Bacterial Communities Are Less Diverse in a Strepsipteran Endoparasitoid than in Its Fruit Fly Hosts and Dominated by Wolbachia

Abstract     Microbiomes play vital roles in insect fitness and health and can be influenced by interactions between insects and their parasites. Many studies investigate the microbiome of free-living insects, whereas microbiomes of endoparasitoids and their interactions with parasitised insects are less explored. Due to their development in the constrained environment within a host, endoparasitoids are expected to have less diverse yet distinct microbiomes. We used high-throughput 16S rRNA gene amplicon sequencing to characterise the bacterial communities of Dipterophagus daci (Strepsiptera) and seven of its tephritid fruit fly host species. Bacterial communities of D. daci were less diverse and contained fewer taxa relative to the bacterial communities of the tephritid hosts. The strepsipteran’s microbiome was dominated by Pseudomonadota (formerly Proteobacteria) (> 96%), mainly attributed to the presence of Wolbachia, with few other bacterial community members, indicative of an overall less diverse microbiome in D. daci. In contrast, a dominance of Wolbachia was not found in flies parasitised by early stages of D. daci nor unparasitised flies. Yet, early stages of D. daci parasitisation resulted in structural changes in the bacterial communities of parasitised flies. Furthermore, parasitisation with early stages of D. daci with Wolbachia was associated with a change in the relative abundance of some bacterial taxa relative to parasitisation with early stages of D. daci lacking Wolbachia. Our study is a first comprehensive characterisation of bacterial communities in a Strepsiptera species together with the more diverse bacterial communities of its hosts and reveals effects of concealed stages of parasitisation on host bacterial communities. Supplementary Information The online version contains supplementary material available at 10.1007/s00248-023-02218-6.


