Many insects that host microbial symbionts can also vector significant plant pathogens (Hogenhout et al., 2008a; Hogenhout et al., 2008b; Scholthof et al., 2011; Mansfield et al., 2012; Weintraub et al., 2019; Shafiq et al., 2020), but investigating the co-occurrence and forces underlying insect symbiont and plant pathogen infection presents a challenge. Some data suggest there could be an energetic cost of hosting symbionts, which might reduce an insect’s plant pathogen vectoring ability, while conversely, the benefits of symbionts might provide energy or metabolic advantages that increase plant pathogen transmission (Heck, 2018; Gonella et al., 2019). In developing a theory to predict the outcomes of these interactions, part of the challenge is their complexity. Insect symbioses tend to be richly multipartite, including many species of bacteria and fungi along with their phages and viruses. This complexity involves many insect body sites (gut, salivary glands, fat body, hemolymph, bacteriomes) and external ecological interactions that mediate symbiont and pathogen exchange. As a final surprising dimension of complexity in the dynamics of insect symbionts and plant pathogens, several studies show that insect symbionts can transfer through plant tissues and plant pathogens can evolve readily into insect symbionts (Chrostek et al., 2017; Martinson et al., 2020).

Hemipterans—especially aphids, whiteflies, psyllids, leafhoppers (Cicadellidae), and planthoppers (Fulgoromorpha)—are the most well-studied vectors of plant pathogens, but few studies have investigated treehoppers (Membracidae). Whereas related Auchenorrhyncha hemipterans typically host two bacteria in their symbiont organs (bacteriomes) that cooperatively synthesize missing essential amino acids not present in their plant-sap diet (Wu et al., 2006; Bennett and Moran, 2013; Douglas, 2016; Mao et al., 2017), there are exceptions. Some species have lost one of the obligate symbionts and gained a replacement symbiont (Sudakaran et al., 2017; Matsuura et al., 2018; Bell-Roberts et al., 2019). Other Hemiptera may have gained numerous symbionts. For example, light microscopy studies show Brazilian membracids host at least 28 diverse symbiotic microbes with up to six microbial species cohabiting a bacteriome (Rau, 1943; Müller, 1962; Buchner, 1965). It is unclear which of these microbes are primary or obligate from the host’s perspective, or secondary (facultative), functioning to enable their hosts to survive biotic or abiotic stresses or thrive in particular niches (Oliver et al., 2012; White et al., 2013; Oliver et al., 2014; Guidolin et al., 2018; Santos-Garcia et al., 2018; Lemoine et al., 2020). Many membracids further depend on additional behavioral symbioses with ants, bees, and wasps (Delabie, 2001; Godoy et al., 2006; Ibarra-Isassi and Oliveira, 2018; Klimes et al., 2018; Canedo-Júnior et al., 2019; dos Santos et al., 2019), which feed on the membracids’ secreted honeydew which, in turn, contains additional symbiotic microbes (Leroy et al., 2011; Fischer et al., 2015; Calcagnile et al., 2019; Shamim et al., 2019). However, there are few molecular studies of membracid symbioses. To date, there is one genomic study of a dual-symbiosis in a membracid (Mao et al., 2017) and one 16S rRNA and microscopy-based study showing three or four interacting microbial symbionts in two membracids (Kobiałka et al., 2019). Both studies, however, focused on membracids from the temperate climates, whereas membracid taxonomy, ecology, and microbiota are richest in the neotropics (Buchner, 1965; Cryan et al., 2000; Cryan et al., 2004; Deitz and Wallace, 2012; Hu et al., 2019).

