High-diversity microbiomes in the guts of bryophagous beetles (Coleoptera: Byrrhidae)

The diversity and role of the gut microbiota of insects is a rapidly growing fi eld of entomology, primarily fueled by new metagenomic techniques. Whereas endosymbionts in the guts of xylophagous or herbivorous insects are well studied, the microbiomes in moss-eating (bryophagous) insects remain uncharacterized. Using the Illumina MiSeq platform, we determined the composition of microbiomes in the gut, abdomen and on the body surface of two bryophagous species: Simplocaria semistriata (Fabricius, 1794) and Curimopsis paleata (Erichson, 1846) (Coleoptera: Byrrhidae). Gut microbiomes differed substantially from abdominal microbiomes in the same individuals, which indicates the need to separate them during dissection. Microbiomes in the gut and abdomen differed markedly from surface microbial assemblages. Gut microbiomes in bryophages had the highest MOTU richness, diversity and relative rarity. The eudominant bacteria in the guts and abdomens of bryophages were Novosphingobium, Bradyrhizobium, Ralstonia and Caulobacter, which are responsible for the detoxifi cation of secondary metabolites or nitrogen fi xation. These are less common in the surface samples and, therefore, likely to be associated with the specifi c ability of bryophages to feed on mosses. * Corresponding author; e-mail: pavel.drozd@osu.cz INTRODUCTION Many insects establish symbiotic interactions with microorganisms in their gut, body cavities, or cells (Dillon & Dillon, 2004). Protista have been identifi ed in the digestive tract of lower termites and wood roaches (Hongoh, 2010); whereas fungi and methanogenic archaea are frequent in the gut of xylophagous and detritivorous beetles and termites (Egert et al., 2005; Brune, 2010). Bacteria are found in the gut community of most insects, including herbivores. Most gut microbes are commensals or parasites; however, some are known to provide benefi cial services to their hosts (Engel & Moran, 2013). They can affect resistance against pathogens or parasites (Hedges et al., 2008; Oliver et al., 2010), intestinal cell renewal and systemic growth (Buchon et al., 2009), production of pheromones (Dillon et Eur. J. Entomol. 116: 432–441, 2019 doi: 10.14411/eje.2019.044


INTRODUCTION
Many insects establish symbiotic interactions with microorganisms in their gut, body cavities, or cells (Dillon & Dillon, 2004). Protista have been identifi ed in the digestive tract of lower termites and wood roaches (Hongoh, 2010); whereas fungi and methanogenic archaea are frequent in the gut of xylophagous and detritivorous beetles and termites (Egert et al., 2005;Brune, 2010). Bacteria are found in the gut community of most insects, including herbivores. Most gut microbes are commensals or parasites; however, some are known to provide benefi cial services to their hosts (Engel & Moran, 2013). They can affect resistance against pathogens or parasites (Hedges et al., 2008;Oliver et al., 2010), intestinal cell renewal and systemic growth (Buchon et al., 2009), production of pheromones (Dillon et ally on the surface and inside cushions of the moss Dicranella heteromalla. Individuals were placed in separate plastic boxes together with moss using sterilized tweezers to avoid contamination. The boxes were stored in a refrigerator until dissection (for a maximum of three days after collection to avoid considerable changes in the composition of bacterial assemblages). In order to obtain surface samples, each beetle was washed by vortexing in a 1.5-mL micro centrifuge tube with 1 mL sterile solution of 1% Tween 80 (Sigma-Aldrich, Saint Louis, USA) and phosphatebuffered saline (PBS) for 30 s at 2100 rpm. The procedure was repeated so that the beetle was completely clean and the second wash was discarded (Bateman et al., 2016). Gut contents and abdominal tissues were separated on paraffi n wax, which was previously sterilized by pouring ethanol on it and igniting it. For further metagenomic analysis, surface washes, gut contents and abdominal tissues of fi ve C. paleata and fi ve S. semistriata individuals were obtained. In addition, we verifi ed the bryophagy of both species by analyzing the gut contents of fi ve individuals of C. paleata and fi ve of S. semistriata that were dissected under a binocular microscope. This revealed that fragments of phylloids and rhizoids formed > 95% of the gut contents and the remaining material probably consisted of soil particles.

