Abstract
Wild Chinese pangolins ingest soil and nest-associated materials while feeding on ants and termites, but these materials are generally absent from captive diets. We assessed whether termitarium soil supplementation was associated with changes in fecal bacterial communities and short-chain fatty acid (SCFA) profiles in captive Chinese pangolins. Four adults participated in a sequential within-subject trial comprising a basal diet, a basal diet supplemented with sterilized termitarium soil, and a basal diet supplemented with unsterilized termitarium soil. Fecal bacterial communities were characterized by 16S rRNA gene sequencing, and SCFAs were quantified by gas chromatography-mass spectrometry. Community composition differed across dietary stages, whereas alpha diversity did not. At the genus level, Bacteroides abundance was lower during the sterilized-soil stage, whereas Burkholderia-Caballeronia-Paraburkholderia abundance was higher during the unsterilized-soil stage than during the basal-diet stage. Relative to the basal-diet stage, both soil-supplemented stages had lower fecal concentrations of acetate, propionate, butyrate, and valerate. Partial redundancy analysis showed an overall association between the SCFA profile and the selected bacterial genera. PICRUSt2-based predictions also differed across dietary stages, particularly for Xenobiotics Biodegradation and Metabolism. Overall, termitarium soil supplementation was associated with variation in fecal bacterial composition, SCFA concentrations, and predicted functional potential in captive Chinese pangolins. These findings provide a basis for further research on the role of soil-associated materials in the dietary management of captive pangolins.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1186/s12866-026-05547-x.
Keywords: Chinese pangolin, Termitarium soil, Fecal microbiota, Short-chain fatty acids, Captivity
Introduction
Captive management of the Chinese pangolin is challenging because its diet in the wild is highly specialized and cannot be fully replicated in rescue centers or zoological institutions. As obligate myrmecophagous mammals, pangolins feed mainly on ants and termites in the wild. Because a stable supply of ants and termites is rarely feasible, captive pangolins are usually maintained on artificial diets [1–3]. Although these diets are necessary for long-term husbandry, gastrointestinal disorders, poor appetite, and low survival have frequently been reported in captive pangolins [1, 4, 5]. These problems raise the possibility that artificial diets may lack not only natural prey but also soil, nest material, and other substances ingested during foraging.
During natural foraging, pangolins ingest soil and nest materials from termite mounds and ant nests together with their prey. These materials are usually absent from artificial diets and may affect mineral intake, microbial exposure, and gastrointestinal conditions. Soil ingestion, or geophagy, has been reported in many mammals and has been linked to mineral supplementation, adsorption of dietary toxins, gastrointestinal regulation, and changes in microbial exposure [6, 7]. Termitarium soil is distinct from surrounding soil because termite activity modifies its physical structure, organic matter content, mineral composition, and microbial communities [8–11]. Termite mounds and nest materials may also contain organic residues and microorganisms associated with the degradation of plant- and insect-derived substrates. Termitarium soil may therefore expose pangolins to substrates and microorganisms that are absent from standard captive diets [12].
Gut microbes contribute to host nutrition, metabolism, immune regulation, and intestinal homeostasis [13, 14]. In mammals, captivity and simplified diets can alter gut microbial communities through changes in diet, environmental exposure, and management conditions [15, 16]. Similar patterns have been reported in pangolins. Gut microbial communities differ between wild, rescued, and captive individuals and also vary among captive pangolin species, indicating contributions from diet, captivity, and host-related factors [17–20]. Short-chain fatty acids (SCFAs), including acetate, propionate, and butyrate, are major products of microbial fermentation and are involved in intestinal barrier function, immune regulation, and host energy metabolism [21–23]. Fecal SCFA concentrations do not directly represent total intestinal production because they are influenced by microbial metabolism, host absorption, fecal water content, and gut transit time [24]. Measuring both fecal bacterial communities and fecal SCFAs can therefore help characterize microbial and metabolic variation across dietary stages. Host-associated microbiomes are also relevant to wildlife conservation, especially for threatened species maintained in captive breeding, rescue, or reintroduction programs [25–27].
Here, we conducted an exploratory longitudinal study to examine stage-associated variation in fecal bacterial composition, fecal SCFA profiles, and predicted functional potential across sequential dietary stages in captive Chinese pangolins. The study included a basal-diet stage followed by stages supplemented with sterilized and unsterilized termitarium soil. We expected the soil-supplemented stages to differ from the basal-diet stage and explored whether the observed patterns differed between the two soil preparations. Fecal bacterial communities were characterized by 16S rRNA gene sequencing, SCFAs were quantified by GC–MS, and predicted functional potential was inferred using PICRUSt2.
Materials and methods
Animals, experimental design, diets, and sample collection
Chinese pangolins (Manis pentadactyla) used in this study were rescued from various locations across Guangdong Province and housed at the Guangdong Provincial Wildlife Rescue Center, China. Upon admission, individuals were quarantined in rescue enclosures for medical treatment until traumatic injuries were resolved and normal appetite and fecal output were restored. Health status was further evaluated using routine hematological indicators before transfer to rearing enclosures.
