Gut microbial diversity in two insectivorous bats: Insights into the effect of different sampling sources

Abstract The gut microbiota is now known as a key factor in mammalian physiology and health. Our understanding of the gut microbial communities and their effects on ecology and evolution of their hosts is extremely limited in bats which represent the second largest mammalian order. In the current study, gut microbiota of three sampling sources (small intestine, large intestine, and feces) were characterized in two sympatric and insectivorous bats (Rhinolophus sinicus and Myotis altarium) by high‐throughput sequencing of the V3‐V4 region of the 16S rRNA gene. Combining with published studies, this work reveals that Gammaproteobacteria may be a dominant class in the whole Chiroptera and Fusobacteria is less observed in bats although it has been proven to be dominant in other mammals. Our results reveal that the sampling source influences alpha diversity of the microbial community in both studied species although no significant variations of beta diversity were observed, which support that fecal samples cannot be used as a proxy of the microbiota in other gut regions in wild animals.

A majority of previous studies of gut microbiota have focused on fecal samples because they are easily accessible. compositions varied between different gut regions and feces of mice, (Gu et al., 2013;Looft et al., 2014;Mu et al., 2017) and pikas (Li, Chen, et al., 2017) possibly because microbiotas play different functional roles in different intestinal compartments and niches. To date, very few such comparisons have been conducted in wild animals (but see Kohl et al., 2017; to test whether fecal samples can be used as a proxy of the microbiota in other gut regions. In the current study, we aim at comparing the composition of the bacterial microbiota of bats from different gut sections. For this aim, we sampled two bat species, Rhinolophus sinicus (Rhinolophidae, Yinpterochiroptera) and Myotis altarium (Vespertilionidae, Yangochiroptera). Both species live in the same cave. Both of them are insectivorous and feed on insects such as Coleoptera (Hu, Yang, Tan, & Zhang, 2012;Zhang, 1985). Here by focusing on these two bat species, we aim to (a) characterize their core microbiota and compare it to that of other bat species and mammals; (b) compare the microbiota of paired small intestine, large intestine and feces to test whether fecal samples can be used as a proxy of the microbiota in other gut regions.

| Species identification
To confirm the species identification of bats based on the morphology in the field, we amplified and sequenced the cytochrome b (cytb) gene for all individuals. Genomic DNA was extracted from the muscle tissue using DNeasy kits (Qiagen). Details of primers, PCR reaction and the thermal profile for cytb have been provided in (Mao, He, Zhang, Rossiter, & Zhang, 2013
Filtered reads were merged using FLASH software (Reyon et al., 2012) with 10 bp of minimal overlapping, 200 bp of maximum overlapping and 20% of maximum mismatch rate. Using QIIME software (version 1.8.0) (Caporaso et al., 2010), sequences with ambiguous bases or homopolymers were discarded and only sequences with the length of over 200 bp and 75% of bases over Q20 were retained. Then, sequences with chimera were detected and removed using USEARCH (Edgar, Haas, Clemente, Quince, & Knight, 2011).
The final valid sequences were used for the downstream analysis in QIIME.
Similar numbers of sequences were generated per sample ranging from 32,161 to 47,840 (Table 1) Figure S1).

| Effects of sampling sources on the microbial diversity
To determine the effect of sampling sources (the small intestine, large intestine, and feces) on microbial diversity, microbiota composition and abundance were compared among different sampling sources in each of two bat species (R. sinicus and M. altarium) using both alpha and beta diversity analyses.
The Observed Species counts were used as the indicator of alpha diversity in the samples, but Chao1 and Shannon indices were also generated for comparisons with the published literatures. To compare microbiota abundance difference, genera with the top 15 abundance were obtained and visualized using barplots. The membership and structure of samples at the top 15 abundance genera were revealed by heatmap plots. For comparisons between three sampling sources, one-way ANOVA was used. For comparisons between two sampling sources (pairwise comparisons), Welch's t test was used.
Significant difference was considered at p < 0.05.
Beta diversity (between sample diversity) was estimated by calculating unweighted and weighted UniFrac distance matrices To test the validity of using fecal samples as a proxy of microbiota in other gut regions, we generate UPGMA (Unweighted Pair Group Method with Arithmetic mean) trees based on unweighted and weighted UniFrac distance matrices across all 18 samples.

| Core microbiota in two insectivorous bats
In this study, we amplified and sequenced the cytochrome b (cytb) gene for all individuals (GenBank accessions:MH325072-MH325077). By performing BLAST searches at NCBI database, we confirmed their taxonomic assignments made based on the morphology in the field. In addition, Maximum Likelihood (ML) tree reconstructed based on cytb sequences revealed that all three R.
sinicus samples used in this study were classified with individuals from East R. s. sinicus (Supporting Information Figure S2).   consistently occurred in all three sampling sources of each species (Figure 2a,b). In the small intestine, the distribution of the three dominant phyla was relatively average in both bat species, whereas in the large intestine of M. altarium and feces of R. sinicus, Proteobacteria occupied over a half (Figure 2a,b). In R. sinicus, other   phyla with more than 1% abundance were Acidobacteria (1.0%) in the small intestine and Actinobacteria (2.0%) in the large intestine.