DNA isolation, PCR amplifi cation and library preparation
Microbial DNA from the surface wash, gut contents and abdominal tissues of each beetle was isolated using the PowerSoil® DNA Isolation Kit (MoBio, Carlsbad, CA) following the manufacturer's standard protocol and then subjected to sequencing library preparation. To ensure the recovery of broad spectra of bacterial and archaeal diversity, we used the universal F515/R806 bacterial ribosomal primers for the MiSeq platform to amplify the V4 region of 16S rDNA (Caporaso et al., 2010b). A two-step PCR was used during library preparation. First-step amplifi cations were done in quintuple reactions using only gene-specifi c primers to avoid PCR artifacts caused by long primers with attached sequencing adapters and identifi ers, as well as the stochasticity in the PCR amplifi cation (Berry et al., 2011;Schmidt et al., 2013). The fi rst-step PCR was performed according to Caporaso et al. (2010b) with minor modifi cations consisting of initial denaturation at 94°C for 3 min; 25 cycles at 94°C for 45 s, 50°C for 60 s, 72°C for 90 s; and a fi nal extension at 72°C for 10 min. The number of cycles was kept low to prevent potential depletion of specifi c primers from the degenerate mixture. Quintuple PCR reactions were pooled, purifi ed (UltraClean® PCR Clean-Up Kit; MoBio) and subjected to a second-step PCR. The latter included 15 cycles of PCR amplifi cation (following the same cycling profi le as in the fi rst-step PCR) with fused Illumina primers containing sequencing adapters and sample-unique multiplex identifi ers necessary for demultiplexing the reads from each sample. Each PCR reaction volume (25 μL) contained 14.25 μL of molecular biology-grade water, 2.5 μL ExTaq 10 × buffer, 0.2 mM dNTPs, 0.8 μM of each primer, 1.25 U ExTaq polymerase (all Clonetech, Mountain View, CA) and 20 ng extracted DNA or 2 μL cleaned PCR product in the second-step PCR. All PCR products were purifi ed and checked using an agarose gel and quantifi ed with the Quant-iT kit (Life Technologies, Carlsbad, CA). Subsequently, equimolar proportions of all samples were pooled to create a fi nal sequencing library at 7.5 ng/μL and submitted for sequencing on the MiSeq platform (Illumina Technologies, San Diego, CA) at the Interdisciplinary Center for Biotechnology Research of the University of Florida, USA. Raw demultiplexed sequencing data with sample annotations are available at the Short Read Archive database (http://www.ncbi.nlm.nih.gov/Traces/sra/) under accession number SRP055203 and further details can be found under the BioProject accession number PRJNA275854. bryophagous insects, even though the utilization of mosses as a source of food is a unique phenomenon due to the specifi c properties of mosses. Bryophytes contain several compounds that reinforce the cell wall and hinder its digestion by inhibiting potential symbiotic organisms ( Vanderpoorten & Goffi net, 2009). Moreover, mosses have evolved effective defenses against herbivores by producing antimycotics (Frahm, 2004) and antibiotics (McCleary et al., 1960), organically soluble fractions (Parker et al., 2007) and high levels of water-soluble phenolic compounds (Davidson et al., 1989;Glime, 2006). It is assumed that all these mechanisms strongly affect the gut microbiota of their occasional consumers, and, as a result, deter the majority of herbivores from feeding on mosses (Gerson, 1969).
The aim of the present study was to determine the molecular operational taxonomic units (MOTUs) of bacteria in the gut and/or abdominal microbiomes in two bryophagous species Curimopsis paleata (Erichson, 1846) and Simplocaria semistriata (Fabricius, 1794) (Coleoptera: Byrrhidae). Adults of S. semistriata are recorded feeding, mating and ovipositing on mats of the moss Dicranella heteromalla (Hedw.) Schimp. and also feeding on the moss Mnium hornum Hedw. (Johnson, 1990). In the genus Curimopsis, all species are thought to be strictly bryophagous based on collecting, rearing or dissections. The dissections revealed fragments of undetermined moss in the alimentary tract of C. paleata (Johnson, 1986). Interestingly, Curimopsis nordensis Tshernyshev, 2013, a new species that occurs in Russian alpine tundra and steppes, also feeds on carrion, so at least some adults of the genus Curimopsis can feed also on non-vegetable matter (Tshernyshev, 2013).
We aim to determine, whether the gut microbiomes in both species, which are in permanent contact with a moss diet, are less diverse and distinctly different in their composition from that of their surface microbiota, which are in permanent contact with bacteria-rich soil. We compare the bacterial assemblages associated with both species and subsequently in different parts of their bodies in terms of their composition, species richness and diversity, overlap and relative rarity. Finally, we aimed to determine the MOTUs signifi cantly associated with gut and/or abdominal microbiomes and discuss their potential role in moss digestion in both of the bryophagous species studied. In addition to the principal analysis, we compare the gut and abdominal microbiomes of the same bryophagous individuals to determine whether it is necessary to separate body parts before metagenomic analyses. This aspect is often neglected in studies on small insects because it is diffi cult, but its omission could confound the results.

