Diet is not the primary driver of bacterial community structure in the gut of litter-feeding cockroaches
Diet is a major determinant of bacterial community structure in termite guts, but evidence of its importance in the closely related cockroaches is conflicting. Here, we investigated the ecological drivers of the bacterial gut microbiota in cockroaches that feed on lignocellulosic leaf litter.
The physicochemical conditions determined with microsensors in the guts of Ergaula capucina, Pycnoscelus surinamensis, and Byrsotria rothi were similar to those reported for both wood-feeding and omnivorous cockroaches. All gut compartments were anoxic at the center and showed a slightly acidic to neutral pH and variable but slightly reducing conditions. Hydrogen accumulated only in the crop of B. rothi. High-throughput amplicon sequencing of bacterial 16S rRNA genes documented that community structure in individual gut compartments correlated strongly with the respective microenvironmental conditions. A comparison of the hindgut microbiota of cockroaches and termites from different feeding groups revealed that the vast majority of the core taxa in cockroaches with a lignocellulosic diet were present also in omnivorous cockroaches but absent in wood-feeding higher termites.
Our results indicate that diet is not the primary driver of bacterial community structure in the gut of wood- and litter-feeding cockroaches. The high similarity to the gut microbiota of omnivorous cockroaches suggests that the dietary components that are actually digested do not differ fundamentally between feeding groups.
KeywordsInsects Gut microbiota Deep sequencing Cockroaches Lignocellulose
Cockroaches are the closest relatives of termites [1, 2]. The intestinal tracts of both insect groups are densely colonized by a symbiotic gut microbiota of bacteria and archaea, and sometimes also unicellular eukaryotes [3, 4, 5]. The gut microbiota of termites and its role in symbiotic digestion have been studied intensively during the past decades (for reviews, see [6, 7, 8]). In all evolutionarily lower termite families, lignocellulose digestion is carried out primarily by a dense assemblage of symbiotic flagellates, which are absent in all cockroaches and higher termites (family Termitidae). In the wood-feeding members of the Termitidae, their key roles in the digestion of cellulose and hemicelluloses were apparently replaced by specific lineages of Fibrobacteres and Spirochaetes [9, 10, 11].
Much less is known about the bacteria colonizing the intestinal tracts of cockroaches and their role in symbiotic digestion. While termites are highly specialized on a lignocellulosic diet, cockroaches are mostly omnivorous scavengers that typically exploit a variety of food sources . Nevertheless, lignocellulosic plant litter and decaying wood present a major food source for many species, and lignocellulose digestion by cockroaches is considered to play a critical role in the turnover of organic matter in forest ecosystems .
In the wood-feeding Parasphaeria boleiriana (Blaberidae: Zetoborinae) and all members of the genera Panesthia and Salganea (Blaberidae: Panesthiinae), which dwell in decaying wood logs [13, 14, 15], xylophagy most likely evolved independently from that in the termite clade . Also many detritivorous cockroaches feed on leaf litter or other diets rich in lignocellulosic substrates . The survival of xylophagous Panesthiinae on pure cellulose has been attributed to the presence of glycoside hydrolases produced by both the host and its gut microbiota ([15, 17]; for a review, see ), but detailed balances of plant polymer degradation in litter-feeding cockroaches are lacking.
Next to diet, niche heterogeneity has been recognized as another important determinant of bacterial community structure in the hindgut of termites. The composition of the termite gut microbiota is not only characteristic for members of different feeding groups  but also differs among the individual compartments of their intestinal tract  and between microhabitats located within the same compartment, such as gut wall, fiber fraction and luminal content [10, 21]. These differences in community structure are usually accompanied by changes in both microenvironmental conditions (pH, oxygen status, and intestinal redox potential) and microbial activities [22, 23, 24, 25].