Introduction
Insects have associations with diverse microbial communities that are important in host biology, host fitness and immunity and can provide protection against pathogens, parasitoids and toxins [1][2][3]. Symbiotic microbes can reside within the digestive tract, in particular the gut lumen [4,5], on the surface of the insect host (ectosymbionts) and within host cells and tissues (endosymbionts) [6,7]. The most common endosymbionts are maternally inherited Wolbachia (Alphaproteobacteria) that occur in over 50% of insects and other arthropod species [8,9]. In many hosts, they manipulate host reproduction to enhance the production of infected females and thereby their prevalence in populations [9,10]. For example, the induction of cytoplasmic incompatibility (CI) results in embryonic mortality when infected males mate with uninfected females or females infected with an incompatible Wolbachia strain; however, this CI is overcome in embryos of females infected with the same or a compatible Wolbachia strain, which can rescue the CI effect [9,10]. Besides reproductive manipulation, Wolbachia strains may be beneficial to their hosts. Some strains confer protection to their hosts against parasites, viruses and other pathogens [11][12][13]. Other Wolbachia strains can synthesise vitamins deficient in host diets; for instance Wolbachia provides B vitamins to the bedbug, Cimex lectularius [14,15]. Wolbachia can also influence the microbiome of hosts [16][17][18]. For instance, Wolbachia alters the relative abundance of bacterial taxa within microbial communities in the parasitoid wasp Nasonia vitripennis [18], the cabbage root fly Delia radicum [19] and adult mosquitoes [16]. Conversely, other symbiotic bacteria can influence Wolbachia prevalence and abundance, for example, Asaia can impede the establishment and stable transmission of Wolbachia in mosquitoes [20].
Many studies of host-microbe interactions have been performed on free-living insects, but less so on parasites [21] such as endoparasitoids that develop entirely within freeliving insects. So far, the microbiomes of Strepsiptera, an endoparasitic insect order, have not yet been investigated [22], likely due to its extraordinary characteristic of extreme endoparasitism. Strepsiptera consists of 630 known species that parasitise hosts belonging to the seven insect orders Blattodea, Diptera, Hemiptera, Hymenoptera, Mantodea, Orthoptera and Zygentoma [23]. All strepsipteran species are obligate endoparasitoids and almost entirely complete their life cycle within their hosts [24,25]. Adult strepsipterans display stark sexual dimorphism, with adult males that have external morphological features of a free-living adult insect, while the females are neotenic (lacking adult features) and fully endoparasitic within their hosts, except for adult females of one strepsipteran family, the Mengenillidae (suborder Mengenillidia) which are free-living [25,26]. Parasitisation of the host occurs via the first instar larvae (planidia), which leave a parasitised host to then enter a new host where they undergo hypermetamorphosis to the fourth larval instar [26,27]. In Mengenillidae, the fourth larval instars of both sexes leave and pupate on the outside of their hosts, while in all other families (all contained within the suborder Stylopidia), both male pupae and neotenic females extrude through the hosts' cuticle [26,27]. Males then emerge from the pupae within the host while females remain fully endoparasitic. Therefore, parasitisation by strepsipterans becomes visible as stylopisation in the later stages of strepsipteran development, while early stages of Strepsiptera may remain concealed in hosts and may only be detected by dissection or molecular tools [28]. Furthermore, strepsipterans are koinobionts which means that the hosts continue to live and feed while these endoparasitoids develop within their insect host [26].
Host-parasite interactions can be shaped by microbes associated with either the host or the parasite [21,29]. Microbes can protect their hosts against parasites, for instance, a bacterial symbiont of aphids, Hamiltonella defensa, can protect its hosts against the parasitoid wasp Lysiphlebus fabarum [30,31]. In contrast, the host microbiome can also aid in the establishment of parasites in their insect hosts, as seen in the interaction between the tapeworm Hymenolepis diminuta and its intermediate host, the grain beetle Tenebrio molitor [32]. Strepsiptera have an intimate relationship with their hosts and depend entirely on their hosts for nourishment [33]. Therefore, endoparasitoids may compete with the hosts' microbiota for resources. Furthermore, host-associated microbes can also influence host immunity. For instance, altering the bacterial communities of the fruit fly Drosophila melanogaster by antibiotic treatment influenced its resistance to the parasitoid wasp Asobara tabida by moderating the encapsulation rate of the parasitoid eggs by the host [34]. Conversely, parasitisation can affect the host microbiome, such as seen in larvae of the two moth species Diatraea saccharalis and Spodoptera frugiperda parasitised by the parasitoid wasp Cotesia flavipes, which changed the bacterial community composition and structure of the moth larvae [35].
While some host species may have few or no microbial associations [36], parasites and parasitoids may have less diverse microbiomes than their hosts due to the relatively small size, their life cycle and exclusive dependence on their hosts for resources. For example, a parasitic plant, the obligate parasite Orobanche hederae, exhibits a reduced microbiome compared to its host plant Hedera [37]. Similarly, the bacterial alpha diversity is lower within the intestinal tapeworm Eubothrium than in its host, the Atlantic salmon [38], and the microbiomes associated with entomopathogenic nematodes used in the biological control of insect pests are of relatively low complexity [39]. Yet, an amplicon sequencing study of one of the smallest insects, the parasitoid wasp Megaphragma amalphitanum, has revealed that it still carries a diverse variety of bacteria, albeit different in composition from other larger parasitoid wasp species [40].
Strepsipteran neotenic females reproduce viviparously and obtain nutrients exclusively from the host hemolymph [33]. Strepsipteran larvae have a gut, and nutrient uptake from the host hemolymph occurs in the midgut; however, after extrusion of the females, the strepsipteran gut is degenerate and filled with hemolymph [41], and nutrient uptake from the host hemolymph occurs via a particular structure, the apron [42].Therefore, due to the complete dependence of Strepsiptera on their hosts, it could be predicted that they have a less diverse microbiome. Such microbiome simplicity in parasites may be a parallel feature to the reduced morphological and genomic characteristics observed in several parasitic, ectosymbiotic and endosymbiotic organisms [26,[43][44][45] including the reduced genomic characteristics exhibited by bacterial endosymbionts [46,47].
Our study focused on Dipterophagus daci, a strepsipteran endoparasitoid of tephritid fruit flies [48] belonging to the family of Halictophagidae [28]. To date, D. daci is the only described strepsipteran endoparasitoid of Diptera (besides other undescribed strepsipteran endoparasitoids of platystomatid flies from Papua New Guinea) and has been reported from 22 species of the tephritid subfamily of Dacini in Australia and the Solomon Islands [28,48,49]. A recent study revealed that the presence of two Wolbachia strains previously detected in flies of seven Australian tephritid species [50,51] was due to concealed parasitisation of these flies with early developmental stages of D. daci [28]. This recent study also concluded that D. daci is the actual host of the two Wolbachia strains wDdac1 and wDdac2, which occur at high prevalence in this host, and D. daci without Wolbachia is found in only about 10% of parasitised flies [28]. The two strains wDdac1 and wDdac2 belong to the Wolbachia supergroup A and have previously been characterised using the Wolbachia surface protein (wsp) gene and five multi locus sequence typing (MLST) loci [50].
Tephritid fruit fly species are diverse and can infest diverse host plants but also different plant parts [52,53]. Furthermore, tephritids have bacterial communities that can vary in diversity and structure depending on life stage, fly species and phylogeny, host plant species, diet and rearing environment [54][55][56]. For example, the bacterial communities of the island fly, Dirioxa pornia, differs from those of Bactrocera species, and this could be due to their different life histories [54]. Furthermore, bacterial communities with diverse compositions were observed among different Bactrocera species suggesting that several factors such as host plant specialisation and domestication play a role in shaping the microbiome of tephritid fruit flies [54].
Our study aimed to explore the diversity and composition of bacterial communities of the strepsipteran D. daci. We hypothesised that due to its endoparasitic life cycle, the bacterial communities of D. daci consist of only few taxa and are distinct from the bacterial communities of its fruit fly hosts. Furthermore, we expected that Wolbachia would dominate the microbiome of D. daci but not of the parasitised fruit fly species. We also tested whether early stages of D. daci parasitisation influenced the fruit fly microbiome, and whether this was influenced by the presence of Wolbachia in D. daci. To address these questions, we performed highthroughput next generation amplicon sequencing analyses of the commonly used and conserved bacterial marker gene, the 16S rRNA gene, of (i) D. daci male pupae, (ii) fruit flies parasitised by early stages of Wolbachia-positive D. daci, (iii) fruit flies parasitised by early stages of Wolbachianegative D. daci and (iv) unparasitised fruit flies.