Sampling and dissection of beetles
Beetles were collected from November to December 2013 in deciduous forests near Ostrava, Czech Republic (49°52´04˝N, 18°14´17˝E). We focused on two species of the family Byrrhidae, a unique group of moss-feeding beetles: C. paleata and S. semistriata. Both are widespread Palearctic species with minute bodies and perennial activity. The beetles were captured individu-

Processing of sequencing data and statistical analysis
Sequencing data were processed using QIIME 1.8.0 (Caporaso et al., 2010a), including quality checking, demultiplexing, read clustering and taxonomic assignments. Forward and reverse reads were joined to create contigs. Afterward, reads were demultiplexed in a parallel way with quality fi ltering that included a maximum unacceptable Phred quality score of 20 and a maximum number of consecutive poor quality base calls of 12 due to lower-quality overlaps of paired-end reads. Resulting reads were clustered into MOTUs using UCLUST (Edgar, 2010) with 97% similarity threshold against the bacterial 16S rRNA reference database Greengenes gg_13_8 release (DeSantis et al., 2006). Finally, we compiled information on read counts for all MOTU clusters from all samples together with taxonomic information into a MOTU table, which was used for comparing and describing the diversity of the samples. To enable comparison of beta diversity at the same sequencing depth rarefaction of resample datasets from all samples was carried out to the lowest observed read count. Resampling was done at a depth of 13,400 sequence reads to allow the inclusion of all samples. Singletons (clusters with only one read in individual samples) were discarded before producing the fi nal dataset.
We analyzed data in R 3.4.3 (R Development Core Team, 2017) and Canoco 5.03 (Ter Braak & Smilauer, 2012). We used the "Rarity" library (Leroy, 2016) to calculate rarity indices for assemblages of bacterial MOTUs in individual samples. First, we calculated rarity weights for each MOTU using the weighting function "W" (Leroy et al., 2012) and the improvements proposed by Leroy et al. (2013). Rarity cut-off points were counted using the Leroy method on a set of actual s pecies assemblages. The obtained rarity weights were used to calculate the index of relative rarity. Extrapolated MOTU richness (estimation of unobserved MOTUs) was based on abundances in subsamples using the abundance-based coverage estimator (ACE) in the "vegan" library (Oksanen et al., 2017). The diversity of assemblages was calculated using Fisher's alpha diversity (Fisher et al., 1943).
We analyzed differences in microbial composition in the gut and abdomen of bryophages using PERMANOVA with strata defi ned by individuals and expressed them using the Jaccard (J) and Renkonen (P) similarity indices (Renkonen, 1938). We used the generalized linear mixed model in the "lme4" library (Bates et al., 2014) with normal distribution of residuals and random effect of individuals to determine the relationship between the species of bryophage and origin of samples as explanatory variables and (a) Fisher's alpha diversity, (b) ACE (abundance-based coverage estimation) of species richness and (c) index of relative rarity as dependent variable. The generalized linear model with a negative binomial distribution ("MASS" library) and likelihood ratio test were used to compare the MOTU richness recorded in bryophages in our dataset with the MOTU richness of various herbivorous beetles  Yun et al. (2014). Rarefaction curves were plotted using the "vegan" library, line plots were created using the "sciplot" library (Morales, 2017) and Venn diagrams were constructed using the "gplots" library (Warnes et al., 2016). We analyzed differences in microbiota in relation to the species of beetles and their particular body part (gut-, abdomen-and surface-associated) using RDA and principal component analysis (PCoA) in Canoco 5.03 and PERMANOVA in R with strata defi ned by individuals. For RDA, the number of depicted MOTUs were reduced to 25 based on best fi t. We determined the particular bacterial MOTUs signifi cantly associated with the gut and/or abdomen of bryophagous beetles using the pairwise Wilcoxon rank sum test for multiple testing and false discovery rate method for correcting p-values using a paired test.

