North American Fireflies Host Low Bacterial Diversity

Although there are numerous studies of firefly mating flashes, lantern bioluminescence, and anti-predation lucibufagin metabolites, almost nothing is known about their microbiome. We therefore used 16S rRNA community amplicon sequencing to characterize the gut and body microbiomes of four North American firefly taxa: Ellychnia corrusca, the Photuris versicolor species complex, Pyractomena borealis, and Pyropyga decipiens. These firefly microbiomes all have very low species diversity, often dominated by a single species, and each firefly type has a characteristic microbiome. Although the microbiomes of male and female fireflies did not differ from each other, Ph. versicolor gut and body microbiomes did, with their gut microbiomes being enriched in Pseudomonas and Acinetobacter. Ellychnia corrusca egg and adult microbiomes were unique except for a single egg microbiome that shared a community type with E. corrusca adults, which could suggest microbial transmission from mother to offspring. Mollicutes that had been previously isolated from fireflies were common in our firefly microbiomes. These results set the stage for further research concerning the function and transmission of these bacterial symbionts.


Introduction
Beetles (order: Coleoptera) are one of the most diverse insect groups, containing nearly 400,000 species [1]. The beetles in the family Lampyridae are commonly known by many names, fireflies, lightning bugs, glow worms, lamp-lighters, and nighttravelers, all describing the bioluminescent lantern that they use to signal to their mates [2]. Fireflies are found on every continent except Antarctica, living in woods, plains, and marshes. North America is home to over 125 firefly species, including Ellychnia corrusca, the Photuris versicolor species complex, Pyractomena borealis, and Pyropyga decipiens [2]. These 4 types of fireflies occur sympatrically, ranging from the eastern coast of Canada, south to Florida, and west to the Great Plains [2,3]. All four of these fireflies are carnivorous or omnivorous as larvae, attacking snails as their prey and/or eating nectar and plant sap [4]. Ellychnia corrusca, Pyractomena borealis, and Pyropyga decipiens are all closely related to each other, and both E. corrusca and Pyropyga decipiens are diurnal with no lantern [2]. Ellychnia corrusca, Pyractomena borealis, and many other North American fireflies use lucibufagins as a chemical defense. Lucibufagins are cardioactive C-24 steroidal pyrenes and a subclass of the bufadienolides [5][6][7][8] that create a disagreeable taste to predators and thereby provide fireflies with protection from bats, birds, and spiders [5]. However, the biosynthetic origin of lucibufagins is not known. At least five genera of fireflies, including the groups in this study, ingest the nectar of the common milkweed plant (Asclepias syriaca L.). Although eating is not required for adult firefly survival, cardenolides (a type of toxic steroid) found on milkweed plants are ingested by adult fireflies and may be used to produce lucibufagins [9].
In this study, we focused especially on two fireflies with unique lifestyles: E. corrusca and the Photuris versicolor species complex. Ellychnia corrusca is nicknamed the winter firefly because it has a winter-spring activity cycle, in contrast to the late spring-summer activity cycles of most other fireflies [2,10]. Ellychnia corrusca larvae in New England (North America) extend their larval stage across 2 years instead of one to ensure that they ingest ample calories, emerging in the late fall of their second year as adults and sexually maturing during the winter to start mating, which lasts until late spring [11]. Normally, adult fireflies do not need to feed because they only survive as adults for 2 to 4 weeks [2,4]. However, E. corrusca firefly adults live Scott R. Smedley Deceased for~9 months, and in MA, USA, they ingest interstitial fluid and sap from maple trees to help them survive during the cold winters [11]. The Photuris versicolor species complex (hereafter: Ph. versicolor) contains > 17 firefly species that are differentiated by their male courtship lantern flashes but that cannot be identified from their morphology with confidence. The Ph. versicolor adult life cycle occurs during the summer months (June to August), but unlike other fireflies, Ph. versicolor females mimic the mating light display of female Photinus fireflies to attract Photinus males, earning Ph. versicolor the nickname "femme fatale" [12]. Once a male Photinus firefly is near, the Photuris versicolor female will attack and kill the male, gaining nutrients and lucibufagins that she uses to protect herself, her eggs, and her pupae, which do not themselves produce lucibufagins [5]. Many E. corrusca populations complete their life cycles before Ph. versicolor becomes seasonally abundant, especially where temperatures are colder. However, Ph. versicolor predation of E. corrusca can occur, particularly in the Southern USA. Ph. versicolor will also attack and eat E. corrusca to gain protective lucibufagins in the lab [2,8], which could give insight into the selective pressures that caused E. corrusca to mate in the winter instead of in the summer.
