10 samples from MM and MF animal groups, and 7 samples from the FF animal group). Most samples were collected in triplicate. A closed-circuit filter recycles the pool's water 10 times a day, and during the research period, the water's temperature hovered at 22–24 °C. A variety of fish species, including crucian carp (Carassius auratus), common carp (Cyprinus carpio), and sharp belly (Hemiculter leucisculus), which made up 6–8% of the animals' body weight, were fed to them every day. During the sample procedure, no antibiotics or drugs were administered to any of the animals, and they were all in good health. Fecal samples were obtained from each animal using operant conditioning techniques during a feeding session in the afternoon (Hao et al. 2019). A trainer gently tickled the animal's anal area and encouraged spinning around the belly to collect fecal samples. In order to get fecal samples, a trainer softly tickled the animal's anal region and encouraged it to spin around its abdomen. After noting the date and animal ID, the samples were immediately frozen and stored at −80 °C for later analysis.

DNA extraction and 16S rRNA gene amplification

The QIAmp PowerFecal Pro DNA Kit (Qiagen, Hilden, Germany) was used to extract DNA from 250 mg of each fecal sample in accordance with the manufacturer's protocols following optimization. The purity and concentration of DNA were assessed utilizing a NanoDrop 2000 spectrophotometer and agarose gel electrophoresis. To amplify the genomic DNA, polymerase chain reaction (PCR) was conducted with forward (338F: 5′-barcode-ACTCCTACGGGAGGCAGCA-3′) and reverse (806R: 5′-barcode-GGACTACHVGGGTWTCTAAT-3′) primers, specifically targeting the hypervariable V3–V4 region of the bacterial 16S rRNA gene, as previously outlined (Shah et al. 2024a, 2024b). Purification and PCR amplification were carried out in accordance with Shah et al. (2024a).

The purified amplicons were sequenced using the Illumina NovaSeq6000 system (Illumina, San Diego, CA, USA) after being adjusted to an equivalent concentration, according to Shah et al. (2024a, 2024b). The raw sequencing data were processed using DADA2 within QIIME2 pipeline (version 2019.10) with default settings (Callahan et al. 2016; Bolyen et al. 2019). Following procedures outlined in recent research (Shah et al. 2024a, 2024b), the resultant reads were then taxonomically identified and matched to the Silva database (version 138.1, 16S rDNA) (Quast et al. 2012).

Bioinformatic analysis of the 16S rRNA

Primarily, R packages (v4.1.2) and QIIME2 (v2019.10) were used to process the 16S rRNA sequencing data (Bolyen et al. 2019). Unspecified alpha diversity indices of the microbial community were computed for each sample utilizing Mothur software (version 1.30.2) (Schloss et al. 2009). As previously mentioned, QIIME2 (v2019.10) was used to assess the samples' rarefaction curves (Bolyen et al. 2019; Shah et al. 2024a, 2024b). To assess the structure of the microbial community, principal coordinate analysis (PCoA) was conducted using the Bray–Curtis dissimilarity metrics, implemented through the Vegan package (version 2.5-3). The LEfSe (Linear Discriminant Analysis Effect Size) method was employed to identify differentially abundant bacterial taxa across groups, utilizing a significance threshold of P < 0.05 and an linear discriminant analysis (LDA) score cutoff of >3.5, as described by Díaz-Sánchez et al. (2019). The analyses were carried out in R (v4.1.2). Statistical comparisons at the genus and phylum levels were performed utilizing the Wilcoxon rank-sum test (Lam and Longnecker 1983), with P-values adjusted for multiple testing using the Benjamini–Hochberg false discovery rate correction (Benjamini and Hochberg 2000).

Untargeted metabolomics analysis

Using a non-targeted UHPLC-MS/MS metabolomics approach, the fecal metabolites in the sex and various social groups were identified. To extract metabolites, 400 µL of a 4:1 v/v methanol/water solution was utilized, and a 100 µL fecal sample was weighed exactly. After settling at −20 °C, the mixture was treated for 6 min at 50 Hz utilizing a high-throughput tissue crusher. Following this, it was vortexed for 30 s and then ultrasonically treated for 30 min at 40 kHz and 5 °C. Samples were initially kept at −20 °C for 30 min to precipitate protein.

The supernatants were placed in sample vials for LC-MS/MS analysis using the Thermo Fisher Scientific UHPLC-Q Exactive System after being centrifuged at 13,000 × g for 15 min at 4 °C. Peak alignment and identification were performed on raw data using Progenesis Q1 2.3 (Nonlinear Dynamics, Waters, USA). The resulting dataset contained information on mass-to-charge ratio (m/z), retention time, and peak intensity. At least 80% of the metabolic features were retained across all sample groups. After filtering, metabolite characteristics were normalized using the sum normalization method, and samples below the quantification threshold were assigned the lowest detectable metabolite value. During quality control (QC), features having a relative standard deviation >30% were eliminated from the dataset using an internal standard. Metabolic features were identified through high-accuracy biochemical databases such as the Human Metabolome Database (HMDB) (https://www.hmdb.ca) and the Metlin Database (https://metlin.scripps.edu), based on mass accuracy, isotope ratio deviations, and MS/MS fragmentation patterns. Multivariate statistical analyses were performed using the Majorbio Cloud Platform (https://cloud.majorbio.com) with the R package ropls (version 1.6.2).

