Diversity and characterization of bacterial communities of five co‐occurring species at a hydrothermal vent on the Tonga Arc

Abstract Host–symbiont relationships in hydrothermal vent ecosystems, supported by chemoautotrophic bacteria as primary producers, have been extensively studied. However, the process by which densely populated co‐occurring invertebrate hosts form symbiotic relationships with bacterial symbionts remains unclear. Here, we analyzed gill‐associated symbiotic bacteria (gill symbionts) of five co‐occurring hosts, three mollusks (“Bathymodiolus” manusensis, B. brevior, and Alviniconcha strummeri) and two crustaceans (Rimicaris variabilis and Austinograea alayseae), collected together at a single vent site in the Tonga Arc. We observed both different compositions of gill symbionts and the presence of unshared operational taxonomic units (OTUs). In addition, the total number of OTUs was greater for crustacean hosts than for mollusks. The phylogenetic relationship trees of gill symbionts suggest that γ‐proteobacterial gill symbionts have coevolved with their hosts toward reinforcement of host specificity, while campylobacterial Sulfurovum species found across various hosts and habitats are opportunistic associates. Our results confirm that gill symbiont communities differ among co‐occurring vent invertebrates and indicate that hosts are closely related with their gill symbiont communities. Considering the given resources available at a single site, differentiation of gill symbionts seems to be a useful strategy for obtaining nutrition and energy while avoiding competition among both hosts and gill symbionts.

Chemoautotrophic symbionts were first discovered in the vestimentiferan tubeworm Riftia pachyptila at hydrothermal vents along the Galapagos Rift in 1981 (Cavanaugh et al., 1981). Since then, taxonomic and biogeographic knowledge of bacterial symbionts of diverse vent organisms including mytilid mussels, provannid snails, alvinocaridid shrimps, and bythograeid crabs has steadily advanced (Duperron et al., 2006;Fujiwara et al., 2000;Goffredi, 2010;Ponnudurai, 2019;Suzuki et al., 2006;Williams, 1980;Won et al., 2003;Zbinden et al., 2008;Zhang et al., 2017). According to previous studies, bacterial symbionts densely populate specific organs and tissues of their hosts. Vestimentiferan endosymbionts occur densely in bacteriocytes within a highly vascularized internal organ, the trophosome (Jones, 1988). Meanwhile, some vent organisms, including mytilids, provannids, and alvinocaridids, contain dense aggregations of endo-and/or episymbionts on the gills or in branchial chambers (Distel et al., 1995;Dubilier et al., 2008;Duperron et al., 2006;Petersen et al., 2010). In terms of the interactions between hosts and symbionts, the representative symbiont of vestimentiferans, Candidatus Endoriftia persephone, has a broad geographic distribution as well as wide ranges of vent habitats and hosts (Di Meo et al., 2000;Perez & Juniper, 2016). In addition, some mytilid mussels show dual symbiosis, having two bacterial symbionts with different metabolic functions (Duperron et al., 2007(Duperron et al., , 2008Jang et al., 2020). Based on these studies, the distribution, occurrence, and transmission of bacterial symbionts are assumed to be influenced by various factors, including habitat features, vent fluid composition, and the geographic distribution of their hosts (Vrijenhoek, 2010). However, previous studies have generally been conducted separately for various host species, preventing comprehensive analyses of the competitive interactions between hosts in various taxonomic groups and their symbionts.
The gill, which is one of the most extensively studied organs in relation to symbiosis, is the major organ of gas exchange and direct uptake of various organic and inorganic components from water, and its basic structure and function are similar across most aquatic animals (Riisgård, 1988;Rivera-Ingraham et al., 2016;Wood & Soivio, 1991). The uptake rates of dissolved gases and chemical compounds are influenced by their concentrations and molecular weights, environmental conditions, and species-specific anatomical features of the gill (Black & McCarthy, 1988;Hayton & Barron, 1990;Jørgensen, 1974;Perry & Laurent, 1993). In chemosynthetic hydrothermal vent environments, gills also function as a point of entry for highly concentrated toxic materials, such as cadmium, copper, mercury, sulfur, and methane, into the internal tissues (Cavanaugh et al., 1981;Serafim et al., 2006;Felbeck, 1981;Lee et al., 2015;Vetter, 1985). Bacterial symbionts related to gills are assumed to play key roles in supporting host metabolism and other physiological functions, such as carbon fixation, detoxification of metals, and oxidation of sulfides and methane (Cavanaugh et al., 1988;Childress et al., 1986;Jannasch, 1985;Ponsard et al., 2013;Powell & Somero, 1986;Zbinden et al., 2015).
The South-West Pacific Area biogeographic province, our study area, covers a large area of hydrothermal vents in the southwestern Pacific Ocean. These vents are relatively young (<10 Mya) and are enriched in CO 2 , SO 2 , H 2 S, Fe, and particularly Hg compared to vent fields in other oceans (Auzende et al., 1988;Lee et al., 2015). The water masses surrounding the area are well mixed by the South Equatorial Current system (Desbruyères et al., 2006;Mitarai et al., 2016). In addition, this region is a marine biogeographic province with the highest biodiversity, as it contains diverse vent invertebrates belonging to various phyla, including mollusks, crustaceans, annelids, echinoderms, and cnidarians, which are abundant at vent sites with active hydrothermal chimneys (Bachraty et al., 2009;German et al., 2011;Thaler & Amon, 2019).
At the vent of our study site (Figure 1) were found under identical environmental conditions (Video S1 and Figure S1). The mechanisms by which these co-occurring species share and partition the given resources at a single vent site remain unknown. In this study, we investigated their coexistence strategies based on gill-associated symbiotic bacteria (gill symbionts). First, we obtained 16S rDNA libraries of gill symbionts from these five sympatric invertebrates and characterized their community compositions. To clarify the codependent relationship between gill symbionts and their hosts, we constructed phylogenetic relationship trees of the dominant gill symbionts and discussed their relationships along with the taxonomic relationships among hosts.