Fruit Fly Collection and DNA Extraction
We sequenced and analysed the bacterial 16S rRNA gene diversity of total genomic DNA extracts of 84 adult male fruit flies and 17 D. daci male pupae (Table S1) [28,50,51]. The 84 adult male fruit flies comprised individuals of seven species including Bactrocera bryoniae (n = 4), Bactrocera decurtans (n = 2), Bactrocera frauenfeldi (n = 11), Bactrocera neohumeralis (n = 22), Bactrocera tryoni (n = 32), Dacus axanus (n = 2) and Zeugodacus strigifinis (n = 11), collected from Queensland in 1998, 2001, 2012, 2013 and 2019 using male lure traps with malathion as part of fruit fly monitoring programs (Table S1) [57]. After emptying of traps, the trapped fruit fly specimens were kept dry and at room temperature for identification and then stored in ethanol at − 20 °C until DNA extraction. The D. daci male pupae were dissected from visibly parasitised (stylopised) male fruit flies of six species (Bactrocera breviaculeus, B. frauenfeldi, Bactrocera mayi, Bactrocera pallida, B. neohumeralis and B. tryoni; Table S1) collected from Queensland in 2019 [28]. The pupae were removed from the cephalotheca extruding from the abdomen of the parasitised fruit flies ( Fig. 1a and b). In contrast to these stylopised individuals, the 84 adult male flies were either parasitised by concealed stages of D. daci (D. daci-positive by PCR) or unparasitised (D. daci-negative by PCR) (Fig. 1c) [28]. Prior to DNA extraction, the male fruit fly specimens and D. daci male pupae were surface-treated with 4% sodium hypochlorite to remove any external microorganisms and then washed with 0.2% Triton-X and rinsed thoroughly using Milli-Q water [51].
Total genomic DNA was extracted from individual fruit fly male abdomens and individual whole D. daci pupae using GenElute DNA Miniprep Kit (Sigma-Aldrich) as per the manufacturer's instructions. The DNA quality was determined using NanoDrop and gel electrophoresis, and the extract was then stored at − 20 °C for subsequent experiments. The fruit fly and D. daci DNA extracts were screened by PCR using specific primers for the wsp and 16S rRNA genes [50,51] and the D. daci cytochrome c oxidase I (cox1) gene [28]. Based on the PCR results, the samples were categorised into four groups: (i) D. daci male pupae (Dd) which were all positive for Wolbachia (n = 17), (ii) fruit flies parasitised by early stages of Wolbachia-positive D. daci (Flies-DdW) (n = 30), (iii) fruit flies parasitised by early stages of D. daci without detectable Wolbachia (FliesDd) (n = 19) and (iv) unparasitised fruit flies (Flies) (n = 35) (Table S1). It is noted that Wolbachia-negative D. daci are relatively rare [28], and, therefore, flies parasitised by early stages of D. daci without detectable Wolbachia were preferentially included in our amplicon sequencing study to obtain a fair representation when compared to flies parasitised by early stages of Wolbachia-positive D. daci.