Four healthy adult Chinese pangolins, including two males and two females, were included in the experiment. Because only a small number of suitable animals were available for this critically endangered species, a sequential within-subject design was used. The same four animals completed all three dietary stages, and each animal served as its own control; therefore, the biological sample size was four animals throughout the experiment. Each pangolin received the following three dietary stages: (1) a basal diet only (control stage, CZ), (2) a basal diet supplemented with 18 g of sterilized termitarium soil (MZ), and (3) a basal diet supplemented with 18 g of unsterilized termitarium soil (BZ). The stages were administered in the same order for all individuals rather than randomized. The composition of the basal diet is provided in Table S1. The experiment followed a 40-day sequential schedule comprising a 10-day basal-diet stage (CZ; 4–13 May 2020), a 10-day sterilized-soil stage (MZ; 14–23 May 2020), a 10-day washout period during which only the basal diet was provided (24 May–2 June 2020), and a 10-day unsterilized-soil stage (BZ; 3–12 June 2020) (Table S1). Fecal samples were collected during the final three days of each dietary stage. The 10-day dietary stages provided a consistent period of exposure before sampling, whereas the 10-day basal-diet washout was included between the two soil-supplemented stages to reduce potential carry-over from the preceding soil treatment. Body mass was recorded during each dietary stage, and individual values are provided in Table S1. During the MZ and BZ stages, 18 g of termitarium soil was provided to each animal per day and mixed with the basal diet. Food residues were checked daily, and the soil supplement was completely consumed (Table S1).
Fresh termitarium soil was collected using sterile sampling bags from termite-associated foraging sites used by wild Chinese pangolins in forested areas of Pengzhai Town, southeastern Heping County, Heyuan City, Guangdong Province, China. Soil samples were immediately transported on dry ice and stored at − 80 °C prior to processing. Sterilization of termitarium soil was performed by autoclaving at 121 °C and 103.4 kPa for 1 h [28], whereas soil used in the unsterilized-soil treatment was not autoclaved. During the experimental period, pangolins were fed according to individual body condition, and water was provided ad libitum. All animals consumed the offered diet, remained normally active, produced formed feces, and showed no adverse clinical signs during the experiment. Because defecation frequency varied among individuals, a total of 30 repeated fecal samples were obtained: 12 during CZ, seven during MZ, and 11 during BZ (Table S1). These samples were repeated subsamples collected from the same four animals and were not independent biological replicates; the biological sample size was four animals at each stage. To minimize environmental contamination, the outer layer of each fecal pellet was removed, and material from the central portion was transferred into sterile collection tubes. Samples were immediately frozen in liquid nitrogen and stored at − 80 °C until further analysis.
DNA extraction, 16S rRNA gene sequencing, and bioinformatic analysis
Total genomic DNA was extracted from fecal samples using a commercial DNA extraction kit (D4015, Omega Bio-Tek, USA) following the manufacturer’s instructions and stored at − 80 °C prior to analysis. A nuclease-free water blank was processed alongside the fecal samples during DNA extraction and PCR as a negative control. No visible amplification was observed, and the blank was not included in library preparation or sequencing. The V3–V4 hypervariable regions of the bacterial 16S rRNA gene were amplified by polymerase chain reaction (PCR) using the universal primers 341 F (5′-CCTACGGGNGGCWGCAG-3′) and 805R (5′-GACTACHVGGGTATCTAATCC-3′). PCR amplification was performed for 32 cycles. PCR products were purified using AMPure XT beads (Beckman Coulter Genomics, USA) and quantified using a Qubit fluorometer (Invitrogen, USA). Amplicon libraries were assessed for fragment size and concentration before sequencing. Paired-end sequencing (2 × 250 bp) was conducted on the Illumina NovaSeq platform.
Raw paired-end reads were demultiplexed according to their barcode sequences. Adapter, primer, and heterogeneity-spacer sequences were removed using Cutadapt (version 1.9), with a minimum adapter overlap of 17 bp and a minimum retained read length of 100 bp. Paired-end reads were merged using FLASH (version 1.2.8), with a minimum overlap of 10 bp, a maximum overlap of 100 bp, and a maximum mismatch ratio of 0.25 [29]. Quality filtering was performed using fqtrim (version 0.94) with a 100-bp sliding window and a mean Phred-quality threshold of 20. Reads were trimmed from the beginning of the first window with a mean quality score below 20. Reads shorter than 100 bp after trimming or containing more than 5% ambiguous bases were removed. Chimeric sequences were identified and removed using VSEARCH (version 2.3.4) with default parameters [30]. ASVs were inferred using DADA2 implemented in QIIME2 (version 2019.7) [31], and singleton ASVs occurring only once across the entire dataset were excluded.
Taxonomic assignment of ASVs was performed using the SILVA ribosomal RNA gene database (release 138) with a confidence threshold of 0.7 [32, 33]. ASVs not assigned to the domain Bacteria were excluded before downstream fecal microbiome analyses, leaving 4,919 bacterial ASVs across the 30 fecal samples. No ASVs in the final taxonomy table were annotated as chloroplasts, mitochondria, or eukaryotes.
Sample-based rarefaction curves were generated from the filtered ASV count table using the rarefy function in the vegan package. For each sample, expected ASV richness was calculated at 120 sequencing depths ranging from one read to the total read count of that sample. The curves were used only to assess sequencing coverage and were not used to normalize the ASV table. Alpha-diversity indices were calculated from the filtered count table, and Chao1, Shannon, and Simpson indices were compared among dietary stages. For beta-diversity analysis, ASV counts were converted to sample-wise relative abundances before Bray–Curtis dissimilarities were calculated. For genus-level analysis, ASVs were aggregated by genus, and the 20 genera with the highest mean relative abundance across all samples were retained. A pseudocount of 1 × 10⁻⁶ was added before normalization to proportions and centered log-ratio transformation.