Statistical analysis

The current study examined the potential connections between gut microbiota and metabolites using the R psych package (Revelle 2020). The Spearman rank correlation was used to filter out microbiota and metabolites having a P-value of <0.05.

Results

Sequencing metadata

Using Illumina MiSeq, 27 fecal samples were sequenced, yielding 1,562,397 (mean: 57,866; range: 47,413–68,594) raw 16S rRNA reads, resulting in 1,424 operational taxonomic units (OTUs). The Shannon rarefaction curves for all samples attained the plateau phase, as determined by read abundance, indicating that nearly all bacterial species were detected (Supplementary Figure S1A). During the study, a total of 1 Kingdom, 26 phyla, 68 classes, 175 order, 304 families, 562 genus, 820 species, and 1,424 OTUs unique features were discovered.

The gut microbiota composition

Three social groups—MM, FF, and MF—were used to classify 1,424 OTUs (Supplementary Figure S1B). Of these, there were 237 OTUs in all groups, 132 in MF, 21 in FF, and 552 in MM (Supplementary Figure S1B).

Alpha and beta diversity of the gut microbiota in the sex and different social groups

Figure 1A–D and Supplementary Table S2, which includes ACE, Chao, Shannon, and sobs, displays the alpha diversity index. The bacterial communities of ACE, Chao, and Sobs showed a substantial alteration between MM and MF (P < 0.05). There was substantial variation in the Shannon index across the 3 social categories. Using the Bray–Curtis (ANOSIM and ADONIS) approaches, the current study examined variations in the composition of fecal microbiota among captive YFPs. Significant alterations in the composition of gut microbiota were detected between the MM, FF, and MF social groups (Fig. 1E, F).

Figure 1.

Alpha diversity in the gut microbiota of the captive YFP was conducted using the following indices at the OTU level: (A) ACE index, (B) Chao index, (C) Sobs index, and (D) Shannon index. MM, FF, and MF groups in the captive YFP were tested for significant differences using the Kruskal–Wallis H test. The PCoA using the Bray–Curtis distance (E) ANOSIM, and (F) ADONIS shows how the structure of the fecal bacterial community of captive YFP changes with sex and different social groups.

Evaluation of species diversity and variances in the gut microbiota

Three social groups' gut microbiotas of captive YFPs were examined, shown in Fig. 2. MM, FF, and MF social groups were identified with the following main phyla: Firmicutes (72%, 95%, and 92%), Actinobacteria (19%, 4%, and 6%), Proteobacteria (3%, 0%, and 1%), Unclassified_k_norank_d_Bacteria (3%, 1%, and 0%), Cyanobacteria (2%, 0%, and 0%), and Fusobacteriota (1%, 0%, and 0%) (Fig. 2A). As shown in Fig. 2C, the Kruskal–Wallis H test indicates that Firmicutes were considerably abundant in the FF social group, but Proteobacteria, Cyanobacteria, and Fusobacteriota were significantly higher in the MM group. The most prevalent bacterial genera of captive YFP in the 3 distinct social groups MM, FF, and MF were Peptostreptococcaceae (18%, 51%, and 54%), Romboutsia (25%, 30%, and 11%), Clostridium_sensu_stricto_1 (9%, 10%, and 14%), Mycobacterium (16%, 3%, and 5%), Ureaplasma (4%, 1%, and 4%), Paraclostridium (5%, 1%, and 6%), Paeniclostridium (0%, 1%, and 0%), and Terrisporobacter (2%, 1%, and 1%) (Fig. 2B). The Kruskal–Wallis H test revealed that the MM group had considerably higher levels of Clostridium_sensu_stricto_13, norank_o__Chloroplast, Macrococcus, and Cetobacterium. In contrast, the MM group had significantly higher levels of Turicibacter and Paeniclostridium, as seen in Fig. 2D. The species composition and community structure research showed substantial alterations in the gut microbiota of captive YFP between social groups and sexes.

Figure 2.

Alterations in the diversity of gut microbiota at the (A and B) phylum and genus levels among the sex and social groups. The top 10 bacterial phyla and genera’ relative abundance is shown in the histogram. The graphic's vertical coordinates illustrate the relative abundance of several bacterial species. The Kruskal–Wallis H test at the (C) phylum and (D) genus levels represent significant differences across the sex and the 3 social groups.