A few membracids in North America are considered significant crop pests—specifically members of tribe Ceresini, such as Spissistilus, Ceresa, and Stictocephala (alfalfa and buffalo treehoppers) which cause swelling on stems or cuts that can facilitate infections in soybean and alfalfa (Meisch and Randolph, 1965; Bailey, 1975); however, in the neotropics, an estimated 18 genera of treehoppers are found as crop pests (Godoy et al., 2006). One study suggests Ceresa in Argentina may vector the witches’ broom phytoplasma (ArAWB) (Grosso et al., 2014). Further studies suggest membracids vector significant plant viruses: Micrutalis (tribe Micrutalini) vectors the viral pseudo-curly top disease (TPCTV) of tomatoes (Mead, 1986; Briddon et al., 1996) and Spissistilus vectors a closely related DNA virus, Grapevine red blotch-associated virus (GRBaV) (Bahder et al., 2016). Membracids’ roles as vectors of plant pathogens are not as well-studied as those of aphids, psyllids, and leafhoppers, which are more commonly found as crop pests. However, their predominance on woody plants makes membracids prime candidates worth investigating as possible vectors for phytoplasmas specific to woody plants such as 16SrIII, 16SrX (apple proliferation), or ESFY (European stone fruit yellows) (Wilson and Weintraub, 2007). Some studies suggest hemipteran secondary symbionts may travel between insects by passage through plants (Chrostek et al., 2017; Li et al., 2018; Pons et al., 2019), including Wolbachia, Rickettsia, Candidatus Cardinium, Serratia, and Symbiopectobacterium (formerly BEV), but these plant–insect–microbe exchanges have not been studied in membracids.

The current study sought to characterize symbionts and potential plant pathogens in membracids, including neotropical species, examining co-occurrence and correlations in abundance as an initial survey of this underexamined group. Our approach used shotgun metagenomics of the membracid bacteriome and surrounding tissues and hemolymph, searched for matches to major bacterial plant pathogens (Mansfield et al., 2012) including phytoplasmas (Hogenhout et al., 2008b), and DNA plant viruses (Shafiq et al., 2020), and an array of bacteriophage (Duron, 2014; Rouïl et al., 2020). Results showed our set of membracids hosted at least nine types of potential plant pathogens, in addition to 12 primary and secondary endosymbionts, along with numerous bacteriophages, and other viruses and parasites. Although our study focused on bacteriomes and surrounding hemolymph, we detected traces of phytoplasmas and spiroplasmas with no indication of plant viruses and we detected significant correlation between subsets of microbes present. Furthermore, we identified divergent Brenneria plant pathogen-like organisms, potential transitional strains of Arsenophonus-like symbionts, Ralstonia and Ralstonia-type phages, Serratia, and APSE-type phages and bracoviruses, indicating membracids may be host to a rich array of symbiont–plant pathogen interactions.

As a step toward discovering how symbionts impact vectoring of plant pathogens in an underexplored group of phloem-feeding insects, we performed this metagenomic sequencing study on membracids. We found, as expected based on historical studies of membracids from Brazil (Rau, 1943; Buchner, 1965), that these insects host a rich collection of primary and putative secondary symbionts, suggesting this group may be a model group for studying complex microbial interactions. Furthermore, from just 16 insect species using community metagenomics and a rapid blast pipeline, we found 12 potential symbiont clades, 9 groups of bacteriophages, 9 putative plant pathogen groups, and many other viruses and parasites, suggesting our approach could be promising if applied on a broader scale. We included membracid root taxa (Aetalion) (Cryan et al., 2000; Cryan et al., 2004; Costa, 2009; Dietrich et al., 2017; Evangelista et al., 2017; Skinner et al., 2020), taxa reported previously as symbiont-rich (Enchophyllum) (Rau, 1943), known virus-vectoring taxa (Micrutalis), and major crop pests (Spissistilus and Ceresa).

