Shades of grey: host phenotype dependent effect of urbanization on the bacterial microbiome of a wild mammal

Abstract

Background: Host-associated microbiota are integral to the ecology of their host and may help wildlife species cope with rapid environmental change. Urbanization is a globally replicated form of severe environmental change which we can leverage to better understand wildlife microbiomes. Does the colonization of separate cities result in parallel changes in the intestinal microbiome of wildlife, and if so, does within-city habitat heterogeneity matter? Using 16S rRNA gene amplicon sequencing, we quantified the effect of urbanization (across three cities) on the microbiome of eastern grey squirrels (Sciurus carolinensis). Grey squirrels are ubiquitous in rural and urban environments throughout their native range, across which they display an apparent coat colour polymorphism (agouti, black, intermediate).

Results: Grey squirrel microbiomes differed between rural and city environments; however, comparable variation was explained by habitat heterogeneity within cities. Our analyses suggest that operational taxonomic unit (OTU) community structure was more strongly influenced by local environmental conditions (rural and city forests versus human built habitats) than urbanization of the broader landscape (city versus rural). The bacterial genera characterizing the microbiomes of built-environment squirrels are thought to specialize on host-derived products and have been linked in previous research to low fibre diets. However, despite an effect of urbanization at fine spatial scales, phylogenetic patterns in the microbiome were coat colour phenotype dependent. City and built-environment agouti squirrels displayed greater phylogenetic beta-dispersion than those in rural or forest environments, and null modelling results indicated that the phylogenetic structure of urban agouti squirrels did not differ greatly from stochastic expectations.

Conclusions: Squirrel microbiomes differed between city and rural environments, but differences of comparable magnitude were observed between land classes at a within-city scale. We did not observe strong evidence that inter-environmental differences were the result of disparate selective pressures. Rather, our results suggest that microbiota dispersal and ecological drift are integral to shaping the inter-environmental differences we observed. However, these processes were partly mediated by squirrel coat colour phenotype. Given a well-known urban cline in squirrel coat colour melanism, grey squirrels provide a useful free-living system with which to study how host genetics mediate environment x microbiome interactions.

Keywords: 16S rRNA gene; Colour polymorphism; Dispersal limitation; Eastern grey squirrel; Gene x environment interactions; Microbial ecology; Null modelling; Plasticity.

Fig. 1

Representative photos of the (left) agouti (zero copies of the melanism causing MC1R-Δ24 mutant allele), (centre) intermediate (one copy of the MC1R-Δ24 mutant allele), and (right) black (two copies of the MC1R-Δ24 mutant allele) eastern grey squirrel (Sciurus carolinensis) coat colour phenotypes

Fig. 2

Bacterial α-diversity in the squirrel microbiome is negatively correlated with melanism at rural sites (p = 0.02), but positively correlated with melanism among squirrels within city sites (p < 0.01). Solid horizontal lines within violins denote the interquartile range

Fig. 3

The first three axes of a principal coordinate analysis ordination of Euclidean distances (centred log-ratio transformed OTU dataset) separating eastern grey squirrel microbiomes. Points coloured by land class and shaped by local environment type (forest versus built-environment). Solid lines denote 95% confidence ellipses around environment type

Fig. 4

Weighted UniFrac principal coordinate analysis ordination of the eastern grey squirrel microbiome separated by environment and coloured and shaped by phenotype with 95% confidence ellipses. Agouti squirrels display greater beta-dispersion (p = 0.05) within cities (right panel) than in rural environments (left panel)

Fig. 5

An interaction between coat colour phenotype and environment affected (A) |mean nearest taxon distances| (box-cox transformed) (city versus rural), and (B) β mean nearest taxon distance (βMNTD_ses) (built-environment versus forest), such that city and built-environment agouti squirrels had values closer to stochastic expectations. Dotted horizontal lines denote values beyond two standard deviations from the null distribution. Solid horizontal lines within violins denote the interquartile range. Horizontal brackets and ‘*’ denote significant differences in βMNTD_ses variance homogeneity using permutation tests (p < 0.05)

Fig. 6

A stacked barplot of bacterial genera indicated by ANCOM-BC analyses to significantly differ in relative abundance between forest and built-environments in the full dataset or among at least one coat colour phenotype. Facetted by coat colour phenotype and environment type