Animal selection

Sampling

Cloacal samples were collected from both the captive and wild EISs in the same manner. First, the skin around the cloacal opening was wiped clean with 70% isopropyl alcohol-soaked gauze (alcohol prep pads, McKesson Medical-Surgical Inc., Richmond, VA, USA). Once the skin was dry, a sterile HydraFlock® swab (sterile collection swab, Zymo Research Corp. Cat. No. C1100-80, Irvine CA, USA) was inserted roughly 2 cm and rotated within the cloacal lumen five times. The cloacal swabs were then placed in a microbial DNA preservative buffer (DNA/RNA Shield TM, Zymo Research Corp. Cat. No. R1108, Irvine, CA, USA), maintained in freezer storage at approximately –16.5 °C for 8–12 weeks, and sent to MiDOG LLC for DNA extraction and sequencing.

Metagenomics analysis methods

Genomic DNA (gDNA) was extracted and purified using a commercial kit (ZymoBIOMICS^® DNA Microprep Kit (D4301, Zymo Research, Irvine, CA). gDNA samples were profiled with shotgun metagenomic sequencing. Sequencing libraries were prepared with the Illumina DNA Prep Kit (Illumina, San Diego, CA) following the manufacturers protocol using 10-base-pair (bp) unique dual indexes. All libraries were quantified with Qubit (Thermo Fisher Scientific) and then pooled together by equal abundance. The final pool was quantified using quantitative Polymerase Chain Reaction (qPCR). The gDNA was sequenced using the Illumina Novaseq X platform, generating 151-bp paired-end reads. The library produced 20 million (M) paired-end reads per sample, which were trimmed with Trimmomatic v.0.33 [25] to remove low quality fractions and adaptors. Quality trimming was performed on the reads by a sliding window with 6-bp window size and a quality cutoff of 20. Reads with a remaining size lower than 70-bp were removed. Low-diversity reads were detected and removed with sdust v.0.1 (GitHub - lh3/sdust: Symmetric DUST for finding low-complexity regions in DNA sequences). Reads that survived quality trimming were mapped against the genome of Drymarchon corais (GCA_043091225.1) using Burrows-Wheeler Alignment (BWA) v.0.7.18 [26] and removed from the pool to eliminate host sequences. The microbial community composition was profiled with Sourmash v 4.8.11 [27] using the k = 51 and scaled = 1000 options to build k-mer signatures from the FASTQ files. These signatures were searched against internally decontaminated databases of publicly available genomes from GTDB and NCBI. These databases included bacterial (n = 64,792), fungal (n = 3,274), DNA viral (n = 23,177), protozoan (n = 485), and metazoan (n = 412; Nematoda and Platyhelminthes only) genomes, where n reflects the number of unique species represented, not the total number of genome assemblies. The Sourmash results were filtered with a coverage threshold of at least 5 kilobase pairs (kbp) to eliminate false positives. The filtered Sourmash results were combined and converted to a phyloseq object for downstream analysis using the R package sourmashconsumr v 0.1.0 [28]. Taxa abundances were calculated using the abundance weighted number of unique k-mers, 51-bp-long DNA sequences derived from the sequencing data that are unique to each taxon in the databases.

Statistical analyses were conducted with R v.4.4.1 [29] using the package PHYLOSEQ v 1.50.0 [30]. Samples were divided into three groups for comparative microbiome analysis: captive snakes with C. serpentis infection (Positive, n = 30), captive snakes without C. serpentis infection (Negative, n = 30), and wild snakes (Wild, n = 20). The relative abundances of bacteria, fungi, protozoa, bacteriophages, and nematodes were recorded across the three groups of cloacal swabs. Alpha diversity metrics were calculated using the Shannon index and number of observed taxa from each group. Pairwise Wilcoxon tests were performed between C. serpentis-positive and C. serpentis-negative snakes, and between C. serpentis-positive and wild snakes, to test for significant differences in total microbial alpha diversity. Beta-diversity was calculated using Bray-Curtis dissimilarity and Permutational Multivariate Analysis of Variance (PERMANOVA) using adonis2 from the vegan v.2.6.10 R package [31] with 999 permutations. Microbial community structure was analyzed using a Canonical Analysis of Principal Coordinates (CAP) to examine the influence of C. serpentis on the cloacal microbiome composition. A linear regression analysis was performed to test possible effects of ambient temperature and snake age on microbial Shannon diversity.