In all cockroaches investigated to date, microenvironmental conditions are rather uniform. The gut content is slightly acidic to neutral and has a negative redox potential [26, 27, 28]. In adult cockroaches, the center of all gut compartments is typically anoxic, but in the gut of early larval stages, suboxic conditions have an impact on microbial community assembly during host development . Hydrogen accumulation has been reported only for the posterior midgut of the omnivorous scavengers Blaberus sp. and Shelfordella lateralis (maintained on formulated rabbit or chicken feed) [26, 30], and for the crop of Panesthia angustipennis (maintained on decaying wood) . Each major gut compartment of the omnivorous S. lateralis, the wood-feeding P. angustipennis, and a detritivorous Panchlora sp. (maintained on refuse pile material of leaf-cutter ants) distinctly differs in structure and composition of its bacterial community [26, 27, 31]. In experiments with germ-free S. lateralis that were inoculated with gut communities from various hosts, similar microbial lineages were selected by the gut environment, irrespective of the inoculum , which suggests a strong selection pressure by the microenvironmental conditions and the functional niches available in the gut.
It remains unclear whether structure and composition of the bacterial gut microbiota of cockroaches are strongly affected by diet. A significant response of the hindgut microbiota to diets with different protein contents was found in the omnivorous Blattella germanica  but contrasts with a resilience to dietary changes reported for Periplaneta americana . In S. lateralis, potential effects of high-protein and high-fiber diets of bacterial community structure were masked by strong individual variations . The high similarity in the bacterial community structures of omnivorous cockroaches and a Panchlora sp. that lives in the refuse piles of fungus-cultivating leafcutter ants suggests the existence of a core microbial community that is independent of a particular diet . However, the number of cockroach species investigated so far is too small to test the effects of host diet on bacterial community structure, and information on representatives that thrive on lignocellulosic plant litter is sorely needed.
We addressed this gap by characterizing the bacterial gut microbiota of cockroaches from the genera Byrsotria, Pycnoscelus, and Ergaula, which represent litter feeders from three subfamilies (Blaberinae, Corydiinae, Pycnoscelinae), are available from commercial breeders, and can be maintained on a diet of dried oak leaves. Using high-throughput amplicon sequencing of the bacterial 16S rRNA genes, we taxonomically analyzed the communities using a phylogenetically curated reference database (DictDb), tailor-made for the accurate identification of bacterial lineages specific to termite and cockroach guts , and compared community structure and composition to previously published datasets of cockroaches from other diet groups. To identify differences in microenvironmental conditions responsible for differences in community structure between compartments, we used microsensors to determine oxygen and hydrogen partial pressure, intestinal pH, and redox potential of the gut lumen along the entire intestinal tract. To determine whether host diet determines bacterial community structure in cockroaches, we identified the core bacterial families in cockroaches with a lignocellulosic diet and compared them to those in omnivorous cockroaches and xylophagous higher termites.
Physicochemical conditions in different gut compartments
The redox potential of the gut contents, measured at the gut center, was highly variable in crop and midgut but more consistent in the hindgut compartment of all species. Although all compartments were anoxic at the gut center (not shown), negative redox potentials (− 100 to − 200 mV) were observed only in B. rothi. In the other species, the values ranged from + 100 to + 200 mV, even in the dilated hindgut. Hydrogen partial pressure was either low (0.3–3.5 kPa) or below the detection limit (< 0.1 kPa in the hindguts of B. rothi). Only B. rothi showed a moderate accumulation of hydrogen in the crop (6–21 kPa). Oxygen partial pressure was below the detection limit at the center of all compartments (not shown).