Differential bacterial biomarkers in the sex and different social groups of captive YFP

The LEfSe analysis, based on an LDA threshold of >3.5 and <P < 0.05, identified significant variations in the relative abundance of microbial taxa among sexes and distinct social groups, as illustrated in Fig. 3A,B. The MM in this study was dominated by the genera Enterococcus and Bacillus, whereas the FF group was dominated by the genera Staphylococcus, Turicibacter, and Clostridium_sensu_stricto_1. The MF group showed a substantial increase in the phylum Fusobacteriota, whereas the genus Cetobacterium was the most significant.

Figure 3.

LEfSe analysis was used to investigate the gut microbiota composition of the captive YFP across different social groups (LDA > 3.5, P < 0.05). (A) The most prevalent taxa in various social groups are displayed by the LDA score histogram. (B) The cladogram shows the gut microbiota's evolutionary distribution.

Fecal metabolome in the sex and different social groups in the captive YFP

The present study identified a total of 542 metabolites in the negative ionization mode and 827 metabolites in the positive ionization mode. Following annotation, 498 metabolites were confirmed in the negative ionization mode and 718 in the positive ionization mode. QC samples demonstrated robust clustering near the origin of the plot, indicating exceptional instrument stability and reproducibility. As illustrated in Supplementary Figure S2A,B, the metabolites of the 3 groups were effectively separated using an unsupervised PCA model. PLS-DA analysis further confirmed distinct metabolic differences among the 3 social groups (Supplementary Figure S2C,D). Model validation demonstrated that the PLS-DA model was well-suited, highly predictive, and provided an excellent fit for the dataset (Supplementary Figure S2E,F). In both negative and positive ion modes, the y-axis intercept was 0.6241, and the Q2 regression line intercept was 0.2146, with R2 consistently exceeding Q2.

Distinct metabolites in the captive YFP among the sex and social groups

Differential metabolites (DMs) associated with sex and distinct social groups were identified based on a significance threshold of P < 0.05, and a variable importance in projection (VIP) score >1.0, as illustrated in Fig. 4A–F. The VIP scores were derived from the OPLS-DA model. Five hundred and twenty-nine DMs were discovered in both the positive and negative ion modes out of 3,544 peaks in total (Supplementary Table S3). As shown in Fig. 4A–C, a total of 363 DMs were identified when comparing MM and FF groups. Among these, 191 DMs, including maslinic acid, boc-val-pro-arg-mca, ptelatoside b, dimp, norophthalmic acid, mangalkanyl glucoside, zalcitabine, trazodone, and pg(18:3(6z,9z,12z)/pgel), exhibited upregulation in the MM group. Conversely, 172 DMs, such as pro leu, glycocholic acid, 25-acetylvulgaroside, 2-(4-chlorophenoxy)-2-methyl-n-(2-phenylimidazo[1,2-a]pyridine-3-yl)propenamide, petrosaspongiolide m, dg(i-19:0/0:0/18:3(10,12,15)-oh(9)), 22-deoxoisocucurbitacin d, boviquinone 4, desglucocheirotoxol, 28-homobrassinolide, deoxycholyphenylalanine, everolimus, ps(pgf1alpha/24:0), and hyodeoxyxholix acid, were downregulated in the FF group. Compared with MM versus MF, 153 DMs were found; among them, 60 DMs (erythromycin, 13,14-dihydro-15-keto-20-ethyl prostaglandin f2alpha, n-desethyl vardenafil, epsilon-viniferin, 3-(3-(n-(2-chloro-3-trifluoromethylbenzyl)(2,2-diphenylethyl)amino)propoxy)phenylacetic acid, mangalkanyl glucoside, pg(18:3(6z,9z,12z)/pge1), typhasterol, and isocupressic acid) were upregulated in the MM and 93 DMs (sulisobenzone, ginsenoside I, sinapic acid sulfate, obscurolide a1, epothilone D, genistein 7-sulfate, combrestatin A4, ganoderic acid F, dg(18:2(9z,12z)/18:3(9z,12z,15z)/0:0), boviquinone 4, dg(18:1(9z)-o(12,13)/0:0/I-19:0), dg(22:5(4z,7z,10z,13z,19z)-o(16,17)/0:0/19:0), cdp-dg(18:0/18:2(9z,12z)), lucidenic acid M, pc(dime(11,5)/20:5(6e,8z,11z,14z,17z)-oh(5)), ganglioside ga2 (d18:1/16:0), 28-homobrassinolide, pi(pgf2alpha/22:4(10z,13z,16z,19z)), 25-acetylvulgaroside, dg(I-19:0/0:0/18:3(10,12,15)-Oh(9)), and cer(d18:1/20:4(6e,8z,11z,14z)-oh(5s))) were downregulated in the MF. Compared with FF versus MF, 262 DMs were identified: among them, 120 DMs (1-[6-(morpholine-4-carbonyl)-2,3-dihydroindol-1-yl]ethanone, triethanolamine, epsilon-viniferin, aniracetam, dg(lte4/a-25:0/0:0), chenodeoxycholic acid phenylalanine, (r)-2-hydroxystearic acid, 22-deoxoisocucurbitacin d, indole-3-acetyl-l-valine, n-lauroyl phenylalanine, petrosaspongiolide m, n-desethyl vardenafil, n(2)-phenylacetyl-l-glutaminate, phenylalanylhydroxyproline, ps(pgf1alpha/24:0), deoxycholylphenylalanine, 2alpha-(3-hydroxypropyl)-1alpha,25-dihydroxy-19-norvitamin d3, lacto-n-biose I, and 2-(4-chlorophenoxy)-2-methyl-n-(2-phenylimidazo[1,2-a]pyridin-3-yl)propanamide) were upregulated in the FF, and 142 DMS (sulisobenzone, ginsenoside I, sinapic acid sulfate, mupirocin, boc-val-pro-arg-mca, combrestatin a4, ser tyr lys asp, trazodone, npc, maslinic acid, and gamma-glutamylglutamine) were downregulated in the MF.