Wild snakes were removed from the dataset and a linear discriminant analysis effect size (LEfSe) test was performed using the R package microbiomeMarker v.1.12.2 [32] to identify microbial taxa that were significantly enriched in C. serpentis-positive vs. C. serpentis-negative captive snakes. Bacterial, fungal, bacteriophage, nematode and protozoan species were determined to be significantly enriched by a linear discriminant analysis (LDA) score ≥ 2.5 and P-value ≤ 0.05. DESEQ2 v.1.50.2 [33] was used to further statistically confirm differential abundance of taxa between C. serpentis-positive and C. serpentis-negative snakes using absolute log2fold change > 1.5 and P-value ≤ 0.05, as it models count data directly, accounts for variance-mean dependence, and performs robust statistical testing with multiple testing correction.

To assess the functional differences in cloacal microbiomes, HUMAnN3 v3.6 [34] was used to quantify the abundances of bacterial gene families and metabolic pathways. The trimmed, host-depleted reads were concatenated and aligned against the UniRef90 and ChocoPhlAn databases using default parameters. Significant differences in pathway abundances between wild, C. serpentis-positive, and C. serpentis-negative snakes were tested for using MaAsLin2 v.1.16.0 R package [35]. Wild snakes were again removed from the study to test for significant pathway differences between captive C. serpentis-positive and negative snakes.

Bacteria

In our study, bacteria represented the dominant kingdom in the cloacal microbiome of EISs. Previous microbiome studies have identified Pseudomonadota (formerly Proteobacteria), Bacteroidota (formerly Bacteroidetes), and Bacillota (formerly Firmicutes) as major core phyla associated with the reptilian gastrointestinal tract [2]. Pseudomonadota was the most abundant bacterial phylum detected in our study, consistent with findings from other reptile gut and/or cloacal microbiome studies utilizing 16S rRNA amplicon sequencing [1,3,6,9,12,22,40]. As a facultative anaerobe, Pseudomonadota is thought to aid in the breakdown of complex sugars, promote nutrient absorption, and provide a protective role in immune responses against infection or inflammation [16,37,41]. Aerobic conditions in the cloaca may lead to this environment being dominated by Pseudomonadota [3,42]. However, a persistent overrepresentation of Pseudomonadota may indicate dysbiosis or underlying gastrointestinal disease [1]. Notably, other studies have shown that the gut bacterial microbiota of some snake species, such as captive Burmese Pythons, are dominated by both Bacteroidota and Bacillota phyla [43]. Our study detected only a small proportion of microbes from the phylum Bacillota. These findings suggest that the cloacal microbiome of EISs may harbor distinct microbial communities upstream in the gastrointestinal tract, which could differ from other snake species.

In our LEfSe analysis, Aeromonas hydrophila (A. hydrophila) and Edwardsiella tarda (E. tarda) were strongly enriched in captive C. serpentis-positive snakes. This result suggests the presence of potential bacterial co-morbidities that may interact with C. serpentis under captive conditions. Both A. hydrophila and E. tarda are frequently implicated as fish pathogens and are widely acknowledged as zoonotic organisms with enteropathogenic potential in humans [44,45]. Aeromonas spp., including A. hydrophila, are known to cause gastroenteritis, wound infections, and bacteremia with virulence factors that facilitate intestinal colonization and invasion [46–48]. Typically, E. tarda presents with human gastroenteritis and in severe cases can lead to bacteremia [49,50]. Although all captive snakes were periodically fed prey items originating from aquatic environments that may serve as sources of Aeromonas and Edwardsiella spp., including capelin (Mallotus villosus), rainbow trout (Oncorhynchus mykiss), and American bullfrog legs (Lithobates catesbeianus), the absence of enrichment in C. serpentis-negative snakes suggests that dietary exposure alone is unlikely to explain the observed association.

