Phylosymbiosis shapes skin bacterial communities and pathogen-protective function in Appalachian salamanders

Abstract

Phylosymbiosis is an association between host-associated microbiome composition and host phylogeny. This pattern can arise via the evolution of host traits, habitat preferences, diets, and the co-diversification of hosts and microbes. Understanding the drivers of phylosymbiosis is vital for modelling disease-microbiome interactions and manipulating microbiomes in multi-host systems. This study quantifies phylosymbiosis in Appalachian salamander skin in the context of infection by the fungal pathogen Batrachochytrium dendrobatidis (Bd), while accounting for environmental microbiome exposure. We sampled ten salamander species representing >150M years of divergence, assessed their Bd infection status, and analysed their skin and environmental microbiomes. Our results reveal a significant signal of phylosymbiosis, whereas the local environmental pool of microbes, climate, geography, and Bd infection load had a smaller impact. Host-microbe co-speciation was not evident, indicating that the effect stems from the evolution of host traits influencing microbiome assembly. Bd infection is correlated with host phylogeny and the abundance of Bd-inhibitory bacterial strains, suggesting that the long-term evolutionary dynamics between salamander hosts and their skin microbiomes affect the present-day distribution of the pathogen, along with habitat-linked exposure risk. Five Bd-inhibitory bacterial strains showed unusual generalism: occurring in most host species and habitats. These generalist strains may enhance the likelihood of probiotic manipulations colonising and persisting on hosts. Our results underscore the substantial influence of host-microbiome eco-evolutionary dynamics on environmental health and disease outcomes.

Keywords: Batrachochytrium dendrobatidis; Batrachochytrium salamandrivorans; community assembly; host-microbiome interactions; phylosymbiosis; probiotics.

Figure 1

Conceptual diagram illustrating how host environment impacts the regional pool of microbes, whereas species-specific host traits filter this pool of microbes.

Figure 2

Sampling scheme. Sampling sites within each of the three localities: Mountain Maryland, George Washington and Jefferson National Forests and Front Royal Conservation Biology Institute, are shown on a topological map (A). Sites are distinguished by point colour within each locality, and habitat types are shown as icons (as in panel B) to the right of each site name. Waffle plots (B) show the number of individuals of each species in each habitat type. Each square represents one individual, coloured by species as in panel E. Batrachochytrium dendrobatidis (Bd) infected individuals are indicated by a cross. Bd prevalence is shown as a bar plot (panel C; proportion of infected individuals) and Bd load is shown as a violin plot (panel D; units of 1000 copies; jittered points show actual values for all infected individuals). Bars and violins are coloured by species as in E. Photographs in E are all the authors’ work except for P. hoffmani, which is available on a creative commons licence (CC-BY-NC, © Josh Emms 2018).

Figure 3

Alpha diversity for all samples. Beeswarm plots show ASV richness for all environmental (left) and salamander skin samples (right), grouped by habitat on the x-axis. Each point represents a single sample and points with similar ASV richness values are separated on the X-axis to minimise overlap. Horizontal bars show the mean for each habitat. Point colour indicates host species and shape indicates locality.

Figure 4

There is a strong signal of phylosymbiosis in Appalachian salamander skin microbiomes. NMDS plots based on Bray-Curtis distance (A–C) for all salamander skin and environmental samples. Each point represents a single sample and points are coloured by host species (A), habitat (B), or locality (C). Point shapes show sample type (i.e. salamander skin or environmental samples). A dated host phylogeny (D) is shown beside neighbour-joining based hierarchical clustering of mean pairwise Bray-Curtis microbiome distance between each salamander species pair (E). Coloured lines link the same species between the two dendrograms. Coloured tip points indicate species, and outlines around salamander images indicate primary habitat. Results of multiple regression on distance matrices (MRM) analysis (F) show the effect of multiple explanatory variables of skin microbiome distance. Bar plots show standardised regression coefficients for host phylogeny, geographic distance, climate distance, environmental microbiome distance, and Bd load using four different skin-microbiome beta-diversity statistics. Stars above each bars indicate significance (P < .001: ^***; P < .01: ^**; P < .05: ^*). In our specificity analysis (G), there were most significantly specific ASVs for host phylogeny, followed by environmental microbiome distance.

Figure 5

Host and environmental specificity across the salamander skin microbiome and in known anti-Bd taxa. The bacterial phylogeny of the skin microbiome is shown to order level, with colour indicating mean specificity index to host phylogeny (A) and environmental microbiome distance (B). A specificity index <0 indicates higher specificity and specificity index >0 indicates higher generalism. The dot plot (C) shows the relationship between these metrics with each point representing a single ASV, coloured by phylum. Known anti-Bd ASVs are shown as diamonds and all other ASVs are shown as circles. Known anti-Bd taxa with the lowest specificity (i.e. most positive specificity index) are present in a wide range of host species and habitats (D). Coloured squares indicate presence in each host species and habitat for the most generalist anti-Bd ASVs.