| Effects of sampling sources on the microbial diversity
In R. sinicus, the large intestine contained the largest number of observed taxa followed by feces (Table 1). Similar pattern was observed in the total and unique number of genera. Specifically, among 108 genera found in all R. sinicus samples after removing four uncultured genera, 88 (21 of them are unique) were in large intestine, 68 (13 are unique) in feces, and 62 (four are unique) in small intestine ( Figure 3a). The small intestine and feces shared the least number of genera and similar number of genera was shared by large intestine and small intestine and by large intestine and feces (Figure 3a). A total of 40 genera (37.0%) were shared among the three sampling sources (Figure 3a).
At 32,010 sequences depth, a significant difference of alpha diversity (observed species) was detected among the three sampling sources (one-way ANOVA test, p = 0.0038) (Figure 3b). In addition, pairwise comparisons of the three sampling sources were also significant (p < 0.05 in all Welch's t test) (Figure 3b). However, at 7,010 sequences depth the difference of alpha diversity among the three sampling sources was not significant (p = 0.411) (Supporting Information Figure S3) and also for the three pairwise comparisons (p > 0.05 in all Welch's t test). These results indicated that sequences depth generated for each sample might affect alpha diversity measurements. Although we did not recover any OTUs whose abundance differed significantly among the three sampling sources (one-way ANOVA test, p > 0.05), we did find some taxa with a relative high abundance in specific sampling sources. Specifically, among genera with the top 15 abundance small intestine has more Vibrio; large intestine has more Serratia, Prevotella 9, and Bacteroides; the feces have more Hafnia and Clostridium (Figure 3c). This abundance difference among different sampling sources was largely due to a specific sample and not consistent across all three samples, as revealed by the heatmap (Figure 3d). For example, more Serratia, Prevotella 9, and In M. altarium, feces have the largest number of observed taxa followed by the small intestine (Table 1)   Interestingly, Proteobacteria is rarely found (<10%) in carnivore species except for domestic dogs (Handl, Dowd, Garcia-Mazcorro, Steiner, & Suchodolski, 2011). Unlike the case of Proteobacteria, the relative abundance of Bacteroidetes (>26%) and Firmicutes (>22%) in R. sinicus and M. altarium is much higher than other insectivorous bats of Phyllostomid (10% Bacteroidetes and <5% Firmicutes) (Carrillo-Araujo et al., 2015). Firmicutes is the only phylum universally shared in mammals (Ley et al. 2008) and in some carnivore species and human Firmicutes has a relative abundance of >60% (Menke et al., 2014(Menke et al., , 2017.

| Effects of sampling sources on microbial diversity
The current study revealed significant differences of the microbial community composition (the alpha diversity) in different sampling sources (the large intestine, small intestine, and feces) in two bat species (R. sinicus and M. altarium) although no significant variations of beta diversity were observed. These results were in line with the previous studies in mice (Gu et al., 2013;Pang, Vogensen, Nielsen, & Hansen, 2012;Weldon et al., 2015), pigs (Looft et al., 2014;Mu et al., 2017), and sheep (Zeng et al., 2017). Observed differences across the sampling sources may be caused by environmental heterogeneity in different intestinal compartments and niches, such as different oxygen exposure, pH, and substrate availability (Hao & Lee, 2004). In addition, functional changes have also been reported between cecal and fecal microbiota in mouse (Tanca et al., 2017). Our study further confirms that fecal samples, although easily accessible, cannot be used as a proxy of the microbiota in other gut regions (Li, Li, et al., 2017b).  Phyllostomid bats that did not reveal significant differences in microbiota composition between three different intestine regions (Carrillo-Araujo et al., 2015). This contrast may be caused by different sampling sources used in comparisons. It is now known that microbiota differ across different functional gut regions (Haange et al., 2012). In Carrillo-Araujo et al. (2015), the intestine was divided into three fractions of similar size (anterior, medium, and posterior). In addition, fecal samples were not included in Carrillo-Araujo et al.
(2015). Significant differences in microbiota composition between three different sampling sources observed in the current study may result from the inclusion of fecal samples. Indeed, as for the microbiota difference between large and small intestine, our results in M. altarium were consistent with Carrillo-Araujo et al. (2015). In addition, sequences depth per sample may also contribute to the contrast between our study and Carrillo-Araujo et al. (2015). The current study analysis based on a low sequences depth (7,010 sequences) in R. sinicus did not reveal a significant difference of microbial community compositions across the sampling sources.
Although two genera were identified to exhibit significant abundance differences between different sampling sources, they were not consistent in two bat species. For example, Vibrio showed a significant abundance difference between small intestine and large intestine in R. sinicus but between small intestine and feces in M.
altarium and a significant abundance difference of Prevotella 9 was only observed between small intestine and large intestine in M. altarium. Thus, our current results cannot draw any conclusions about the effect of sampling source on microbial community abundance.

| CON CLUS ION
This study has characterized the microbiota of three sampling sources (the small intestine, large intestine, and feces) in two insectivorous bat species. Our study adds to the list of a growing number of studies on the gut microbiota in bats. Our results revealed that the sampling source influences the alpha diversity of microbial community and suggest that fecal samples cannot be used as microbial inventories in other gut regions. In the future, more number of individuals will be needed to test this suggestion. Recent studies have shown that the sex of hosts may affect the gut microbial diversity (e.g. Fierer, Hamady, Lauber, & Knight, 2008;Markle et al., 2013). In this study, all R. sinicus individuals are male and all M. altarium are females. Thus, we did not compare the difference of microbial diversity between these two bat species. Future investigations will continue to assess the relative effects of genetic divergence of hosts, sex, the gut region, and diet on gut microbial communities.

ACK N OWLED G M ENTS
This work was supported by the National Natural Science Foundation of China (No. 31570378) to X. Mao.

CO N FLI C T O F I NTE R E S T
None declared.

DATA ACCE SS I B I LIT Y
All the raw Illumina sequencing reads have been deposited to the NCBI Sequencing Read Archive (SRA) database under the accession ID SRP144939.