Conserved microbiota among young Heliconius butterfly species

Introduction

Insects have been co-evolving with their bacterial symbionts for at least 250 million years, developing fascinating complex interactions (Wernegreen, 2002; Toft & Andersson, 2010; Brucker & Bordenstein, 2012). A growing appreciation for these relationships is a result of the realization of the great impact that microbiota can have on the physiology, health, reproduction, behavior, growth, thermal tolerance and longevity of the host (Cox-Foster et al., 2007; Russell et al., 2009; Vásquez et al., 2012). Nonetheless, we still lack a fundamental understanding of the microbial communities associated with insects, the degree of host specialization, and the overall functional importance of these communities to the ecology of their hosts.

Heliconius is one of the best known modern insect radiations. This insect genus is composed of over 40 species and hundreds of geographic population variants. The radiation is characterized by the evolution of several lineage-specific innovations including an ecological shift to pollen feeding. Heliconius is unique among butterflies in their ability to collect and utilize pollen as a protein source (Harpel et al., 2015). The transition to pollen feeding is hypothesized to be important in the butterflies’ ability to synthesize toxic compounds and to enable a very long adult life (Gilbert, 1972). Pollen feeding is also associated with a rapid increase in brain size (Sivinski, 1989; Montgomery, Merrill & Ott, 2016) and the evolution of a suite of complex behaviors. Some of these traits include trap-line feeding, gregarious roosting, and elaborate mating strategies (Brown Jr, 1981). Moreover, pollen feeding can strongly shape the gut microbiota, the metabolism and various physiological traits, which affects the reproductive biology and life history traits of insects (Gilbert, 1972; Engel, Martinson & Moran, 2012). Recent studies suggest the role of bee gut bacteria in the metabolism of pollen and pollen-derived substrates, including flavonoids and outer pollen wall components, attributed to Lactobacilli and Bifidobacteria (Kešnerová et al., 2017). Interestingly, Lactobacilli have been suggested to prevent spoilage of the pollen that wild bees provision for their young by inhibiting the growth of fungi (McFrederick, Vuong & Rothman, 2018).

In recent times a study on ants from the same colonies, species, genera and tribes sharing trophic niches have shown similarities between their microbial communities (Anderson et al., 2012). Our study wanted to test if different pollen-feeding Heliconius butterflies species, sexes and populations present a similar microbiota due to a common feeding strategy. To date, there has only been a previous survey of microbial diversity in Heliconius (Hammer, McMillan & Fierer, 2014), which focused on the microbial similarities and differences between caterpillars and adults of H. erato. Although this study identified several bacterial OTUs shared between larvae and adults, it also found a profound turnover in the bacterial communities between life stages. While adult H. erato didn’t show any sexual differences in microbial composition, wild individuals were dominated by Proteobacteria (74%), followed by Firmicutes (13%), Bacteroidetes (9%), Tenericutes (3%). This result roughly agrees with results reported by Yun et al. (2014), which characterized the gut microbial community across 218 species of insects representing 21 orders.

We hypothesize that Heliconius butterflies sharing the same niches and diet may a similar microbiota. With our work we characterized the bacterial community in a diverse group of adult butterflies with our work we characterized the bacterial community on six common Heliconius species found in forests around Gamboa, Panama, to test how these communities varied among individuals, species, populations and sexes. These diverse group of adult butterflies of the same genus included both closely related species, and species that diverged early in the radiation, ∼12–15 million years ago (Kozak et al., 2015). Considering the diverse ways in which bacteria can influence insect evolution, differences between species could be intertwined with speciation and/or host-plant adaptation. Similarly, differences between male and female butterflies could be caused by bacteria induced pheromones or other sex specific chemical compounds produced by specific microbial communities as in locusts and Drosophila (Dillon, Vennard & Charnley, 2002; Sharon et al., 2013; Leftwich et al., 2017).

