Cloacal swabs are unreliable sources for estimating lower gastro-intestinal tract microbiota in chicken

Background The gastrointestinal microbiota in chicken ( Gallus gallus domesticus ) has a central role in health and performance. The ceca are a vital site of functional activity, but assessing cecal microbiota in longitudinal studies remains a challenge. The cecal communities are broadly similar to large intestine samples. Cloacal sampling, due to their proximity to the large intestine, is an alternative, non-invasive method used for assaying and monitoring disease-causing agents, and maybe a viable option for longitudinal studies. Results We collected paired cecal content, and cloacal swab samples from twenty randomly selected commercial broilers raised on two dietary treatments. The microbiota of each sample was assessed using 16S rRNA V4 hypervariable region sequencing on an Illumina MiSeq platform and analyzed using the MOTHUR pipeline. Analysis of fourteen paired samples resulted in 1603 OTU’s assigned to 82 Families. Eleven families were shared between the cecal and cloacal samples, with seven and eleven families unique to cecal content and cloacal swabs, respectively. Paired t-test and Wilcoxon Signed-Rank test showed significant differences in the Chao1 index between the cecal content and cloacal swabs (p-value = 0.000845 and p-value = 0.001397, respectively). However, the Inverse Simpson species diversity estimator was not different using the Wilcoxon Signed-Rank test (p-value = 0.3258) and a paired t-test (p-value = 0.3864). β-diversity between the cloacal swabs and cecal microbiota also showed significant differences based on PERMANOVA (p-value = <0.001), HOMOVA (p-value <0.001), and Weighted Unifrac (WSig = <0.001) testing. not approximate either the α or β diversity of cecal samples, based on a

paired sample analysis. The high variability of cloacal microbiota has been reported previously, and this study provides additional evidence of the randomness of cloacal microbiota in contrast to cecal microbiota. Our findings indicate that cloacal samples are not suitable for longitudinal studies of gut microbiota patterns. High inter-individual variation of cloacal swab data warrants further assessment of their reliability as a targeted diagnostic method.
1 Background Chicken (Gallus gallus domesticus) is the source of the most consumed animal protein globally at nearly twice the amount of pork and beef combined [1]. Because of this, there is a great emphasis on improving poultry health and performance [2][3][4][5]. Notably, the role of gut microbiota in improving performance [6][7][8], welfare [9], and health [4,[10][11][12][13][14], is a topic of intense interest. The gut microbiota is studied intensively in broilers; an NCBI PubMed Central search for "Poultry Gut Microbiota" yielded 2586 research articles within the last five years.
The gut microbiota, an ecological community of commensal and non-commensal microorganisms [15,16], is found throughout the entire length of the broiler's gastrointestinal tract (GIT). Although most research concentrates on the organs within the lower sections: the small intestine (duodenum, jejunum, and ileum), large intestine, cecum, and cloaca. The ceca, a pair of blind sacs, are especially important as the site of functional activity relevant to microbial communities and species studied in performance and health [7,17]. The ceca retain nearly 10 11 microbial cells per gram and are an important location for fluid resorption via the translocation of urea from the urodeum and the fermentation of carbohydrates [17][18][19][20][21]. As a consequence, the ceca are the most sampled gut segment in chicken gut microbiota studies [7,22]. A standard experimental method of microbiota analysis in poultry involves the invasive sampling of the ceca, following euthanasia, which prevents longitudinal studies of the same experimental animals.
Cloacal (or vent) swabs are an alternative, non-invasive method used on domestic, migratory, or endangered bird species [23] where invasive sampling may not be permitted. Due to the non-invasive aspect, cloacal swabs are used frequently for assaying and monitoring agents such as Salmonella spp [23], Avian Influenza [24][25][26], Coccidiosis [27], and Campylobacter coli [28]. Importantly, these swabs were analyzed using real-time PCR or microorganism specific plating methods and not for total microbiota analysis.
Therefore, their suitability of cloacal swabs for assessing gut microbiota is not apparent.
Due to the ubiquity of cloacal swabbing, mainly for diagnostics, it is critical to determine if and how representative cloacal microbiota are of cecal microbiota. Cecal microbiota in chicken show broad similarities with lower large-intestinal microbiota [29], and the cloaca abuts the large-intestine. If cloacal microbiotas approximate the cecal microbiota, it would enable non-invasive longitudinal studies. On the other hand, if cloacal microbiota is not a reliable proxy for cecal microbiota occurrence and abundance, then its utility for assessing avian microbiota would be limited. To resolve the reciprocity of cecal and cloacal microbiotas, we used a paired sample approach to compare cecal and cloacal microbiota communities sampled from the same individuals. Based on previously published works about fecal microbiota, we hypothesized that cloacal microbiota is not representative of cecal microbiota from the same individuals. We used 16S rRNA sequence-based analysis of and β diversity of the communities between the two sampling methods. Here we report that cloacal swabs are unreliable representatives of the presence-absence of taxa, as well as and β diversity.