Comparison of the bacterial assemblages of the species of bryophagous beetle studied
We classifi ed the sequences into 402 MOTUs belonging to 22 phyla, 55 classes, 91 orders and 182 families of bacteria, and one MOTU belonging to Archaea. On average, we recorded 78.5 MOTUs per sample. Gammaproteobacteria were the most abundant class on the surfaces of both S. semistriata (68%) and C. paleata (49%), whereas Alphaproteobacteria were the most abundant class in the gut and abdomen of both, S. semistriata (65%, 39% respectively) and C. paleata (62%, 55% respectively). We recorded great differences in the composition of bacterial microbiomes based on species (PERMANOVA: df = 24; F = 1.68; P < 0.001; R 2 = 0.040) and greater differences for particular body parts (df = 24; F = 8.02; P < 0.001; R 2 = 0.385).
At the order level, the most abundant MOTUs on the surface of beetles were the Enterobacteriales for S. semistriata (34%) and Pseudomonadales for C. paleata (28%). The surface assemblages of bacteria on both species of beetles were similar with the most dominant MOTUs Pseudomonas (24%), Burkholderiaceae (16%) or Pedobacter (12.5%). In the abdomen, the most abundant order was Sphingomonadales in both of the species of beetles (26.5%, 21% respectively). The bacterial assemblages in the abdomens of these beetles were also similar in composition and dominated by the MOTUs Novosphingobium (24%), Bradyrhizobium (20%), Ralstonia (14%) or Caulobacter (12.5%). In contrast, the bacterial assemblages in the guts of the two species differed. That in the gut of S. semistriata was dominated by Enterobacteriales (34.5%) and Rickettsiales, namely Rickettsia (14.5%), while that in C. paleata was dominated by another Rickettsiales, Wolbachia (17%) and by Entomoplasmatales (23%). A detailed composition at the order level is provided in Fig. 1 and the genera most strongly associated with both species of beetles are depicted in Fig. 2. The bacterial assemblages associated with Curimopsis paleata and Simplocaria semistriata did not differ either in terms of Fisher's alpha diversity (df = 22, F = 0.36, P = 0.519), species richness (df = 22, F = 0.42, P = 0.481) or relative rarity (df = 22, F = 0.20, P = 0.623).

