Wild black bears harbor simple gut microbial communities with little difference between the jejunum and colon

The gut microbiome (GMB), comprising the commensal microbial communities located in the gastrointestinal tract, has co-evolved in mammals to perform countless micro-ecosystem services to facilitate physiological functions. Because of the complex inter-relationship between mammals and their gut microbes, the number of studies addressing the role of the GMB on mammalian health is almost exclusively limited to human studies and model organisms. Furthermore, much of our knowledge of wildlife–GMB relationships is based on studies of colonic GMB communities derived from the feces of captive specimens, leaving our understanding of the GMB in wildlife limited. To better understand wildlife–GMB relationships, we engaged hunters as citizen scientists to collect biological samples from legally harvested black bears (Ursus americanus) and used 16S rRNA gene amplicon sequencing to characterize wild black bear GMB communities in the colon and jejunum, two functionally distinct regions of the gastrointestinal tract. We determined that the jejunum and colon of black bears do not harbor significantly different GMB communities: both gastrointestinal sites were dominated by Firmicutes and Proteobacteria. However, a number of bacteria were differentially enriched in each site, with the colon harboring twice as many enriched taxa, primarily from closely related lineages.

Alpha and beta diversity. Overall, the top linear mixed effect model (LMM) for Faith's phylogenetic diversity (here after Faith's PD) included no interactions (Supplementary Table S1) and only age-class significantly influenced Faith's PD (β sub = 0.3, β adu = −0.07, χ 2 = 8.62, p = 0.014; Table 1); for means of each alpha diversity index, see Supplementary Figure S1 and Supplementary Table S2. Contrasts of estimated marginal means (EMMs) for age-class revealed that Faith's PD was driven by differences between subadults and adults only (p = 0.05; Table 2). We found no significant differences in either Shannon diversity or inverse Simpson diversity between gastrointestinal tract sites, sexes, or among age-classes in the top LMM models (Table 1).
In terms of beta diversity, there were significant effects from gastrointestinal tract site, age-class, and sex; however, the effects had very little explanatory power regarding variation in weighted UniFrac distances Scientific Reports | (2020) 10:20779 | https://doi.org/10.1038/s41598-020-77282-w www.nature.com/scientificreports/ (gastrointestinal: R 2 = 0.08, p = 0.03; Age-class: R 2 = 0.03, p = 0.05; sex: R 2 = 0.005, p = 0.007). Ordination plots show colon communities were more conserved. We also observed substantial overlap in microbiome composition between the two sites (Supplementary Figure S2); yet, there appeared to be no substantiated clusters on PCoA plots for gastrointestinal site or age-classes (Supplementary Figure S2). Indeed, jejunum and colon appeared to harbor phylogenetically similar GMBs, showing very little difference in intra-and inter-variation in either phylogenetic representation or relative abundance (Fig. 1A). Weighted UniFrac distances also showed little difference within and between sexes (Fig. 1B). For age-classes, we detected the greatest variation within adults (Fig. 1C). PERMDISP results were not significant for any of the three factors.
Community composition of the gastrointestinal tract sites. We identified three major phyla (Firmicutes, Proteobacteria, Actinobacteria) in the jejunum and three major phyla (Firmicutes, Proteobacteria, and Epsilonbacteraeota) in the colon ( Fig. 2A). We defined major taxa as representing > 1% of amplicon sequence variants (ASVs). In addition to the three major phyla identified in each gastrointestinal tract site, we identified 21 minor phyla that were present at < 1% relative abundance. Firmicutes and Proteobacteria were the only major phyla detected in all samples, with Firmicutes being the most dominant phylum in both the jejunum community (71% ± 34% SD) and colon community (60% ± 33% SD), whereas Proteobacteria was the second most dominant phylum in the jejunum community (24% ± 30% SD) and the colon community (33% ± 30% SD). Actinobacteria (1.6% ± 2.9% SD) was a third phylum that was unique to the jejunum, while Epsilonbacteraeota (5.4% ± 10% SD) was a third phylum unique to the colon. All other phyla were minor in both gastrointestinal communities. At the genus level, we identified 21 major genera across both gastrointestinal sites. The jejunum harbored 13 major taxa, six of which were unique to the jejunum. The colon harbored 15 major taxa, eight of which were found solely in the colon communities (Fig. 2B). Sarcina (17% ± 33% SD) dominated the jejunum, followed by Lactobacillus (12% ± 26% SD), and Escherichia-Shigella (12% ± 20% SD). Escherichia-Shigella dominated the colon community (27% ± 27% SD), followed by Sarcina (15.8% ± 28% SD).