Sample Collection
From each dietary treatment (T1, T2), we randomly sampled ten broilers. Within each diet treatment group, an equal number of individuals were sampled (five). We moved the randomly selected birds to a clean room for cloacal swab collection, euthanasia, and postmortem sample collection from the ceca. For cloacal swabbing, we used a Puritan PurFlock Ultra Sterile mini-tip Flock swab with a sterile container (Puritan, ME, USA) to sample the cloacal microbiota from live birds following a modified protocol originally reported by Vo & Jedlicka [30]. First, the exterior surface of the cloaca was wiped with a cotton ball sprayed with 70% Ethanol. The PurFlock swab was gently inserted approximately 22 mm into the cloaca, a depth just beyond the length of the swab tip. The swab was rotated five times in slow clockwise motion around the cloaca, applying moderate pressure so that the swab-tip maintained contact. Additionally, we rolled the swab-tip so that the entire surface of the swab was coated with cloacal material. Following sample collection, the swab was inserted into the supplied sterile container, immediately placed on ice after collection, and transferred to a -80 o C freezer until further processing.
After completing the cloacal swab sample collection, individual broilers were euthanized by CO 2 exposure, followed by cervical dislocation. The animal use protocol (AUP) and procedures employed were ethically reviewed and approved by the Institutional Animal

Bioinformatic Pipeline for Microbiota Evaluation
Resultant .fastq files from sequencing were processed using MOTHUR software v. 1.39.5 [32]. Briefly, paired-end reads were joined using the make.contigs command. We aligned the sequences to the SILVA database v. 132 [33] and removed chimeric sequences using the UCHIME program v. 4.2.40 [34]. Low abundance operational taxonomic Units (OTU's) and Singletons were removed from analysis with the split.abund command using cutoff = To compare β diversity, we used the Permutational Multivariate Analysis of Variance (PERMANOVA) using the "Adonis" function of the Vegan package with 9999 permutations [46,47]. In addition to PERMANOVA, we compared β diversity in MOTHUR using AMOVA and unifrac.weighted [48]. The weighted unifrac test was applied to investigate the probability that two or more communities have the same structure by chance. These three species-level non-parametric tests were computed using the Yule and Clayton measure of dissimilarity average phylogenetic distances [49]. The statistical significance of comparisons was assessed at α = 0.05.

Sampling Location Yields a Variability in Sequencing Depth
The raw data from sequencing generated a total of 560,935 reads, with an average of 20,033 reads per sample. Total read depth per sample was limited to an arbitrary minimum of 7,435 to ensure adequate read depth in any given sample [50], and thus was the cutoff for inclusion in further analysis. Out of the 40   However, it is noteworthy that Peptostreptococcaceae is only present in four out of fourteen cloacal swab samples. These top three families found in cloacal data were present in all fourteen cecal content samples.
A PCoA (Fig. 3) was calculated to compare community member composition within the cecal content and cloacal swab methods. The cecal content samples cluster tightly together, whereas the cloacal swab samples show high variability while still encompassing the cecal content samples. This high variability is not surprising given the total number of families represented in the cloacal swabs. Overall, the ordination pattern of these paired samples shows broad-ranging differences between the two sampling approaches.
Finally, when comparing the cumulative distribution functions filtered for > 0% of the relative abundances between cecal and cloacal microbiotas, we found major differences in

