Microbes colonize a poison frog embryo after hatching and are vertically transmitted during tadpole transport.

(A) Embryos in the vitelline envelope (dashed line) were manually separated from jelly using tweezers. The vitelline envelope (dashed line) containing the embryo was transferred to sterile water and opened to free the embryo. The embryo was washed in fresh sterile water before homogenization. (B) Detection/non detection of 16S rRNA across developmental stage. DNA was isolated from whole embryos and jellies of different developmental stages and tested for bacterial presence using a broad range PCR for near-complete 16S rRNA gene (see also Fig. S1). (C) After hatching, siblings of a clutch were either (1) not transported (middle arrow), (2) transported by their biological parent (upper arrow), or (3) transported by a foster poison frog of a different species (Oophaga. sylvatica) (lower arrow). (D) We performed 16S v4 specific amplicon sequencing on swabs from the transporting frogs and the skins of the transported tadpoles and used Sourcetracker to identify the sources of taxa that had been acquired by tadpoles. The function was trained on communities of adult Rv and Os that had served as caregivers. Source proportions of both species (Os: orange dots and Rv: blue dots) were determined for each tadpole (N = 24), resulting in 2 data points per tadpole. Proportions were then grouped to display either (1) proportions of the transporting species in transported tadpoles (Rv proportions in tadpoles transported by Rv and Os proportions in tadpoles transported by Os) (dark green), or (2) proportions of the non-transporting species on transported tadpoles (indicating Rv proportions in tadpoles transported by Os and Os proportions in tadpoles transported by Rv) (light green), or (3) proportions of both species in non-transported tadpoles (indicating Rv and Os proportions in non-transported tadpoles) (white). Proportions were compared with a Kruskal-Wallis test with Benjamini-Hochberg correction. Source proportions were square root transformed for plotting.

Tadpoles skin microbiome is shaped by their environment and is more diverse in Rv and Af compared to Ll tadpoles.

(A) We compared the skin microbiome of three anuran species: two species of poison frogs inhabiting different habitats that transport their offspring (Rv and Af) and a leptodactylid frog (Ll) that deposits its eggs in water without transporting the tadpoles. (B) Alpha diversity measures (observed ASV richness, Shannon diversity and evenness) for tadpoles (T) and adults (A) of each species were compared. Differences were determined with an ANOVA or Kruskal Wallis test, significance (p < 0.01) is indicated by **. (C) Comparison of ASV richness, Shannon diversity, and evenness of communities associated with poison frog tadpoles (Af or Rv) and non-poison frog species (Ll). (D) Comparison of Shannon diversity and evenness of communities associated with the aquatic habitats of Af, Rv, and Ll. Bars in boxplots represent median values. The dataset was separately rarefied to the lowest read depth of each comparison.

Species, life stage and parental care affect clustering of microbial communities.

(A) In a Principal Coordinate Analysis constructed with Bray-Curtis distances (axis 1 and 2 on the left, axes 2 and 3 on the right) tadpoles cluster significantly differently from each other, adults, and their aquatic environment. Significances were determined with a PERMANOVA followed by a pairwise adonis posthoc test. Points in ordination plots represent the communities of each sample, circles represent confidence ellipses. (B) Principal Coordinate Analysis constructed with Bray-Curtis and Unifrac distances for adults, tadpoles, and aquatic environment of each species. (C) Number of core species (prevalence > 75%, relative abundance > 0.1%) shared between adults, tadpoles and the respective aquatic environment of each species.

Tadpole transport influences community structure.

(A) Tadpoles collected from the back of their caregiver (“transported”) and reared in artificial cups for one month were compared to six-week-old tadpoles that hatched from eggs transferred to artificial cups and did not experience transport by adult frogs (“non-transported”). (B) Principal Coordinate Analysis constructed using unweighted Unifrac distances, transported tadpoles cluster significantly differently from non-transported tadpoles. Significances were determined with a PERMANOVA followed by a pairwise adonis post hoc test. (C) Venn diagram comparing unrarefied ASVs agglomerated on a genus level between transported tadpoles, non-transported tadpoles, and the aquatic environment. (D) Bubble diagram depicting the presence (circle) or absence (dot) of 10 ASV found to be shared between parents and transported tadpoles as well as their possible source (“transporting parent”, “nursery water”, “refill water” or “untested”). Non-transported tadpoles that died prior to sampling are indicated by ‘X’.