Predicted functional profiles were inferred from the ASV abundance table and representative ASV sequences using PICRUSt2 (version 2.5.1) [34]. Sequence placement used the default reference phylogeny containing approximately 20,000 full-length bacterial and archaeal 16S rRNA gene sequences from the Integrated Microbial Genomes database. Hidden-state prediction was performed using maximum parsimony, with a minimum aligned fraction of 0.7. ASVs with NSTI values greater than 2 were excluded; --min_reads and --min_samples were both set to 1. Other PICRUSt2 parameters were retained at their default settings. Of the 4,919 ASVs evaluated, three exceeded the NSTI threshold and were excluded. The retained 4,916 ASVs had a median NSTI of 0.136 (range, 0.0001–1.863). Predicted KEGG Ortholog abundances were summarized into KEGG Level 1–4 categories using the KO1–4 mapping files provided with EasyMicrobiome (version 1.21).
Quantification of short-chain fatty acids in fecal samples
Fecal short-chain fatty acids (SCFAs) were quantified using a previously described GC–MS protocol with minor modifications [35]. Briefly, approximately 20 mg of fecal material was placed into a 2 mL centrifuge tube containing grinding beads. Internal standards (acetic acid-d4, propanoic acid-d2, isovaleric acid-d2, hexanoic acid-d2, and octanoic acid-d2) were added, followed by aqueous isobutanol. Samples were homogenized at 50 Hz for 360 s and centrifuged at 12,000 rpm for 5 min. The supernatant was collected and subjected to alkaline treatment and chloroform extraction. After derivatization with isobutyl chloroformate, SCFAs were extracted using n-hexane and analyzed by gas chromatography–mass spectrometry (GC–MS). GC–MS analysis was performed using an Agilent HP-5MS column. The injector temperature was set to 260 °C, and the injection volume was 1 µL. Mass spectrometric detection was conducted in selected ion monitoring (SIM) mode.
A separate fecal aliquot was weighed before and after freeze-drying to determine moisture content, and SCFA concentrations were normalized to fecal dry mass and expressed as nmol/g dry feces. Analytes were identified by comparison of retention times and SIM fragment ions with authentic standards and quantified by the internal-standard method using isotope-labelled standards. Chromatographic resolution was evaluated using a mixed standard containing 15 SCFAs, which showed adequate separation of the target analytes. Data were acquired and processed using MassHunter software (version B.08.00, Agilent Technologies, USA).
Statistical analyses
All analyses were performed in R (v4.2.2). The individual animal was considered the experimental unit, and the 30 fecal samples were treated as repeated measurements from four animals rather than as independent biological replicates. Benjamini–Hochberg false discovery rate correction was applied to multiple comparisons, with q < 0.05 considered significant.
Stage effects on Chao1, Shannon, and Simpson diversity, CLR-transformed genus abundances, log1p-transformed SCFA concentrations, and predicted KEGG Level 2 abundances were evaluated using linear mixed-effects models fitted by restricted maximum likelihood. The general model was specified as Outcome ~ Stage, with dietary stage as a fixed effect and individual as a random intercept. SamplingDay represented the chronological day of fecal collection within each dietary stage. Temporal dependence among consecutive samples was modeled using a continuous-time autoregressive correlation structure, corCAR1(~ SamplingDay | Individual/Stage). Pairwise contrasts were estimated from marginal means and are reported with effect estimates, 95% confidence intervals, P values, and adjusted q values. Model assumptions were assessed using normalized residual-versus-fitted plots, normal Q–Q plots, and residual autocorrelation functions.
ASV community composition and predicted KO functional composition were assessed using Bray–Curtis dissimilarities. Stage effects were tested by PERMANOVA using the model Bray–Curtis dissimilarity ~ Stage, with 9,999 permutations restricted within individual; observations were permuted only among samples from the same animal and were not exchanged among animals. PERMDISP was used to assess differences in multivariate dispersion [36, 37]. PCoA was used for visualization. Genus-level SIMPER analysis was used descriptively to identify the main contributors to pairwise Bray–Curtis dissimilarities.
Partial redundancy analysis was used to examine the multivariate association between the six SCFAs and the Hellinger-transformed relative abundances of Streptococcus, Bacteroides, Romboutsia, Clostridium_sensu_stricto_1, and Cellulosilyticum. The genus abundance matrix was treated as the multivariate response, and the six SCFAs were included as explanatory variables. SCFA concentrations were log1p-transformed and standardized, and individual was included as a conditioning variable. The model was specified as genus abundance matrix ~ six SCFAs + Condition(Individual). Model significance was assessed using 9,999 permutations restricted within individual; samples were permuted only within the same animal and not among animals. Variance inflation factors were examined for collinearity among the SCFA predictors.
