DISCUSSION
Biological understanding of vertebrates is incomplete without a thorough knowledge of the microbiome. One of the specific goals of evolutionary biology is to discover and describe biodiversity (see (Hird, 2017)); herein, 1740 samples from 779 wild birds expand what we know about both birds and their associated microbes. These samples, from a total of seven body sites, provide a fundamental description of the microbiomes of diverse and previously undescribed body sites in over 200 species. The cloacal and intestinal microbiome results support many previous studies that show a dominance of Proteobacteria and Firmicutes (Figure 2, Supplemental Table S2, reviewed in (Grond, Sandercock, Jumpponen, & Zeglin, 2018)).
The avian blood microbiome has only been characterized in chickens, where it was shown to be comprised of 60.58% Proteobacteria, 13.99% Bacteriodetes, 11.45% Firmicutes, 10.21% Actinobacteria, and 1.96% Cyanobacteria (Mandal et al., 2016). Our blood samples exhibited over 20% more Proteobacteria, with correspondingly smaller percentages of the remaining phyla (Fig. 2, Supplemental Table S2). The differences between the two findings may be unsurprising given that wild birds tend to have higher percentages of Proteobacteria (Grond et al., 2018). Clostridium was more abundant in the blood than in the cloaca, intestines, and buccal cavity (Fig. 6a, b, c) and is known to cause infections in birds (Crespo, Fisher, Shivaprasad, Fernández-Miyakawa, & Uzal, 2007). Janthinobacterium was more abundant in the blood than in the cloaca and buccal cavities (Fig. 6a, b). This genus has previously been identified in the cloaca of shorebirds (Santos et al., 2012). Known to contain pathogenic species, Mycobacterium was more abundant in the blood than in the cloaca, intestines, gizzard, and buccal cavity (Fig. 6a, b, c, d) , potentially showing that some birds were infected in this study or that non-pathogenic species are part of the healthy microbiota (Dhama et al., 2011). Phenylobacterium was also more common in the blood than in the cloaca, intestines, gizzard, and buccal cavity and has been previously shown to be in the uropygial glands of house sparrows with malaria (Fig. 6a, b, c, d) (Videvall et al., 2021).Legionella was more abundant in the blood than in the cloaca and the buccal cavity (Fig. 6a, b). This genus has been found in the cloaca of birds and is known to cause opportunistic infections in humans (Santos et al., 2012; Fields, Benson, & Besser, 2002). Parvibaculum (more abundant in the blood than in cloaca, intestines, gizzard, and buccal cavity and more abundant in the buccal cavity than in the cloaca (Fig. 6a, b, c, d, e)), Planctomyces(more abundant in the blood than in the cloaca, intestines, and buccal cavity (Fig. 6a, b, c)) and Sediminibacterium (more abundant in the blood than in the cloaca, intestines, gizzard, and buccal cavity (Fig. 6a, b, c, d)) were abundant in this study but are not commonly described members of the avian microbiota.
The buccal microbiome contained over 50% Proteobacteria and smaller percentages of Firmicutes, Actinobacteria, Tenericutes, and Bacteriodetes (Fig. 2, Supplemental Table S2). These results are similar to what was found in the oral microbiome of the Great Tit (Kropáčková et al., 2017) but are quite different from those found in the Cooper’s hawk, which contains higher relative abundance of Firmicutes than the quantities of Actinobacteria, Bacteriodetes, and Tenericutes (Taylor et al., 2019). Further sampling is required to determine how uniform the oral microbiome is in wild birds. Hylemonella was more abundant in the buccal cavity than in the blood, gizzard, intestines, and cloaca (Fig. 6a, e, f, g). This genus has been identified in the skin microbiome of vultures, raising the possibility of preening transferring skin and buccal microbes (Zepeda Mendoza et al., 2018). Gallibacterium was more abundant in the buccal cavity than in the blood, gizzard, intestines, and cloaca (Fig. 6a, e, f, g) and has been identified in the respiratory tract of healthy and unhealthy birds (Bisgaard, 1977), (Mushin, Weisman, & Singer, 1980). Leucobacter was more abundant in the buccal cavity than in the blood or cloaca in this dataset (Figure 6a, e) and has been identified in the feces of swiftlets (Sien, Lihan, Yee, Chuan, & Koon, 2013). Rothia was more abundant in the buccal cavity than in the gizzard or cloaca (Figure 6e, g) and is also a common member of the human saliva microbiome (Tsuzukibashi et al., 2017).