Community structure of homologous gut compartments
Properties of the iTag libraries of the individual gut compartments obtained from different host species. Diversity indices for crop (C), midgut (M), and hindgut (H) are based on OTUs, classification success is based on the number of assigned reads at different taxonomic levels
Classification success (%)
Several core bacterial families made up a major part of the bacterial communities, especially in the hindgut. Here, the relative abundance of 18 core bacterial families (Additional file 2: Table S3) ranged between 46.0% (P. surinamensis) and 98.4% (Panesthia angustipennis). The different lineages of the polyphyletic Porphyromonadaceae together comprised the most abundant bacterial family in both the midgut and hindgut of lignocellulose-feeding cockroaches, covering on average 22 and 23% of the bacterial community, respectively. However, Porphyromonadaceae_1 were more abundant in the midgut, while Porphyromonadaceae_2, as well as previously undescribed members binned to Porphyromonadaceae, “Cluster V”, and “Gut group” were more abundant in the hindgut. More lineages that accounted for 22% of the bacterial hindgut community fell within the Ruminococcaceae, most of which had no cultured representatives (e.g., “insect cluster”, “gut cluster” and “uncultured”). Members of the genus Ruminococcus were abundantly represented in the hindgut of Ergaula capucina and Byrsotria fumigata, while Papillibacter was present in all hindguts except Pycnoscelus surinamensis. Lachnospiraceae made up on average 12 and 13% of the bacterial community in the crop and midgut of the lignocellulose feeders. In the hindgut, they were represented by several major lineages, such as sequences from “gut cluster 13” (without cultured representatives but related to Butyrivibrio crossotus) or “Incertae sedis 30” (with the cultured representative Clostridium phytofermentans). Large fractions of the midgut community of Panesthia angustipennis (17%) and Salganea esakii (10%) were made up of Ca. Arthromitus. Endomicrobiaceae were found in very low abundance (≤0.8%) in the hindgut communities of all lignocellulose feeders.
Effect of diet on gut community structure
To account for the tendency of members of the same bacterial family to carry out similar metabolic processes, and to rule out the possibility that bacterial lineages co-evolving with both termites and cockroaches were too dissimilar to be captured at the genus level, we extended the core taxon analysis to the family level. Most of the 18 bacterial families that were consistently represented in cockroaches with a lignocellulosic diet (i.e., wood- and litter-feeding cockroaches) were represented also in all omnivorous cockroach species (Fig. 6b). On average, the members of these families made up more than 90% of the bacterial community in the wood-feeding Panesthiinae, 60% in litter-feeding cockroach species, and 77% in omnivorous cockroach species. Their high relative abundance of core families from cockroaches in fungus-feeding termites was to be expected, but their representation in the other feeding groups was substantially higher than at the genus level. Again, the lowest proportion of reads assigned to bacterial core families from cockroaches was observed in the bacterial community in wood-feeding termites. Although their relative abundance was highest in fungus- or soil-feeding termite species, 13 out of the 18 core families present in wood- and litter-feeding cockroaches were represented also in litter- or humus-feeding termites, more than in any other termite feeding group. Litter-feeding cockroaches and litter- or humus-feeding termites were also similar in terms of taxon richness, i.e., the total number of bacterial genera (126 and 136) and families (58 and 66) represented in the respective communities, which were higher than in any of the other feeding groups. The relative abundance of the lignocellulose-feeding cockroach core families across all cockroaches ranged from 63 to 96%; within the cockroaches, the relative abundance was lowest in the litter-feeding cockroaches.
When we tested the representation of the 11 bacterial families that made up the core community in wood-feeding termites, we found that between six and nine of them were present also in cockroaches of all feeding groups (Fig. 6c). However, they represented a much smaller part of the total diversity and relative abundance of the respective communities.
Overall, the hindgut bacterial communities of cockroaches with a lignocellulosic diet featured core bacterial taxa different from those of wood-feeding termites. In some cases, similar core patterns on the family level between the different host feeding groups were due to the abundance of different genus-level lineages within the same family. For instance, Lachnospiraceae contributed, on average, 13 and 25% of the bacterial community in lignocellulose-feeding cockroaches and soil-feeding termites, respectively. However, while the undescribed “Gut cluster 13” within this family was among the most dominant genus-level groups in both host groups, soil-feeding termites additionally featured Ca. Arthromitus in high relative abundance (Additional file 2: Table S4). The Rikenellaceae were represented by Alistipes II in fungus-feeding termites, Alistipes IV in lignocellulose-feeding cockroaches, and Alistipes III and IV in omnivorous cockroaches. Notably, the Acholeplasmataceae, represented in most cockroaches by the genus Acholeplasma with up to 1.4% of the bacterial community, was completely absent in all higher termites and Cryptocercus punctulatus.