We found a total of 23 ASVs that were differentially represented between the jejunum and colon at genus level (LDA score ≥ 3.65, p < 0.05; Table 3). In the jejunum, we identified seven differentially abundant bacteria. All were minor genera belonging to the three major phyla (4 Proteobacteria, 2 Actinobacteria, 1 Firmicutes; Table 3), though none could be classified below the level of order.
We identified five unique and enriched major bacterial taxa in the colon: Turicibacter (8.6% ± 19% SD), Helicobacter (5.3% ± 10% SD), one unidentified genus within the family Enterobacteriaceae (1.2% ± 3% SD), and two unidentified genera in the family Peptostreptococcaceae (1.2% ± 1.9% SD; Table 3). Further, although found in both the jejunum and colon, Escherichia-Shigella, Clostridium sensu stricto1, and nine unidentified minor genera within the three major phyla (Firmicutes, Proteobacteria, and Epsilonbacteraeota) were significantly more abundant in the colon. Although colon samples harbored approximately twice as many enriched bacterial taxa as the jejunum, the majority of the differential colonic membership comprised unidentified genera within the same lineages occurring across the taxonomic hierarchy of the three major phyla (e.g., family, order, class). Conversely, we found a greater degree of phylogenetic variability among enriched bacteria in the jejunum compared to the colon (indicated by black stars in Fig. 3). Further, GMB taxonomic structure from the kingdom level to genus in the jejunum featured a greater abundance of minor taxa contributing to overall GMB composition, www.nature.com/scientificreports/ whereas the colon was dominated by fewer taxa primarily from the major phyla ( Fig. 3). LEfSe results revealed no significantly enriched bacteria between age-classes.

Discussion
To our knowledge, our research was the first to compare microbial communities in the jejunum and colon in a wild carnivore population. Here, we compared GMB structure between two operationally distinct regions of the gastrointestinal tract and among three age-classes of American black bears. Similar to previous studies on captive specimens 12 , black bear GMB membership and alpha diversity were influenced, in part, by age-class, with subadults' GMBs harboring greater phylogenetic diversity compared to adults (Fig. 2). Similar to other bear species, the colonic GMB of wild American black bears was dominated by the phyla Firmicutes and Proteobacteria [40][41][42][43][44][45] .
Although weighted UniFrac distances were significantly different between gastrointestinal sites, sexes, and among age-classes, these factors appeared to be minor drivers of GMB composition. Weighted UniFrac PCoA plots also showed no distinct clustering or discernible patterns (Supplementary Figure S1). Given the different physiologies and micro-environments of the jejunum and colon, with fat, carbohydrates, and protein absorption occurring in the jejunum, and fermentation primarily occurring in the colon, the lack of discernible differences as measured in for distance matrices and alpha diversity was unexpected, especially considering that differences in community structure and membership have been observed previously across gastrointestinal sites in other omnivorous species 37 . However, perhaps due to site-specific environmental differences (i.e., host digestive processes and associated nutrient availability, we did observe that the colon contained high levels of differently enriched taxa from unknown but closely related lineages of bacteria (Table 3). Three differently enriched taxa in the colonic GMB community (Turicibacter, Clostridium sensu stricto 1, and Escherichia/Shigella) were also previously found to be major genera in captive Asiatic black bear 44 and giant panda 38 GMBs. By contrast, jejunum GMBs featured more minor taxa and phylogenetic branching contributing to jejunum GMB community composition (Figs. 2, 3).   28 , and the overall simplicity of the ursine gastrointestinal tract (i.e., short length and lack of vestigial organs such as appendix and cecum) may keep the jejunum and colon in a constant state of disturbance and thus prevent the development of discernible differences in community structure. Specifically, as generalist consumers with a simplistic gastrointestinal tract and rapid digestion, the climax community of the black bear GMB could be dominated by bacteria that function as pioneer species in other hosts with more complex guts and longer retention (i.e., succession) times. Although bacterial colonization is prompt, rapid transit time from consumption to defecation could facilitate constant and rapid