Beetles feed on many different substrates, following herbivorous [13], omnivorous [14], xylophagous [15], detritivorous [16], and carnivorous diets [17]. Beetles that belong to similar taxonomic families but that have different diets have distinct microbial communities [17][18][19]. Diets lacking in nutrients often cause insects to rely on nutritional symbionts to provide vital nutrients [20], as in dung [21], carabid [14], and carrion beetles [22]. Although there is a vast amount of research studying beetle microbiomes, only a handful of studies consider firefly-associated microbes. In those studies, several Mollicute bacteria were isolated from the fireflies E. corrusca and Ph. versicolor, but this does not give insight into the composition of their microbiome beyond these strains [23][24][25][26][27][28][29][30]. To fill this gap, we used 16S rRNA community amplicon sequencing to survey the microbiomes of E. corrusca, Photuris versicolor, Pyractomena borealis, and Pyropyga decipiens fireflies. Our results show that fireflies have simple, species-specific microbiomes and generate hypotheses about how diet and seasonality may drive firefly microbiome structure and function.

Sample Collection
Live E. corrusca and Photuris versicolor fireflies were collected by S. Smedley during the winter of 2016-2017 and spring/summer of 2017 from residential and camping/forest areas within Vernon, Bolton, and Andover, CT, USA. Pyractomena borealis larvae and adults were raised in the laboratory by S. Smedley

Sample Preparation and DNA Extraction
All fireflies were surface-sterilized using 3 rounds of a 10-second submersion in 70% ethanol, followed by a 10-second submersion in phosphate-buffered saline (PBS) [31]. Surface-sterilized fireflies were dissected into two separate tissues: a gut sample and the remaining carcass "body" sample. DNA from the tissue dissections were extracted using a bead beating and chloroformisopropanol protocol as described in [32], except using 0.7 g of 1 mm and 0.3 g of 0.1 mm silica/zirconium beads and 5 cycles of bead beating and chilling on ice. Negative controls containing only the DNA extraction reagents were processed alongside each batch of tissue samples. The DNA concentration of each extract and negative control was determined using the Qubit dsDNA high-sensitivity assay protocol and a Qubit 3.0 fluorimeter (Invitrogen, Carlsbad, CA).

16S V4 rRNA PCR Screen and Community Amplicon Sequencing
DNA samples were PCR amplified using primers 515F and 806R (targeting the bacterial 16S rRNA gene V4 region) to determine the presence of bacterial DNA [33]. Ten nanograms of template DNA were added to 5-μl Green GoTaq Reaction Mix Buffer (Promega, Madison, WI, USA), 1.25 units of GoTaq DNA Polymerase (Promega, Madison, WI, USA), 10 μmol of each primer, and 300 ng/μl BSA (New England BioLabs Inc. Ipswitch, MA), to which nuclease-free H 2 O was added to a volume of 25 μl. Thermocycling conditions (BioRad, Hercules, CA), were 3 min at 95°C, 30 cycles of 30 sec at 95°C, 30 sec at 50°C, and 60 sec at 72°C, followed by a 5 min cycle at 72°C and then an indefinite hold at 4°C. Gel electrophoresis was used to confirm the expected band size of 300-350 bp.