In this study, we compared the physicochemical conditions and bacterial microbiota in the individual gut compartments of litter-feeding cockroaches. Our results confirm previous findings from wood-feeding and omnivorous species, which detected strong differences between foregut, midgut, and hindgut [26, 27, 30], and support the notion that abiotic and biotic conditions in the gut microenvironment are major drivers of bacterial community structure in cockroach guts. Our analysis of abundance and distribution of family-level lineages among five lignocellulose-feeding cockroach species revealed highest similarity among homologous gut compartments, particularly the hindgut. However, the proportion of core taxa shared between wood- and litter-feeding species was much higher in omnivorous cockroaches than in wood-feeding termites, which indicates that diet is not a major driver of community structure in cockroach guts. This conclusion is corroborated by the absence of lineages implicated in fiber digestion in wood-feeding termites from wood- and litter-feeding cockroaches.
Differences in physicochemical conditions
Physicochemical conditions in the gut of litter-feeding cockroaches (this study) do not differ fundamentally from those in omnivorous and wood-feeding species [26, 27, 28, 30]. In all species investigated, the crop is moderately acidic, which has been attributed to a putative fermentation of ingested sugars by microorganisms already in the early studies of Blattella germanica and Periplaneta americana by Wigglesworth , and was later substantiated by the accumulation of lactate and acetate in this compartment [26, 27, 41]. This matches the large populations of lactic acid bacteria in the crop of P. americana documented by Kane and Breznak  and the predominance of Streptococcaceae or Lactobacillaceae in the crop of wood-feeding and litter-feeding cockroaches (, this study, Additional file 2: Table S1).
The increase in pH along the midgut, with a maximum in the posterior region, is most likely caused by host secretions, including the excretory fluid of the Malpighian tubules, whose nitrogenous components are expected to provide substantial buffering capacity . It is not clear whether the low concentrations of lactate in midgut and hindgut are due to an absorption equilibrium between host and symbionts, as postulated for Periplaneta americana , or to a high turnover of the lactate pool, as demonstrated for lower termites [44, 45].
Although the major gut compartments of all cockroach species investigated to date are anoxic at the gut center ([26, 27, 30], this study), the redox potential of the gut contents differs substantially between species. The omnivorous S. lateralis , the litter-feeding Byrsotria rothi (this study), and the wood-feeding Panesthia angustipennis  show negative redox potentials (below − 100 to − 200 mV) at the center of all gut compartments (except for the rectum of B. rothi). In the litter-feeding Ergaula capucina and Pycnoscelus surinamensis, the redox potential was generally positive, decreasing from 150 to 300 mV in the crop to a range of 0 to 100 mV in the anterior hindgut.
The difference in redox profiles indicate differences in the redox-active metabolites in the respective compartments. Only two of the cockroach species examined to date, Panesthia angustipennis  and Byrsotria rothi (this study), accumulate hydrogen in the crop. This is in agreement with the negative redox potential of the crop observed in both species and the absence of hydrogen-consuming processes (viz., methanogenesis and reductive acetogenesis) in the crop of cockroaches . By contrast, in the omnivorous cockroaches Blaberus sp. and Shelfordella lateralis, hydrogen accumulation is restricted to the midgut or anterior hindgut, reaching magnitudes of 29 and 24 kPa, respectively [26, 30]. In all cases, the hydrogen partial pressures observed in the respective compartments range between 20 and 30 kPa, which surpass the values reported for certain wood-feeding termites [45, 46].
As in omnivorous and wood-feeding species [26, 27, 30], hydrogen concentrations in the hindgut paunch of the litter-feeding species were close to or below the detection limit, which indicates that hydrogen-consuming processes (i.e., methanogenesis and homoacetogenesis) provide a strong hydrogen sink in all cockroaches investigated to date.
Microenvironmental conditions determine community structure
The similarity of the microenvironmental conditions in the hindgut compartments of all cockroaches investigated to date matches the high similarity of the bacterial communities of most cockroach species ([31, 11], this study). Considering our observation that a major proportion of the total bacterial community in the hindgut of wood- and litter-feeding cockroaches consists of bacterial taxa that belong to bacterial families that are consistently represented in the hindgut of all species investigated (core families; Fig. 5), the hindgut compartment must provide essentially the same ecological niches for its microbiota, irrespective of the diet of the host.