To assess the influence of unequal numbers of repeated samples among stages, sensitivity analyses were conducted using one mean value or compositional profile per animal and stage, yielding 12 animal-stage observations. Animal-stage univariate outcomes were analyzed using linear models with stage and individual included as fixed terms (outcome ~ Stage + Individual), with individual serving as a blocking factor. ASV and KO profiles were averaged within animal and stage and renormalized before Bray–Curtis analysis. Animal-stage PERMANOVA, PERMDISP, and partial RDA used all 1,295 non-identity stage-label permutations within individual. An additional partial RDA model based on the first two SCFA principal components was used to reduce collinearity among the SCFA predictors.
Results
Fecal bacterial diversity across dietary stages
Across samples, 56,962–79,919 valid reads were retained, representing 72.69%–96.13% of the raw reads. Q20 and Q30 values ranged from 95.25% to 98.65% and from 86.79% to 95.48%, respectively, and GC content ranged from 51.08% to 53.46% (Table S2). Sample-based rarefaction curves approached a plateau, indicating that sequencing depth was sufficient to recover most of the observed ASV richness within the samples (Fig. 1A).
Fig. 1.

Stage-associated variation in fecal bacterial diversity in captive Chinese pangolins. A Sample-based rarefaction curves showing expected ASV richness across sequencing depths. B–D Chao1, Shannon, and Simpson alpha-diversity indices. E PCoA and (F) NMDS based on Bray–Curtis dissimilarities. Ellipses indicate 95% confidence intervals. Individual points represent repeated fecal samples collected from four animals. CZ, basal diet; MZ, basal diet supplemented with sterilized termitarium soil; BZ, basal diet supplemented with unsterilized termitarium soil
Alpha-diversity indices, including Chao1 richness, Shannon diversity, and Simpson diversity, did not differ among dietary stages, and this conclusion was unchanged in the animal-stage analysis (Fig. 1B–D; Table S3). Bray–Curtis-based PCoA and NMDS showed partial separation of fecal bacterial communities among stages (Fig. 1E, F). PERMANOVA indicated a stage-associated difference in ASV community composition in the repeated-sample analysis (R² = 0.15, pseudo-F = 2.309, P < 0.001), while PERMDISP provided no evidence of unequal multivariate dispersion (P = 0.2223). The stage association remained significant in the animal-stage analysis (R² = 0.232, pseudo-F = 1.362, P = 0.0046; PERMDISP P = 0.0934; Table S4).
Taxonomic composition and stage-associated genus-level patterns
Across all fecal samples, the fecal microbiota of captive Chinese pangolins was dominated by Firmicutes, Proteobacteria, and Bacteroidota, which together accounted for 97.92% of the total relative abundance (Fig. 2A). Firmicutes was the most abundant phylum across dietary stages, with a mean relative abundance of 76.35%. At the genus level, the dominant taxa included Clostridium_sensu_stricto_1, Streptococcus, Escherichia-Shigella, Cellulosilyticum, and Romboutsia, with mean relative abundances of 24.24%, 14.87%, 8.20%, 7.83%, and 4.44%, respectively (Fig. 2B).
Fig. 2.

Fecal bacterial community composition in captive Chinese pangolins. A, B Relative abundances of fecal microbial taxa at the phylum (A) and genus (B) levels. C–E SIMPER analysis showing the top genera contributing to Bray–Curtis dissimilarity between dietary stages. F–J Differential abundances of representative genera based on CLR-transformed data. *q < 0.05; **q < 0.01. Small points represent repeated fecal samples, larger shaped points represent animal-stage means, and lines connect means from the same animal across dietary stages. CZ, control; MZ, sterilized termitarium soil; BZ, unsterilized termitarium soil
SIMPER analysis identified the main genus-level contributors to Bray–Curtis dissimilarities between dietary stages (Fig. 2C–E). Streptococcus and Clostridium_sensu_stricto_1 were major contributors to the differences between the control and soil-supplemented stages, while Cellulosilyticum, Escherichia-Shigella, Bacteroides, and Proteus also contributed to between-stage dissimilarities. Comparisons between the sterilized and unsterilized soil stages were mainly characterized by differences in Clostridium_sensu_stricto_1, Streptococcus, and Cellulosilyticum. The leading genus-level contributors were similar in the animal-stage analysis (Table S5).
Stage effects were evaluated for 20 genera, with complete omnibus and pairwise results provided in Table S6. After FDR correction, significant omnibus stage effects were detected for Bacteroides (q = 0.0246) and Burkholderia-Caballeronia-Paraburkholderia (q = 0.0057). Pairwise comparisons showed that the CLR-transformed abundance of Bacteroides was lower during the sterilized-soil stage than during the control stage (estimate = − 2.357, 95% CI: −3.618 to − 1.097, q = 0.0225). In contrast, Burkholderia-Caballeronia-Paraburkholderia was higher during the unsterilized-soil stage than during the control stage (estimate = 2.107, 95% CI: 1.192 to 3.022, q = 0.0047). No other pairwise comparison remained significant after FDR correction. Effect directions were concordant in 58 of the 60 genus-by-contrast comparisons in the animal-stage analysis, although no genus-level comparison remained significant after FDR correction.
Fecal short-chain fatty acid profiles and their multivariate association with fecal bacterial composition
In the mixed-effects analysis, significant omnibus stage effects were detected for acetate (q = 0.0009), propionate (q = 0.0175), butyrate (q = 0.0003), and valerate (q = 0.0053). Each of these SCFAs was lower during both soil-supplemented stages than during the control stage, and all eight corresponding CZ–MZ and CZ–BZ comparisons remained significant after FDR correction (q = 0.0007–0.0418; Fig. 3A, B, E, F; Table S7). No significant differences were detected between the sterilized- and unsterilized-soil stages. Isobutyrate and isovalerate showed no significant omnibus stage effects (Fig. 3C, D). In the animal-stage analysis, the directions of all 18 pairwise contrasts were consistent with those in the repeated-sample analysis, although none remained significant after FDR correction (minimum q = 0.0608; Table S7).
