Deconstructing the Bat Skin Microbiome: Influences of the Host and the Environment

Abstract

Bats are geographically widespread and play an important role in many ecosystems, but relatively little is known about the ecology of their associated microbial communities and the role microbial taxa play in bat health, development, and evolution. Moreover, few vertebrate animal skin microbiomes have been comprehensively assessed, and thus characterizing the bat skin microbiome will yield valuable insight into the variability of vertebrate skin microbiomes as a whole. The recent emergence of the skin fungal disease white-nose syndrome highlights the potentially important role bat skin microbial communities could play in bat health. Understanding the determinant of bat skin microbial communities could provide insight into important factors allowing individuals to persist with disease. We collected skin swabs from a total of 11 bat species from the eastern United States (n = 45) and Colorado (n = 119), as well as environmental samples (n = 38) from a subset of sites, and used 16S rRNA marker gene sequencing to observe bacterial communities. In addition, we conducted a literature survey to compare the skin microbiome across vertebrate groups, including the bats presented in this study. Host species, region, and site were all significant predictors of the variability across bat skin bacterial communities. Many bacterial taxa were found both on bats and in the environment. However, some bacterial taxa had consistently greater relative abundances on bat skin relative to their environments. Bats shared many of their abundant taxa with other vertebrates, but also hosted unique bacterial lineages such as the class Thermoleophilia (Actinobacteria). A strong effect of site on the bat skin microbiome indicates that the environment very strongly influences what bacteria are present on bat skin. Bat skin microbiomes are largely composed of site-specific microbiota, but there do appear to be important host-specific taxa. How this translates to differences in host-microbial interactions and bat health remains an important knowledge gap, but this work suggests that habitat variability is very important. We identify some bacterial groups that are more consistent on bats despite site differences, and these may be important ones to study in terms of their function as potential core microbiome members.

Keywords: 16S rRNA; bat ecology; host-associated bacteria; microbial ecology; microbiome; molecular ecology; white-nose syndrome.

Figure 1

Relative abundance of bacterial classes in environment and bat skin samples. The microbial composition by class of bat skin across all species included in the study. Only the top 15 OTUs are represented, which comprise, on average, 89% of all OTUs in the sample.

Figure 2

Differences in community composition of all bat samples by species. An NMDS ordination of a Bray-Curtis dissimilarity matrix of bacterial communities across all bats in the sample set, colored by species (stress = 0.18). A PERMANOVA test showed differences between species were significant (R²: 0.15, p-value < 0.001).

Figure 3

(A,B) Differences in bacterial community composition of all bat samples by sample site. NMDS ordination of all bats from Figure 2 colored by site (stress = 0.18). A PERMANOVA analysis indicates that site is a significant driver of community composition (R²: 0.26, p-value < 0.001). At right, the Bray-Curtis dissimilarity between sites plotted as a dendrogram.

Figure 4

Comparisons of beta diversity of bat skin by region from paired samples. An NMDS ordination of paired samples (soil and bat swabs collected from the same location) sampled for this data set, colored by region (stress = 0.18). A PERMANOVA analysis indicates that region is significant in determining the bacterial community composition (R²: 0.16, p-value < 0.001).

Figure 5

(A,B) Differences by site in Myotis lucifugus. Within a single species sampled across multiple states (M. lucifugus), site is very important in determining the beta-diversity of the bacterial community, as visualized in an NMDS ordination (stress = 0.14). (PERMANOVA R²: 0.33, p-value < 0.001). The clustering of the Virginia samples (upper left, in pink) and the New York samples (center, in blue) also shows the regional signal. At right, a dendrogram of the Bray-Curtis dissimilarity matrix for these sites.

Figure 6

Dissimilarity analysis of bat skin bacteria between and within host species and sites. Boxplots showing bacterial community composition dissimilarities between and within samples from the same host species (Host) or sample site (Site). The differences between samples from the same sites and different sites is greater than that between and within species. Boxes represent first quartile medians and third quartile values, and lines represent minimum and maximum values. Dissimilarities were calculated using a Bray-Curtis dissimilarity matrix from square-root transformed OTU relative abundances.

Figure 7

(A–C) Comparisons of paired environment and bat samples by alpha and beta diversity metrics. (A) Beta diversity of paired environmental samples (blue) and bat skin samples (red) were significantly different (R²: 0.033, p-value < 0.001) as shown in an NMDS ordination (stress = 0.20). (B) Shannon diversity and (C) richness were not significantly different between the environment and bat skin, with many of the dominant taxa shared between groups. (Kruskal-Wallis test of significance p > 0.05).

Figure 8

Heatmap of shared OTUs between bats and environment by class. A heatmap of the relative abundances if the top ten most common bacterial classes found on paired samples of bat skin and their local environment. Blue indicates high abundance taxa, white indicates moderately abundant taxa, and red indicates less abundant taxa. While the most and least abundant taxa are generally shared between the environment and the bat, bacteria in the classes of Thermoleophilia and Bacilli appear in higher abundances on the bat than in the environment.