| DNA extraction, library preparation and pyrosequencing
We dissected gill tissues from two frozen individuals per invertebrate host species. To remove any contaminants from ships, laboratories, or humans, the dissected gill tissues were rinsed with 70% ethanol once, and then washed five times with 1× phosphate-buffered saline (PBS). After it had been washed, genomic DNA was extracted using the FastDNA SPIN Kit for Soil (MP Biomedicals) following the manufacturer's instructions.
To prepare the metabarcoding libraries, first, the V1-V3 region of the bacterial 16S rRNA gene was amplified using the fusion primer

| Data pre-processing and OTU identification
All pyrosequencing results were subjected to Good's coverage estimation to determine the sequencing depth using CLcommunity software version 3.46. Subsequently, data pre-processing of raw reads was conducted following the methods of Jeon et al. (2013). First, low-quality reads (average Q score < 25 or read length < 300 bp) were discarded and the specific bacterial reads for each host sample were sorted using their unique barcodes. The barcode, linker, and PCR primer sequences were trimmed from both ends of the reads using pairwise sequence alignment and the hmm-search program in the HMMER 3.0 package (Eddy, 2011), and chimeric sequences were removed using UCHIME (Edgar et al., 2011). Based on the clusters of trimmed sequences, which allowed no more than two mismatched bases, representative reads were selected for correcting homopolymer errors (Jeon et al., 2013). The selected representative reads were defined as OTUs and were classified using the EzTaxon-e database. Then, taxonomic ranks were defined based on similarity values (x), as follows: x ≥ 97% for species; 97% > x ≥ 94.5% for genus; 94.5% > x ≥ 86.5% for family; 86.5% > x ≥ 82% for order; 82% > x ≥ 78.5% for class; and 78.5% > x ≥ 75% for phylum (Tindall et al., 2010).

| Characterization of gill symbiont communities
Species-level OTUs were used for subsequent analyses. Species richness and diversity were estimated with the Chao1 and Shannon indices using the Cluster Database at High Identity with Tolerance (CD-HIT) method in CLcommunity ver 3.46 (ChunLab Inc.).
To clarify the taxonomic relationships among the dominant gill symbionts of sympatric hosts, species-level OTUs accounting for more than 1% of each gill symbiont community were selected. These OTUs and bacterial 16S rDNA sequences (420-493 bp) associated with chemosynthetic environments were retrieved from GenBank and aligned using the Geneious Alignment method implemented in Geneious Prime v2020.0.4 (Biomatters) and further corrections were made through visual inspection. Then a neighbor-joining tree was constructed using MEGA X (Kumar et al., 2018) with the pdistance model and bootstrap resampling (1,000 replicates).