Fig. 3.

Stage-associated variation in fecal SCFAs and their multivariate association with fecal bacterial composition. A–F Fecal SCFA concentrations across dietary stages, expressed as nmol/g dry feces. G Partial RDA showing the overall association between the SCFA profile and five selected genera after accounting for individual identity. *q < 0.05; **q < 0.01; ***q < 0.001. Small points represent repeated fecal samples, larger shaped points represent animal-stage means, and lines connect means from the same animal across dietary stages. CZ, control; MZ, sterilized termitarium soil; BZ, unsterilized termitarium soil
Partial RDA indicated a significant overall multivariate association between the six-SCFA profile and the five selected genera in the repeated-sample analysis (R² = 0.317, pseudo-F₆,₂₀ = 2.918, P = 0.0042; Fig. 3G; Table S8). The animal-stage model based on all six SCFAs was not significant (R² = 0.463, pseudo-F₆,₂ = 2.285, P = 0.2068) and showed high collinearity among the SCFA predictors. A reduced animal-stage model based on SCFA PC1 and PC2, which together explained 95.37% of the SCFA variation, also did not reach the conventional significance threshold (R² = 0.286, pseudo-F₂,₆ = 3.509, P = 0.0517; Table S8).
Predicted functional potential of fecal microbial communities
PICRUSt2-based predictions indicated modest stage-associated variation in predicted functional potential. At KEGG Level 1, Metabolism was the dominant predicted functional category across all dietary stages, followed by Genetic Information Processing and Environmental Information Processing (Fig. 4A). PCoA based on Bray–Curtis dissimilarities of predicted KO profiles showed partial separation among dietary stages (Fig. 4B). In the repeated-sample analysis, dietary stage explained 12.2% of the variation in predicted KO composition (R² = 0.122, pseudo-F = 1.880, P = 0.0438), with no evidence of unequal multivariate dispersion among stages (PERMDISP P = 0.7098). The stage association was not retained in the animal-stage analysis (R² = 0.198, pseudo-F = 1.111, P = 0.2361; PERMDISP P = 0.3248; Table S9).
Fig. 4.

PICRUSt2-predicted functional potential of the fecal microbiota. A Relative abundances of predicted KEGG level 1 functional categories. B PCoA of predicted functional profiles based on Bray–Curtis dissimilarity. C–E Predicted abundances of selected KEGG Level 2 categories. *q < 0.05; **q < 0.01; ***q < 0.001. Small points represent repeated fecal samples, larger shaped points represent animal-stage means, and lines connect means from the same animal across dietary stages. CZ, control; MZ, sterilized termitarium soil; BZ, unsterilized termitarium soil
Stage effects were evaluated for 40 predicted KEGG Level 2 categories, with complete results provided in Table S10. In the repeated-sample analysis, Xenobiotics Biodegradation and Metabolism was the only category with a significant omnibus stage effect after FDR correction (q = 0.0454), and its predicted abundance was higher during the unsterilized-soil stage than during the control stage (q = 0.0363). In the animal-stage analysis, Cell Motility and Environmental Adaptation showed significant omnibus stage effects (both q = 0.0297). Both categories had higher predicted abundances during the sterilized- and unsterilized-soil stages than during the control stage (pairwise q = 0.0310–0.0492), with no significant difference between the two soil-supplemented stages. No other KEGG Level 2 category remained significant after FDR correction.
Discussion
Termitarium soil supplementation was associated with variation in fecal bacterial community composition and SCFA concentrations in captive Chinese pangolins. Although alpha diversity remained stable, beta diversity differed among dietary stages. Acetate, propionate, butyrate, and valerate concentrations were lower during both soil-supplemented stages than during the basal-diet stage. In the wild, pangolins ingest soil and nest-associated materials while feeding on ants and termites. These findings suggest that termitarium soil may represent an ecologically relevant dietary component associated with fecal microbial and metabolic variation in captivity.
Termitarium soil supplementation was associated with a change in fecal bacterial community structure without a corresponding change in bacterial richness or evenness. Dietary stage accounted for approximately 15% of the Bray–Curtis variation, and the association remained detectable after repeated samples were aggregated within animal and stage. Similar differences between alpha- and beta-diversity responses have been reported following dietary and environmental changes in mammals [15, 17, 38], indicating that community composition can change through shifts in taxon relative abundance without a broad gain or loss of bacterial diversity. In the present study, Bacteroides decreased during the sterilized-soil stage, whereas Burkholderia-Caballeronia-Paraburkholderia increased during the unsterilized-soil stage. These taxon-specific differences indicate that the separation among dietary stages reflected a redistribution of bacterial genera rather than a uniform change across the community. The increase in Burkholderia-Caballeronia-Paraburkholderia during the unsterilized-soil stage may also reflect the presence of soil-associated bacterial material in feces. However, the genus-level differences were not retained in the animal-stage analysis, and no direct comparison between the two soil-supplemented stages remained significant after FDR correction. The community-level variation therefore cannot be attributed to a single genus or to the effects of soil sterilization. Direct characterization of the soil microbiota and physicochemical properties will be required to distinguish changes in the resident fecal community from the presence of soil-associated bacterial material in feces.