Bacterial 16S rRNA Gene Amplification and Sequencing
The DNA extracts were submitted for 16S rRNA gene amplicon sequencing on an Illumina MiSeq platform at the Western Sydney University Next Generation Sequencing Facility. Primers 341F (5′ CCT ACG GGNGGC WGC AG) and 805R (5′ GAC TAC HVGGG TAT CTA ATC C) were used to amplify the V3-V4 region of the 16S rRNA gene with a total read length of 2 × 301 bp. A bacterial mock community (Microbial Community DNA Standard, ZymoBiomics) provided by the sequencing facility was included.

Sequence Analysis
After sequencing, the sequence reads were pre-processed, quality filtered and analysed using Quantitative Insight into Microbial Ecology (QIIME 2, v. 2019.7). Raw demultiplexed Illumina fastq sequence (Phred33 applied for quality control) and mapping files were imported into QIIME 2 for downstream processes. The manifest file was created by concatenating the forward and the reverse sequences. The DADA2 pipeline was used for denoising, quality filtering, dereplication and chimera removal [58]. Quality analysis was performed by trimming the primers and truncating the reads using the commands -p-trim-left-f 17, -p-trim-left-r 21, -p-trunc-len-f 290 and -p-trunc-len-r 210. A naive Bayes classifier was trained using the Greengenes 99% sequence similarity threshold for calling operational taxonomic units (OTUs) at the V3-V4 region of the 16S rRNA gene. Amplicon sequence variants (ASVs) from DADA2 were used for taxonomic classification at a 99% similarity threshold using QIIME 2 q2-feature-classifier plugin [59] and sample taxonomic composition, and structure was visualised using QIIME 2 bar plot and plotted in R version 3.6.3 (R core Team, 2020, https:// www.R-proje ct. org/). The core-metrics-phylogenetic pipeline was used to construct the phylogenetic tree. A rarefaction curve was used to assess adequate sampling of the microbial communities. Based on the rarefaction curve, the overall alpha and beta diversity analyses were performed at a sampling depth of 6120, and at 1000 upon filtering out the Wolbachia reads, to avoid biases using the q2-diversity plugin. We estimated the alpha diversity among the four groups of samples using Shannon diversity index and Pielou's evenness. Beta diversity was assessed using weighted UniFrac distance (phylogenetic relationships and relative abundance) and Bray-Curtis distance (relative abundance) to determine the microbial community variation in the four sample groups (Dd, FliesDdW, FliesDd and Flies) with pairwise comparisons (PERMANOVA) using qiime diversity beta-group-significance in QIIME 2 (v. 2019.7). Beta diversity results were also visualised using principal coordinates analysis (PCoA) plots in R. To confirm that the Wolbachia ASV of our study corresponded to the Wolbachia previously characterised from D. daci, we compared it in a multiple sequence alignment using CLUSTALW together with 16S rRNA gene sequences of wDdac1 and wDdac2 extracted from genome reads obtained from the Wolbachia-positive sample B. frauenfeldi 485 as part of a whole genome sequencing project [28,43] and with a cloned Wolbachia 16S rRNA gene (GenBank accession KC775794) sequence obtained from B. neohumeralis [51].

Differential Relative Abundance Analysis
To determine whether early stages of D. daci parasitisation had an impact on the microbiome of the host fruit fly, we compared the relative abundance of bacterial taxa in the fruit flies parasitised by early stages of D. daci without detectable Wolbachia (FliesDd) to the unparasitised fruit flies (Flies) ( Table S1). Similarly, we assessed whether parasitisation by early stages of Wolbachia-positive D. daci had an impact on the host fruit f ly microbiome by comparing the FliesDd samples to the FliesDdW samples (Table S1). For these comparisons, we used the original taxonomic assignments of ASVs (at 99% identity) with the Wolbachia reads excluded. OTU datasets generated in QIIME and summarised at genus level were imported into Phyloseq for downstream analysis. The differential relative abundance was then performed in edgeR [60].

