Limited Evidence for Microbial Transmission in the Phylosymbiosis between Hawaiian Spiders and Their Microbiota

Abstract

The degree of similarity between the microbiotas of host species often mirrors the phylogenetic proximity of the hosts. This pattern, referred to as phylosymbiosis, is widespread in animals and plants. While phylosymbiosis was initially interpreted as the signal of symbiotic transmission and coevolution between microbes and their hosts, it is now recognized that similar patterns can emerge even if the microbes are environmentally acquired. Distinguishing between these two scenarios, however, remains challenging. We recently developed HOME (host-microbiota evolution), a cophylogenetic model designed to detect vertically transmitted microbes and host switches from amplicon sequencing data. Here, we applied HOME to the microbiotas of Hawaiian spiders of the genus Ariamnes, which experienced a recent radiation on the archipelago. We demonstrate that although Hawaiian Ariamnes spiders display a significant phylosymbiosis, there is little evidence of microbial vertical transmission. Next, we performed simulations to validate the absence of transmitted microbes in Ariamnes spiders. We show that this is not due to a lack of detection power because of the low number of segregating sites or an effect of phylogenetically driven or geographically driven host switches. Ariamnes spiders and their associated microbes therefore provide an example of a pattern of phylosymbiosis likely emerging from processes other than vertical transmission. IMPORTANCE How host-associated microbiotas assemble and evolve is one of the outstanding questions of microbial ecology. Studies aiming at answering this question have repeatedly found a pattern of "phylosymbiosis," that is, a phylogenetic signal in the composition of host-associated microbiotas. While phylosymbiosis was often interpreted as evidence for vertical transmission and host-microbiota coevolution, simulations have now shown that it can emerge from other processes, including host filtering of environmentally acquired microbes. However, distinguishing the processes driving phylosymbiosis in nature remains challenging. We recently developed a cophylogenetic method that can detect vertical transmission. Here, we applied this method to the microbiotas of recently diverged spiders from the Hawaiian archipelago, which display a clear phylosymbiosis pattern. We found that none of the bacterial operational taxonomic units is vertically transmitted. We show with simulations that this result is not due to methodological artifacts. Thus, we provide a striking empirical example of phylosymbiosis emerging from processes other than vertical transmission.

Keywords: Hawaiian arthropods; endosymbionts; host filtering; microbiota; phylosymbiosis; vertical transmission.

FIG 1

Phylosymbiosis in the microbiota of Hawaiian spiders. (a) Phylogenetic tree of the host Ariamnes spiders obtained using ddRAD markers. Interspecific nodes are supported at 100%, whereas most intraspecific nodes have bootstrap supports greater than 70%. (b) Microbiota dendrogram obtained from the Bray-Curtis dissimilarities as a measure of beta diversities comparing the Swarm composition between spiders’ microbiotas (similar patterns were obtained with 97% OTUs [data not shown]). On both trees, tips are colored according to the geographic area of the host spiders, and internal branches are also colored according to the area if all their descendant tips come from the same area; otherwise, they are in black. Colored links indicate which microbiotas belong to which spiders. The Mantel test between the host phylogenetic distances and the microbiota Bray-Curtis dissimilarities indicated a significant correlation (R = 0.32, P = 0.0001). In addition, white dots on the Ariamnes phylogenetic tree indicate nodes having a significant clade-specific phylosymbiosis (56). (c) Map of the Hawaiian archipelago where the Ariamnes spiders were sampled.

FIG 2

HOME results on OTU alignments from the Ariamnes microbiota. (a) Estimated substitution rate as a function of the estimated number of switches for the different OTUs. Each point corresponds to an empirical OTU and its shape corresponds to the type of OTU (Swarm or 97% OTUs) and the representative sequence (the most [1st] or the second [2nd] most abundant sequence per host individual). OTUs corresponding to endosymbionts (Wolbachia, Rickettsia, and Rickettsiella) are in purple. The only two OTUs that rejected the null hypothesis of independent evolution are in orange. (b) Percentage of OTUs rejecting the null hypothesis of independent host-microbial evolution. The two OTUs that rejected the null hypothesis of independent evolution (columns i and iii) belong to the genera Bacillus and Erythrobacter, respectively, occurred in 12 and 51 host individuals, respectively, have 1 and 13 segregating sites, respectively, and have an estimated number of host switches of 12 and 21, respectively (very high compared to the number of hosts where they occurred): they are likely false positives.

FIG 3

Evaluating the effect of substitution rates and preferential host switches on the rejection of the null hypothesis of host-independent evolution. Percentage of simulated OTU alignments that rejected the null hypothesis of independent host-microbial evolution, according to the different evolutionary scenarios simulated on the 63-tip Ariamnes tree: strict vertical transmission (ξ = 0), vertical transmission with ξ host switches (ξ > 0), or independent evolution. Four cases were tested: high substitution rate (a and c) relative to low substitution rate (b and d) and the possibility of host-phylogenetic preferences in the simulated switches (a and b) or geographic preferences (c and d). Simulated values h = 0 and g = 1 correspond to uniformly distributed host switches, whereas the values h > 0 and g < 1 indicate a certain degree of preferential host switches.