Experimental work further supports the possibility of synergistic interactions between Aeromonas and Cryptosporidium. In a murine coinfection model, A. hydrophila and C. parvum resulted in mutually prolonged colonization within the spleen and intestines [51]. This data suggests that Aeromonas and Cryptosporidium may enhance their pathogenic potential and create a more significant infectious burden than either organism would cause alone. The detection of E. tarda alongside A. hydrophila in C. serpentis-positive snakes may reflect gastrointestinal dysbiosis associated with captivity and promote conditions where multiple opportunistic pathogens can coexist and interact, thereby exacerbating the disease process and posing potential zoonotic risks [20,52].

In the present study, there was a significant difference in microbial beta diversity between all three study groups. Similarly, differences in beta diversity between wild and captive populations have been reported across multiple reptile species [4,22,42,53]. In our findings, wild EISs exhibited greater Shannon diversity of cloacal microbiota taxa compared to their captive counterparts. Greater species Shannon diversity observed in wild snakes may result from increased environmental microbial exposure, a more varied diet, and other environmental factors (temperature, humidity, brumation etc.). It is important to note however that environmental microbial communities were not evaluated in this study. Anthropogenic influences, such as human disturbance, may also contribute to microbiome differences between groups, as suggested in Sceloporus lizards [7]. Future studies that analyze environmental microbiota will be important for understanding how habitat, captivity, and host health interact to shape microbial community composition.

Fungi

Research on the fungal component of the reptilian gut microbiome remains limited, and studies of the cloacal microbiome are even more sparse, with most work focusing on yeasts as potential pathogens and their role as environmental reservoirs using the internal transcribed spacer (ITS) region [54–56]. In our study, LEfSe analysis showed a significant enrichment of Trichosporon asahii (T. asahii) in the cloacal microbiome of C. serpentis-positive snakes. Notably, T. asahii occurred exclusively in captive C. serpentis-positive snakes, suggesting a potential association between both pathogens. Immunocompromised animals are more prone to invasive yeast infections [57], making this association biologically plausible. In EISs, C. serpentis infection has been associated with stress-induced immunosuppression [19,58,59], a phenomenon similarly documented in humans with HIV [55,56].

Trichosporon yeasts are widespread in soil, can cause life-threatening systemic infections in immunocompromised hosts, and are resistant to multiple antifungal medications [55,56,60,61]. Trichosporonosis in reptiles is most often reported as opportunistic shell infections in chelonians but has also been associated with fatal disseminated disease in Banded Rock Rattlesnakes (Crotalus lepidus klauberi), as well as lesions affecting the skin, lungs, gut, liver, kidney, eyes and oral cavity of tortoises, lizards, snakes, and crocodilians [61,62]. In our study, T. asahii was restricted to captive C. serpentis-positive snakes, likely reflecting pathogen-driven dysbiosis, host immunosuppression, or environmental exposure (e.g., water dishes, humid hides, substrate, etc.). Its absence in wild and C. serpentis-negative snakes suggests it may opportunistically colonize disrupted microbial environments. Additionally, Trichosporon species have been isolated from 9% of healthy squamates, with molecular analyses identifying T. asahii, T. inkin, and T. cutaneum [62]. These findings suggest Trichosporon spp. may have a role as both commensal flora and opportunistic pathogens under stress or immunosuppression.