Sequence Read Analysis
The 101 sequenced 16S rRNA gene amplicon libraries (Table S1)  After quality control and filtering, we obtained a total of 2,274,402 sequence reads, with a mean sequence read number of 22,519 per sample (between 42 and 120,845 sequence reads per sample). After normalising the sequence read number at a sampling depth of 6120 to minimise biases, we excluded one fruit fly specimen that contained fewer than 6120 sequences (one D. axanus Flies sample with 42 sequence reads) from the subsequent analysis (Table S1). A total of 1808 ASVs were identified in this study (Table S2).

Bacterial Community of D. daci
The bacterial community of D. daci pupae was dominated by the class Alphaproteobacteria, accounting for 79.2% of the total sequence reads (Fig. 2a, Table S3). Other classes include Gammaproteobacteria (16.2%), Bacilli (2.1%), Deltaproteobacteria (0.6%), Flavobacteria (0.66%), Bacteroidia and various other classes with a combined relative abundance of < 1% (Fig. 2a, Table S3). The ASV with the highest relative abundance was one Wolbachia 16S rRNA gene sequence accounting for 78.7% of all sequence reads and was present in all 17 D. daci pupae (Fig. 2b, Table S4). The 16S rRNA gene sequences from wDdac1 and wDdac2 obtained from a previous whole genome sequencing project did not vary in the V3-V4 region (402 bp) and were identical to the dominant Wolbachia ASV obtained in this study (Fig. S1). An additional 15 Wolbachia ASVs, all singletons, consisted of sequences with up to 2 mismatches to the dominant ASV. The Wolbachia 16S rRNA gene sequence previously obtained in a molecular cloning experiment from the Wolbachia-positive B. neohumeralis was also identical to the wDdac1 and wDdac2 16S rRNA gene sequences albeit in another region (349 bp) (Fig. S1) further confirming that the two Wolbachia strains cannot be differentiated in the V3-V4 region. Other genera that were relatively abundant included Serratia (5.6%), Trabulsiella (2.4%), Enterobacter (1.6%), one unknown Pasteurellales ASV (2.4%), one unknown Enterobacteriaceae ASV (1.4%) and Lactococcus (1.09%) (Fig. 2b, Table S4).
Alpha diversity analysis revealed low Shannon diversity and Pielou's evenness indices in D. daci, whereas both indices were higher for flies (Kruskal-Wallis, p < 0.05, Fig. 3a Table S6 and S7). PERMANOVA analyses based on Bray-Curtis results also revealed the distinct clustering of D. daci bacterial communities from those of host fruit flies ( Table 1).

Comparison of Bacterial Communities Among Fruit Fly Species
The weighted UniFrac analysis showed no distinct clustering pattern in the fruit fly species (Fig. 4a, Table S8).
The most striking difference was the low relative abundance of the one unknown Pasturellales ASV in Z. strigifinis (1%) compared to the other fruit fly species which contained this bacterium at relative abundances ranging from 13 to 35% (Fig. 4c, Table S11). Interestingly, Acinetobacter bacteria were relatively more abundant in Z. strigifinis (21%) compared to B. bryoniae (0.2%), B. frauenfeldi (8.9%), B. neohumeralis (0.01%) and B. tryoni (1.4%) (Fig. 4c, Table S11). PERMANOVA pairwise analyses based on both weighted UniFrac and Bray-Curtis results showed significant differences (p < 0.05, PERMANOVA) in bacterial communities of B. frauenfeldi compared to the other fruit fly species (Table 2). This could be attributed to the one unknown Enterobacteriaceae ASV that had a relative abundance of 15.1% in B. frauenfeldi,  while the other fruit fly species had relative abundances of this unknown Enterobacteriaceae ASV ranging from 2.05 to 9.7% (Fig. 4c, Table S11). No significant differences were observed in comparisons among the bacterial communities of B. bryoniae, B. neohumeralis and B. tryoni ( Table 2).