Figure 4.

Potential differentially expressed metabolites (DMs) were identified in captive YFP. (A–C) Volcano plots illustrate DMs in MM versus FF, MM versus MF, and FF versus MF comparisons. The x-axis (log₂FC) represents fold change in metabolite expression across different sex and social groups, while the y-axis (−log₁₀) shows the statistical significance (P-value) of metabolite differential expression. Elevated values correspond to greater significance in expression differences. (D–F) Cluster analysis was conducted to assess DMs across the 3 social groups (MM vs. FF, MM vs. MF, and FF vs. MF). On the right, the VIP bar chart highlights the key metabolites contributing to group differences, while the left panel represents a clustering dendrogram, showing similarities in metabolite expression patterns. Metabolites with closely related branches exhibit similar expression trends. The length of the bars reflects the extent of each metabolite's contribution to group differences, with a default threshold of 1. Greater differences between groups correspond to higher values. A color bar represents the significance of metabolite differences, where darker shades indicate increasing −log₁₀ values and lower P-values. Statistical significance is represented as follows: *P < 0.05, **P < 0.01, and ***P < 0.001.

Analysis of metabolic pathway enrichment

Second-grade KEGG pathways were used to classify metabolites, and the most commonly searched terms for “metabolism” were energy, cofactors, vitamins, amino acid, lipid, other amino acids, biosynthesis of terpenoids and other secondary metabolites, polyketide metabolism, and carbohydrates (Supplementary Figure S3). The current analysis finds significant metabolic pathways within the KEGG database, with statistical relevance indicated by a P-value of <0.05, as illustrated in Fig. 5A–C; Supplementary Table S4. According to the enrichment analysis, the MM and FF groups’ metabolic pathways differed significantly in d-amino acid metabolism, glutamatergic synapse, bile secretion, protein digestion and absorption, ABC transporters, arginine biosynthesis, glycine, serine and threonine metabolism, GABAergic synapse, alanine, aspartate and glutamate metabolism, primary bile acid biosynthesis, glycerophospholipid metabolism, choline metabolism in cancer, proximal tubule bicarbonate reclamation, lipoic acid metabolism, biosynthesis of plant secondary metabolites, kaposi sarcoma-associated herpesvirus infection, Taurine and hypotaurine metabolism, and arginine and proline metabolism. The MM and MF groups had different metabolic pathways kaposi sarcoma-associated herpesvirus infection, Mineral absorption, ABC transporters, Protein digestion and absorption, glycerophospholipid metabolism, retrograde endocannabinoid signaling, d-amino acid metabolism, aminoacyl-tRNA biosynthesis, carbohydrate digestion and absorption, arginine and proline metabolism, brassinosteroid biosynthesis, epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor resistance, alanine, aspartate and glutamate metabolism, NF-kappa B signaling pathway, pathogenic Escherichia coli infection, and autophagy—other. Bile secretion, alanine, aspartate and glutamate metabolism, citrate cycle (TCA cycle), glutamatergic synapse, GABAergic synapse, arginine and proline metabolism, choline metabolism in cancer, glutathione metabolism, diabetic cardiomyopathy, amebiasis, shigellosis, carbon fixation pathways in prokaryotes, Hypoxia-Inducible Factor-1 signaling pathway, sphingolipid signaling pathway, primary bile acid biosynthesis, biosynthesis of cofactors, ABC transporters, glycine, serine and threonine metabolism, and biosynthesis of plant secondary metabolites were all markedly changed in the FF and MF groups, as illustrated in Fig. 5A–C; Supplementary Table S4.

