Comparing the gut microbiome along the gastrointestinal tract of three sympatric species of wild rodents

Host and gut content sampling

A total of 94 individuals (42 A. speciosus, 9 A. argenteus, and 43 M. rufocanus) were captured from four sites within the Kamikawa Chubu national forest in the central area on the island of Hokkaido, Japan (Supplementary Table S1), and a total of 280 gut content (from the small intestine, cecum, and colon) and fecal matter (from the rectum) samples were collected for microbiome analysis (Supplementary Table S2). Based on 16S rRNA amplicon sequencing using Illumina Miseq, a total of 12,286,171 paired-end reads were obtained after quality filtering and chimeric sequence removal. There was an average of 43,879 reads per sample, although it varied among species and gut region (Supplementary Table S3).

Within host species/among gut region gut microbiota alpha diversity

Alpha diversity of the gut microbiota in the small intestine was significantly lower than the rectum, colon, and cecum in all three host species based on Shannon diversity, Faith’s PD, evenness, and number of ASVs as expected (GLME: all p < 0.01; Fig. 1, Supplementary Fig. S2, Supplementary Tables S4–S7). There was no difference in alpha diversity among the cecum, colon, or rectum within any species (GLME: all p > 0.05; Fig. 1, Supplementary Fig. S2, Supplementary Tables S4–S7). Males had significantly higher alpha diversity within all gut regions of A. speciosus while female A. argenteus had significantly higher alpha diversity as compared to males (GLME, all p < 0.02; Supplementary Tables S4–S7). There was no effect of sex on gut microbiota alpha diversity in any gut region of M. rufocanus (GLME: all p > 0.05; Supplementary Tables S4–S7) while age had no effect in any gut region of any rodent species (GLME: all p > 0.05; Supplementary Tables S4–S7).

Among host species alpha diversity

Myodes rufocanus had significantly higher alpha diversity in all four gut regions as compared to both A. speciosus and A. argenteus based on all four diversity measurements (GLME: all p < 0.01; Fig. 1, Supplementary Fig. S2, Supplementary Tables S8–S11) except for Faith’s PD of the small intestine (GLME: A. speciosus: b = − 0.057, SE = 0.095, p = 0.55; A. argenteus: b = − 0.146, SE = 0.154, p = 0.346; Fig. 1, Supplementary Table S9). There were fewer significant differences in alpha diversity between A. speciosus and A. argenteus as expected with the colon exhibiting differences based on Faith’s PD and evenness, as well as in the small intestine and cecum for Shannon diversity and evenness (GLME: all p < 0.05; Fig. 1, Supplementary Fig. S2, Supplementary Tables S8–S11). There were no significant differences in alpha diversity within the rectum between Apodemus spp., nor was there an effect of age or sex on any alpha diversity measurement in any among species analysis (Fig. 1, Supplementary Fig. S2, Supplementary Tables S8–S11).

Within host species/among gut region beta diversity

When testing for the effect of gut region on microbiome beta diversity within each species when all gut regions were included for PERMANOVA, we found gut region had a highly significant effect in all three host species regardless of distance metric (PERMANOVA: all p < 0.01; Supplementary Tables S12–S14). Field site also significantly impacted beta diversity in all three rodent species (PERMANOVA: all p < 0.03; Supplementary Tables S12–S14) though the effect size was smaller than it was for gut region except for A. argenteus according to both Jaccard and Bray–Curtis distance metrics (Supplementary Tables S12–S14). Age significantly impacted beta-diversity in both A. speciosus and M. rufocanus according to all four diversity metrics (PERMANOVA: all p < 0.01; Supplementary Tables S12, S14) while sex was significant for all except weighted UniFrac in all three species (PERMANOVA: p = 0.055 to 0.266; Supplementary Tables S12–S14) as well as unweighted UniFrac in M. rufocanus (PERMANOVA: R² = 0.011 F = 1.6, p = 0.071; Supplementary Table S14).

To further explore changes in the gut bacterial community structure along the GIT, we utilized pairwise PERMANOVAs to determine if each gut region harbored a unique bacterial community. Consistent with our hypothesis, we found that gut region had a highly significant effect when the small intestine was compared to the cecum, colon, or rectum in all three host species regardless of beta diversity metric (PERMANOVA: all p < 0.01; Supplementary Tables S15–S17). Furthermore, the effect size was much larger than it was for pairwise comparisons among the three regions of the lower GIT as we expected (Supplementary Tables S15–S17). Indeed, gut region was not always distinguishable among the cecum, colon, and rectum. Specifically, in pairwise comparisons among the three gut regions in A. speciosus, gut region had a significant effect based on Bray–Curtis and weighted UniFrac (PERMANOVA: all p < 0.05; Supplementary Table S15), but not Jaccard or unweighted UniFrac (PERMANOVA: all p > 0.05; Supplementary Table S15). No significant effect was found when the same regions were compared in A. argenteus (PERMANOVA: all p > 0.05; Supplementary Table S16) while in M. rufocanus, gut region significantly impacted beta diversity when the colon and cecum (PERMANOVA: R² = 0.04, F = 2.26, p = 0.041) as well as the colon and rectum (PERMANOVA: R² = 0.04, F = 2.35, p = 0.41) were compared based on weighted UniFrac alone (Supplementary Table S17). Our PCoA plots showed similar results as samples from the small intestine clustered separate from the others and a large degree of overlap occurred in the clustering of the cecum, colon, and rectum, but not entirely (Figs. 2, 3, Supplementary Figs. S3, S4).