All samples that had a gel band of the expected size were prepared for community amplicon sequencing of the 16S rRNA gene V4 region using an Illumina MiSeq at the University of Connecticut Microbial Analysis, Resources and Services (MARS) facility. Approximately 30 ng of DNA from each sample was added to a 96-well plate containing 10 μmol each of the forward and reverse Illuminabarcoded versions of primers 515F and 806R, 5 μl AccuPrime buffer (Invitrogen, Carlsbad, CA), 50 mM MgSO 4 (Invitrogen, Carlsbad, CA), 300 ng/μl BSA (New England BioLabs Inc. Ipswitch, MA), a 1 μmol spike-in of both non-barcoded primers 515F and 806R, and 1 unit AccuPrime polymerase (Invitrogen, Carlsbad, CA), to which nuclease-free H 2 O was added to a volume of 50 μl. Reaction mixes were separated into triplicate reactions (each with a volume of 16.7 μl) in a 384-well plate using an epMotion 5075 liquid handling robot (Eppendorf, Hamburg, Germany). The resulting 384-well plate was then transferred to a thermocycler (Eppendorf, Hamburg, Germany), which used the following conditions: 2 min at 95°C, 30 cycles of 15 sec at 95°C, 60 sec at 55°C, and 60 sec at 68°C, followed by a final extension for 5 min at 68°C and then an indefinite hold at 4°C. After PCR, triplicate reactions were re-pooled using the epMotion and DNA concentrations were quantified using a QIAxcel Advanced capillary electrophoresis system (QIAgen, Hilden, Germany). Samples that had concentrations > 0.5 ng/μl were pooled using equal weights of DNA to create the final sequencing libraries. Libraries were then beadcleaned using Mag-Bind RXNPure plus beads (OMEGA, Norcross, GA) in a 1:0.8 ratio of sequencing library to bead volume. Cleaned library pools were adjusted to a concentration of 1.1 ± 0.1 ng/μl, which was confirmed using the Qubit dsDNA high-sensitivity assay on a Qubit 3.0 fluorimeter (Invitrogen, Carlsbad, CA). Microbial community sequencing on an Illumina MiSeq (Illumina, San Diego, CA) was completed in 2 batches, the first composed of 52 samples and the second composed of 195 samples.

Post-sequencing and Bioinformatic Analyses
Our DNA sequencing produced 245 16S rRNA community amplicon sequencing datasets, including 22 negative controls. Reads were analyzed using R v3.5.3 [34] and the dada2 v1.11.1 [35] pipeline for amplicon sequence variants (ASVs) (https://benjjneb.github.io/dada2/tutorial.html, accessed: November 11, 2017). Read counts ranged from 2 to 888,868 (Suppl. File S1). Metadata files for the samples were imported into phyloseq v1.26.1 [36], creating a phyloseq R object that was used for subsequent analyses. Reads that were not classified as belonging to the kingdom Bacteria using the SILVA database v128 were removed [37,38]. ASVs that matched to mitochondria were then removed separately, because SILVA included them in the kingdom Bacteria. Samples were screened for contamination using the decontam v1.2.1 [39] prevalence protocol with a default threshold value of 0.1. No reads were flagged as contamination, and 17 samples with 0 reads were removed, resulting in 1675 unique ASVs (Suppl. File S1). Negative control samples were not considered further. All samples in the dataset were then rarefied to 10,000 reads and read counts were converted to relative abundances. The final phyloseq object contained 133 samples from 20 Photuris versicolor adults, 4 Pyropyga decipiens adults, 7 Pyractomena borealis larvae, 1 Pyractomena borealis adult, 95 Ellychnia corrusca adults, and 6 E. corrusca eggs. Despite amplification during the initial PCR screen, of the initial 223 insect samples, 74 samples were removed due to low read counts (53 E. corrusca adult samples, 3 E. corrusca egg samples, 9 Photuris versicolor samples, 2 Pyropyga decipiens samples, 4 Pyractomena borealis adult samples, and 3 Pyractomena borealis larvae samples). These low read counts may have been due to (1) high amounts of host DNA acting as a PCR inhibitor; (2) there being a minimal firefly microbiome, leading to limited template concentrations; or (3) other technical issues such as inefficient PCR amplification using primers that contained the Illumina barcodes compared to our initial PCR screen using non-barcoded primers.
Alpha diversity was measured using the phyloseq "plot_richness" command, and beta diversity was measured using weighted and unweighted UniFrac distance metrics. Weighted and unweighted UniFrac distances were calculated, ordinated, and viewed using the "distance," "ordinate," and "plot_ordinate" phyloseq commands, respectively. PERMANOVA statistical tests were calculated using vegan v 2.5-4 [40]. Although weighted UniFrac (WUF) and unweighted UniFrac (UUF) distances were used for each test, to keep the text concise, only one test is listed in the text and the complimentary values are presented in Supplemental Tables S1-S4.
The most abundant Mollicute ASVs from our dataset included four Mesoplasma ASVs, two Spiroplasma ASVs, and one Entomoplasma ASV. A phylogenetic tree was constructed to show the relationships between the 16S rRNA sequences of Mollicutes from this study and those that had been previously isolated from fireflies. Reference Mollicute sequences from taxa belonging to the same genera as our firefly ASVs were selected from the SILVA database, especially those Mollicutes that had been isolated from fireflies and other beetles. Additional Mollicute reference sequences for fireflies that were not represented in the SILVA database were downloaded from NCBI. Sequences were aligned using MUSCLE v3.8.31 [45] and trimmed to the same length. The phylogenetic tree, rooted by the 16S rRNA genes from Bacillus subtilis and Mycoplasma haemominutum, was calculated using a GTRGAMMAI substitution model and 500 bootstrap replicates in RAxML v8.2.11 [46]. The topology of the tree created using these selected sequences agreed with that constructed using all Mollicute sequences in SILVA, ensuring that taxon selection did not bias our phylogenetic analysis.