| Diversity of gill symbionts from co-occurring invertebrate hosts
A range of 4023-7170 reads of the V1-V3 region on the bacterial 16S rRNA gene were obtained from the gill tissues of three mollusks ("B." manusensis, B. brevior, Al. strummeri) and two crustaceans (R. variabilis, Au. alayseae), with Good's coverage values >97%, which indicates sufficient sequencing depth to cover the microbial communities (Table 1; individual variations shown in Figure S2 and Table S2). The 436 total operational taxonomic units Crustacean hosts had more OTUs than mollusk hosts (Table 1).
In particular, the blind crab Au. alayseae was associated with a large number of OTUs (more than 300 OTUs) relative to other species.
However, in all hosts, a small number of specific OTUs accounted for more than 80% of total reads, while most OTUs had abundances of <1%. In B. brevior, a single OTU, BBG_OTU1, represented 99.3% of the total reads (Table A1).  (Brown, 2010;Paster & Dewhirst, 2000;Stokke et al., 2015), we were unable to enhance the discussion of these four OTUs because they showed phylogenetic uncertainties on trees based on 16S rDNA partial sequences (data not shown).
In the two crustacean hosts investigated, the complexity of the gill symbiont community exceeded that of the mollusks (Figure 2;   diverse γ-proteobacteria found in vent crustaceans or environmental samples, with the exception of ASG_OTU2.

| Diversity of sulfur oxidizers in the gills of Tongan invertebrates
Based on metabolite uptake experiments, it has been proposed that Tongan hydrothermal vent ecosystems are supported by sulfuroxidizing bacteria as primary producers and inorganic sulfur compounds as their main energy source (Dubilier et al., 1998;Henry et al., 2008;Suzuki et al., 2006). In this study, the gill symbiont communities of Tongan invertebrates were mainly composed of γproteobacterial sulfur oxidizers, including Cocleimonas, Leucothrix, and Candidatus Ruthia/Thioglobus (Figure 3; Table A1). Furthermore, in crustaceans, campylobacterial Sulfurovum species, which are also sulfur-oxidizing bacteria, were dominant.
In hydrothermal vent ecosystems, γ-proteobacteria are the most commonly observed sulfur-oxidizing symbionts of invertebrate hosts (Apremont et al., 2018;Forget & Kim Juniper, 2013;Goffredi, 2010;Spiridonova et al., 2006). We observed γ-proteobacterial OTUs in all five co-occurring invertebrates from the Tonga Arc. Interestingly, none of the hosts shared species-level OTUs that accounted for more than 1% of total reads for their symbiont community (Table A1). Based

F I G U R E 4
Neighbor-joining tree based on the 16S rDNA of the Campylobacteria related with chemosynthetic environments. Red and blue letters indicate gill symbiont OTUs from Rimicaris variabilis and Austinograea alayseae, respectively, identified in this study. OTU names are shown in Table A1. Sequences of isolated bacterial species are shown in bold. Sequences retrieved from GenBank are presented with their host, collection ocean, and GenBank accession no. Bootstrap values >60% are given above the nodes
The campylobacterial Sulfurovum is known as a representative sulfur-oxidizing epibiont of chemosynthetic ecosystems. It grows chemolithoautotrophically using sulfur or thiosulfate as an electron donor and oxygen or nitrate as an electron acceptor (Inagaki et al., 2004). Although only four Sulfurovum species have been described to date, many 16S rDNA sequences closely related to Sulfurovum have been detected in marine sulfidic environments worldwide (Giovannelli et al., 2013;Huber et al., 2007;Inagaki et al., 2004;Mino et al., 2014). In this study, AAC_OTU2 of Sulfurovum clade I showed greater than 97% similarity with 14 other 16S rDNA sequences detected from diverse hosts in various regions in the global ocean ( Figure 4). This result indirectly indicates that campylobacterial species with these sequences are distributed globally and have weak host preference. Moreover, a similar symbiotic relationship was revealed for members of Sulfurovum clades II and III, as shown in Figure 4. Based on these results, we further consider studies at the genome level to understand low variations within 16S rDNA sequences among Sulfurovum strains from different oceans and hosts.

