Bacterial communities of diverse Drosophila species: ecological context of a host-microbe model system

Abstract

Drosophila melanogaster is emerging as an important model of non-pathogenic host-microbe interactions. The genetic and experimental tractability of Drosophila has led to significant gains in our understanding of animal-microbial symbiosis. However, the full implications of these results cannot be appreciated without the knowledge of the microbial communities associated with natural Drosophila populations. In particular, it is not clear whether laboratory cultures can serve as an accurate model of host-microbe interactions that occur in the wild, or those that have occurred over evolutionary time. To fill this gap, we characterized natural bacterial communities associated with 14 species of Drosophila and related genera collected from distant geographic locations. To represent the ecological diversity of Drosophilids, examined species included fruit-, flower-, mushroom-, and cactus-feeders. In parallel, wild host populations were compared to laboratory strains, and controlled experiments were performed to assess the importance of host species and diet in shaping bacterial microbiome composition. We find that Drosophilid flies have taxonomically restricted bacterial communities, with 85% of the natural bacterial microbiome composed of only four bacterial families. The dominant bacterial taxa are widespread and found in many different host species despite the taxonomic, ecological, and geographic diversity of their hosts. Both natural surveys and laboratory experiments indicate that host diet plays a major role in shaping the Drosophila bacterial microbiome. Despite this, the internal bacterial microbiome represents only a highly reduced subset of the external bacterial communities, suggesting that the host exercises some level of control over the bacteria that inhabit its digestive tract. Finally, we show that laboratory strains provide only a limited model of natural host-microbe interactions. Bacterial taxa used in experimental studies are rare or absent in wild Drosophila populations, while the most abundant associates of natural Drosophila populations are rare in the lab.

Figure 1. Composition and distribution of dominant bacterial taxa within 20 natural populations of Drosophila.

A. Pooled samples across all species, diets and locations. “Other taxa” represents 34 families with an average abundance of <0.05% and 18 orders with an average coverage of <1%. B. Venn diagram representing the presence of these taxa within the 20 Drosophila populations. The numbers in the circles indicate how many populations contain at least one member of each of the three dominant bacterial taxa. Note that the Enterobacteriaceae and the Lactobacillales are almost universally found, each being found in 18 and 17 different populations, respectively. 10 populations contain all three dominant bacterial taxa. C. Relative abundance of bacterial orders within 20 wild Drosophila populations. Dark red indicates 100% of sample is composed of that order and white indicates 0% (exact scale at bottom). Note that each population is dominated by either the Enterobacteriales (all family Enterobacteriaceae), the Rhodospirillales (all family Acetobacteraceae), or the Lactobacillales. Diet Key: FRU = Fruit; FLW = Flower; MSH = Mushroom; CCT = Cactus. Location Key: CAL = Northern California; SEY = Seychelles; HAW = Hawaii; TWN = Taiwan; AUS = Australia; MAL = Malaysia; NY = New York; MEX = Mexico. Library identifiers are given in Table 1.

Figure 2. Rarefaction analysis of observed richness within Drosophila.

All calculations were performed using mothur . OTUs were defined at the 3% divergence threshold using the average neighbor clustering algorithm. Library identifiers are given in Table 1. Note the different scales of the Y-axis in panels A and B. A. Rarefaction analysis of wild populations of Drosophila. B. Rarefaction analysis of laboratory collected samples.

Figure 3. Principle component analysis of the natural Drosophila microbiome.

All sequences were aligned and trimmed as described in the text. A single rooted tree for each PC analysis was generated using FastTree . PC analysis was done with the FastUniFrac web application . A: Comparison of the Drosophila microbiome with respect to diet type. All 20 naturally collected samples are included along with the laboratory samples from adult Drosophila feeding on rich Bloomington media (Text S1). B: Comparison of the natural Drosophila bacterial microbiome and the mammalian bacterial microbiome. D. melanogaster data is from Corby-Harris et al., 2007 . Selected mammalian orders are from Ley et al., 2008a .

Figure 4. Composition of OTUs.

A. Composition of OTUs within naturally collected flies with respect to diet type. Asterisks indicate OTUs which derive more than 95% of their sequences from a single diet type. OTU names for the four largest OTUs are given. OTUs with fewer than 5 sequences are omitted. B: Composition of all OTUs with respect to sampling environment (i.e. laboratory or wild environment). OTU names and the absolute number of sequences from lab and wild populations, respectively, are given for the four largest OTUs. OTUs with fewer than 5 sequences are omitted.