Material and Methods

Heliconius butterfly collection

Our experimental approach, done in Gamboa, Panama in 2011 (9°07′11.8″N 79°42′05.6″W), the same location as Hammer, McMillan & Fierer (2014), was to rear and house caterpillars and adult butterflies under controlled experimental conditions in outside insectaries on their natural host plants. Captive rearing had an effect on the host microbial community in adult H. erato, presumably due to differences in the types of host plants available (Hammer, McMillan & Fierer, 2014). However, standardized conditions allowed us to somewhat control environmental variation and better understand any species-specific and/or sex-specific differences in the bacterial communities. Of every (sub) species three males and three females were collected (except H. ismenius which had two individuals of both sexes). Host plants were reared next to the rainforest where the butterflies naturally occur. After emerging, the adult butterflies lived for 2 days and were fed using plants they also feed on in the rainforest, and supplemental sugar water—a common dietary supplement used in butterfly gardens. These plants are also visited by wild butterflies and other insects when stored outside the cage. The butterfly species of our study are shown in: Heliconius cydno chioneus, H. melpomene malleti (not native to Panama), H. melpomene rosina, H. hecale melicerta, H. ismenius boulleti, H. doris viridis (red form) and H. sara magdalena (Fig. 1). For their phylogenetic relationships see phylogeny present Figs. 2 and 3, which is based on (Kozak et al., 2015).

Results

Microbiota comparisons among Species

While the conserved microbiome between the six sequenced species consists of only three shared OTUs, a closer look at the core microbiome within species revealed a highest core microbiome of 11 OTUs in H. doris followed by H. sara with 7 OTUs (Fig.S2). Alpha-diversity did not show significant differences in number of OTUs between Heliconius species (Fig. 2, Table S3). Nonetheless, even with inter-individual variation, bacterial diversity was not distributed randomly among species. Indeed, community structure was evaluated as a beta-diversity measure between species, considering the abundance and composition of microbes. Using a weighted Unifrac distance-matrix, we determined that butterfly species explained 13% of the variation in beta-diversity (Fig. 3 ANOSIM R = 0.1321, P = 0.029, perm = 999), despite the dominant 10 OTUs. In fact, PERMANOVA testing Pseudo-F: 1.50, DF = 1.51, indicated significant differences in the bacterial structure according to species, (p-value = 0.001). Although beta-diversity doesn’t tell which bacteria cause the difference, a log-likelihood ratio parametric test we revealed 40 OTUs that changed significantly in abundance between species (Fig. 4A). Certain groups of microorganisms such as Lactobacillales and Pseudonocardia showed distinct abundance differences between species. While Lactobacillales OTUs were less abundant in H. doris and H. ismenius, Pseudonocardia OTUs were found more abundant in H. doris and H. m. rosina (Fig. S3). We also found that H. hecale, H. cydno and H. ismenius share unique taxa that are not shared with other butterflies (including Erwinia). Finally, some OTUs were more abundant in some species: Bacillus in H. m. rosina and H. m. malleti, Enterobacteriaceae in H. cydno, Entomoplasmatales and Acinetobacter in H. ismenius, Fructobacillus in H. hecale, Lactococcus in H. doris, and Pseudomonas in H. sara (Figs. 1A, 1B, Fig. S3).

Discussion

Our study revealed that the Heliconius adult butterfly microbial richness is relatively low, compared to herbivorous insects feeding in recalcitrant food components and mammals (Yun et al., 2014). This is exemplified by at the overall number of OTUs, which ranged from 29 to 186 OTUs, and from the presence of only 10 dominant OTUs across the whole microbiome of the guts of our 38 individuals. However, the low diversity found in our study is in line with a recent study on H. erato, where an average of 43 OTUs were found of which 12 OTUs have been found to dominate the microbiome (abundance >1%) in nine wild individuals (Hammer, McMillan & Fierer, 2014). Interestingly, the low microbiome complexity doesn’t seem a unique characteristic of Heliconius butterflies. The same simple microbiota composition was also found in Spodoptera littoralis insects (Chen et al., 2016) as well as in different species of mosquitoes from Kenya (Osei-Poku et al., 2012). Our results are thus generally in agreement with the current microbiome literature of insects. As a matter of fact, gut bacterial species richness in insects is considered to be low, except for wood eating insects such as termites (Colman, Toolson & Takacs-Vesbach, 2012; Yun et al., 2014). Our study focused only on the gut microbiota of the butterflies and no data have been collected from the plants they feed or the sugar water that was utilized as food supplement. Our study was designed to provide the first characterization of the microbiome in a group of closely related butterfly species. Thus, due the lack of data from food source we cannot determine the effect of pollen feeding and sugar water on the observed conserved microbiota.