turnover of microbial membership, preventing succession from progressing beyond a pioneer stage and favoring generalist/opportunistic microbial species. Furthermore, the black bear's simple gastrointestinal tract affords no refuge where microbes can escape the constant turnover. The cecum and appendix are important morphological features whose function appear to be linked to immune system maintenance in mammals 46,47 , involving preservation of microbial biofilms in a location that is relatively sheltered from the main flow of intestinal contents 46 . This function is suggested to be important for the recovery of the microbiome following microbial loss due to diarrheal illnesses 46,48,49 and as a reservoir for microbes 46 , a concept corroborated by recent studies 37,50 . For example, Greene and McKenney 50 opportunistically sampled GMBs from the appendix, cecum, and colon of deceased captive aye-ayes and found that the appendix harbors distinct microbial diversity and composition, whereas the cecum and colonic communities were more homogenous 50 . Although species with physiologically more complex gastrointestinal tracts might harbor more stable microbial communities, carnivores such as black bears possess neither a cecum nor an appendix and thus lack a potential microbial reservoir to buffer microbial communities from the constant disturbance associated with a simple gastrointestinal tract and rapid food passage transit times.
The presumed turn-over of microbial communities within the gastrointestinal tract of black bears might serve as an evolutionary advantage, not in association with specialized digestion of specific foods, but regarding adiposity. In humans and mice, dysbiosis (i.e., an imbalance in microbial membership, often brought about through dietary shifts and often presumed to be detrimental) and low GMB diversity can lead to increased capacity for energy harvest and obesity 16,51,52 . Although obesity has been linked to health aliments in humans 53,54 , the need for black bears to undergo hyperphagia (i.e., physiological Stage IV) to rapidly gain weight is paramount for reproductive success and survival during torpor (i.e., physiological Stage I). For example, previous research shows that the seasonal composition and structure of GMBs in brown bears differ between physiological stages, with gut microbiota promoting energy storage during hyperphagia 40 . Thus, a constant state of microbial disturbance within black bear GIT could lead to more rapid weight gain. www.nature.com/scientificreports/ Although we did not observe any differentially enriched bacteria among any age-classes, the overall observed increase in phylogenetic diversity could provide evidence of GMB colonization across the gastrointestinal tract as bear age. For instance, Faith's PD was the lowest in yearlings, peaked in subadults, and subsequently decreased in adults (Fig. 1C). Notably, we also detected greater variability in Faith's PD among subadults (Supplementary Figure S1), which could suggest subadults were in an intermediate succession states as previously observed in captive lemurs 12 . Additionally, we observed 5 major phyla, including greater proportions of Helicobacter and Escherichia-Shigella, both known potential pathogens, in yearlings (Fig. 1C,D). Both number of phyla and the relative abundance of Helicobacter gradually decrease in subsequent age-classes. Our results could suggest that as black bears mature, their bodies are better able to filter or select for specific microbes, i.e., via the immune system.
Interestingly, we also discovered the presence of the minor genus Ursidibacter (family Pasteurellaceae) in several jejunum and colon samples in the present study. Ursidibacter (family: Pasteurellaceae) was first sequenced from oral-cavity swabs collected from both wild and captive polar bears, and from captive brown bears 55 . Almost all members of the family Pasteurellaceae appear to be closely coupled to a single vertebrate host and are believed to be adapted to specific habitats 56 . Pasteurellaceae are generally pathogenetic or commensal bacteria, unable to survive in external environments, and have been found in the upper respiratory tract, throat, reproductive tract, and the gastrointestinal tract of vertebrate hosts 56 . To our knowledge, Ursidibacter has not been reported in other Ursid GMB communities since it was first sequenced in 2015, which may result from the range and bias of reference databases/versions used for bioinformatic analysis. Like the genus Prevotella in non-human primates 57,58 , Ursidibacter may have co-evolved and radiated with the family Ursidae.