Putative plant pathogens included the ‘soft rot’ group of Enterobacteriales, Brenneria sp. and Pectobacterium sp., as well as other gammaproteobacteria such as Enterobacter sp., Pantoea agglomerans, and Serratia sp. Amongst these, Brenneria was most remarkable: Brenneria spp. are not previously known to be transmitted by insects, yet we found sequence matches that occurred at high coverage in two samples (Harmonides sp. membracids). Despite solid phylogenetic placement next to plant pathogenic Brenneria strains, the membracid Brenneria-like 16S rRNA gene sequences were highly divergent, suggesting an increased evolutionary rate, as is common in endosymbionts. However, without further study, we can only speculate on the features of this new Brenneria variant based on the group in which it is found. Importantly, Brenneria are relatives to three of the ‘top 10’ ranked plant pathogens Erwinia, Dickeya, and Pectobacterium (Mansfield et al., 2012) and are pathogens causing numerous diseases (cankers) of woody plants, including the deep bark canker of walnut (Brenneria rubrifasciens) and acute oak decline (Brenneria goodwinii) (Hauben et al., 1998; Bakhshi Ganje et al., 2021). They are noted for producing several unique compounds, such as the red pigment rubrifacine that may contribute to its virulence by inhibiting electron transport in mitochondria. Brenneria species also use sucrose to synthesize levan-type fructans for storage and defense (Liu et al., 2017), which may be of interest in hemipterans whose phloem diet is dominated by sucrose (Shaaban et al., 2020). The Brenneria strain did not, however, closely group with the broadly symbiotic group, Symbiopectobacterium, which includes recently evolved symbionts that independently colonized tissues of various arthropods and nematodes from plant pathogenic ancestors (Martinson et al., 2020; Vallino et al., 2021). Thus, the Brenneria-like sequences may reflect another independent case of a transition from plant pathogen to insect symbiont. Conversely, these sequences could simply be plant pathogens vectored by membracids.

Correlation analyses showed positive associations between the Brenneria and Pectobacterium strains and an insect microvirus, Sulfuriferula, and a parasitoid, but there was no other statistically supported association, suggesting no obvious interaction between primary or secondary symbionts and these Pectobacteriaceae. The Serratia strain found in this study was also correlated with a virus (Indivirus) and parasitoids. In contrast to the Brenneria strains, the Serratia strain showed high similarity (98% 16S rRNA) to Serratia rubidaea, a widespread plant pathogen. Notably, some Serratia species seem to circulate in both plants and hemipterans (Pons et al., 2019), for example, Serratia symbiotica, which likely helps its host digest plant proteins by secreting proteases (Skaljac et al., 2019). However, the Serratia strain occurred in one sample as a short scaffold and so any further analysis would require more data. Pantoea, a different enterobacterial plant pathogen that can be found in various settings including in insects (Walterson and Stavrinides, 2015), was found here only at very low coverage as a short ~85 bp match; therefore, it was not analyzed further. Similarly, while we found a sequence with 98.3% 16S identity to Ralstonia solanacearum, one of the top 10 plant pathogens, we found this in only a single sample (MemA Micrutalis calva from Texas). This Ralstonia was positively correlated with Ralstonia-type phages, as might be expected; however, Ralstonia-type phages were also found in two other U.S. samples (MemE Enchenopa binotata from Illinois and MemM Spissistilus festinus 2 from California), suggesting possible undetected Ralstonia in these samples. Phages of plant pathogens are increasingly becoming of interest for possible biocontrol of bacterial plant disease (Buttimer et al., 2017; Álvarez et al., 2019).

Phytoplasma species are wall-less phloem-infecting plant pathogenic bacteria that require both hemipteran insects and plants in their life cycles and occur in a wide range of woody plants; as such, they might be expected to occur in membracids. Their transmission and life cycle traits, and interactions with existing symbiont have been characterized in leafhoppers (Cicadellidae) and planthoppers (Hogenhout et al., 2008b; Ishii et al., 2013; Weintraub et al., 2019). Because Membracidae is a clade nested within the polyphyletic Cicadellidae (Skinner et al., 2020), authors have speculated that membracids might be expected to be important phytoplasma vectors (Wilson and Weintraub, 2007). Phytoplasmas appear most abundant in tropical and subtropical regions and one study from South America indicated a phytoplasma occurred in the membracid Ceresa (Grosso et al., 2014). Thus, it was surprising to find no 16S rRNA matches to Phytoplasma in these membracids, despite an initial search database of >5000 Phytoplasma 16S genes. Although we found eight samples with scaffolds having short matches to the large Phytoplasma genomic databases (including 1820 genomes), these are not strongly convincing that these membracids vector phytoplasmas. Although the path of these bacteria is through the stylet, intestine, hemolymph, and salivary glands (Weintraub et al., 2019), we expected that our bacteriome-focused dissections and sequencing depth would incidentally include phytoplasmas if they are present. With the caution that Phytoplasma vector status is ambiguous in these data, we note that there were some significantly positive associations in abundance between phytoplasmas and Bombella sp., phage, and other symbionts, suggesting perhaps ecologically common sources of these bacteria or bacterial fragments. Within order Hemiptera, Spiroplasma species reported thus far only from leafhoppers (Weintraub et al., 2019) where they can be either plant pathogens or be vertically transmitted as secondary symbionts, but we found short and low-similarity genome matches to this group as we did for phytoplasmas.