The two phyla, Proteobacteria and Firmicutes, that dominate the microbiome in our study are the same dominant ones found in two recent insect microbial survey studies from (Hammer, McMillan & Fierer (2014) and Yun et al. (2014). A closer look at the overall microbial diversity highlight a total of seven phyla in our study compared to the 13 phyla in the Lepidoptera by Yun et al. (2014). However, we identified for the first time the presence of one rare OTU belonging to the phyla Chlamydiae, which was found in five butterflies representing four species, but was absent in the study of Yun et al. (2014) and has not been reported in Lepidoptera before. This finding could be very important due to the fact members of the Chlamydiae are known to be symbionts or pathogens in insects and other Eukaryotes (Moran, McCutcheon & Nakabachi, 2008). Thus, future studies on the effect of the presence of Chlamydiae in butterflies could reveal important insights into the interaction of this bacteria with the physiology or life history of these insects.

When we compare our study to the only other work published on Heliconius butterflies by Hammer, McMillan & Fierer (2014), we found a strong overlap between the two datasets. The only discrepancy is represented by the much higher presence of Bacteroidetes (9%) in Hammer, McMillan & Fierer (2014) compared to our data (0.02%). One possible explanation of this difference could be that while Bacteroidetes are almost nonexistent in our recently eclosed butterflies (2 days old) they become abundant after few days of pupal emergence represented by the 4-day-old butterflies studied by Hammer, McMillan & Fierer (2014). In concordance with the results by Hammer, McMillan & Fierer (2014) this may indicate that metamorphosing insects increase their microbiota diversity over time, after eclosing, as gut walls become a more stable surface for microbial colonization.

Despite 10 dominant OTUs occupying all species in this study, only three were found present in all butterflies, suggesting the microbiome is conserved among species and has a low diversity. Despite this dominant trend, we found that butterflies species explined 13% of beta-diversity, with 40 low abundant OTUs differing significantly between species. We also found that the two population of Heliconius melpomene (H. m rosina and H. m. malleti) separate slightly in beta-diversity analyses (Fig. 3). The observation that differences between Heliconius species were restricted to rare OTUs can be rather important since rare taxa have been shown to be important in other biological systems. Distinct studies in soils (Hausmann et al., 2016), human skin, human gut and the coral holobiont (Shade et al., 2014; Godoy-Vitorino et al., 2017) have suggested that phylotypes with very low abundance (rare taxa) are likely to be functionally relevant.

The few differences in beta-diversity suggest that the dominant microbiota is conserved between sexes. Sex differences are restricted to only 13 rare OTUs, which showed significant differences in abundance between males and females (p < 0.05). These rare bacteria represent the only differences in microbial patterns associated with butterfly’s sex, which could represent, as mentioned above, candidate rare taxa with likely important functions. Indeed, it is possible that these rare microorganisms are important for responding to changes in environmental conditions, and act as a reservoir that help to stabilize and buffer the dominant microbial community (Shade et al., 2014; Jousset et al., 2017).

When comparing abundances between species and sexes, one should also consider individual microbial variation. In our study, we found a large variation between individuals, as shown by the 11 butterflies dominated by Proteobacteria while another 25 butterflies were dominated by Firmicutes. Similar pattern of variation at the individual level was also observed by Hammer, McMillan & Fierer (2014), and in other studies where intraspecific variability in microbial composition has also been associated with diet- and gut-associated microbiota (Priya et al., 2012; Staudacher et al., 2016).

Diverse aspects can influence the microbiome composition and abundance of specific groups between the different Heliconius butterflies. Dietary factors may represent one of the main elements that modulate the gut microbiota of butterflies, as already demonstrated in other organisms (Colman, Toolson & Takacs-Vesbach, 2012). However, Heliconius areunique among butterflies due the fact they feed on pollen. Their simple diet (sugar and protein) can partially explain their low gut diversity. Moreover, the flowers could function as breeding and sharing grounds of microorganisms for the insects that will visit them, as supported by the broad similarities among pollinating insects (Ushio et al., 2015). This interaction can homogenize the microorganisms’ intake associated to food, especially in groups of individuals that inhabit the same location and thus visit the same flowers. Possibly such interaction could also extend to birds and insects’ pollinators (Vannette & Fukami, 2017).