With the increasing vulnerability of large carnivore populations to human-mediated environmental change worldwide, emphasis on bettering management and conserving these charismatic and ecologically significant species is paramount. Studying carnivore GMBs provides an opportunity to identify effects of anthropogenic pressures on carnivores not solely from a behavioral perspective or as a result of direct persecution but also in revealing how carnivores respond physiologically to human pressures (e.g., habitat degradation, access to processed/human foods) and the potential consequences of those physiological responses to the health of carnivores. Moreover, by better understanding the GMBs of generalist hosts and how the GMB responds to environmental change, wildlife managers can consider the value of incorporating strategies to promote holobiont conservation (e.g., the conservation of both host and GMB diversity) 59 when striving to create effective management plans for species coping with human-mediated environmental change. To better understand the potential co-evolutionary history of Ursidae and their GMBs, future research could expand upon our study by investigating (1) whether dynamic, low-diversity GMBs may provide an evolutionary advantage for wildlife species with unique life histories such as black bears with future studies which incorporate multi-omics (i.e., metabolomic, metatranscriptomic) and (2) the bacterial lineage of Ursidibacter.

Methods
Study area and sample collection. We received an exemption from review by the Northern Michigan University's Institutional Animal Care and Use Committee because samples were collected opportunistically from dead bears that were legally harvested by hunters, separate from this study. We collected samples with permission from individual hunters/guides under a Michigan DNR-Wildlife Division-Scientific Collector's Permit (#SC 1613). We sampled black bears across the Upper Peninsula (UP) of Michigan, USA (7° 00′-45° 09′ N, 90° 18′-84° 37′ W). The UP consists of primarily coniferous and deciduous hardwood forests with intermittent conifer swamps, wetlands, shrub patches, and agriculture with elevation ranging between 170 to 600 m above sea level. Average daily temperatures during the September-October sampling frame varied from a low of − 1.6 °C to a high of 22 °C. We collected colon and jejunum content from legally harvested black bear and one roadkill bear (n = 35) within 30 min of death during the annual fall harvest season (September 10-October 26, 2018). Each guide/ hunter was provided pre-packaged, color coordinated sampling kits. All kits were packed under sterile conditions in the lab, and hunters did not open the kits until after they had harvested the bear. Prior to collection, we met with each hunter/guide face to face individually to review the protocol as literature in citizen science have established face-to-face interactions are known to increase motivation and participation in online projects 60 . Further, Budde et al. 61 demonstrated that a combination of technical instruction (i.e., provided kits with clearly worded and color-coded protocols/materials) and instruction (i.e., face-to-face meetings) combine synergistically to significantly reduce error rates.
After harvesting the bears, guides/hunters donned nitrile gloves and used the sterile tongue depressor to collect colon samples from the anus and immediately placed the sample into centrifuge tubes containing 7 mL of 95% ethanol. After guides/hunters field dressed the bears by opening the abdomen and mobilizing the stomach and intact intestines, they donned new sterile nitrile gloves to prevent contamination of samples with human/ environmental microbes, used the provided 16-inch string to sample the gut microbiome at a standardized distance from the pyloric sphincter and used a clean knife to make a small incision in the jejunum. Finally, squeezed the intestine to extrude gut content straight into a pre-labelled sample tube containing 7 mL of 95% ethanol (see Supplementary Figure S3 for hunter instructions). All samples were stored at room temperature until microbial DNA was extracted (~ 50 days). Sex was recorded and the Michigan Department of Natural Resources provided age data from teeth they collected per their harvest registration protocols (Supplementary Table S3). Using this data, bears were categorized into age-classes (i.e., yearling = 1, subadult = 2-3, adult ≥ 4).

DNA isolation and sequencing.