Soil supplementation was associated with coordinated changes in fecal SCFA concentrations and bacterial community composition. Acetate, propionate, butyrate, and valerate concentrations were lower during the soil-supplemented stages in the mixed-effects analysis, and the animal-stage analysis showed the same direction of change. SCFAs are major products of microbial fermentation and contribute to host energy metabolism and intestinal homeostasis [22, 23]. Butyrate, in particular, supports intestinal epithelial barrier function and mucosal immune regulation [39, 40]. Previous studies in pangolins have shown that dietary soil and chitin can affect digestibility and fecal characteristics [12], while captivity-related differences in the microbiome include predicted capacities for chitin degradation and SCFA synthesis [17]. However, direct information on fecal SCFA profiles in wild pangolins or pangolins consuming natural prey remains limited. Because fecal SCFA concentrations reflect the balance among microbial production, intestinal absorption, gastrointestinal transit, and fecal output [24, 41, 42], the lower concentrations may result from changes in microbial fermentation or host intestinal handling of these metabolites. Partial RDA further showed that the overall SCFA profile covaried with the composition of the selected genera, linking the metabolic and compositional changes observed across dietary stages. This association was weaker in the animal-stage analysis and requires confirmation in a larger cohort.
PICRUSt2 analysis suggested that the functional potential of the fecal bacterial community varied across dietary stages. Xenobiotics Biodegradation and Metabolism differed in the repeated-sample analysis, whereas Cell Motility and Environmental Adaptation remained significant in the animal-stage analysis. These categories are broadly related to the processing of environmental compounds and bacterial responses to changing conditions and may reflect shifts in bacterial traits associated with the introduction of soil-derived material. A previous study in Sunda pangolins showed that diets containing soil and chitin affected apparent digestibility and fecal characteristics [12], indicating that environmental dietary components can influence digestive processes in captive pangolins. Because these functional profiles were inferred from 16S rRNA gene data, they represent predicted functional potential rather than direct evidence of microbial activity. Metagenomic and metabolomic analyses are needed to determine whether these predicted differences correspond to changes in microbial metabolism.
This study has several limitations. The small cohort and fixed treatment sequence limited the separation of stage-associated variation from temporal and sequence effects, while repeated sampling increased within-animal resolution rather than biological replication. The administered soils were not characterized microbiologically or physicochemically, and changes caused by autoclaving were not quantified, precluding identification of the soil components associated with the observed fecal patterns. In addition, DNA-based profiling could not distinguish viable cells from nonviable cells or relic DNA, and PICRUSt2 provided predictions of functional potential rather than direct measures of microbial activity. Finally, fecal bacterial and SCFA profiles alone cannot establish effects on digestion, intestinal physiology, or animal health. Future studies should incorporate larger randomized designs, direct characterization of soil properties and microbial viability, and complementary physiological measurements.
Conclusion
Termitarium soil supplementation was associated with shifts in fecal bacterial community composition and lower concentrations of acetate, propionate, butyrate, and valerate in captive Chinese pangolins, while alpha diversity remained stable. Predicted functional profiles also varied across dietary stages, with differences in Cell Motility and Environmental Adaptation detected in the animal-stage analysis. These findings provide a basis for further evaluation of termitarium soil as an ecologically relevant dietary component in captive pangolin nutrition and microbial ecology.
Supplementary Information
Acknowledgements
We thank the staff and graduate students of the Wildlife Conservation Research Team of the Guangdong Academy of Forestry for the collection of samples.
Authors’ contributions
Y.H. conceived and designed the study. Z.Z., K.W., F.A. and H.Y. performed the animal experiments and sample collection. Z.Z. carried out the bioinformatic, statistical, and SCFA analyses. Z.Z. and K.W. interpreted the results and drafted the manuscript. Y.H. supervised the study and critically revised the manuscript. All authors read and approved the final manuscript.
Funding
This work was supported by the National Key R&D Program of China (Grant No. 2025YFC3507700).
Data availability
Raw sequencing reads have been deposited in the NCBI Sequence Read Archive under BioProject accession number PRJNA890888.
Declarations
Ethics approval and consent to participate
This study was approved by the Experimental Animal Management and Ethics Committee of the Guangdong Academy of Forestry (approval No. 202000510) and was conducted in accordance with the Regulations on the Administration of Laboratory Animals of the People’s Republic of China and relevant guidelines for the care and use of laboratory animals. The study involved dietary supplementation and non-invasive collection of freshly voided fecal samples from live animals. The approved protocol specifically included the sequential dietary supplementation with sterilized and unsterilized termitarium soil, together with the non-invasive collection of freshly voided fecal samples. No animals were anesthetized, euthanized, or sacrificed for this study. The animals remained under routine care at the Guangdong Provincial Wildlife Rescue Center and were monitored daily for food intake, activity, fecal appearance, and overt clinical abnormalities.
Consent for publication
Not applicable.