The gizzard samples were composed mostly of Firmicutes, Proteobacteria, and Cyanobacteria (Fig. 2, Supplemental Table S2). This is not consistent with the only published study on gizzard microbiomes that were mainly composed of Bacteroidetes, Cyanobacteria, Planctomycetes, Verrucomicrobia, and Alpha and Gammaproteobacteria (García-Amado et al., 2018). The liver and spleen microbiomes were both dominated by Firmicutes and Proteobacteria (Fig. 2, Supplemental Table S2); this is similar to the microbiome of wild mouse spleens which are also composed primarily of Firmicutes and Proteobacteria (Ge, Guo, Ge, Yin, & Yin, 2018). The relative abundances of these two phyla were very different in the liver and spleen, as compared to the blood. Spleens filter blood, so similarities between these two sample types may be expected. However, the vast differences between them show that these environments may be hospitable to different communities of bacteria, although this requires further confirmation, as our spleen and liver sample sizes were quite low.
The majority of our samples were of the blood, buccal, gizzard, intestines, and cloaca. These sample types were significantly different from each other in all three beta diversity metrics measured (p <0.001) and the variation explained by each of these metrics was relatively high (6.7%- 17%). This shows that the types of taxa in each sample type are different from each other and that taxa those that are more phylogenetically distinct are more abundant. These measurements are lower than some other bird body site studies (Grond, Guilani, & Hird, 2020), but perhaps to be expected as this study incorporates dozens to hundreds of bird species that may have distinct microbiomes.
Many ornithologists are interested in studying the microbiome without harming the bird and therefore many have asked whether non-destructive sampling (e.g., oral swabs, cloacal swabs, feces) is adequate to describe the gut microbiome (Videvall, Strandh, Engelbrecht, Cloete, & Cornwallis, 2018). At the ASV level, our cloacal and intestine samples had substantial, and roughly equivalent, unique components; however, at the level of the sequencing reads, almost all the diversity was shared (Fig. 5b). Therefore, the unique ASVs contribute far less to the total microbiota than the shared ASVs. Compositionally, ten genera were more abundant in the cloaca than in the intestines: Acinetobacter , Aerococcus ,Cloacibacterium , Cupriavidus , Limnohabitans ,Micrococcus , Propionibacterium , Rheinheimera ,Staphylococcus , and Stenotrophomas (Figure 5a). Notably,Cloacibacterium has been found in avian blood (Mandal et al., 2016),Micrococcus in healthy conjunctiva and nasal passages (Silvanose, Bailey, Naldo, & Howlett, 2001), and Staphylococcus can cause infections in birds (Hermans, Devriese, De Herdt, Godard, & Haesebrouck, 2000). Six genera were more abundant in the intestines than in the cloaca: Balneimonas , Enterococcus ,Lactobacillus , Lactococcus , Psychrobacter , andRickettsiella (Figure 5a). Enterococcus has been found in the cloaca of birds (Jørgensen et al., 2017) andPsychrobacter has been found in the throats and guts of birds (Kämpfer et al., 2015), (Kämpfer et al., 2020).Lactobacillus and Lactococcus have been previously identified in the cloaca of birds (Allegretti et al., 2014; Gunasekaran, Trabelcy, Izhaki, & Halpern, 2021), but we found them at higher abundances in the intestines, which is similar to previous comparative studies (Hird et al., 2015; Capunitan et al., 2020). Enterococcus was more abundant in the intestines than the cloaca, gizzard, blood, and buccal cavity and more common in the blood than the gizzard (Figs. 5a, 6c, d, f, i). This genus has been found in the cloaca of chickens (Jørgensen et al., 2017).Lactobacillus similarly was more common in the intestines than the cloacal, gizzard, blood, and buccal cavity (Figs. 5a, 6c, f, i). This matched a finding in parrots that found Lactobacillus in their cloaca (Allegretti et al., 2014). In a comparison of body site microbiota in ostriches, several families that include our differentially abundant intestinal microbes were also significantly higher at internal gastrointestinal sites, as compared to the cloaca (Lactobacillaceae ,Streptococcaceae , Enterococcaceae ; Videvall et al., 2018).