By contrast, the bacterial communities in the crop and midgut of each wood-feeding and litter-feeding species differ substantially from that of the hindgut, showing highest similarity to the corresponding compartment in the same host (Figs. 3 and 5 ;. Also in a detritivorous Panchlora sp. that lives in the refuse pile of leaf-cutter ants, each gut compartment harbors a distinct bacterial community that differs substantially from that of the fungal gardens or the waste material deposited by the ants . The representation of bacterial lineages from the same core families in the homologous gut compartments of different cockroaches and the strong correspondence of certain bacterial lineages with specific physicochemical parameters are best explained by the stochastic uptake of bacteria from the environment and the subsequent selection and proliferation of certain bacterial lineages from the inoculum. The inoculation occurs either with the food or by coprophagy, which seems to be common among cockroaches [47, 48, 49] and is essential for the normal development of the first instar in Blattella germanica .
The gut microenvironment has been identified as a strong selective factor shaping the compartment-specific bacterial communities also in higher termites . Despite ample opportunity for a vertical transmission of gut bacteria via proctodeal trophallaxis (feeding on the hindgut content of nestmates), even the termite gut microbiota comprises numerous bacterial lineages that were obviously acquired by horizontal transfer from other species or from the environment . It remains to be investigated whether the highly abundant Dysgonomonadaceae (Bacteroidales termite cluster V), Rikenellaceae, and Ruminococcaceae are part of an ectosymbiotic community on intestinal thelastomatid nematodes (pinworms), as shown in P. angustipennis . Overall, the presence of numerous gut-specific clades among the microbiota of both termites and cockroaches also supports the hypothesis that most microbial ecosystems are dominated by specialist taxa .
Host diet and putative cellulose digestion
One major hypothesis on the assembly of intestinal communities concerns the role of the host diet, which should select for bacterial taxa specialized on the degradation of its recalcitrant constituents or the supplementation of deficient components. While different artificial diet regiments have been shown to change the hindgut community in the omnivorous Blattella germanica and the litter-feeding Pycnoscelus surinamensis [33, 54], no such effect has been observed in Shelfordella lateralis . Also in Periplaneta americana, the core gut community appears to be stable and resilient to changes in diet . Our results for litter-feeding cockroaches confirm the general similarity of the bacterial communities in the hindguts of all cockroaches and their difference from those in termites . The slight differences between the hindgut communities of wood- and litter-feeding cockroaches and omnivorous species were much less pronounced than those between termites from different diet groups (Fig. 8).
The two most dominant phyla in the hindgut communities of wood- and litter-feeding cockroaches (Firmicutes and Bacteroidetes) predominate not only the gut communities of all cockroaches but also those of fungus-feeding termites . This matches the surprisingly high similarity in the overall gut community structure between these only distantly related host groups, which has been explained by a convergent adaptation of the microbiota to the protein-rich diet common to cockroaches and macrotermitine termites . The response of the hindgut community in P. surinamensis to a diet supplemented with different proportions of Termitomyces fungus supports this hypothesis .
The core families present in the hindgut of all lignocellulose-feeding cockroaches are abundantly represented (45–98%) among the bacterial communities of all cockroaches, irrespective of their feeding group, and even more abundant in omnivorous than litter-feeding species (Fig. 6b). Their low abundance in the hindgut of wood-feeding termites underscores that their presence is not determined by the lignocellulosic diet. Bacterial lineages representing the lignocellulolytic community associated with the wood fibers in higher termites, such as uncultured members of the “Treponema I” clade and the Fibrobacteria [10, 56], were not represented at all or encountered in low relative abundance only in the gut of Ergaula capucina (1% unclassified Fibrobacterales). However, it has been shown in the lower termite Reticulitermes flavipes that low-abundant bacteria in particular may drive diet-induced changes in gut community composition .
It is well documented that termites efficiently degrade the cellulose contained in their diet but hardly any of the lignin [58, 59, 60]. Except for the work on Periplaneta americana , such data is lacking for cockroaches. Although the litter-feeding cockroach species examined in this study were maintained on an entirely lignocellulosic diet, we observed that Pycnoscelus surinamensis prefers the softer leaf lamina over the more recalcitrant petiole and veins. Based on the similarity between the hindgut microbiota of panesthiine cockroaches (which dwell in decaying wood) and fungus-cultivating macrotermitine termites, it has been proposed that wood-feeding Panesthiinae digest wood-degrading fungi rather than the wood itself . This implicates not only cellulose and hemicelluloses but also fungal proteins and other microbial biomass as important dietary components. Based on our present results, this hypothesis can be extended to litter-feeding cockroaches, which underscores the need to analyze the dietary components that are actually digested by members of the different feeding groups.