Figure 5.

Enrichment analysis of metabolic pathways based on differentially abundant metabolites across different groups: (A) MM versus FF, (B) MM versus MF, and (C) FF versus MF. *P < 0.05, **P < 0.01, and ***P < 0.001.

Association between the fecal metabolome and gut microbiome

A Spearman correlation analysis was conducted to investigate the association between 20 significantly altered metabolites and the microbiota, as illustrated in Fig. 6. The findings showed a substantial positive correlation (P < 0.05) between Paeniclostridium and pe(pge1/22:2(13z,16z)). Terrisporobacter were substantially positively associated (P < 0.05) with adrenic acid while substantially negatively associated (P < 0.05) with dibutyl sebacate. Romboutsia and Peptostreptococcacea were significantly positively (P < 0.05) linked 7-ketolithocholic acid, cholic acid 1-(4-o-beta-d-glucopyranosyl-3-methoxyphenyl)-3,5-dihydroxydecane, pc(22:5(7z,10z,13z,16z,19z)/18:1(9z)), adrenic acid and tetracosahexaenoic acid, n-3, while these metabolites were significantly negatively correlated (P < 0.05) with Actinomyces, Ureaplasma, Enterococcus, Streptococcus, unclassified_k__norank_d__Bacteria, Mycobacterium, unclassified_c__Bacilli, ZOR0006, norank_o__Chloroplast, Paracoccus, Cetobacterium and Macrococcus. Ureaplasma, unclassified_k__norank_d__Bacteria, Mycobacterium, Clostridium_sensu_stricto_1, and unclassified_c__Bacilli were significantly positively (P < 0.05) linked with ganoderic acid V, cytochalasin B, and taurocholic acid, while dibutyl sebacate metabolite were significantly positively linked with Streptococcus, unclassified_k__norank_d__Bacteria, Mycobacterium, unclassified_c__Bacilli, norank_o__Chloroplast, Paracoccus, Cetobacterium, Macrococcus and Paraclostridium. Unclassified_f__Clostridiacea and Clostridium_sensu_stricto_13 were substantially negatively connected (P < 0.05) with tetracosahexaenoic acid and 17-octadecynoic acid.

Figure 6.

Multi-omics analyses revealed significant associations between gut microbiota and metabolites across different sex and social groups in the captive YFPs. Heatmap showing the link between the gut bacteria and metabolite clusters. Different colors represent the value of the Spearman correlation coefficient. Blue denotes a negative connection, whereas red denotes a positive one. The study reveals a statistically significant correlation between metabolite clusters and gut microbiota, as indicated by the P-values (*P < 0.05, **P < 0.01, and ***P < 0.001).

Discussion

Social group membership and sex are key factors influencing microbiota diversity in wildlife. Previous studies have shown that group membership significantly impacts microbiota composition, with animals in the same social group sharing more similar microbiotas due to common environmental exposures and behaviors (Springer et al. 2017; DeCandia et al. 2021). In primates such as chimpanzees (Moeller et al. 2016) and baboons (Tung et al. 2015), social interactions like grooming are strongly linked to microbiota similarity, independent of genetic relatedness. This pattern is also observed in geladas (Trosvik et al. 2018), where shared social environments and behaviors contribute to microbiota convergence. Host sex further influences microbiota composition. Sex differences in gut microbiota have been noted in several species, including spotted hyenas (Rojas et al. 2020) and primates (Leclaire et al. 2014), often contributed to hormonal, immune, and physiological differences. In cetaceans, Popov et al. (2024) found no significant influence of sex on the gut microbiota of captive dolphins. However, contrasting studies, such as those by Pacheco-Sandoval et al. (2022), suggest that sex-related differences in intestinal microbiota do exist among marine mammals, potentially attributable to physiological variations between males and females. In contrast, the role of social rank in shaping microbiota is less consistent. While social status affects the microbiota in meerkats and sifakas (Leclaire et al. 2017; Greene et al. 2019), it has no significant effect in juvenile spotted hyenas (Rojas et al. 2020) or cheetahs (Wasimuddin et al. 2017). These variations may reflect differences in social dynamics and the specific developmental stages of the animals studied. In lines with these findings, our study on captive YFP showed that both sex and social group membership significantly affected microbiota diversity, with distinct patterns observed in MM, FF, and MF groups. These results further support the idea that social structure and sex are crucial determinants of microbiota composition in captive animals.