Competing Interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Yang CW, Chen S, Chang CY, Lin MF, Block E, Lorentsen R, et al. History and dietary husbandry of pangolins in captivity. Zoo Biol. 2007;26:223–30. 10.1002/zoo.20134. [DOI] [PubMed] [Google Scholar]
- 2.Cabana F, Plowman A, Van Nguyen T, Chin SC, Wu SL, Lo HY, et al. Feeding Asian pangolins: an assessment of current diets fed in institutions worldwide. Zoo Biol. 2017;36:298–305. 10.1002/zoo.21375. [DOI] [PubMed] [Google Scholar]
- 3.Wang XM, Janssens GPJ, Xie CG, Xie BW, Xie ZG, He HJ, et al. To save pangolins: a nutritional perspective. Animals. 2022;12:3137. 10.3390/ani12223137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Clark L, Van Thai N, Phuong TQ. A long way from home: the health status of Asian pangolins confiscated from the illegal wildlife trade in Viet Nam. In: Workshop on Trade and Conservation of Pangolins Native to South and Southeast Asia. 2009;111.
- 5.Hua L, Gong S, Wang F, Li W, Ge Y, Li X, et al. Captive breeding of pangolins: current status, problems and future prospects. ZooKeys. 2015;507:99–114. 10.3897/zookeys.507.6970. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Krishnamani R, Mahaney WC. Geophagy among primates: adaptive significance and ecological consequences. Anim Behav. 2000;59:899–915. 10.1006/anbe.1999.1376. [DOI] [PubMed] [Google Scholar]
- 7.Young SL, Sherman PW, Lucks JB, Pelto GH. Why on earth? Evaluating hypotheses about the physiological functions of human geophagy. Q Rev Biol. 2011;86:97–120. 10.1086/659884. [DOI] [PubMed] [Google Scholar]
- 8.Brauman A, Bignell DE, Tayasu I. Soil-feeding termites: biology, microbial associations and digestive mechanisms. In: Abe T, Bignell DE, Higashi M, editors. Termites: Evolution, Sociality, Symbioses, Ecology. Springer; 2000;233–59. 10.1007/978-94-017-3223-9_11. [DOI]
- 9.Fall S, Hamelin J, Ndiaye F, Assigbetse K, Aragno M, Chotte JL, et al. Differences between bacterial communities in the gut of a soil-feeding termite (Cubitermes niokoloensis) and its mounds. Appl Environ Microbiol. 2007;73:5199–208. 10.1128/AEM.02616-06. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Holt JA, Lepage M. Termites and soil properties. In: Abe T, Bignell DE, Higashi M, editors. Termites: Evolution, Sociality, Symbioses, Ecology. Springer; 2000;389–407. 10.1007/978-94-017-3223-9_18. [DOI]
- 11.Jouquet P, Traoré S, Choosai C, Hartmann C, Bignell D. Influence of termites on ecosystem functioning. Ecosystem services provided by termites. Eur J Soil Biol. 2011;47:215–22. 10.1016/j.ejsobi.2011.05.005. [DOI] [Google Scholar]
- 12.Cabana F, Tay C. The addition of soil and chitin into Sunda pangolin (Manis javanica) diets affect digestibility, faecal scoring, mean retention time and body weight. Zoo Biol. 2020;39:29–36. 10.1002/zoo.21517. [DOI] [PubMed] [Google Scholar]
- 13.Round JL, Mazmanian SK. The gut microbiota shapes intestinal immune responses during health and disease. Nat Rev Immunol. 2009;9:313–23. 10.1038/nri2515. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Zmora N, Suez J, Elinav E. You are what you eat: diet, health and the gut microbiota. Nat Rev Gastroenterol Hepatol. 2019;16:35–56. 10.1038/s41575-018-0061-2. [DOI] [PubMed] [Google Scholar]
- 15.McKenzie VJ, Song SJ, Delsuc F, Prest TL, Oliverio AM, Korpita TM, et al. The effects of captivity on the mammalian gut microbiome. Integr Comp Biol. 2017;57:690–704. 10.1093/icb/icx090. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Guo W, Mishra S, Wang C, Zhang H, Ning R, Kong F, et al. Comparative study of gut microbiota in wild and captive giant pandas (Ailuropoda melanoleuca). Genes. 2019;10:827. 10.3390/genes10100827. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Liu C, Hu J, Wu Y, Irwin DM, Chen W, Zhang Z, et al. Comparative study of gut microbiota from captive and confiscated-rescued wild pangolins. J Genet Genomics. 2021;48:825–35. 10.1016/j.jgg.2021.07.009. [DOI] [PubMed] [Google Scholar]
- 18.Jiao W, Liu L, Zeng Z, Li L, Chen J. Differences in gut microbes in captive pangolins and the effects of captive breeding. Front Microbiol. 2022;13:1053925. 10.3389/fmicb.2022.1053925. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Dai Z, Xie B, Xie C, Xiang J, Wang X, Li J, et al. Comparative metagenomic analysis of the gut microbiota of captive pangolins: a case study of two species. Animals. 2025;15:57. 10.3390/ani15010057. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Xiang C, Su L, Han M, Liang J, Hou F, Liao J. Comparative analysis of gut microbiota and metabolome in captive Chinese and Malayan pangolins. Front Microbiol. 2025;16:1599588. 10.3389/fmicb.2025.1599588. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Flint HJ, Scott KP, Duncan SH, Louis P, Forano E. Microbial degradation of complex carbohydrates in the gut. Gut Microbes. 2012;3:289–306. 10.4161/gmic.19897. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.den Besten G, van Eunen K, Groen AK, Venema K, Reijngoud DJ, Bakker BM. The role of short-chain fatty acids in the interplay between diet, gut microbiota, and host energy metabolism. J Lipid Res. 2013;54:2325–40. 10.1194/jlr.R036012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Koh A, De Vadder F, Kovatcheva-Datchary P, Bäckhed F. From dietary fiber to host physiology: short-chain fatty acids as key bacterial metabolites. Cell. 2016;165:1332–45. 10.1016/j.cell.2016.05.041. [DOI] [PubMed] [Google Scholar]