The presence of closely related bacterial lineages in the hindgut of all cockroaches, irrespective of phylogenetic position or feeding group, strongly suggests that the gut habitat, rather than host diet, plays a critical role in constraining the structure of microbial communities in cockroaches. Future studies will have to describe further mechanisms of selection in the cockroach gut environment and assign functional roles to individual members of the gut microbial communities.
Sampling and dissection
Cockroaches of the species Ergaula capucina (Corydiidae, Corydiinae) (formerly Polyphagidae ;, Byrsotria fumigata and Byrsotria rothi (Blaberidae, Blaberinae), and Pycnoscelus surinamensis (Blaberidae, Pycnoscelinae) were purchased from a commercial breeder (J. Bernhardt, Halsbrücke, Germany, http://www.schaben-spinnen.de). All colonies were then maintained in ventilated polypropylene containers (length 27 cm, width 20 cm, height 10 cm) continuously in the dark at 25 °C on the same diet of dried oak leaf litter and water for at least 2 months. In all cases, successful molting and maturation of freshly hatched cockroaches over several instars indicated that the colonies could be sustained on this particular diet. Only adult female insects were selected for the experiments.
To assess physicochemical conditions in each gut compartment, intestinal oxygen and hydrogen concentrations, pH, and redox potential were measured with microelectrodes (50-μm tip diameter; Unisense, Aarhus, Denmark). Oxygen and hydrogen microsensors were calibrated as described previously  using N2, synthetic air (21% O2), and a H2/N2 mixture (5% H2). The pH microelectrode was calibrated with commercial pH standard solutions (pH 4.0, 7.0, and 10.0). The redox microelectrode was calibrated with saturated solutions of quinhydrone in pH standards (pH 4.0 and 7.0). For pH and redox microelectrodes, the electric potential was measured against an Ag/AgCl reference electrode. For the measurements, the guts were dissected, placed in glass-faced chambers, fixed with insect pins to a bottom layer of silicone, and covered with air-saturated Insect Ringer’s solution (7.5 g NaCl, 0.35 g KCl, and 0.21 g CaCl2 per liter) . It was not possible to obtain all parameters from the same gut preparation.
Cockroaches were dissected, and the guts were separated into crop, midgut, and hindgut compartments as previously described . The gut compartments of three adult females of each species were placed separately in 2-ml tubes containing 750 μl sodium phosphate buffer (120 mM; pH 8.0) and homogenized with a polypropylene pestle. DNA was extracted and purified using a bead-beating protocol . Extraction success was monitored by observing DNA integrity on an agarose gel. DNA quality was checked via spectrophotometric evaluation of absorption at 230, 260, and 280 nm (NanoDrop 1000 Spectrophotometer, Thermo Scientific, Waltham, USA) and fluorometric quantification (Qubit Fluorometer, Thermo Scientific, Waltham, USA). The V3-V4 region of the 16S rRNA genes was amplified using the universal bacterial primers 343Fmod and 784Rmod  and tagged with sample-specific hexameric barcodes . Purified PCR products were normalized to equimolar amounts, pooled and commercially sequenced (2 × 350 nt paired-end sequencing) on an Illumina MiSeq platform (GATC Biotech, Konstanz, Germany).
The iTag libraries obtained in this study and previously published datasets obtained from termites [19, 13] were processed as previously described . Briefly, paired-end reads with a minimum length of 250 bp and a maximum expected error of 0.5 were assembled into contigs and quality-trimmed (no homopolymers > 10 nucleotides, no ambiguities, average phred score > 25 on a moving window of five nucleotides), and the barcode and primer sequences were removed using mothur . Sequences in each sample were clustered at a threshold of 99% similarity with dnaclust  and de-replicated and aligned with the mothur aligner. The original contigs (before quality trimming) of the samples obtained in this study were deposited in the sequence read archive (SRA) of the National Center for Biotechnology Information (NCBI, Bioproject PRJNA448568).