The dominance of Firmicutes in the gut microbiota of YFPs (72–95% across groups) mirrors findings in aquatic cetaceans, including captive YFPs and blue whales (Guass et al. 2016; Wan et al. 2016), where Firmicutes are typically associated with energy harvest and fermentation (Flint et al. 2012). For example, in humans, higher Firmicutes-to-Bacteroidetes ratios are linked to increased caloric extraction from food (Turnbaugh et al. 2007). This functional similarity could reflect adaptations to a captive diet, potentially designed to optimize energy metabolism. Specifically, Actinobacteria represented 19%, 4%, and 6% in the MM, FF, and MF groups, respectively, while Proteobacteria accounted for 3%, 0%, and 1% in the same groups. Notably, the relative abundances of these phyla were higher than expected given the controlled diet and uniform environmental conditions, suggesting that factors beyond diet and environment—specifically, sex and social structure—may influence microbiota composition. Proteobacteria, a phylum known for its sensitivity to environmental factors, including diet, has been associated with microbial instability in various mammals (Shin et al. 2015). Its relatively higher presence in the MM group could indicate a more unstable microbiome, as Proteobacteria is often linked to disease states such as metabolic disorders and intestinal inflammation (Morgan et al. 2012; Fei and Zhao 2013). Interestingly, the lower relative abundance of Proteobacteria in both FF and MF groups, despite having the same diet and environmental exposure, suggests that sex and social structure could play a role in maintaining a more stable microbial environment. This implies that social dynamics between the sexes, particularly in mixed-gender groups, may foster conditions that promote a more balanced microbiota. Higher levels of Actinobacteria in the MM group might reflect a more resilient microbiota, as this phylum is generally considered stable and beneficial. However, the significant decrease in Actinobacteria in the FF and MF groups, particularly in the latter, suggests that MF social interactions may create conditions that alter the microbial composition. The study observed that Cyanobacteria were more abundant in MM compared with FF and MF groups. Cyanobacteria are recognized for producing a variety of harmful secondary metabolites that have a significant effect on health. These include endocrine, hepatotoxic, carcinogenic, neurotoxic, and immunomodulatory effects (Adamovsky et al. 2015; Palikova et al. 2015; Moosova et al. 2019). The presence of such toxins can pose serious risks to the health of organisms. For instance, cyanobacteria-induced toxicity has been reported in captive bottlenose dolphins (Tursiops truncatus) in the Florida Keys, emphasizing the potential harm they can cause to marine mammals (Lydon et al. 2022). These findings highlight the need for further investigation into the relationship between social groupings and the composition of microbial communities, as the presence of harmful bacteria like Cyanobacteria could have significant implications for both individual health and group dynamics. The relative richness of Fusobacteriota was observed to be increased in the MM compared with both the FF and MF groups. Fusobacteria were reported in stranded short-finned pilot whales and Indo-Pacific humpback dolphins (Sanders et al. 2015; Bai et al. 2021a). Stranded animals are often exposed to heightened stress which could contribute to an increased abundance of Fusobacteria.

In our study, Peptostreptococcaceae were found to be less abundant in the MM compared with the FF and MF groups. The higher richness of Peptostreptococcus within the gastrointestinal tract of animals may reveal that Peptostreptococcus plays a significant role in their gut microbiota, potentially as a commensal bacterium (Murdoch 1998). However, it is also important to consider that Peptostreptococcus species have been reported to cause a clinical disease in both humans and marine animals (Murphy and Frick 2013; Nielsen et al. 2013), suggesting that under certain conditions, such as stress or compromised health, this bacterium could behave opportunistically. In the present study, Romboutsia species were found to be more prevalent in the MM and FF compared with the MF group. Romboutsia are renowned for their metabolic functions, which include the usage of metabolic end products, anaerobic respiration, and the fermentation of single amino acids (Gerritsen et al. 2019). Previous research has shown that Romboutsia species tend to be more abundant in animals experiencing mild and chronic stress, with their presence being positively associated with depression-like behaviors and anxiety in mice (Sun et al. 2019). This finding aligns with our earlier study, which also observed an increase in Romboutsia abundance in the mother–calf group during stress, compared with the mother-separated group (Shah et al. 2024a). Together, these results suggest a potential link between Romboutsia species and stress-related behaviors, potentially contributing to the physiological and psychological effects observed under such conditions. In this study, Clostridium_sensu_stricto_1 was found to be more prevalent in the MF group compared with the MM and FF groups. Cetaceans, including Pacific white-sided dolphins (Lagenorhynchus obliquidens), beluga whales (Delphinapterus leucas), common bottlenose dolphins (T. truncatus), and short-finned pilot whales, usually contain Clostridium_sensu_stricto in their gut microbiota (Bai et al. 2021a, 2021b). Clostridium, a prevalent genus in cetacean gut microbiomes, has been linked to potential pathogens in unhealthy cetaceans, despite some studies suggesting low virulence (Erwin et al. 2017; Marón et al. 2019). The differential abundance of Clostridium in the MF group may therefore suggest a unique microbial response to the social interactions or stressors specific to this group, which could influence their overall health and microbial composition. In this study, Mycobacterium species were found to be more abundant in the MM compared with the FF and MF groups. Infection by Mycobacterium species has frequently been reported in both wild and captive marine mammals, with studies documenting cases of infection in various cetacean species (Bowenkamp et al. 2001; Wünschmann et al. 2008). The higher prevalence of Mycobacterium in the MM group may suggest a potential association between this microbial shift and specific social or environmental factors within this group, which could influence susceptibility to infection.