The commands used for all analyses are attached as Suppl.

Firefly Microbiomes Are Typically Dominated by Single Taxa
We characterized the bacterial communities in 133 firefly gut and body dissections (Suppl. File S1) using community amplicon sequencing of the 16S rRNA gene. All firefly microbiomes had low α-diversity (Suppl. Fig. S1). Mean Shannon diversity scores for the firefly microbiomes ranged from 0.70 to 2.52, and alpha diversities of Photuris versicolor and E. corrusca microbiomes differed from each other, as did those from Pyropyga decipiens and E. corrusca (Dunn test, p = 0.001 and p = 0.003, respectively). A heatmap of the top 26 bacteria genera found in the microbiomes reflects these low levels of α-diversity, with most firefly microbiomes dominated by a single taxon but with minute amounts of other taxa also present ( Fig. 1). Unsupervised clustering of these data grouped E. corrusca samples tightly together at the left side of the heatmap, while the samples from other species clustered together on the right. Samples from different tissue dissections and sexes did not cluster together for any species of firefly.
Using these top 26 bacteria genera, we identified taxa whose relative abundances were correlated with each other to infer potential interactions between them (Suppl. Fig. S2). Of the 26 bacteria genera, the relative abundances of 14 genera did not correlate with those of another genus in our dataset, and the relative abundances of 11 genera were positively correlated with those of another genus. The relative abundances of Ralstonia and Cupriavidus, found in the Pyropyga decipiens samples, were strongly and positively correlated with each other, as were the relative abundances of Tanticharoenia and Gluconobacter, found in E. corrusca eggs, and the relative abundances of Gordonia and Tsukamurella, found in Photuris versicolor. Only the relative abundances of Mesoplasma and Acinetobacter were negatively correlated with each other, and these taxa were not found together in any sample.
In the heatmap, many samples clustered together that were dominated by single taxa. Based on these clusters, we defined community types defined by the genus or genera of bacteria that were present in these samples with relative abundances ≥ 30% (Fig. 2). Ellychnia corrusca adults were assigned to 7 distinct community types: T-1 (Mesoplasma), T-2 (Salmonella), T-3 (Mesoplasma and Salmonella), T-4 (Serratia), T-5 ( R i c k e t t s i a ) , T -6 ( P s e u d o m o n a s ) , a n d T -7 (Rickettsiella). Ellychnia corrusca egg microbiomes were all assigned to T-13 (Gluconobacter), except for a single egg microbiome that belonged to T-2 (Salmonella), the same as some E. corrusca adult microbiomes. Photuris versicolor microbiomes were assigned to 3 community types: T-8 (Acinetobacter), T-9 (Yersinia), and T-6 (Pseudomonas). Pyractomena borealis and Pyropyga decipiens were assigned to T-1 0 ( E m p e d o b a c t e r ) and T-12 (C u p r i a v i d u s ) , respectively.

Firefly Species Harbor Unique Microbial Communities
The four sampled firefly taxa had distinct bacterial communities (Weighted Unifrac [WUF] PERMANOVA: R 2 = 0.215, p = 0.001; Fig. 3a, Suppl. Fig. S3, Suppl. Table S1). In a PCoA ordination of WUF beta diversity distances, E. corrusca samples clustered together, spanning from left to right (Fig. 3). Pyropyga decipiens samples clustered together with some E. corrusca samples and those from Photuris versicolor on the left side of Fig. 3a, and Ph. versicolor samples grouped together down the left side and at the bottom, along with the Pyractomena borealis samples. Species-specific clustering can be seen in the UUF plot, with E. corrusca having the most variance (Suppl. Fig. S3B). The separation between the E. corrusca and Ph. versicolor microbiomes was maintained in a better-balanced comparison when the few Pyropyga decipiens and  Table S2).