Comparison of community structure
Aligned sequences were screened, degapped, and clustered into operational taxonomic units (OTUs) at 97% sequence similarity. OTU sampling coverage was estimated from rarefaction curves . Expected richness , diversity , and evenness  of the communities were calculated for each sample. The OTUs were assigned to taxonomic groups using the Ribosomal Database Project (RDP) naïve Bayesian classifier implemented in mothur with a confidence threshold of 80% in combination with a manually curated reference database (DictDb v. 3.0; ). The libraries were subsampled to the size of the smallest sample (53,896 reads per sample for the comparison between the nine samples in this study; 1643 reads for the comparison between all hindgut communities from 28 hosts). Community structure was compared using the taxonomy-dependent Bray-Curtis metric (based on the classification results), a statistic used to quantify the compositional dissimilarity between two different samples, based on counts in each sample , and using the phylogeny-dependent weighted UniFrac algorithm  embedded in mothur. The high dimensionality of the pairwise dissimilarity scores was then compressed to two dimensions via non-metric multidimensional scaling (NMDS) using the vegan package in R . Covariance between community structure, gut compartment, and physicochemical parameters were determined by permutational multivariate analysis of variance (PERMANOVA) and visualized by canonical correspondence analysis (CCA) using the adonis function, both implemented in the vegan package (for details, see Additional file 2: Table S2).
Analysis of core microbial taxa
To identify core microbial lineages, all unclassified reads and all reads in taxa represented by fewer than ten reads were removed from the dataset. Bacterial genera that were present in at least 70% of all samples from a group of insect hosts or from a specific compartment were considered core genera of this group. For bacterial families, this threshold was set to 100%. The similarity on the family level between the gut communities of the five cockroach species with a lignocellulosic diet was determined using the Morisita-Horn index  and visualized using an arc diagram, implemented in the vegan and arcdiagram  packages in R, respectively.
We thank Katja Meuser for excellent technical assistance and Karen Brune for proof-reading the manuscript.
NL conceived the study, performed the experiments, analyzed data, and wrote the manuscript. AM conceived the study, supervised the experiments, and analyzed data. AB conceived the study, wrote the manuscript and secured funding. All authors read and approved the final manuscript.
This work was supported by the Collaborative Research Center SFB 987 (Microbial Diversity in Environmental Signal Response) of the German Research Foundation (DFG), the LOEWE Center for Synthetic Microbiology (SYNMIKRO), Marburg, and the Max Planck Society (MPG). None of the funding bodies was involved in the design of the study, the collection, analysis, or interpretation of data, or in writing the manuscript.
Ethics approval and consent to participate
Consent for publication
The authors declare that they have no competing interests.
- 2.Legendre F, Nel A, Svenson GJ, Robillard T, Pellens R, Grandcolas P. Phylogeny of dictyoptera: dating the origin of cockroaches, praying mantises and termites with molecular data and controlled fossil evidence. PLoS One. 2015;10:e0130127. https://doi.org/10.1371/journal.pone.0130127.CrossRefPubMedPubMedCentralGoogle Scholar
- 7.Brune A, Ohkuma M. Role of the termite gut microbiota in symbiotic digestion. In: Bignell DE, Roisin Y, Lo N, editors. Biology of termites. Dodrecht: Springer; 2011. p. 439–75.Google Scholar
- 12.Bell W, Roth L, Nalepa C. Cockroaches: ecology, behavior, and natural history. Baltimore: Johns Hopkins University Press; 2007.Google Scholar