Influence of Early Stages of D. daci Parasitisation on Bacterial Communities in Fruit Fly Hosts
A comparison of the fruit flies parasitised by early stages of D. daci without detectable Wolbachia (FliesDd) and unparasitised flies (Flies) was performed to determine the impact of early D. daci parasitisation on the fruit fly bacterial diversity. The OTU datasets used were retrieved from QIIME and summarised to genus level. This comparison revealed an impact of early parasitisation by D. daci on the relative abundance of Pseudomonadota (formerly Proteobacteria) and Bacillota (formerly Firmicutes) in fruit flies (Fig. 5a, Table S12). The relative abundance of nine bacterial genera including Proteus, one unknown Enterobacteriaceae ASV, Klebsiella, one unknown Acetobacteriacea ASV,  Fig. 5 Scatter plot of the bacterial taxa with differential relative abundance in A fruit flies parasitised by early stages of Dipterophagus daci without detectable Wolbachia (FliesDd) compared to unparasitised fruit (Flies) and in B fruit flies parasitised by early stages of Wolbachia-positive Dipterophagus daci (FliesDdW) compared to fruit flies parasitised by early stages of D. daci without detectable Wolbachia (FliesDd). A log fold change of logFC > 0 indicates that the abundance of the genera increased, whereas logFC < 0 indicates that the abundance of the genera decreased. The taxa with significantly different relative abundances are coloured by phylum Ochrobactrum, Morganella, Providencia, three unknown Pasteurellales ASVs and Enterococcus were increased in FliesDd, while three bacterial genera (Enterobacter, Citrobacter and one unknown Halomonadacea ASV) decreased in FliesDd (Fig. 5a, Table S12).
Similarly, we compared the relative abundance of bacterial taxa between flies parasitised by D. daci with and without detectable Wolbachia. Sequences used in this analysis were corrected for Wolbachia and normalised to a sequencing depth of 1000, based on the minimum number of reads after excluding Wolbachia. We found that the relative abundance of 11 genera comprising Proteus, Providencia, Dysgonomonas, Morganella, one unknown Acetobacteriacea ASV, two unknown Pasteurellales ASVs, Vagococcus, Serratia, one unknown Enterobacteriaceae ASV, Staphylococcus and Enterobacter were decreased in FliesDd, while the relative abundances of Klebsiella, Trabulsiella, Myroides and Citrobacter were increased (Fig. 5b, Table S13).

Discussion
We used 16S rRNA gene amplicon sequencing to characterise the bacterial communities involved in the interactions between tephritid fruit flies and the strepsipteran endoparasitoid D. daci. With this we have performed, according to our knowledge, the first comprehensive characterisation of bacterial communities in a species of the endoparasitic insect order Strepsiptera [23]. The bacterial communities of D. daci were dominated by Wolbachia; however, this dominance was not observed in fruit flies parasitised by early stages of Wolbachia-positive D. daci (and Wolbachia was completely absent in flies not parasitised by D. daci), supporting previous findings that D. daci is the host of Wolbachia in this host-parasitoid interaction [28]. We found that the bacterial communities of D. daci are not as diverse but distinct when compared to the more diverse bacterial communities of the fruit fly hosts. Furthermore, early stages of D. daci parasitisation and presence of Wolbachia in D. daci altered the microbiome of parasitised fruit flies. We also found that the bacterial communities of Z. strigifinis were distinct from the bacterial communities of the Bactrocera species and this may be linked to their different ecologies, with Z. strigifinis developing in cucurbit flowers, whereas the analysed Bactrocera species develop in fruit [53,61].

Dipterophagus daci Has a Less Diverse Microbiome
The most abundant bacterial phylum in D. daci was Pseudomonadota comprising 96.2% of the total bacterial sequence reads, followed by Bacillota and Bacteroidota at a substantially lower relative abundance. A high relative abundance of Pseudomonadota and Bacillota has previously been detected in fruit flies [54,62] and other insect species [55,63]; however, the relative abundance of Pseudomonadota in D. daci pupae found in our study was generally higher and mostly just consisted of Wolbachia. This indicates that bacterial communities in D. daci are not very diverse, which is perhaps due to its parasitic life cycle. The strepsipteran D. daci and all other Strepsiptera are almost fully endoparasitic in their host and depend exclusively on the host for nourishment [24,26,33]. The high presentation of Pseudomonadota in D. daci was due to Wolbachia, a member of the Alphaproteobacteria, in combination with Gammaproteobacteria, Bacilli, Deltaproteobacteria, Bacteroidia and Flavobacteriia at substantially lower relative abundance. It needs to be noted that we were only able to characterise the bacterial communities of D. daci in isolation from its host by carefully dissecting pupae out of the cephalotheca extruding from the abdomen of parasitised fruit flies, followed by surface treatment to minimise contamination. We do not know how bacterial communities in D. daci change throughout its development. Given the endoparasitic life cycle of D. daci, it is likely that exposure to environmental bacteria is limited, which could impact the observed low levels of bacterial diversity in D. daci. Therefore, bacterial symbionts detected in D. daci pupae are either maternally inherited or horizontally acquired from the host fly or from the environment during the short period that planidial larvae search for new hosts. It is perhaps less likely that bacteria acquired by adult D. daci males are then paternally transmitted. We did not obtain free living males of D. daci as they would require different sampling techniques such as light trapping or collection of adult males emerging from parasitised flies and are, therefore, more difficult to collect than parasitised flies.