Heliconius butterflies exhibit complex behavior such as host-plant choice and oviposition, which can explain difference between species (Supple et al., 2014). For example, H. sara and H. doris are two species that display gregarious oviposition behavior, resulting in larvae growing near each other, usually on a single leaf (Beltrán et al., 2007). All the above life history traits can play a role in the physicochemical characteristics in the gut compartments of different butterfly species (whether being pH, O₂ levels or substrate availability), which can influence their microbiome variation.

From our data, Lactobacillales, emerged as some of the few taxa that significantly changed in abundance between species. This group of microorganisms have previously been identified in the bee gut microbiota and associated with the metabolism of pollen and pollen-derived substrates, including flavonoids (Kešnerová et al., 2017). Moreover, Lactobacillales seem to provide anti-fungal effects in pollen provisions (McFrederick, Vuong & Rothman, 2018) and play a key role in the bee health and sociability (Rangberg et al., 2015). These lactic acid bacteria have also found to be dominant members of the microbiota of other insects, including some species of Hymenoptera (McFrederick et al., 2013). Among the other important functions suggested for Lactobacilli, the effect on the regeneration of the gut’s epithelia (Jones et al., 2013), and the possible influence in mating preferences by changing the levels of cuticular hydrocarbon sex pheromones (Matos & Leulier, 2014) have been reported. However, the exact mechanisms with which the Lactobacilli achieve this, remain to be elucidated (Sharon et al., 2010). Although the data in the current study cannot provide direct evidence of the functional capabilities of Lactobacillales, they are likely to have similar functions such as pollen metabolism and gut health in Heliconius butterflies.

Finally, one of the most intriguing results emerged from our data is the presence of an intra-cellular bacterium from the phylum Tenericutes, Spiroplasma sp. This is the first time that this facultative symbiont is described in Heliconius. Neither Muñoz et al. (2010) in 307 wild-caught H. erato chestertonii individuals, nor Hammer, McMillan & Fierer (2014) with 63 H. erato individuals, report this symbiont. However, we found that the raw data from Hammer, McMillan & Fierer (2014) had sequences belonging to the class Mollicutes to which Spiroplasma belongs. We believe that these sequences most likely are Spiroplasma sp., although the classification is less certain due to the 151 bp sequence length compared to our 252 bp. More precisely, we found that the wild caught H. erato male WM5 from Hammer, McMillan & Fierer (2014) presented 33% of its reads homologues to Spiroplasma. It is known that Spiroplasma can infect insects in general, and also Nymphalidae, and is known to kill males in Danaus chrysippus (Jiggins et al., 2000). It is also known that Spiroplasma can enhance survival, as in Drosophila hydei by protecting from attack by parasitic wasps (Xie, Vilchez & Mateos, 2010). If and how these symbionts interfere with the Heliconius life cycle or behavior remains to be investigated.

Conclusions

We found that the Heliconius microbial community, similar to other pollinator insects, is characterized by a low species richness, dominated mainly by Proteobacteria and Firmicutes (Hammer, McMillan & Fierer, 2014; Yun et al., 2014). This low diverse microbiota contrast with a high intra-species variation in microbiome composition. These two results seem to lead to the conclusion that the microbial composition of each individual recapitulate the distinct interaction that each butterfly might have with the surrounding environment. Nonetheless, our data show significant differences between species (13% of beta-diversity, and 40 OTUs) and sexes (13 OTUs). These differences emerged from the low abundant (rare) microbial taxa, suggesting that a deeper sequencing strategy with an Illumina approach might allow to resolve these differences with more confidence and power. We believe that the rare microbiota could actually be very important for the microbial community stability under changing conditions as seen in other host-microbiome systems and should deserve a future investigation. Finally, the presence of the intracellular symbionts like Spiroplasma or Chlamydiae, should be the focus of a functional study to better understand the effect of this bacteria in the butterfly’s behavior, life history and evolution. Altogether, our study represents a step forward into the description of the microbial diversity in a very charismatic group of butterflies not yet very well studied.

Supplemental Information