We identified the bacterial taxa whose relative abundance differed between E. corrusca and Photuris v e r s i c o l o r f i r e f l i e s u s i n g D E S e q . A l t h o u g h Pseudomonas ASVs were found in both fireflies, one Pseudomonas ASV was more abundant in Ph. versicolor and two other Pseudomonas ASVs were more abundant in E. corrusca (Fig. 3B). Two Mesoplasma A S V s , S a l m o n e l l a , S e r r a t i a , R i c k e t t s i a , and Rickettsiella were all more abundant in E. corrusca than in Ph. versicolor, and Tsukamurella, Gordonia, Curvibacter, and Sphingobacterium were all more abundant in Ph. versicolor than in E. corrusca (Fig. 3b).
Gut and Body Microbiomes Are Distinct, but These Differences Are Firefly-Specific The above analysis indicated that the four sampled firefly groups all hosted distinct microbiomes (Fig. 3). However, further analyses using these same data also indicated that different firefly tissues might also host distinct microbiomes, in a species-specific manner (WUF PERMANOVA for the interactions between species and tissue: R 2 = 0.048, p = 0.003; Suppl. Fig. S3, Suppl. Table S1). In contrast, a parallel analysis using only samples for which sex was determined indicated that microbiomes did not differ between the sexes, regardless of species or tissue type (all PERMANOVAs testing for differences between sexes: p > 0.05; Suppl. Fig. S5, Suppl. Table S3). In the analysis described above with Pyractomena borealis and Pyropyga decipiens samples removed to avoid possible artifacts due to unbalanced sample sizes, firefly tissue microbiomes again differed in a species-specific manner (WUF PERMAONVA for interactions between E. corrusca and Photuris versicolor gut and body microbiomes: R 2 = 0.053, p = 0.001; Suppl. Fig. S4, Suppl. Table S2).
Because these PERMANOVAs indicated that tissue microbiomes differed in a species-specific manner, we repeated our analysis for each species separately. Ellychnia corrusca eggs, gut, and body microbiomes differed from each other (WUF PERMANOVA: R 2 = 0.122, p = 0.001), with most gut and body samples clustered together in the PCoA and most egg samples clustered separately (Fig. 4a, Suppl. Table S4, Suppl. Fig. S6). When the egg samples were removed from the analysis, E. corrusca gut and body microbiomes differed from each other only when using the UUF distance metric (UUF PERMANOVA: R 2 = 0.072, p = 0.001; WUF: Suppl. Table S4). Wolbachia had low relative abundance in all E. corrusca samples but was the only Fig. 1 Heatmap of the 26 most abundant bacterial genera found in the studied fireflies. To the left, a dendrogram represents the Euclidian distances between the relative abundances of reads assigned to each bacterial genus, which are labeled on the right. The top dendrogram clusters the relative abundances of these genera in each firefly sample using ward.D distances. On the bottom, 3 different rows indicate the host taxon, tissue type, and sex for each sample, indicated by the different colors in the key to the left of the figure. The topmost color key represents the relative abundance of reads in each sample that were assigned to each bacterial genus, with white and dark blue representing 0 and 100% relative abundance, respectively. n = 133 bacterial genus that was more abundant in E. corrusca eggs than in adults. Mesoplasma, Pseudomonas, Acinetobacter, and several other genera were all more abundant in E. corrusca adults than in eggs (Fig. 4b). Ellychnia corrusca Fig. 2 Firefly microbial community types. The X axis indicates the number of samples that were assigned to each community type. Colors indicate firefly taxa, as indicated by the key. Note that some gut and body samples originate from the same individual firefly and so are counted twice a b Fig. 3 a PCoA of weighted UniFrac distances between microbial communities for all four firefly taxa. Species are differentiated using colors. P. borealis = Pyractomena borealis, P. decipiens = Pyropyga decipiens. n = 133. b Over-and underrepresented ASV sequences in E. corrusca and Photuris versicolor, adults, including both gut and body samples. The X-axis indicates the genera of bacteria whose relative abundances differed between firefly taxa, and the Y-axis indicates the log2 fold change in relative abundance between samples, where the higher numbers indicate overrepresentation in Ph. versicolor and negative numbers indicate overrepresentation in E. corrusca. Colors indicate phyla. n = 114 eggs and adults also had distinct community types, with E. corrusca egg microbiomes mainly assigned to T-13 (Gluconobacter), and only a single egg microbiome assigned to T-3 (Salmonella) like those of E. corrusca adults (Fig. 2). In contrast, Photuris versicolor gut and body microbiomes more strongly differed from each other (WUF PERMANOVA: R 2 = 0.340, p = 0.001; Fig. 4c, Suppl. Fig. S7, Suppl. Table S4), with Pseudomonas, Acinetobacter, Leucobacter, a b d c Fig. 4 a PCoA of weighted UniFrac distances between microbiomes of E. corrusca eggs and adults. Egg, gut, and body samples are differentiated using colors. n = 100. b Over-and underrepresented ASV sequences in E. corrusca egg and adult samples. The X-axis indicates the genera of bacteria whose relative abundance differed between E. corrusca eggs and adults, and the Y-axis indicates the log2 fold change in these relative abundances between tissues, where the higher