Recently, Zierer et al. (2018) demonstrated that the fecal metabolome can provide valuable insights into gut microbial activity and its association with host metabolism. Building on this concept, our study utilized fecal metabolomics to explore the microbial functions within different social groups and sexes of captive YFP. Our findings revealed significant DMs across the various social and sex-based groups, highlighting the intricate relationships between the gut microbiome, host metabolism, and social dynamics in the captive YFP. Maslinic acid, a lipids and lipid-like molecule, has been shown to possess neuroprotective properties, particularly against apoptotic cell death and Aβ protein-induced neurotoxicity (Lozano-Mena et al. 2014; Yang et al. 2015). The MM group showed an upregulation of maslinic acid compared with the FF group. This suggests that social dynamics might play a role in regulating the levels of maslinic acid, potentially influencing neuroprotection. The MM group showed an upregulation of norophthalmic acid metabolite, which is known to function as an intracellular antioxidant (Dulka et al. 2017). The MM group showed a significant increase in the upregulation of Trazodone, an organoheterocyclic compound, compared with the FF group. Trazodone has previously been reported to be used for the treatment of depression and anxiety in both humans and animals (Chea and Giorgi 2017). This suggests that social dynamics, particularly gender-based differences, may influence the expression of trazodone, potentially affecting its role in managing mood disorders. Many animals' bile contains large amounts of glycocholic acid, a bile acid complex that is produced in the liver from bile acids and glycine. It is important for increasing bile production, accelerating fat breakdown, and increasing lipase activity (Feng et al. 2014). Recent studies have revealed its therapeutic properties, including antipyretic, anti-inflammatory, antibacterial, and antioxidant effects (Li et al. 2025). Interestingly, glycocholic acid was downregulated in the FF group compared with the MM, suggesting that social dynamics may influence its expression and potential therapeutic effects. Petrosaspongiolide M was substantially raised in the FF compared with the MM and MF groups. This substance has been demonstrated to dramatically reduce eicosanoid and TNFα in rats and mice, respectively, as well as acute inflammatory reaction brought on by Freund's adjuvant (Posadas et al. 2003). This suggests that social dynamics may influence the expression of petrosaspongiolide M and its anti-inflammatory effects in the FF groups. Everolimus is a phenylpropanoids, and polyketides metabolite and were downregulated in the FF compared with the MM group. Everolimus has been shown to have detrimental effects on depressive and anxiety-like behavior in mice; it also promotes neurogenesis and improves memory performance (Russo et al. 2016). Hyodeoxycholic acid, a lipid and lipid-like molecule, was downregulated in the FF compared with the MM group. This compound has been shown to significantly suppress intestinal cholesterol absorption (Shih et al. 2013), indicating that the downregulation observed in females could have implications for cholesterol metabolism. The MM group showed an increase in the upregulation of epsilon-viniferin, a phenylpropanoid and polyketide metabolite, compared with the MF group. This compound is known for its anti-inflammatory and antioxidant properties (Beaumont et al. 2022). The increased levels in the MM group suggest that social interactions within this group may enhance metabolic pathways related to oxidative stress and inflammation. The study found that the MM group showed a downregulation of the metabolite epothilone D, a phenylpropanoids and polyketides metabolite, compared with the MF group. Epothilone D has been revealed to promote spinal cord repair and neurological recovery (Ruschel and Bradke 2018). The study found that the MM group showed a downregulation of ganoderic acid F metabolite compared with the MF group. This compound has been reported to possess a range of therapeutic activities, including antibacterial, antiviral, antitumor, hepatoprotective, antioxidative, antihypertensive, and antiaggregation effects (Geng et al. 2020). The reduced levels of ganoderic acid F in the MM group may indicate a potential alteration in metabolic pathways related to these health benefits. Aniracetam was higher in the FF than in the MF group. Aniracetam is known to improve learning and memory in animals (Yu and Cai 2003). Deoxycholylphenylalanine was higher in the FF than the MF group and has been shown to reduce bile acid formation, affecting lipid digestion and cholesterol breakdown in mice (Rafferty et al. 2023). Maslinic acid was lower in the MF compared with the FF group and is known for its pharmacological effects, including antitumor, anti-inflammatory, cardioprotective, antiviral, antimalarial, neuroprotective, and antioxidant activities (Yap and Lim 2015).