Whereas most viruses of plants have single-stranded RNA genomes and therefore would not be detected in this DNA sequence-based study, we searched for a wide range of DNA plant viruses, expecting to potentially discover some of these, especially those vectored by hemipterans including treehoppers (Mead, 1986; Briddon et al., 1996; Bahder et al., 2016; Varsani et al., 2017; Shafiq et al., 2020). We did not find any geminiviruses, including those related to Topocurvirus which includes the Micrutalis-vectored pseudo-curly top virus, TPCTV, or Grablovirus which includes the Spissistilus-vectored Grapevine red blotch-associated, GRBaV. Although most Geminiviridae are persistent or semipersistent and circulative in their hosts (Shafiq et al., 2020), therefore potentially found in the hemolymph or tissue surrounding the bacteriomes in this study, these are ssDNA viruses that form a dsDNA intermediate in the plant host but may not form a dsDNA phage in the insect, unless they are propagative. For most geminiviruses, it is not clear if they are propagative in the host. We also did not detect Caulimoviridae, which are dsDNA reverse transcribing viruses mostly transmitted by a range of hemipterans (Shafiq et al., 2020), although it is unclear how many of these viruses are circulating and propagative in the insects, suggesting perhaps these plant viruses could be vectored by these treehoppers but not easily detected by these methods.

Among the most abundant presumed secondary symbionts, we found these membracids to be dominated by the genera Arsenophonus, Rickettsia, Sodalis, and Bombella, each of which has members that can be found within plants or causing pathogenicity to plants (Crotti et al., 2016; Chrostek et al., 2017; Gonella et al., 2019). In general, secondary symbionts are diverse functionally, often enabling their hosts to survive a wide range of biotic or abiotic stresses (Gottlieb et al., 2008; Oliver et al., 2012; White et al., 2013; Oliver et al., 2014; Su et al., 2015; Sudakaran et al., 2015; Guidolin et al., 2018; Santos-Garcia et al., 2018; Lemoine et al., 2020). The prevalence, abundance, and phylogenetic analyses presented herein provide some hints, and many open questions, about how these bacteria function in these samples. The high abundance of Arsenophonus in these data has several possible explanations: its 16S rRNA gene occurs in multiple copies per genome (Sorfová et al., 2008) rather than as a single copy as for many endosymbionts including primary symbionts Sulcia and Nasuia, and remarkably Arsenophonus might occur as an abundant hypersymbiont living nested within the cells of the primary symbiont Sulcia (Kobiałka et al., 2016), in which case it could occur at high copy number. Regardless, the abundant and phylogenetically dispersed place of most of these Arsenophonus-like sequences suggest the pattern typical of facultative symbionts (Nováková et al., 2009). The lack of similarity to plant-pathogenic Arsenophonus-like organisms (P. fragariae and Ca. Arsenophonus phytopathogenicus, formerly ‘SMC proteobacteria’) suggests it is unlikely that these strains play this role. Similarly, the lack of similarity to the two probable obligate Arsenophonus-like organisms, Aschnera and ALO-3 (Duron, 2014; Santos-Garcia et al., 2018), suggests no evidence for this role in the sampled membracids. However, the discovery of one variant that clusters at the root of the adelgid symbiont clade (Ca. Hartigia pinicola) as sister to outgroup pathogens suggests a potentially distinct or perhaps parasite-to-commensal transitional function in this organism. Additionally, the typical Arsenophonus strains appear to have at the root of the tree a variant (accession KF751212.1) symbiotic in Stomaphis spp., which are hemipterans specializing on stems and roots of trees, raising the question of this diet as ancient in the Arsenophonus hosts. Ultimately, multi-locus Arsenophonus phylogenomics will be important in uncovering these relationships more accurately, particularly because the 16S rRNA tends to multi-copy in this group (Sorfová et al., 2008) along with comparative omics analysis and detailed microscopy to understand these symbionts, particularly given the observation that Arsenophonus can live within the cells of the primary symbiont Sulcia (Kobiałka et al., 2016).