numbers indicate overrepresentation in eggs and negative numbers indicate overrepresentation in adults. Colors indicate phyla. n = 100. c PCoA of weighted UniFrac distances between microbiomes of Photuris versicolor gut and body samples. Gut and body samples are differentiated using colors. n = 20. d Over-and underrepresented ASV sequences of Ph. versicolor gut and body samples. The X-axis indicates the genera of bacteria whose relative abundance differed between Ph. versicolor gut and body samples, and the Y-axis indicates the log2 fold change in these relative abundances between tissues, where the higher numbers indicate an overrepresentation in the gut samples. Colors indicate phyla. n = 20 Elizabethkingia, Curvibacter, and Achromobacter all being more abundant in Ph. versicolor guts than in bodies (Fig.  4d). Pyropyga decipiens gut and body microbiomes did not differ from each other (WUF PERMANOVA: R 2 = 0.312, p = 0.667; Suppl. Table S4), and Pyractomena borealis gut and body microbiomes differed only when using the UUF distance metric (UUF PEMANOVA: R 2 = 0.616, p = 0.008; Suppl. Table S4). These conclusions should be considered preliminary because of the small sample sizes available for both Pyropyga decipiens and Pyractomena borealis.

Fireflies Host an Abundance of Potentially Symbiotic Mollicutes
Mollicutes were the most abundant bacteria in our 133 firefly samples, consistent with the handful of Mollicutes that had previously been isolated from different firefly species [23][24][25][26][27][28][29][30]. We therefore created a phylogenetic tree to discover how our most abundant Mollicute ASVs were related to the sequences of these known firefly Mollicute isolates (Fig. 5). The most prevalent Spiroplasma ASV (ASV Spiroplasma1), found in many of the E. corrusca samples, was similar to the 16S rRNA sequence of S. corruscae, a species that was first isolated from E. corrusca [25]. ASV Spiroplasma2 was found in one Photuris versicolor sample and was similar to the 16S rRNA sequence of S. ixodetis, a male-killing agent in butterflies [47]. Spiroplasma ixodetis has not been found in fireflies to date. The four Mesoplasma ASVs, mainly found in E. corrusca, are all similar to the 16S rRNA sequences for M. corruscae and Entomoplasma ellychniae, which had both been isolated previously from E. corrusca [26,28,30]. Although originally separated using cell morphology and culture media requirements, the genera Mesoplasma and Entomoplasma are paraphyletic [48]. Our phylogenetic tree showing that both M. corruscae and E. ellychniae cluster tightly together, as seen by Gasparich et al. [48], who suggest that the genus Mesoplasma should be changed to Entomoplasma. This would mean that M. corruscae would be reclassified as a member of the genus Entomoplasma, along with all other Mesoplasma species. The single Entomoplasma ASV, found in Pyractomena borealis adults, closely resembled the 16S rRNA sequence for E. somnilux, which had been previously isolated from Pyractomena angulata [24]. This phylogenetic tree shows that the Mollicutes sequences detected in study are highly similar to strains that had been previously isolated from fireflies, except for that related to S. ixodetis. Fig. 5 Phylogenetic tree of the Mollicute 16S rRNA gene ASVs in our dataset (in bold) and reference 16S rRNA gene sequences, particularly those isolated from fireflies and beetles (indicated by a^). The tree was constructed using RAxML with 500 bootstraps and rooted using Bacillus subtilis and Mycoplasma haemominutum. Bootstrap values of 60-79 and 80-100 are indicated by * and **, respectively. Numbers in parentheses indicate the number of samples in which that ASV was found, and colors group sequences from the same genera. NCBI accession numbers are shown to the right of each reference sequence. Me = Mesoplasma, My = Mycoplasma

Discussion
Our results show that all of our sampled fireflies have low-complexity microbiomes. The alpha diversity of our firefly microbiomes was very low, and all samples had only a few taxa present in high abundance (Fig. 1). A previous survey also suggested that beetles had lowdiversity microbiomes, with carnivorous and herbivorous species hosting the lowest diversity [17]. Our data is consistent with this, as Photuris versicolor fireflies are carnivorous and E. corrusca ingest plant sap. Kolasa et al. [17] also describe large variation in microbiome composition within a host beetle species, which is consistent with our firefly microbiome data. However, the Shannon diversities measured for our firefly microbiomes (means ranging from 0.7 to 2.52) are lower than those reported in that beetle survey [17]. The low number of correlations between the relative abundances of firefly taxa (Suppl Fig. S2) is consistent with this predominance of simple microbiomes, because (by definition) few interactions can occur when only a single bacterium is abundant in a microbiome, and correlations will inevitably be weak when there is high variability between microbial communities. This variance and low diversity could be explained by microbiomes being acquired via neutral community assembly. Fireflies may be exposed to environmental bacteria by chance, and these bacteria may then multiply and form a stable microbiome. This would explain why there is high variance between the microbiomes from individual insects of the same firefly species.