Diversity of bacterial assemblages in guts and abdomens of bryophages
As indicated by the fi rst two axes of the PCoA, accounting for 58% of the variability in the data, assemblages on the body surfaces differed strongly from those in their abdomens and guts, which occur in a separate and distant cluster. The microbiota in the gut and abdomen was quite similar in both species (Fig. 3) generally, but there was a substantial difference between the assemblages in the guts and abdomens at the level of individuals (df = 16; F = 5.63; P < 0.001), with J = 21.8% average MOTU overlap and mean P = 0.437 (Fig. 4). Fisher's alpha diversity index was signifi cantly dependent on the origin of the sample (df = 23; F = 10.98; P < 0.001). The highest overall (gamma) diversity was recorded in the gut microbiome (n = 241 MOTUs for S. semistriata, n = 215 MOTUs for C. paleata). As shown in Fig. 5a, the gut microbiome also had the highest average richness (96.22 MOTUs per sample) and the abdomen the lowest (63.89 MOTUs per sample). Fish-er's alpha diversity also refl ects this pattern (Fig. 5b). Rarefaction curves showed that the species richness recorded in gut samples was higher (i.e., in terms of the number of MOTUs) than recorded for other samples. Estimated species richness differed signifi cantly also based on the origin of the sample (df = 23; F = 6.82; P = 0.002). The species richness of the gut microbiota of beetles was higher than that recorded for the surface and abdomen microbiota (Fig.  6a). In addition, the species richness recorded in the gut of bryophagous beetles was signifi cantly higher than that recorded for herbivorous beetles (df = 16; LR = 12.96; P < 0.001; Fig. 6b).   The highest propor tion of origin-specifi c MOTUs was recorded in the gut (20.10%), followed by on the surface (13.76%) and the lowest in the abdomen (10.32%) (Fig.  7a). Consequently, index of relative rarity of assemblages differed signifi cantly (df = 23; F = 9.88; P < 0.001), with the highest index values (many MOTUs with low frequency in the dataset) recorded for gut microbial assemblages (Fig. 7b). The eudominant taxa, Bradyrhizobium, Caulobacter, Novosphingobium and Ralstonia, were not only signifi cantly more abundant in the abdomen than the gut, but also signifi cantly more abundant in the gut than on the surface (for all: U abd/gut = 36, P abd/gut = 0.012; U gut/sur = 36, P gut/sur = 0.012).

High richness and low similarity of gut bacterial assemblages
Comparable metagenomic studies report relatively few microbial species in most insect guts (Engel & Moran, 2013) and many previous identifi cations based on 16S rRNA gene sequences reveal fewer than 20-30 bacterial MOTUs per insect taxon (Dillon & Dillon, 2004;Robinson et al., 2010;Wong et al., 2011). We recorded great overall richness in bacterial MOTUs, particularly in the guts of bryophages. The overall high number of MOTUs detected may refl ect the greater accuracy of the metagenomic approach used. For studies comparing bacterial richness, culture-independent DNA metabarcoding is more appropriate than conventional cultivation methods with subsequent taxonomic and/or molecular identifi cations (Broderick et al., 2004;Vaz-Moreira et al., 2011). A comparison with the results reported for herbivorous beetles (Yun et al., 2014) revealed that in bryophagous beetles there were more MOTUs per individual. For this comparison, we used only those MOTUs recorded in the gut of bryophages. A comparable metagenomic study of herbivorous beetles by Kelley & Dobler (2011) indicate that the gut of Cryptocephalus spp. (Chrysomelidae) harbours only 15-30 bacterial MOTUs per species. In contrast, Montagna et al. (2014) report that Cryptocephalus sp. averaged 86.3 MOTUs per individual, which is consistent with our results. A higher MOTU richness is reported for some non-herbivorous  Both seasonality and the substrate from which the beetles were sampled may substantially infl uence richness and diversity of bacterial assemblages. We collected our specimens at the beginning of winter. Although the effect of timing remains to be determined, bacterial diversity in the phyllosphere increases in response to drought and heat (Peñuelas et al., 2012). Similarly, insect gut microbiota changes with the season (Behar et al., 2008). We collected samples from the moss D. heteromalla, which forms low cushions barely a few millimeters above the soil. Soil bacterial assemblages are very complex (Borneman et al., 1996) and determine the composition and structure of gut microbial assemblages in insects living in direct contact with the soil (Huang & Zhang, 2013). Consistently, high bacterial richness of up to 695 MOTUs, are reported in the detritivorous beetles Onthophagus sp. (Yun et al., 2014).
The bacterial microbiome in the guts of beetles has the greatest proportion of specifi c MOTUs and the highest index of relative rarity compared to that recorded in other body parts, indicating that many bacterial MOTUs in the guts of bryophagous beetles differed among individuals. In some insects, gut bacterial communities vary among individuals within a species and consist mainly of bacteria without specifi c adaptation(s) to life in the gut of their host species (Cariveau et al., 2014), which indicates that the diet of the host might have some infl uence (Broderick et al., 2004). Nevertheless, a relatively static community is also documented (Tang et al., 2012). Physicochemical conditions in gut compartments, such as pH, redox potential, or availability of particular substrates, may select for particular species. Thus, ev en when acquired independently during each generation, gut communities are not expected to be random assemblages of bacteria derived from the food or local environment (Engel & Moran, 2013). Instead, the high abundance and ubiquitous presence of soil bacteria, such as Variovorax, Pedobacter, Bacillus, Pseudomonas and Erwinia on the surface of the bodies of S. semistriata and C. paleata could be explained by the proximity to the soil of mosses such as D. heteromalla. Indeed, the bacteria that are predominant on the surface of the beetles are also widespread in the rhizosphere of plants (Mahaffee & Kloepper, 1997) and, in the case of Pseudomonas and Erwinia are also endophytic bacteria of many species of plants (Cankar et al., 2005).