In all three host species, field site had a larger effect than gut region based on R² values for pairwise comparisons among the three regions of the lower GIT (Supplementary Tables S15–S17). For example, when comparing the gut microbiota of the colon and cecum in A. speciosus based on weighted UniFrac dissimilarity, the effect of site (PERMANOVA: R² = 0.083, F = 2.217, p = 0.013; Supplementary Table S15) was more than twice as large as gut region (PERMANOVA: R² = 0.031, F = 2.449, p = 0.04; Supplementary Table S15). When comparing the small intestine to the cecum, colon, or rectum, the effect of gut region was much larger than field site for both A. speciosus and A. argenteus regardless of dissimilarity metric (Supplementary Tables S15 and S16). However, in M. rufocanus, site had a slightly larger effect than gut region for Jaccard (e.g. PERMANOVA of small intestine—cecum; gut region, R² = 0.067, F = 4.09, p = 0.001; site, R² = 0.108, F = 2.206, p = 0.001; Supplementary Table S17) and Bray–Curtis dissimilarities, but the opposite was true for unweighted UniFrac while site had no effect according to weighted UniFrac distance (e.g. PERMANOVA of small intestine—cecum: gut region, R² = 0.062, F = 1.381, p = 0.177; Supplementary Table S17). Age and sex widely effected the gut microbiota beta diversity in both A. speciosus and M. rufocanus, especially when comparing the three regions of the lower GIT, but no effect was found for A. argenteus (Supplementary Tables S15, S17).

Among host species beta diversity

Host species had a significant effect on beta diversity for the small intestine, cecum, colon, and rectum for Jaccard and Bray–Curtis as well as unweighted and weighted UniFrac distances when all three species were included (PERMANOVA: all p < 0.05; Supplementary Table S18). Field site also had a significant impact on all gut regions according to Jaccard, Bray–Curtis and unweighted UniFrac (PERMANOVA: all p < 0.05; Supplementary Table S18) although the effect size was several times smaller than host species. Sex and age had no effect in any gut region (PERMANOVA: all p > 0.05; Supplementary Table S18). The PCoA plots confirmed these findings as there was clustering according to host species within each gut region (Fig. 2, 4, Supplementary Figs. S3, S5). For the small intestine, however, there was a large overlap for weighted UniFrac as well as a small sub-clustering for both A. speciosus and M. rufocanus that could not be explained by site, age, or sex (Fig. 4).

To test if there were larger differences in the bacterial community structure between M. rufocanus and either Apodemus spp. than between A. speciosus or A. argenteus, between species pairwise analyses were utilized (Supplementary Tables S19–S21). We found that host species had a significant effect when the small intestine, cecum, colon, and rectum were compared between M. rufocanus and both species of Apodemus (PERMANOVA: all p < 0.05; Supplementary Tables S19, S21). Furthermore, the effect size was larger when comparing the lower GIT such as the cecum between A. speciosus and M. rufocanus (PERMANOVA: weighted UniFrac, R² = 0.23, F = 11.786, p = 0.001; Supplementary Table S19) than when comparing the small intestine (PERMANOVA: weighted UniFrac, R² = 0.174, F = 17.242, p = 0.001; Supplementary Table S19). When compared between A. speciosus and A. argenteus, host species was significant for Jaccard, Bray–Curtis, and unweighted UniFrac for all gut regions (all p < 0.01), but only the cecum based on weighted UniFrac (PERMANOVA: R² = 0.096, F = 3.77, p = 0.003; Supplementary Table S20). Importantly, the effect size was smaller than it was when comparing either Apodemus species to M. rufocanus (Supplementary Tables S19–S21). Field site was also significant for most pairwise comparisons (Supplementary Tables S19–S21) though notably the R² value was smaller than it was for host species in comparisons between M. rufocanus and either species of Apodemus (Supplementary Tables S19,S21). The opposite was true when comparing A. speciosus and A. argenteus as site had a slightly larger R² value than host species (Supplementary Table S20). Sex and age were rarely significant in any of the pairwise analyses (Supplementary Tables S19–S21).