We also detected Ophidiomyces ophiodiicola (O. ophiodiicola), the agent of snake fungal disease (SFD), in all study groups. SFD is an emerging infectious disease in North American snakes, which causes dermal crusts that can progress to ulceration, invasion of deeper tissues, and fatal systemic infection [63,64]. O. ophiodiicola infected Northern Cottonmouths (Agkistrodon piscivorus) had a median survival of 90 days and 40% mortality in experimental trials [63]. O. ophiodiicola also induces skin microbiome dysbiosis [65], suggesting that gut or cloacal colonization may result in similar pathogen-driven community shifts. Although cloacal sites were disinfected with 70% isopropyl alcohol-soaked gauze prior to sample collection, the possibility of contamination during swabbing cannot be entirely ruled out. More invasive sampling approaches, such as biopsy, may be necessary to definitively confirm its presence within the gastrointestinal tract.

Multiple other opportunistic fungi were detected including Diutina catenulata, Purpureocillium lilacinum, Aspergillus spp., Alternaria alternata, and Aureobasidium melanogenum. Diutina catenulata (formerly Candida catenulata), a common yeast in human and animal gut microbiomes, was identified across all study groups [54]. In contrast, Purpureocillium lilacinum, a saprophytic fungus and known opportunistic pathogen [66], was detected in captive C. serpentis-negative and wild snakes but was rare in C. serpentis-positive individuals. Both organisms were likely acquired through diet or environmental exposure [56,67,68], with reduced abundance of Purpureocillium lilacinum in infected snakes possibly due to disease-associated microbiome changes. Aspergillus spp., Alternaria alternata, and Aureobasidium melanogenum were detected only in wild snakes, consistent with increased exposure to environmental fungal spores. Their absence in captive EISs may be due to reduced exposure in controlled environments with air filtration systems. Collectively, these findings suggest that the presence and distribution of opportunistic fungi in the cloacal microbiome may be influenced by a combination of environmental exposure, diet, and disease status.

Protozoa

Only 2 of 30 known C. serpentis-positive captive snakes were actively shedding oocysts at the time of cloacal swab sampling. Low-to-intermittent oocyst shedding is well documented in snakes with gastric cryptosporidiosis and complicates prevalence estimates when single time-point cloacal swabs are used [19,69,70]. In addition to intermittent shedding, sampling methodology may also have contributed to the low prevalence in our captive positive samples. Recent comparative work in EISs shows that diagnostic yield for C. serpentis varies by specimen type and test. Gastric biopsy outperformed cloacal swabbing and C. serpentis-specific probe hybridization qPCR outperformed Hsp70 PCR for detecting low-level or intermittent oocyst shedding [58]. Gastroscopy with targeted gastric biopsy performed three days after a meal remains the most sensitive sample type and C. serpentis-specific probe hybridization qPCR paired with histologic analysis of gastric mucosa remains the most sensitive test type for detection of C. serpentis [19,71]. Using serial cloacal or gastric swabs and sensitive assays such as C. serpentis-specific probe hybridization qPCR to improve detection may be a viable alternative if gastric biopsy is unable to be performed [58].

The restriction of Cryptosporidium chipmunk LX-2015 (also called chipmunk genotype I) and Cryptosporidium hominis (C. hominis) to wild snakes in our dataset suggest environmental spillover rather than an established snake infection. Both taxa are recognized human pathogens, with C. chipmunk LX-2015 being confirmed as zoonotic [72,73], and C. hominis being the dominant anthroponotic species [74,75]. Their detection in wild snakes could reflect ingestion of infected prey (e.g., small mammals or reptiles), exposure to contaminated surface water, or transient gut passage without gastrointestinal colonization. These results highlight the role of wildlife–environment–human interfaces in potential transmission pathways.

Protozoa were not detected in the negative captive group. However, because C. serpentis sheds intermittently and primarily affects the stomach, continued monitoring with repeated samples is recommended for colony management and conservation programs.