The comparison of body sites also identified many additional taxa common to bird microbiomes and/or which are known pathogens:Campylobacter(Kapperud & Rosef, 1983; Hird et al., 2018),Cloacibacterium(Mandal et al., 2016).Comamonas (Kropáčková et al., 2017), Enhydrobacter(Kreisinger, Čížková, Kropáčková, & Albrecht, 2015), Methylotenera(Boukerb et al., 2021),Pseudomonas (Oprea, Crivineanu, Tudor, łOGOE, & Popa, 2010), Psychrobacter(Kämpfer et al., 2015), (Kämpfer et al., 2020).Sphingobacterium(Gunasekaran et al., 2021),Streptococcus(Devriese et al., 1994),Curtobacterium(Giorgio, De Bonis, Balestrieri, Rossi, & Guida, 2018), Kocuria(Braun, Wang, Zimmermann, Boutin, & Wink, 2018), Brevundimonas(Giorgio et al., 2018), Kingella (Foster et al., 2005), Micrococcus(Silvanose et al., 2001),Staphylococcus(Hermans et al., 2000),Lactococcus(Gunasekaran et al., 2021).Methylobacterium was found to be more abundant in the blood and buccal cavity more than the gizzard and more abundant in the buccal cavity than in the intestines and cloaca (Figure 6d, e, f, g). This is noteworthy as methylobacterium is known to be a contaminant in kits (Salter et al., 2014).
Although alpha diversity did not vary significantly across anatomical sites, we did identify specific microbial taxa that were differentially abundant between sites. (Fig. 3). Together with the beta diversity results, this shows that while the sample types contain similar levels of diversity, the composition of those communities is different across the body sites.
The microbiome is a trait of the host that may not be independent of the underlying phylogeny. Strictly speaking, phylogeny describes the evolutionary history of organisms. For a variety of reasons, phylogeny captures more than just evolutionary history and more closely related organisms frequently have more similar traits (“phylogenetic signal”). Therefore, comparisons of microbial “traits” across species need to control for phylogeny of the hosts.
Bird weight and one third of the alpha diversity measurements contained significant phylogenetic signal, as assessed by Pagel’s Lambda (Table 1: two blood, one buccal, two intestine). When subsetting to include only Passeriformes, only three of the 15 tests contained significant phylogenetic signal: two buccal and one cloacal (although the three gizzard tests were likely affected by extremely low sample size). This shows an inconsistent or low level of association between phylogeny and microbiota richness and diversity, meaning that factors beyond phylogeny impact the microbial communities. This is similar to previous comparative work that found a White Noise model (of no phylogenetic signal) may fit the avian microbiome better than a neutral model or a model that includes selection (Capunitan et al., 2020).
How does size of a host influence the richness and diversity of the microbiota? Birds can be conceptualized as “islands” containing communities of microorganisms and their composition could potentially be driven by the Theory of Island Biogeography. We found eight significant correlations between bird size and microbiota, when using phylogenetic comparative methods (Table 1). In the full dataset, the blood samples show a significant negative correlation between their microbial diversity and the average host weight for all three of the diversity metrics (p <0.001), indicating that larger birds actually have significantly lower diversity in their blood than smaller birds.
Because bird orders diversified quickly, there can be different associations of host traits within and across orders (Harmon et al., 2010). Therefore, we restricted analyses to a single order (Passeriformes) and reran the PGLS analyses. In the subset data, all three of the microbiome metrics for the cloacal samples exhibited significant negative correlations (p <0.05) (Table 1, Supplemental Fig. S3). The Theory of Island Biogeography predicts that larger islands will house more diversity; our results appear to show the opposite: either a significant negative correlation or no correlation at all.
In broader terms, our results show that as host “islands” increase in size, the number of microbial taxa immigrating decreases and/or the number of extinction events grows larger, indicating that regulatory mechanisms associated with host body size may be influencing the Species-Area Relationship. For example, avian microbiomes may be more prone to colonization by bacteria that inhibit conspecific proliferation, the immune systems of the host may be more sensitive to new “immigrant” members of the microbiome, different anatomical sites may be limited by nutrients, or other possibilities may be working to prevent a significant and large difference in alpha diversity due to host average weight. However, and importantly, even when thep -values showed there was a negative or positive significant correlation between average host weight and alpha diversity, the value was small in every case. This suggests that there are only minor significant changes in the alpha diversity corresponding to average host weight. Taken together, our findings do not support an unexamined application of traditional Species-Area Relationships to the avian microbiome.