The Microbiome of D. daci is Dominated by Wolbachia and Distinct from the Fruit Fly Hosts' Bacterial Communities
Wolbachia is a common maternally inherited endosymbiont of insects and other arthropods that can manipulate host reproduction to increase its prevalence in host populations [8,64,65]. In several host species, Wolbachia provides fitness benefits which can also maintain this endosymbiont in host populations [66]. For several insect species, it has been found that, when present, Wolbachia can dominate bacterial communities within hosts [67,68]. Our findings of the dominance of Wolbachia in bacterial communities within D. daci (but not in the bacterial communities within fruit flies) further confirms that the two Wolbachia strains first detected in fruit flies [50,51] are actually associated with D. daci and had previously been detected in these fruit flies because of parasitisation by concealed early stages of Wolbachia-positive D. daci [28]. Additionally, alpha diversity analysis revealed low Shannon diversity and Pielou's evenness values in D. daci bacterial communities (a consequence of the Wolbachia dominance) while this was not observed in fruit flies parasitised by early stages of Wolbachia-positive D. daci.
Previous analyses have found that Wolbachia occurs at high prevalence in D. daci [28], and D. daci is depauperate in mitogenome diversity across large parts of its geographic distribution [28,43]. Because of maternal co-inheritance with mitochondria, Wolbachia may have caused a selective sweep of mitochondria due to either reproductive manipulation or beneficial host fitness effects. Previous analyses of whole genome sequenced specimens have not detected the presence of Wolbachia genes involved in reproductive manipulations, and, therefore, it is likely that Wolbachia confers a fitness benefit to D. daci [28], for example, by providing a key nutrient and/or supporting immunity; however, this will need further investigation. Most strepsipteran life stages are fully endoparasitic except for the free-living first instar larvae (planidia) and adult males and are therefore fully dependent on the host for nourishment [42]. The host may not always provide all the essential nutrition, and therefore endoparasitoids may form beneficial interactions with maternally inherited endosymbionts like Wolbachia. In the bedbug, Cimex lectularius, a Wolbachia supergroup F strain, provides B vitamins which are deficient in the bedbug's diet [14]. Similarly, a Wolbachia supergroup A strain provides D. melanogaster with metabolic support in periods of nutritional stress [69], and Wolbachia supergroup B strains have been associated with synthesis of biotin and riboflavin to increase host fitness in the small brown planthopper Laodelphax striatellus and the brown planthopper Nilaparvata lugens [15]. Furthermore, Wolbachia supergroup A strains provide protection against pathogens such as RNA viruses in several insect species such as Drosophila [11,13,70] and mosquitoes [71,72]. Throughout its entire development, D. daci is exposed to the fruit flies' immune system and the host's viruses. It has recently been found that the tephritid host species of D. daci have a very high incidence and prevalence of insect-specific RNA viruses [73] with vertical and horizontal transmission modes [74]. Future research should investigate how these viruses interact with fruit fly hosts and D. daci.
Furthermore, the weighted UniFrac and Bray-Curtis beta diversity analyses revealed that the bacterial community of D. daci was distinct from the bacterial communities of its fruit fly host species. This may be due to the phylogenetic distance between the strepsipteran and its fruit fly hosts, or the differences in host life cycle and diet. It also indicates that D. daci and the fruit fly hosts do not share microbiome components.