Samples could be assigned to one of thirteen community types, defined by the taxa in these samples with > 30% abundance. Different firefly taxa rarely shared the same community type, and most community types were defined by a single taxon, with only one community type having two taxa that were each > 30% abundant (Fig. 2). The presence of multiple community types in each firefly species raises questions about whether the bacteria are either (1) resident members replicating within the firefly or (2) transient members that replicate more slowly than the rate of expulsion from the firefly [49]. Although such transient microbiomes are not likely to be conserved between fireflies, it is possible that they do have some effect on their hosts (positive or negative). Community types vary within the same firefly species, suggesting that these microbiomes are transient, but some bacteria were found in extremely high relative abundances (Fig. 1) that could suggest a stable resident microbiome of unknown function. It is possible that each community type provides a different function for the firefly or that the multiple, highly abundant bacteria in different community types, such as Mesoplasma, Salmonella, and Pseudomonas, could provide similar functions for their hosts. However, this may be unlikely for Mollicute bacteria due to their reduced genomes.
Fireflies from different taxa also had distinct microbiomes (Fig. 3a). Ellychnia corrusca and Photuris versicolor had distinct bacterial communities, and samples from the same species clustered together in PCoA plots of beta diversity (Suppl. Fig. 4). Differences between the microbiomes of these two groups might be due to their different lifestyles. North American E. corrusca are active in winter and may feed on tree sap [10,11], and so E. corrusca microbial communities may therefore be acquired via the ingestion of these fluids. Ellychnia corrusca also live for~9 months. This much longer life span compared to other the firefly taxa in this study could explain the greater diversity of community types in E. corrusca. Unlike E. corrusca, Ph. versicolor females are predatory and active in the summer. Insect gut microbiomes often vary based on differences in their host's habitat and diet (e.g., omnivory vs. carnivory) [18], and such differences might underlie the differences that we observed between E. corrusca and Ph. versicolor microbiomes. Both E. corrusca and Ph. versicolor hosted Pseudomonas ASVs that were similar to those previously found on trees and in soils, which could be explained by E. corrusca and Ph. versicolor living in soil, on tree bark, and on plant leaves [2], where Pseudomonas is common. Pseudomonas is also commonly found in beetles from the family Staphylinidae, which are primarily predatory, but not in other nonpredatory beetles [17]. This could suggest a relationship between diet and microbiome composition.
All of the E. corrusca and Ph. versicolor samples used in this study were collected in CT, USA, and all Pyractomena borealis samples were lab raised. Pyropyga decipiens samples were collected~184 miles apart but had similar microbiome diversities compared to fireflies collected in Connecticut, which suggests that Pyropyga decipiens microbiomes may not differ between geographic locations. Whether the microbiomes of E. corrusca, Ph. versicolor, and Pyractomena borealis vary in other geographic locations needs future research.