Within host species/among gut region microbiota taxonomic composition

By comparing relative abundances of bacterial genera along the GIT in each host species when all GIT regions were included using LEfSe analysis, we found Ruminococcaceae NK4A21 group had significantly higher relative abundance in the rectum of A. speciosus and Treponema 2 was higher in the cecum of M. rufocanus (Supplementary Tables S22, S24). Four genera in the small intestine, 13 in the cecum, five in the colon, and 17 in the rectum of A. argenteus were found to exhibit significantly higher abundance, suggesting highly differential microbiome communities along the length of the GIT (Supplementary Tables S22, S24). It must be noted that the relatively larger number of significant differences in microbial abundances in A. argenteus may be a type 1 error due to the small sample size (nine individuals).

To develop a clearer picture regarding differential relative abundance of the various bacterial genera within the four gut regions, pairwise LEfSe analysis was conducted (Supplementary Tables S23, S25–S30). We were specifically interested if relative abundances in the small intestine were distinctly different than the cecum, colon, or rectum, as well as if there was a high degree of similarity throughout the lower GIT. We found a large number of bacterial genera with significantly different relative abundances between the small intestine and the lower GIT in all three host species with more genera exhibiting higher relative abundance in the lower GIT as compared to the small intestine than vice versa (Supplementary Table S23). Some bacterial genera such as Lactobacillus and Veillonella were found to have higher relative abundance within the small intestine in all three host species regardless of which region of the lower GIT it was compared to (Fig. 5, Supplementary Tables S25, S27, S29). Others such as Leptotrichia had higher relative abundance within the small intestine as compared to the cecum, colon, or rectum in both Apodemus spp. but not in M. rufocanus (Fig. 5, Supplementary Tables S25, S27, S29). Many exhibited a host species-specific trend such as the higher relative abundance of Helicobacter in the small intestine of M. rufocanus as compared to the lower GIT (Fig. 5, Supplementary Table S29). Similarly, some genera were found to have higher relative abundances throughout the lower GIT as compared to the small intestine in all three host species such as Oscillibacter and Ruminiclostridium (Fig. 5, Supplementary Tables S25, S27, S29). However, most bacterial genera either exhibited higher relative abundance throughout the lower GIT (compared to the small intestine) in a single host species such as Harryflintia in M. rufocanus or no clear trend was found (Fig. 5, Supplementary Tables S25, S27, S29).

Relatively few microbial genera exhibited differential relative abundance when comparing the cecum, colon, and rectum, agreeing with our hypothesis (Supplementary Tables S23, S26, S28, S30). Disagreeing with our predictions, in both A. speciosus and M. rufocanus relative abundances were found to be more similar between the rectum and cecum than the rectum and colon where 7 and 12 (A. speciosus and M. rufocanus respectively) microbial genera were found to have significantly higher relative abundance in the rectum as compared to the colon (Supplementary Tables S23, S26, S30). Notably, there was little consistency in terms of which bacterial genera were found to exhibit higher relative abundance in each of the lower gut regions (Supplementary Tables S26, S28, S30). However, higher relative abundance of Oscillibacter was found in the cecum of both A. speciosus, and M. rufocanus as compared to the colon and rectum (Supplementary Tables S26, S30), but not in A. argenteus (Supplementary Table S28). Furthermore, Ruminococcus 1 and Pygmaiobacter in the rectum of M. rufocanus (Supplementary Table S30), Ruminococcaceae NK4A214 in the rectum of A. speciosus (Fig. 5, Supplementary Table S26), and Rodentibacter in the rectum of A. argenteus (Supplementary Table S28) were found to have higher relative abundance as compared to the colon or cecum.

When exploring the effect of host species on the relative abundance of bacterial genera within each gut region using LEfSe analysis with all three host species included (i.e. non-pairwise), we found a large number of genera exhibiting host species-specific higher relative abundance (Supplementary Tables S31, S33). To determine if the relative abundance of bacterial taxa were more similar between A. speciosus and A. argenteus than between M. rufocanus and either species of Apodemus, a pairwise LEfSe analysis between each host species for each gut region was utilized (Supplementary Tables S32, S34–S37). The largest number of variations in relative abundances were between M. rufocanus and both species of Apodemus as predicted (Supplementary Tables S32, S34–S37). Far fewer genera were found to have higher relative abundance in either A. speciosus or A. argenteus when compared to each other (Supplementary Tables S32, S34–S37). Some bacterial genera showed a similar trend when comparing M. rufocanus to A. speciosus or A. argenteus. For example, there was higher relative abundance of Ruminococcaceae NK4A214 throughout the lower GIT of M. rufocanus (Fig. 5, Supplementary Tables S35–S37) as well as Ureaplasma within the small intestine when compared to either A. speciosus or A. argenteus (Supplementary Table S34). Similarly, Lachnospiraceae UCG_006 was found to have higher relative abundance throughout the lower GIT in both A. speciosus and A. argenteus as compared to M. rufocanus (Fig. 5, Supplementary Tables S35–S37).

Source: Ecology - nature.com