Little overlap in the microbiomes in the gut and abdomen
The bacterial assemblages in the guts and abdomens of individual bryophagous beetles differed substantially. This indicates the possibility of confounding two distinct microbial niches if gut and abdomen are not separated during dissection, as is the case of most studies on such small insects. Despite the careful dissection of the guts and abdomens of beetles, we cannot completely exclude the possibility of cross-contamination. Thus, the recorded dissimilarity of the assemblages in the gut and abdomen could be even more signifi cant, further emphasizing the importance of their segregation during dissection.

Ecology and function of potential symbionts
Although symbiotic bacteria are often acquired via the diet or soil, they can be involved in digestion and other processes (Kelley & Dobler, 2011). Proteobacteria and Firmicutes are often the predominant bacterial phyla in the majority of insect guts (Vasanthakumar et al., 2008;Douglas, 2011;Colman et al., 2012), which is consistent with our fi ndings. The eudominant bacterial genera significantly more associated with the gut and abdomen of both byrrhids were Bradyrhizobium, Caulobacter, Novosphingobium and Ralstonia. These bacteria were not only more abundant in the gut than on the surface of their body, but also more abundant in the abdomen than in the gut, indi- cating a more intimate association with the organisms and possible involvement in the metabolic processing of bryophytes. Bradyrhizobium, the commonest soil bacterium fi xing nitrogen in legumes (Klimaszewski et al., 2013), has been repeatedly associated with Sphagnum moss (Bragina et al., 2012) and reported in the guts of Cerambycidae (Grünwald et al., 2010), Staphylinidae (Klimaszewski et al., 2013) and Tortricidae fed on artifi cial diets (Landry et al., 2015). Many other MOTUs recorded in the gut and abdomen belong to Bradyrhizobiaceae, which are typical rhizosphere bacteria fi xing nitrogen ( Some of these microorganisms can provide their host with nitrogen, which is typically defi cient in plant materials (Benemann, 1973). Rapid passage through the midgut may reduce the ability to extract nitrogenous compounds from food. Therefore, an association with nitrogen-fi xing bacteria may be especially benefi cial. Some genera present in the microbiota of bryophages, such as Bradyrhizobium, fi x nitrogen. Closely related to the nutritional role of symbiotic microorganisms is their ability to detoxify (Engel & Moran, 2013). Bryophytes contain a broad spectrum of secondary metabolites that protect them from being eaten (Gerson, 1982). For example, symbionts of the bryophagous Peloridiidae are thought to play a role in excretion or detoxifi cation of ingested moss (Kuechler et al., 2013). Similarly, the microorganisms found in the bryophagous beetles studied, such as Novosphingobium and Ralstonia can degrade phenols and aromatics.

CONCLUSION
The recorded MOTU diversity in the microbiomes in the gastrointestinal tract of bryophagous beetles was high. Substantial differences in the composition of the assemblages recorded in the gut and abdomen indicate the importance of separating body parts before any metagenomic analysis. High bacterial diversity may stem from the particular diet of bryophagous beetles, as the most abundant MOTUs recorded are involved in metabolic processes as-sociated with digestion; alternatively, they may refl ect the greater possibilities of metagenomics. Further studies are required to distinguish bacteria, which are either incorporated only transiently through the diet or are ingested accidentally with soil particles, from those involved in bryophyte digestion. This vast bacterial assemblage could potentially serve as a source of enzymes degrading a variety of chemical compounds present in mosses. Moreover, further research is likely to help us understand the unique ability of bryophagous beetles to digest such a specifi c and generally deleterious source of food, moss.