Our finding of distinct strains of the symbiotic acetic acid bacteria (Acetobacteraceae).

Bombella (formerly Candidatus Parasaccharibacter apium), Asaia, and Saccharibacter floricola, is a novel finding and of special interest in the sugar metabolism of these insects which secrete sugary honeydew to engage trophobiosis with ants and bees. Saccharibacter floricola (Jojima et al., 2004) specifically is bee-associated (Smith et al., 2020). Asaia and Bombella spp. occur in tropical plants and can be plant growth promoters (Crotti et al., 2016), whereas in insects they can stimulate the innate immune system, increase the rate of larval development, and provide insecticide resistance (Chouaia et al., 2010; Chouaia et al., 2012; Mitraka et al., 2013; Comandatore et al., 2021). They can transmit horizontally and vertically, crossing from the gut to the hemolymph and eggs. As symbionts, they likely play a major role in metabolizing sugars to acids (Dong and Zhang, 2006; Crotti et al., 2010); thus, we hypothesize this group to be potentially very important in the observed strong associations between ants and bees and the seven samples of membracids in which we found them. Based on the level of sequence divergence between the membracid Bombella isolates and other Bombella sp. together with the fact that most of the Bombella sequences were similar or identical, we speculate that there may be horizontal transfer of Bombella either between these membracids and their tending hymenopteran insects, or with plants.

Other symbionts exhibited patterns typical for their respective groups, as they occur in other Hemiptera. For example, Rickettsia was dispersed amongst samples and the phylogenetic tree as a typical facultative, partially vertically transferred symbiont, with similar sequences within a host-species. In most cases, Rickettsia is considered parasitic, but there is also evidence that some strains may confer survival benefits (Hendry et al., 2014). Sodalis showed a similar pattern in these data, with phylogenetically dispersed strains, consistent with multiple environmental to secondary symbiont transitions. However, two 16S rRNA gene sequences matching Sodalis, both from the membracid Membracis tectigera (BM13-2), displayed significant sequence divergence, which could suggest a transition in these strains from secondary to primary endosymbiont (Toju et al., 2010), or possibly pseudogenization of 16S rRNA gene copies. Although we found several matches to other symbionts (Wolbachia, Burkholderia, Hamiltonella, Gullanella, and Sulfuriferula sp.), the most notable of these was a high coverage 99.02% 16S rRNA match to a Burkholderia strain that is endophytic, living within the tissues of palm leaves. The strain, initially named Burkholderia sp. JS23, was re-named Chitinasiproducens palmae (Madhaiyan et al., 2020). Its occurrence at high coverage in this membracid sample is a mystery, but interestingly, this same Burkholderiaceae clade includes the Mycoavidus bacteria, which are endohyphal bacteria of the fungus Mortierella elongata (Ohshima et al., 2016).