Richness and diversity differences between cecal and cloacal samples
We compared microbial species richness and diversity of the cecal content and cloacal swabs using the Chao1 and Inverse Simpson estimators (Fig. 4). Both the Wilcoxon Signed-Rank test (W = 4, p-value = 0.000845) and paired t-test (t = 4.042, p-value = 0.001397) showed highly significant differences in the Chao1 index between the cecal content and cloacal swabs, with the highest richness observed in the cecal samples. A higher Chao1 value indicates a higher number of low abundance taxa, e.g., singletons [51,52]. The higher value in cecal samples, suggests that more rare taxa were captured in cecal samples. However, the Inverse Simpson species diversity estimator was not different between cloacal and cecal samples based on a Wilcoxon Signed-Rank test (W = 36, pvalue = 0.3258) and a paired t-test (t = 0.89623, p-value = 0.3864). Similar to the Chao1 findings, the cecal content had higher microbial diversity, compared to the cloacal swabs.
As the Inverse Simpson index estimates the richness weighted by the proportional abundance of taxa present within samples, the non-significance suggests that the two types did not differ in their internal weighted abundances.
To assess whether cecal or cloacal swabs captured differences between dietary treatments, we performed richness and diversity analyses, comparing the two diets. Next, we investigated whether differences in the diet treatments (T1 and T2) elicit differences in communities (β diversity) inferred using cecal versus cloacal samples. We found that neither sampling method detected differences in β diversity between the diets.
The results from PERMANOVA showed that the cecal content was not different between Further comparison of the dietary treatments within each method and between methods using the KS, Lepage, and Cucconi tests yielded no significant differences (Table 2). We found non-significant results using the same three tests for the comparison of dietary treatments within each sampling method. These results are expected as we found no major difference in the structure and abundances between the treatments using either cecal content and cloacal swabs.

Discussion
In this study, we showed that the microbiota identified from cloacal swabs are not representative of the cecal microbiota, and therefore not a suitable approach to sampling the microbial communities of the lower gastrointestinal tract. This result was highly surprising, given that the cecal and large intestine microbiotas are alike by week five in chicken [29]. Not only were the cloacal communities limited in their resemblance to cecal communities, the patterns of presence-absence as inferred by richness estimates were also significantly different. These findings suggest that there is a high degree of stochasticity to taxa sampled from the cloaca. Our results show similarities to the findings of Videvall et al. [53], who compared cloacal swabs and fecal samples in the ostrich (Struthio camelus) to analyze the lower GIT microbiota community and demonstrated the inaccuracy of fecal and cloacal swabs to portray the microbiota communities of the lower GIT organs. The broad-ranging differences between cloacal and cecal microbiota mirror the patterns seen with fecal microbiota in chicken. Fecal samples show qualitative similarities with quantitative differences compared to GIT [27,54]. Hieke et al. [14] also showed fecal samples are not representative of cecal communities in young layer-type chicken.