The study found that DMs significantly influenced KEGG pathways across sexes and social groups. DMs like zalcitabine, oxoglutaric acid, and tetracycline were upregulated in the MM group, while glycocholic acid, deoxycholic acid, cholic acid, and taurolithocholic acid 3-sulfate were downregulated in the FF group. These DMs significantly altered glycerophospholipid metabolism, which is crucial for membrane protein recognition and signal transduction. Dysregulation of glycerophospholipid metabolism is associated with metabolic disorders like cardiovascular disease, hypertension, diabetes, obesity, neurological diseases, and cancer (Qi et al. 2023). When comparing the MM and MF groups, the DMs L-alanine, L-proline, and L-isoleucine were significantly higher in the MM group, affecting protein digestion and absorption pathway. This pathway is important for providing essential amino acids and peptides, which are needed for biological processes like development and growth (Zhao et al. 2024). When comparing the FF and MF groups, the DMs citric acid and oxoglutaric acid were significantly higher in the MF group, affecting the citrate cycle (TCA cycle). The citrate cycle is important for producing key components needed to make nonessential amino acids, lipids, nucleotides, and porphyrins (Owen et al. 2002). Any disruption in this cycle can lead to major changes in cell metabolism, beyond just affecting ATP production (Ryan et al. 2021).

The current study used the Spearman correlation analysis to determine a strong association between fecal microbiota and metabolites. This approach was employed to assess metabolite–microbial relationships in captive YFP across different social and sex groups. The findings provide insights into how social structure and sex might influence the composition of gut microbiota and associated metabolites in this species. The results of this study revealed a significant positive association between Paeniclostridium and pe(pge1:22:2(13z,16z)). Paeniclostridium species are known to be widespread in nature, displaying a variety of physiological characteristics. The pathogenic potential of Paeniclostridium is largely attributed to its production of β-toxin, which exhibits high biological activity and specificity, contributing to the virulence of these bacteria (Gryaznova et al. 2021). These microorganisms are frequently isolated in association with diseases that mimic other forms of clostridiosis, such as sudden death in sheep, acute abomazitis in cows and lambs, hemorrhagic enteritis, and gangrenous lesions in the reproductive tracts of newborn cows (Vidor et al. 2015). The positive correlation between Paeniclostridium and suggests pe(pgel1/22:2(13z,16z)) that this metabolite may play a role in the ecological niche or pathogenicity of Paeniclostridium, potentially influencing its presence or virulence in certain conditions. In the current study, Terrisporobacter was significantly positively correlated with adrenic acid and negatively correlated with dibutyl sebacate. This bacterium is associated with host health, as its higher abundance has been shown to benefit host survival, likely through its production of short-chain fatty acids (SCFAs) in the carbohydrate metabolism pathway. SCFAs, including butyrate, propionate, and acetate, play a critical role in maintaining gut health by supporting the intestinal barrier, reducing inflammation, and serving as an energy source for colonocytes (Wang et al. 2022). These findings suggest that Terrisporobacter may contribute to the overall health of the YFP, underscoring its potential importance in gut microbiome dynamics.

In conclusion, this study demonstrates the significant influence of sex and social grouping on the gut microbiome and its metabolites in captive YFP. The findings show that different social groupings (MM, FF, and MF) lead to distinct variations in the alpha and beta diversity of the microbiome. Specific bacterial groups, such as Firmicutes, Proteobacteria, and Fusobacteriota, were more abundant in certain social groups, highlighting the role of social interactions in shaping microbial communities. Additionally, changes in the microbiome were linked to alterations in key metabolic pathways, including bile secretion, glycerophospholipid metabolism, and carbohydrate digestion. A significant correlation between microbiota composition and metabolic activity further underscores the interconnection between microbial communities and host metabolism. Overall, these findings underscore the importance of considering social dynamics and sex in the management and welfare of captive YFP, offering insights for improving their care by understanding the complex interplay between microbiota, metabolism, and social interactions.

Authors’ contributions

The study was conceived by S.A.U.R.S., B.T., and Y.H., with sample collection carried out by C.W. and F.F. Data analysis was conducted by S.A.U.R.S., M.A., and B.T. The original manuscript draft was written by S.A.U.R.S. Y.H. and D.H. supervised the study and provided funding. Y.H., D.H., J.Z., G.N., and K.W. edited and reviewed the manuscript. The final manuscript has been thoroughly reviewed and approved by all the authors.

Funding

This study was funded by the National Key R&D Program of China (2021YFD1200304; 2022YFF1301604).

Ethics statement

The Institute of Hydrobiology and the Research Ethics Committee of the Chinese Academy of Sciences approved all of the procedures used in the present study (IHB/LL/2022-YFF1301604), guaranteeing that they complied with Chinese regulations and ethical standards for wildlife research.

Data availability

The datasets utilized in this study are included in the manuscript and its Supplementary Material.

Supplementary material

Supplementary material can be found at https://academic.oup.com/cz.