For some firefly taxa, eggs, guts, and bodies had distinct microbiomes. Ellychnia corrusca adult and egg microbiomes differed from each other, with samples clustering separately in PCoAs (Fig. 4a, Suppl. Fig. S6), but an E. corrusca egg also had the same community type (T-2; Salmonella) as some adults. Although it is unknown which firefly laid this egg, these results suggest that it might be possible for Salmonella to be transmitted vertically between E. corrusca adults and eggs. The mechanisms that E. corrusca may use to acquire these microbes from the environment are unknown, but there are other insects that rely on horizontal transmission for their symbionts, such as the bean bug Riptortus pedestris, which acquires its Burkholderia symbiont from soil each generation using a mucus-filled organ that the Burkholderia specifically penetrates and colonizes [50]. Although E. corrusca is not known to have an organ that selects for a specific bacterial symbiont, Mesoplasma was found in many samples of E. corrusca. It is possible that E. corrusca may select this bacterium from the environment or that Mesoplasma selects E. corrusca as a preferred host. Ellychnia corrusca is the most sampled firefly species in our study and hosts microbiomes belonging to the most community types. This increased sampling may therefore have increased the number of community types detected in this organism relative to the other species in our study.
The microbiomes of the male and female Photuris versicolor did not differ from each other, suggesting that the microbiome is not acquired from the firefly prey because only female Ph. versicolor are carnivorous, although adult Ph. versicolor males do eat milkweed nectar. Photuris versicolor gut and body microbiomes also differed from each other, with gut microbiomes having higher amounts of Pseudomonas and Acinetobacter than body microbiomes (Figs. 4c and 4d). These ASVs may be acquired from the firefly's environment, because both ASVs were similar to other strains of bacteria that are common in soil and on plants. The bacteria present in the Ph. versicolor gut may be derived from eating other fireflies or nectar, and thus the Pseudomonas ASVs that are common in Ph. versicolor guts may originate from their last meal. The Ph. versicolor microbiome may therefore be transient and assembled from the diet microbiome. Alternatively, it is possible that the Ph. versicolor gut microbiome is not transient, and the gut provides a niche that is favorable for the growth of Pseudomonas and Acinetobacter microbes that might originate from soil, plants, or the diet microbiome and subsequently colonize the Ph. versicolor gut.
The origin of the body microbiome remains unclear. Body dissections include all other parts of the firefly except for the gut, meaning that other non-gut tissues and hemolymph from inside of the firefly could have been colonized with bacteria. Alternatively, adult firefly abdomens are surrounded with sclerotized segments called ventrites and tergites on their ventral and dorsal sides, respectively [2]. These ventrites and tergites have small gaps between them that allow for movement. Although our firefly samples were ethanol-washed before dissection, this sterilization may not be absolutely perfect, and the body microbiome may therefore include bacteria that are trapped within these segments [51] that were acquired during movement on tree bark or in the leaf litter, where many bacteria are found.
Mollicutes have been isolated from several different firefly species, but their prevalence and function are unclear [23][24][25][26][27][28][29][30]. We detected 4 Mesoplasma ASVs in North American fireflies, mainly in E. corrusca, that all closely resembled M. corruscae and Entomoplasma ellychniae, which were both isolated previously from E. corrusca. As mentioned in the results, the tight clustering of M. corruscae and Entomoplasma ellychniae in Fig. 5 implies that both belong to the same genus. Gasparich et al. [48] proposed reassigning all Mesoplasma species to the genus Entomoplasma, including changing Mesoplasma corruscae to Entomoplasma corruscae. Members of the currently defined genus Mesoplasma frequently occur on plants without causing disease, and therefore it could be horizontally acquired by E. corrusca when they feed on milkweed nectar or nutrient-poor tree sap to aid in survival during cold winter months [10,11]. All the fireflies in this study are omnivorous or carnivorous predators as larvae. Other than adult female Ph. versicolor species, adult fireflies do not need to eat to survive, although some fireflies might acquire bacteria while feeding as adults. Pyractomena borealis larvae are known to ingest tree sap, just as E. corrusca larvae do [4], and were colonized by Entomoplasma somnilux. Although Entomoplasma somnilux has only been found in fireflies, it is possible that it was acquired via the Pyractomena borealis diet. Mollicutes and other bacteria may be acquired by a larval firefly's diet of tree sap and remain colonized through the adult stage. However, it is unknown whether bacteria survive firefly metamorphosis from larvae to adults. If not, the presence or absence of Mollicute in these adult fireflies might be explained by the presence of their diet. Future work will be required to determine the function, if any, of these Mollicutes in their firefly hosts.
Our results provide the first description of the microbiomes found in E. corrusca, Photuris versicolor, Pyropyga decipiens, and Pyractomena borealis. These microbiomes have low alpha diversities and are species-specific. Some firefly species also have distinct egg, gut, and body microbiomes. Future research will determine the function of these microbiomes and how they are acquired from their environment or transmitted between hosts.