A total of 66 samples were collected from black bears across all ageclasses to be used for 16S sequencing from seven yearlings (all yearlings had both jejunum and colon samples), 15 subadults (13 subadults had both jejunum and colon samples), 11 adults (nine adults had both colon and jejunum samples), and two bears from unknown age-classes (both with jejunum and colon samples; metadata for all individuals are available in Supplementary Table S3). We extracted microbial DNA from jejunum and colon samples using DNeasy PowerSoil Kits (QIAGEN, Hilden, Germany), following the manufacturer's protocol with the addition of (1) a heat-step of 10 min at 65 °C at the beginning of the protocol to breakdown proteins and (2) a second elution as the final step of extraction, as previously described by McKenney et al. 62 . We quantified DNA yields using a NanoDrop 2000c (ThermoFischer Scientific, Massachusetts, USA) and stored extractions at − 20 °C. After all extractions were complete, standardized DNA aliquots were shipped to Argonne National Laboratory (Lemont, IL, USA) for PCR amplification and paired-end DNA sequencing of the 16S rRNA V4 gene region, according to methods described by Caporaso et al. 63 .
Bioinformatic analysis. Multiplexed EMP-paired-end sequence reads were imported into Quantitative Insights Into Microbial Ecology (QIIME2, version 2019.4) 64 and demultiplexed. Sequences were joined, denoised, filtered to remove chimeras and residual Phix reads, and dereplicated; ASVs were called using the DADA2 QIIME2 plugin 65 , and sequence lengths were truncated to 150 bp.
Taxonomic classification. We used the SILVA 99 database for V4 region (version 132) 66 to assign taxonomic classification in QIIME2, using a trained Naïve Bayes sklearn classifier 67 to classify organisms at the genus level. Sequences were aligned with MAFFT 68 , a plugin for phylogenetic diversity analysis, which removes highly variable positions in the process. We further filtered samples to remove chloroplast, mitochondrial, and Archaea sequences, as well as unidentified microbial DNA unidentified below kingdom. Statistical analysis. All statistical analyses and visualizations were performed using R (version 3.6.1) 69 and Rstudio (version 1.2.1335) 70 . We investigated the alpha diversity of GMB communities per gastrointestinal tract site through analysis of the Shannon and inverse Simpson diversity indices, and of Faith's PD. We use linear mixed effects models (LMM) for analyses to determine the relationship between Faith's PD and gastrointestinal tract sites. We included gastrointestinal tract site, sex and age-class (yearling = 1, subadult = 2-3, adult ≥ 4) as categorical fixed effects; alpha diversity indices as the response variables; and individual as a random effect. In each model, we checked residuals to confirm model requirements (e.g., normality, homoscedasticity, residuals). To determine whether interactions between main effects should be considered, we fit models with the maximum likelihood (ML) and compared likelihood ratio tests, which performs Wald Chi-squared tests for LMM. Final models were fit with restricted maximum likelihood (REML). We also determined the significance of main effects and interactions within the top models using Wald Chi-squared tests. Faith's PD and inverse Simpson diversity values were log-transformed prior to analysis, to accommodate their skewed values. If main effects or Scientific Reports | (2020) 10:20779 | https://doi.org/10.1038/s41598-020-77282-w www.nature.com/scientificreports/ interactions were significant, we acquired estimated marginal means (EMMs) of pairwise comparisons for post hoc testing with Tukey adjustment. For beta diversity, we quantified compositional dissimilarity between gastrointestinal tract sites, sex, and age-class using the quantitative phylogenetic weighted UniFrac distance and plotted these data on principle coordination analysis (PCoA) ordination plots. We performed permutational multivariate analysis of variance (perMANOVA) 71 on each distance matrix with a strata for subject ID. We implemented an analysis of multivariate homogeneity (PERMDISP) 72 , an analogue of Levene's test for homogeneity of variance, again with a strata for subject ID, to test for significant differences in sample heterogeneity between gastrointestinal tract sites. Finally, we used Linear discrimination analysis Effect Size (LEfSe) in the Galaxy online tool (https ://hutte nhowe r.sph. harva rd.edu/galax y) to identify any ASVs that were differentially represented between gastrointestinal tract sites. We designated a logarithmic Linear Discriminate Analysis (LDA) score of 2.0 as the cut-off for LEfSe analysis 73 . We considered a p value threshold of 0.05 significant for each test performed. See Zenodo Digital Repository https ://doi.org/10.5281/zenod o.40604 80 for QIIME2 pipeline and code for statistical analysis.

Data availability
The data and coding reported in this paper are available and will be archived in the Zenodo Digital Repository https ://doi.org/10.5281/zenod o.40604 80.