- 14.Pellens R, Grandcolas P, da Silva-Neto ID. A new and independently evolved case of xylophagy and the presence of intestinal flagellates in the cockroach Parasphaeria boleiriana (Dictyoptera, Blaberidae, Zetoborinae) from the remnants of the Brazilian Atlantic forest. Can J Zool. 2002;80:350–9. https://doi.org/10.1139/z01-230.CrossRefGoogle Scholar
- 18.Watanabe H, Tokuda G. Cellulolytic systems in insects. Annu Rev Entomol. 2010;55:609–32. https://doi.org/10.1146/annurev-ento-112408-085319.CrossRefPubMedGoogle Scholar
- 22.Brune A, Emerson D, Breznak JA. The termite gut microflora as an oxygen sink: microelectrode determination of oxygen and pH gradients in guts of lower and higher termites. Appl Environ Microbiol. 1995;61:2681–7.Google Scholar
- 27.Bauer E, Lampert N, Mikaelyan A, Köhler T, Maekawa K, Brune A. Physicochemical conditions, metabolites and community structure of the bacterial microbiota in the gut of wood-feeding cockroaches (Blaberidae: Panesthiinae). FEMS Microbiol Ecol. 2015;91:1–14. https://doi.org/10.1093/femsec/fiu028.CrossRefPubMedGoogle Scholar
- 36.Mikaelyan A, Köhler T, Lampert N, Rohland J, Boga H, Meuser K, et al. Classifying the bacterial gut microbiota of termites and cockroaches: a curated phylogenetic reference database (DictDb). Syst Appl Microbiol. 2015;38:472–82. https://doi.org/10.1016/j.syapm.2015.07.004.CrossRefPubMedGoogle Scholar
- 37.Chao A. Nonparametric estimation of the number of classes in a population. Scanadinavian J Stat. 1984;11:265–70.Google Scholar
- 39.Legendre P, Legendre L. Numerical ecology, 2nd Edition. Amsterdam: Elsevier; 1998.Google Scholar
- 46.Köhler T, Dietrich C, Scheffrahn RH, Brune A. High-resolution analysis of gut environment and bacterial microbiota reveals functional compartmentation of the gut in wood-feeding higher termites (Nasutitermes spp.). Appl Environ Microbiol. 2012;78:4691–701. https://doi.org/10.1128/AEM.00683-12.CrossRefPubMedPubMedCentralGoogle Scholar
- 49.Rosas T, García-Ferris C, Domínguez-Santos R, Llop P, Latorre A, Moya A. Rifampicin treatment of Blattella germanica evidences a fecal transmission route of their gut microbiota. FEMS Microbiol Ecol. 2018. https://doi.org/10.1093/femsec/fiy002.
- 52.Murakami T, Onouchi S, Igai K, Ohkuma M, Hongoh Y. Ectosymbiotic bacterial microbiota densely colonize the surface of thelastomatid nematodes in the gut of the wood-feeding cockroach Panesthia angustipennis. FEMS Microbiol Ecol. 2019;95:fiy238. https://doi.org/10.1093/femsec/fiy238.
- 56.Tokuda G, Mikaelyan A, Fukui C, Matsuura Y, Watanabe H, Fujishima M, et al. Fiber-associated spirochetes are major agents of hemicellulose degradation in the hindgut of wood-feeding higher termites. Proc Natl Acad Sci. 2018; 115:E11996–12004. https://doi.org/10.1073/pnas.1810550115.CrossRefGoogle Scholar
- 58.Breznak JA, Brune A. Role of microorganisms in the digestion of lignocellulose by termites. Annu Rev Entomol. 1994;39:453–87. https://doi.org/10.1146/annurev.en.39.010194.002321.CrossRefGoogle Scholar
- 59.Esenther GR, Kirk TK. Catabolism of aspen sapwood in Reticulitermes flavipes. Ann Entomol Soc Am. 1974;67:989–91.Google Scholar
- 60.Griffiths BS, Bracewell JM, Robertson GW, Bignell DE. Pyrolysis–mass spectrometry confirms enrichment of lignin in the faeces of a wood-feeding termite, Zootermopsis nevadensis and depletion of peptides in a soil-feeder, Cubitermes ugandensis. Soil Biol Biochem. 2013;57:957–9. https://doi.org/10.1016/j.soilbio.2012.08.012.CrossRefGoogle Scholar
- 64.Schloss PD, Westcott SL, Ryabin T, Hall JR, Hartmann M, Hollister EB, et al. Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl Environ Microbiol. 2009;75:7537–41. https://doi.org/10.1128/AEM.01541-09.CrossRefPubMedPubMedCentralGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.