Geographic variation in bacterial assemblages on cane toad skin is influenced more by local environments than by evolved changes in host traits

Abstract

Bacterial assemblages on amphibian skin may play an important role in protecting hosts against infection. In hosts that occur over a range of environments, geographic variation in composition of bacterial assemblages might be due to direct effects of local factors and/or to evolved characteristics of the host. Invasive cane toads (Rhinella marina) are an ideal candidate to evaluate environmental and genetic mechanisms, because toads have evolved major shifts in physiology, morphology, and behavior during their brief history in Australia. We used samples from free-ranging toads to quantify site-level differences in bacterial assemblages and a common-garden experiment to see if those differences disappeared when toads were raised under standardised conditions at one site. The large differences in bacterial communities on toads from different regions were not seen in offspring raised in a common environment. Relaxing bacterial clustering to operational taxonomic units in place of amplicon sequence variants likewise revealed high similarity among bacterial assemblages on toads in the common-garden study, and with free-ranging toads captured nearby. Thus, the marked geographic divergence in bacterial assemblages on wild-caught cane toads across their Australian invasion appears to result primarily from local environmental effects rather than evolved shifts in the host.

Keywords: Bufo marinus; Invasive species; Rapid evolution; Skin bacteria.

Fig. 1.

Genera with relative abundance >2% per toad group from skin swab samples. Values are averages within each toad group. Toad groups represent those present in the common-garden experiment offspring, relocated captive toads at Middle Point, and wild toads.

Fig. 2.

Alpha diversity values in skin bacterial communities of captive and wild cane toads across Northern Australia. Boxplots show the median, interquartile range, reasonable range of the data, and outliers (open circles). Vertical lines separate data between wild toads, captive relocated toads, and common-garden toad offspring.

Fig. 3.

Results of pairwise contrasts of diversity metrics (presented in Table 2) of bacterial ASVs on cane toad skin in Australia. Results of alpha diversity analyses are above the diagonal, and beta diversity comparisons are below the diagonal. Sample sizes are provided in Table 1 and differ between alpha and beta diversity metrics. Dark-grey-shaded squares are pairwise comparisons not focused on in this study. Squares of comparisons of interest are separated into four quadrants, one for each diversity metric. Filled triangles indicate significant differences between groups, with color-coding by study question. Color-coding aids in identifying comparisons relevant for the questions of interest but is not necessary for interpretation. Significant pairwise differences between groups were rare for alpha diversity metrics (GLMs or LMs), but almost ubiquitous in beta diversity metrics (PERMANOVAs). Common-garden offspring of different ancestry had the most similarities in diversity metrics. ∼ indicates instances where the OTU comparison had an opposing significance (< or >0.05) to the ASV comparison. Asterisks denote pairwise differences in dispersion. All global analyses of dispersion were significant. WA: toads captured near the invasion front in Western Australia. This figure is available in table format in the supplementary information. Note that Shannon diversity and Pielou's evenness for ASV data did not have a significant interaction between origin site and captivity, so we do not report results of pairwise contrasts for the ASV data of those metrics.

Fig. 4.

Venn diagrams representing core skin bacterial communities (ASVs with 100% prevalence) on groups of cane toads (sample sizes per group are given in Table 1). Comparisons are among (A) wild sites and (B) toads at Middle Point. Common-garden offspring are further compared with toads from their respective ancestral sites and wild toads from Middle Point for each of (C) Innisfail, (D) Tully, and (E) Western Australia. Numbers and shading in the diagrams indicate number of microbes in the segment. Values in parentheses are the total number of ASVs present in samples from that group of toads. Below each diagram are percentages of reads represented by core microbes shared between groups of interest (average±s.d.). These overlaps of interest are highlighted in the diagrams with bold outlines. Relocated parent toads were excluded from Venn diagrams due to their small sample sizes.