High variability of cloacal microbiota
While the factors influencing fecal microbiota differences from cecal communities (external conditions, environmental microbiota) are expected, the cloacal swab dissimilarities and variability are more surprising. It is not clear if the cloaca of chicken is colonized, unlike other parts of the GIT. While numerous surveys of cloacal microbiota exist in the literature, in wild birds, the cloacal microbiota is often the only locus for characterizing gut microbiota as euthanasia may not be an option. However, our results show that the taxonomic composition and community profiles obtained from cloacal swabs can be highly random, with little consensus even when collected under controlled conditions. The high-interindividual differences in cloacal microbiomes were also reported for barn swallows [55]. Barn swallows have different social structures and sex-based behavioral differences and make direct comparisons with chicken difficult, but the poor reproducibility of cloacal microbiota is, nonetheless, a notable similarity. We found lower richness and diversity of microbial taxa in the cloaca, compared to the cecal microbiota.
Van Veelen et al. [56] showed lower richness and diversity of cloacal microbiota but surmised that top-down regulation by the host's genetics drives this pattern. However, our data showing significantly higher richness in the ceca, suggests that host genetics are not driving lower richness or diversity in the cloaca. The variability of cloacal swab data was revealed only in contrast with the paired cecal datasets.
On the other hand, Hird et al. [57] found that cloacal microbiomes differed among species of ducks, and by Influenza infection status. In this case, the interspecies differences may be driving the resolution of differences among species. Furthermore, as they did not characterize cecal microbiota, it is not possible to determine how the cloacal data compared to cecal data. Our analysis leads us to advocate extreme caution when inferring lower GIT microbiota patterns from cloacal swabs of birds.
Cloacal swabs are used routinely to assess infection status in domesticated, pet, and wild bird species [58][59][60]. In the majority of these cases, targeted assays (RT-PCR) used swab samples for the detection of pathogenic species. In these cases, the sensitivity of the assays provides valuable information for treatment or containment of pathogens, especially in poultry operations. While our data show high variability in the representation of taxa in cloacal samples, the sensitivity of RT-PCR approaches may allow lower detection thresholds. However, the reciprocity of taxon representation with cloacal 16S rRNA sequencing and targeted PCR methods needs to be established experimentally.

Resolution of microbial community differences between diets
In our analysis of microbiota between the two dietary treatments, we found that neither cecal nor cloacal samples were able to differentiate between diets. Both these sample types appear to be equivalent in their inability to differentiate between diets. However, we emphasize that this equivalency exists aside from the fact that the cecal and cloacal communities were highly dissimilar. Also, while the cecal samples were similar due to the overlapping distributions between diets, the similarity of cloacal swabs is driven by the high variability across all cloacal samples ( Figure 5B). Additionally, the housing environment, rather than dietary protein source, is known to be a more significant factor driving cecal microbiota differences. Hubert et al. [61] reported that birds raised in the same housing environment, regardless of dietary protein source, had similar cecal microbiota. In this present study, all the chickens were raised in the same barn (across replicate pens), where they were provided with the same bedding material and water source. Therefore, the high variability among cloacal samples, all collected in a controlled environment, represents, in our opinion, the high variability inherent to cloacal samples.

Conclusions
In this study, we showed that cloacal swabs do not faithfully approximate either the α and β diversity of cecal samples, based on paired samples. Therefore, cloacal swabs are unsuitable for assessing lower GIT microbiota in birds. While the high variability of cloacal microbiota has been reported previously, our study provides experimental evidence to capture the randomness of cloacal microbiota, concerning the consistency of cecal samples

Competing Interests
The authors declare that they have no competing interests.

Funding
Not applicable.

Authors Contributions
The experiment was conceived and designed by TW and GA. Samples were collected by TW. Wet lab work was performed by TW. TW and GA performed the data analysis. TW and GA wrote the manuscript. All authors read and approved the final manuscript.

Acknowledgments
The authors would like to acknowledge Mohamed Ibrahim Magdy, James Alfieri, Dr. Shawna Marie Hubert, and Dr. Anne-Sophie Charlotte Hieke for assisting with collecting tissue samples. The authors would also like to acknowledge James Alfieri for troubleshooting portions of the R code used in this data analysis.  The relative abundance of families observed for each cloacal swab samples that have>= 7435 reads; n = 14. The vertical black line is a visual separator so that treatment 1 samples are on the left, and treatment 2 is on the right.

Authors' Information
29 Figure 1 The relative abundance of families observed for each cloacal swab samples that have>= 7435 reads; n = 14. The vertical black line is a visual separator so that treatment 1 samples are on the left, and treatment 2 is on the right.      measurements for alpha diversity of cecal content samples by Treatment. Figure   5B. Boxplot of the Observed and Inverse Simpson (InvSimpson) measurements for alpha diversity of cloacal swab samples by Treatment.  measurements for alpha diversity of cecal content samples by Treatment. Figure   5B. Boxplot of the Observed and Inverse Simpson (InvSimpson) measurements for alpha diversity of cloacal swab samples by Treatment.

Supplementary Files
This is a list of supplementary files associated with the primary manuscript. Click to download.
SI.zip SI.zip