Viruses

The virome analysis revealed the presence of bacteriophages as the only detectable viral group. Notably, wild snakes exhibited a reduced bacteriophage diversity relative to their captive counterparts. One plausible explanation for this observation is that wild snakes may harbor a greater proportion of viral dark matter, which would include novel or uncharacterized viruses that remain undetectable with current reference databases. An additional consideration is that metagenomic sequencing in this study was limited to DNA viruses, thereby excluding RNA viruses from detection and potentially underestimating the overall viral diversity present in our samples.

Nematodes

Nematode diversity was limited and no reptile-specific pathogens were identified. Steinernema diaprepesi, an entomopathogenic nematode, was consistently detected across all groups. This association suggests it may represent an environmental contaminant (possibly from substrate or consumed prey item) rather than a host-associated parasite [76]. Wild snakes harbored the greatest diversity of nematodes, consistent with broader environmental exposure compared to the restricted conditions of captivity.

Sampling methodology and study limitations

Sampling methodology can influence microbial diversity. Previous studies have shown that microbial composition varies along the gastrointestinal tract [1,12] and aerobic conditions in the cloaca may influence the microbial community present [3,42]. Cloacal swabs, while providing a broad overview of gut bacterial diversity, may not fully represent hindgut communities. For example, in Northern Cottonmouths (Agkistrodon piscivorus), nearly 50% of the microbial composition in the cloaca was dominated by Pseudomonadota and reflected similar bacterial diversity in the small intestines [12]. Conversely, fecal samples are often more representative of hindgut communities and may exhibit higher alpha diversity, as observed in sea turtles where Bacillota and Bacteroidota predominated, while Pseudomonadota dominated cloacal microbiomes [3]. In Sceloporus virgatus lizards, fecal samples were enriched with Bacillota and Bacteroidota, while cloacal swab samples exhibited a higher abundance of Pseudomonadota [14]. Furthermore, cloacal and fecal sampling methods revealed significant differences between captive and wild Fijian crested iguanas (Brachylophus vitiensis), with each sample type exhibiting distinct bacterial community structure and diversity [4]. Given these considerations, future minimally invasive studies on EIS should incorporate fecal sampling alongside cloacal swabs to obtain a more comprehensive view of gut and cloacal microbial diversity and abundance. The combined approach could reveal microbial communities and patterns that might otherwise be underrepresented when relying solely on cloacal swabs.

Several limitations should be considered when interpreting these findings. First, environmental samples such as soil, water, or enclosure surfaces were not collected alongside cloacal swabs. The cloaca is directly exposed to the external environment, making it plausible that environmental microbes may influence the cloacal microbial community. Without environmental sampling, however, the contribution of environmental microbes to the cloacal microbiome in wild and captive EIS remains inferential and warrants further investigation in future studies. However, a study in Texas horned lizards (Phrynosoma cornutum) reported that environmental microbes did not significantly contribute to the cloacal microbiome community [22]. Dietary microbiota sampling of captive prey items and evaluating their contribution to the captive EIS cloacal microbiome may also be of benefit [22]. Additionally, a limitation to microbiome analysis using shotgun metagenomic sequencing is the availability of reference genomes to classify DNA reads. While genomic resources for bacteria are relatively robust, the same is not true for the protozoan and metazoan species. For example, the genome for Cryptosporidium varanii has not been sequenced, and these samples could not be evaluated for C. varanii by this methodology. Similarly, at the time of this writing, there is no published genome of the host Drymarchon couperi and the genome of Drymarchon corais was used as a substitute to remove host reads from the dataset. The reader should interpret the data with this in mind, but as metagenomic sequencing continues to be a growing field, genomic databases continue to expand and taxonomic resolution will improve. In our study, a substantial portion of sequencing reads could not be taxonomically classified. This underscores the importance of acknowledging and reporting this undescribed DNA, or “dark matter”, in the samples (see supplemental data) to enable future studies and comparisons. The authors encourage similar studies to report detected dark matter in their samples and pursue follow-up experiments using the sequence data deposited with this manuscript.