| Higher diversity of bacterial communities in crustaceans
In all Tongan invertebrates, more than 90% of bacterial OTUs in gills had abundances <1% (Figure 2). If bacterial communities of gills are mainly affected by their external environments, the level of rare OTUs should be similar in all hosts. However, the bacterial community diversities, along with the number and read ratio of rare OTUs, were extremely elevated in crustacean hosts (Table 1, Figure 2). Based on previous studies with dominant OTUs, differentiation of bacterial communities seems to be affected by symbiont forms, endosymbiont for mollusks versus episymbiont for crustaceans (Apremont et al., 2018;Duperron et al., 2007).
Although this is one of the important factors for understanding bacterial communities, it is not sufficient to explain rare OTUs in this study.
In terms of animal behavior, grooming activities performed by crustaceans would have positive effect on bacterial fouling (Gebruk et al., 2000;Thurber et al., 2011). For example, according to an interesting behavioral study of the squat lobster, Kiwa puravida, its cheliped-waving increases in close proximity to seeps discharging methane-rich fluids, which is assumed to be a strategy to ensure a supply of chemical resources for the episymbionts covering its cheliped setae (Thurber et al., 2011). Generally, the necessity and usefulness of symbioses are approached from the perspective of hosts, rather than that of symbionts. Probably, from the bacterial symbionts' viewpoint, mobile organisms may be considered better hosts than sessile ones (Micheli et al., 2002;Van Dover & Fry, 1989;Van Dover et al., 1988).

| Competitive interaction between sympatric bathymodiolin mussels
Sympatric organisms occupying the same ecological niche generally have strategies to avoid competition for food resources and habitat (Baumart et al., 2015;Friedlaender et al., 2015). In chemosynthetic ecosystems, two bathymodiolin mussel species are occasionally found at the same sites, and such species show different gill symbiont compositions (Table 2). Interestingly, in all three cases presented in Table 2, one of the two sympatric hosts had a single symbiont metabolic type, thiotrophic in vents and methanotrophic in seeps, while the other had dual symbiont types, either methanotrophicthiotrophic or carboxydotrophic-thiotrophic (Duperron et al., 2007;Jang et al., 2020). Generally, bathymodiolin mussels depend on their gill symbionts for nutrition (Duperron, 2010). Considering given TA B L E 2 Comparison of symbiont types among sympatric bathymodiolin mussels

| Minor gill symbionts of Tongan invertebrates
Most studies of bacterial community structures have focused on the predominant species in particular environments and hosts.
Recently, however, a few studies have discussed the importance of rare bacteria in various natural communities, such as those in human organs, polluted soils and water, and biogas plants (Ainsworth et al., 2015;Sachdeva et al., 2019). Similarly, in chemosynthetic environments, the main research targets are thiotrophic and methanotrophic bacteria, but little is known about other rare bacteria. In this study, we observed a minor proportion of two bacterial groups, the β-proteobacteria (3.53% for RVB_ OTU1) in R. variabilis and α-proteobacteria (7.76% for AAA_OTU1, 5.03% for AAA_OTU2) in Au. alayseae (Table A1). Although the functions of these gill symbionts remain unclear, we can assume that they cohabitate with their hosts and/or other bacteria to obtain nutrients and act as regulators of physiological processes (Dubilier et al., 2008;Duperron, 2010).
This study is the first comparison of gill symbiont communities of co-occurring invertebrates living at a single vent site of the Tonga Arc. The results indicate that hosts are closely related with their gill symbiont communities. Thus, each host species has certain lifestyle traits, that is, it may be either sessile or mobile, filter-feeding or predatory, and competitive or cooperative, leading to the formation of a specific symbiotic relationship between the host and symbiont.
Eventually, such host-symbiont specificity would potentially reduce competition, thus promoting the coexistence of densely populated co-occurring hosts.
Previous studies have focused on certain tissue types of specific taxa. Therefore, the process by which symbiotic relationships are formed between hosts and symbionts and the strategies used

ACK N OWLED G M ENTS
We would like to thank the captain and crew of the R/V Sonne and the technical team of ROV ROPOS for their invaluable sampling efforts. This expedition was carried out through an R&D project titled "Exploration of Seafloor Hydrothermal Deposits in Tongan Waters" (PM57063). This study was also supported by the Ministry of Ocean

CO N FLI C T O F I NTE R E S T
None declared. Writing-original draft (equal). The OTU name consists of four parts: the first letter of the generic name of the host + the first letter of specific epithet of the host + the first letter of the class name of the gill-symbiont + a number assigned in descending order of read count at the level of bacterial class.