Variable Bacterial Communities in Fruit Fly Host Species
Our analyses of the bacterial communities in fruit flies with the presence of several Enterobacteriaceae taxa (one unknown Enterobacteriaceae ASV, Acinetobacter, Providencia, Enterobacter, Klebsiella, Citrobacter and Serratia) confirmed their importance in tephritid fruit fly microbiomes as found in previous studies [54,56,75]. However, we also found an abundance of bacterial taxa such as Vagoccoccus, one unknown Pasteurellales ASV, Trabulsiella, one unknown Desulfovibrionaceae ASV and Dysgonomonas, which were different bacterial community members when compared to previous studies on tephritid fruit flies. This difference could be attributed to our sample collection and handling procedures (samples were collected in male lure traps with an insecticide and kept dry and at room temperature until identification). Additionally, for our study, we specifically selected individuals that were parasitised by early stages of D. daci and this could also have resulted in a sampling bias.
Tephritid fruit fly species exhibit diverse life histories and host plant preferences [52,53,61], and these can affect their microbiomes [53,54,76]. The Shannon diversity and Pielou's evenness showed significant difference in B. frauenfeldi and Z. strigifinis bacterial communities compared to the other fruit flies. Additionally, the Bray-Curtis PCoA revealed that bacterial communities associated with Z. strigifinis were distinct from those of Bactrocera species, possibly suggesting a fly genus effect, albeit we only included one Zeugodacus species in our study. Furthermore, there could be a host plant effect as Z. strigifinis is a pest of Cucurbitaceae flowers while B. tryoni, B. neohumeralis, B. frauenfeldi and B. bryoniae infest fruits [53,61]. The weighted UniFrac analysis, however, did not show any distinct clustering, suggesting that the variation between the bacterial communities of Z. strigifinis and the Bactrocera species may only be in the relative abundance of the bacterial taxa that may have similar function. This difference may be due to the unknown ASVs of Pasteurellales and Acinetobacter. In addition to the different bacterial communities observed in Z. strigifinis, PERMANOVA revealed that B. frauenfeldi were also different in bacterial community structure when compared to B. bryoniae, B. tryoni and B. neohumeralis. Bactrocera tryoni and B. neohumeralis are closely related sibling species [77]; hence, this may explain the similarity of their bacterial communities [54], while it is unclear why B. bryoniae grouped with these two species.

Dipterophagus daci Parasitisation Alters Structure of Bacterial Communities
We observed a significant decrease in the relative abundance of nine bacterial genera in fruit flies parasitised by 1 3 D. daci without Wolbachia, while three bacterial genera increased in their relative abundance, suggesting that D. daci parasitisation affects the relative abundance of bacterial taxa in host bacterial communities. Furthermore, despite the low relative abundance of Wolbachia in fruit flies parasitised by Wolbachia-positive D. daci, we found a decrease in the relative abundance of 11 bacterial genera in flies parasitised by Wolbachia-positive D. daci, while the relative abundance of four bacterial genera increased. This suggests that parasitisation by D. daci and presence of Wolbachia in D. daci affect bacterial communities in flies. This is in line with other research that has shown that microbes can influence host-parasite interactions [19,29,32,35,78]

Conclusions
According to our knowledge, our study is the first comprehensive characterisation of the bacterial communities of a strepsipteran using a next generation amplicon sequencing approach. We demonstrated that bacterial communities of D. daci are not very diverse, dominated by Wolbachia and distinct from those of its host fruit fly species, and this could be attributed to the differences in host life cycles, life histories and phylogeny. Further studies should investigate the role of the two Wolbachia strains in D. daci, in particular as it is clear from previous genome analyses that they lack the capacity to manipulate host reproduction yet have an overall high prevalence in D. daci [28]. Furthermore, we observed variability in the relative abundance of bacterial taxa across fruit fly species, irrespective of parasitisation by D. daci, suggesting that phylogeny, host plant preference and host plant use play a role in shaping bacterial communities in fruit flies [54]. In addition, early stages of D. daci parasitisation affected the relative abundance of bacteria in microbial communities of host fruit flies. Hence, parasitisation can also shape the microbiome of insects and should therefore be considered in host-microbiome studies.
Acknowledgements The authors would like to thank Caroline Janitz from the Western Sydney University Next-Generation Sequencing Facility for the Miseq sequencing. We also thank Melissa Starkie and Caterina Torrisi from the Queensland Department of Agriculture and Fisheries for their help with fruit fly collection and identification and two reviewers for their constructive feedback on our manuscript.
Author Contribution STK, MR and JLM conceived the study. The research was designed by STK, MR and JLM; STK performed the experiments and data analysis with input from MR and JLM; JR and SC provided the insect specimens. STK and MR wrote the manuscript with input from all authors.
Funding Open Access funding enabled and organized by CAUL and its Member Institutions This research was supported by the Australian Research Council (ARC) Industrial Transformation Training Centre (ITTC) Fruit Fly Biosecurity Innovation (IC150100026). STK was supported by an ARC ITTC scholarship and the 2018 E.A. Southee Award of the Hawkesbury Foundation.

Data Availability
The raw bacterial 16S rRNA gene sequence reads have been deposited in GenBank Sequence Read Archive (GenBank accession SAMN26586984-SAMN26587084).

Competing Interests
The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.