As expected, all sampled membracids hosted Sulcia, which almost certainly serves to synthesize amino acids missing from the bugs’ phloem diet, but four samples were missing the betaproteobacteria co-symbiont Nasuia, that normally cooperatively synthesizes the remaining amino acids (Bennett and Moran, 2013; Douglas, 2016; Mao et al., 2017). Given the scope of the present study, it remains unclear whether other bacteria or yeast have become replacement symbionts in samples that are missing Nasuia. However, it is noteworthy that we discovered four distinct sequences matching Ophiocordyceps/Hamiltonaphis-like symbionts, three of which occurred in species that were missing Nasuia (BM11 Aetalion reticulum, MemA Micrutalis calva, and BM53 Guyaquila tenuicornis). Numerous studies from aphids, leafhoppers, and other hemipterans suggest that yeast-like symbionts are common and can emerge as co-symbionts or replacement symbionts (Fukatsu and Ishikawa, 1996; Suh et al., 2001; Sacchi et al., 2008; Nishino et al., 2016; Meseguer et al., 2017; Matsuura et al., 2018).

In addition to the Ralstonia phage discussed previously, two other groups of phages were highly abundant: the Hamiltonella-type APSE phages and Wolbachia phage WO, and in general, we found correlation in abundances between bacteria and their presumptive phages, as expected. The APSE phage, well-studied for its parasitoid-protective toxin in Hamiltonella defensa (Oliver et al., 2012; Oliver et al., 2014; Su et al., 2015), particularly in aphids and whiteflies (Rouïl et al., 2020), is commonly integrated into Arsenophonus through lateral gene transfer (Duron, 2014). The infrequency of Hamiltonella and widespread occurrence of APSE in these membracids suggest perhaps Arsenophonus serves as the APSE host in these species. Despite this potential protection from parasitoids conferred by the APSE phage, these membracids showed signs of parasitoid infection, particularly in the level of wasp-specific bracoviruses, especially Cotesia-type bracovirus. These dsDNA viruses act as mutualists for their wasp hosts, contributing to immune suppression of the parasitized insect once injected along with the wasp’s eggs. While bracoviruses are best-known from braconid wasps specifically parasitizing Lepidoptera, discoveries of these virus sequences in Hemiptera previously (Peng et al., 2011; Cheng et al., 2014) and in the present study raise interest in further study of these viruses and related Polydnaviridae in Hemiptera. The high bracovirus levels in these data could also arise from bracovirus horizontal gene transfers into the membracids, as has been observed in other hosts (Cheng et al., 2014; Chevignon et al., 2018). The low levels of parasitoid DNA compared with parasitoid virus DNA in these data could also be explained by our focus on bacteriome tissue, from which most parasitoid wasp DNA, if present, would likely be missed. Finally, despite the opposite biological effects of these two abundant viruses, APSE and bracoviruses, in our data, rather than being negatively correlated, they were significantly positively correlated.

In conclusion, genomic sequence analysis of this kind cannot directly predict insect vector capacity nor microbe pathogenicity; however, these genomic analyses can be invaluable for uncovering previously overlooked microbial associations. Although membracids have been long studied for their exceptional morphological traits, such as the elaborate pronotum, there is scant data on their microbial associations and vectoring potential. These results showed a rich array of microbes and viruses, including plant pathogens and potential allies, painting a preliminary picture of some critical taxa and interactions worth further research. Furthermore, this study generated a large amount of assembled genomic data with thousands scaffolds that are long enough for future in-depth analysis of gene content. We suggest future studies should investigate the prevalence, function, and mechanisms of these potentially interacting microbes, and potential vectored microbes identified here, such as the Brenneria-like bacteria, Serratia, Ralstonia, mycoplasmas or spiroplasmas, and various associated phages. Specifically, co-occurrence patterns leave uncertainty as to role and function of these microbes which should be addressed with phylogenomic analyses, FISH and TEM microscopy, comparative genomics and studies to assess possible HGTs and pseudogenes, and of course wherever possible also controlled infection experiments.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: NCBI accession: PRJNA733715, SAMN19458266-SAMN19458285.

AB led the design of the study, developed bioinformatics code and pipelines, and drafted the manuscript. MP led the work and field collection in Brazil and assisted with student and laboratory infrastructure and specimen cataloging. MS led the library preparation on Brazilian specimens and assisted in writing the manuscript. All authors contributed to the article and approved the submitted version

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.