- 24.Vogt JA, Wolever TMS. Fecal acetate is inversely related to acetate absorption from the human rectum and distal colon. J Nutr. 2003;133:3145–8. 10.1093/jn/133.10.3145. [DOI] [PubMed] [Google Scholar]
- 25.Trevelline BK, Fontaine SS, Hartup BK, Kohl KD. Conservation biology needs a microbial renaissance: a call for the consideration of host-associated microbiota in wildlife management practices. Proc Biol Sci. 2019;286:20182448. 10.1098/rspb.2018.2448. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.West AG, Waite DW, Deines P, Bourne DG, Digby A, McKenzie VJ, et al. The microbiome in threatened species conservation. Biol Conserv. 2019;229:85–98. 10.1016/j.biocon.2018.11.016. [DOI] [Google Scholar]
- 27.Zhu L, Wang J, Bahrndorff S. Editorial: the wildlife gut microbiome and its implication for conservation biology. Front Microbiol. 2021;12:697499. 10.3389/fmicb.2021.697499. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Wolf DC, Skipper HD. Soil sterilization. In: Weaver RW, Angle JS, Bottomley PS, editors. Methods of Soil Analysis: Part 2, Microbiological and Biochemical Properties. Soil Science Society of America; 1994;41–51.
- 29.Magoc T, Salzberg SL. FLASH: fast length adjustment of short reads to improve genome assemblies. Bioinformatics. 2011;27:2957–63. 10.1093/bioinformatics/btr507. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Rognes T, Flouri T, Nichols B, Quince C, Mahé F. VSEARCH: a versatile open source tool for metagenomics. PeerJ. 2016;4:e2584. 10.7717/peerj.2584. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJ, Holmes SP. DADA2: high-resolution sample inference from Illumina amplicon data. Nat Methods. 2016;13:581–3. 10.1038/nmeth.3869. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P, et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. 2013;41:D590–6. 10.1093/nar/gks1219. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Yilmaz P, Parfrey LW, Yarza P, Gerken J, Pruesse E, Quast C, et al. The SILVA and All-species Living Tree Project (LTP) taxonomic frameworks. Nucleic Acids Res. 2014;42:D643–8. 10.1093/nar/gkt1209. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Douglas GM, Maffei VJ, Zaneveld JR, Yurgel SN, Brown JR, Taylor CM, et al. PICRUSt2 for prediction of metagenome functions. Nat Biotechnol. 2020;38:685–8. 10.1038/s41587-020-0548-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Furuhashi T, Sugitate K, Nakai T, Jikumaru Y, Ishihara G. Rapid profiling method for mammalian feces short-chain fatty acids by GC-MS. Anal Biochem. 2018;543:51–4. 10.1016/j.ab.2017.12.001. [DOI] [PubMed] [Google Scholar]
- 36.Anderson MJ. A new method for non-parametric multivariate analysis of variance. Austral Ecol. 2001;26:32–46. 10.1111/j.1442-9993.2001.01070.pp.x. [DOI] [Google Scholar]
- 37.Anderson MJ. Distance-based tests for homogeneity of multivariate dispersions. Biometrics. 2006;62:245–53. 10.1111/j.1541-0420.2005.00440.x. [DOI] [PubMed] [Google Scholar]
- 38.Delsuc F, Metcalf JL, Wegener Parfrey L, Song SJ, Gonzalez A, Knight R. Convergence of gut microbiomes in myrmecophagous mammals. Mol Ecol. 2014;23:1301–17. 10.1111/mec.12501. [DOI] [PubMed] [Google Scholar]
- 39.Peng L, Li ZR, Green RS, Holzman IR, Lin J. Butyrate enhances the intestinal barrier by facilitating tight junction assembly via activation of AMP-activated protein kinase in Caco-2 cell monolayers. J Nutr. 2009;139:1619–25. 10.3945/jn.109.104638. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Furusawa Y, Obata Y, Fukuda S, et al. Commensal microbe-derived butyrate induces the differentiation of colonic regulatory T cells. Nature. 2013;504:446–50. 10.1038/nature12721. [DOI] [PubMed] [Google Scholar]
- 41.Rahat-Rozenbloom S, Fernandes J, Gloor GB, Wolever TMS. Evidence for greater production of colonic short-chain fatty acids in overweight than lean humans. Int J Obes. 2014;38:1525–31. 10.1038/ijo.2014.46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Morrison DJ, Preston T. Formation of short-chain fatty acids by the gut microbiota and their impact on human metabolism. Gut Microbes. 2016;7:189–200. 10.1080/19490976.2015.1134082. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Raw sequencing reads have been deposited in the NCBI Sequence Read Archive under BioProject accession number PRJNA890888.
