Sponges were collected from the Formosa Ridge cold seep ( Figure 1A ) during the scientific cruise of Research Vessel (R/V) KEXUE organized by the Institute of Oceanology, Chinese Academy of Sciences in 2021. The cold seep covers a long circular area (W-E: ~200 m; N-S: ~150 m) (Cao et al., 2021) and a sponge ground was found on the rocks surrounding this area ( Figure 1B ). We brought back rocks from the seafloor and collected the samples on board ( Figure 1C ). Upon arrival at the deck, the samples were rinsed several times with 0.22-µm membrane-filtered seawater to remove loosely attached microbes and debris. Small sections of clean tissues were sub-sampled and placed into separated sterile plastic tubes and instantly frozen at -80 °C for molecular analyses, while the majority of tissues were stored in ethanol (80%) as vouchers for morphological identification. Our sponge samples were collected from three stations: station 267 (119.285636°E, 22.115354°N; 1128 m depth; 13 June 2021), station ZD-10 (119.285152°E, 22.115256°N; 1164 m depth; 18 June 2021), and station ZD-7 (119.284662°E, 22.114870°N; 1208 m depth; 14 June 2021). Among these, station 267 was located closest to the seepage and belongs to the gas flares area. Station ZD-7 was the farthest from the active seepage and had shell debris covering the seabed. Station ZD-10 was a biologically rich area with flourishing benthic communities. Additional information about the Formosa Ridge cold seep, including the methane and hydrogen sulfide concentrations, can be found in the study by Cao et al. (2021).

A total of six sponge species were identified according to both morphological and molecular characteristics, with two species having two biological replicates and four species having three biological replicates. To perform morphological identification, the sponge spicules were isolated by digesting a small piece of sponge tissue using concentrated nitric acid. Subsequently, the spicules were examined using both a light microscope (LM) and a scanning electron microscope (SEM). For molecular data, DNA libraries were prepared from individual genomic DNA samples and subsequently, sequencing was performed using the Illumina PE150 strategy. The sequencing reads were quality-filtered using Trimmomatic and assembled using Megahit. The assembly contigs were then used to extract the 18S ribosomal RNA gene sequences of the sponge host. The 18S rRNA gene sequences of 16 individual samples are presented in Supplementary Table S1 .

Microbial DNA was extracted from sponge samples using the E.Z.N.A.^® Soil DNA Kit (Omega Bio-tek, Norcross, GA, U.S.) following the manufacturer’s protocols. The V1-V9 regions of the bacterial 16S rRNA gene were amplified by PCR (95°C for 2 min, followed by 27 cycles at 95°C for 30 s, 55°C for 30 s, and 72°C for 60 s, and a final extension at 72°C for 5 min) using primers 27F 5’-AGRGTTYGATYMTGGCTCAG-3’ and 1492R 5’-RGYTACCTTGTTACGACTT-3’ ligated with an eight-base barcode unique to each sample. PCR reactions were performed in triplicate 20 μL mixture containing 4 μL of 5 × FastPfu Buffer, 2 μL of 2.5 mM dNTPs, 0.8 μL of each primer (5 μM), 0.4 μL of FastPfu Polymerase, and 10 ng of template DNA. Amplicons were extracted from 2% agarose gels and purified using the Axy Prep DNA Gel Extraction Kit (Axygen Biosciences, Union City). SMRT bell libraries were prepared from the purified amplicons by blunt ligation according to the manufacturer’s instructions (Pacific Biosciences). Purified SMRT bell libraries from the pooled and barcoded samples were sequenced on a single PacBio Sequel II cell by Shanghai Biozeron Biotechnology Co. Ltd (Shanghai, China).

PacBio raw reads were processed using the SMRT Link Analysis software version 9.0 to obtain demultiplexed circular consensus sequence (CCS) reads with the following settings: minimum number of passes = 3 and minimum predicted accuracy = 0.99. Raw fastq files were first demultiplexed using in-house perl scripts according to the barcode sequences information for each sample with the following criteria: (i) The 250bp reads were truncated at any site receiving an average quality score <20 over a 10 bp sliding window, discarding the truncated reads that were shorter than 50bp; (ii) exact barcode matching, 2 nucleotide mismatch in primer matching, and reads containing ambiguous characters were removed; and (iii) only sequences that overlapped longer than 10 bp were assembled according to their overlap sequence. Reads which could not be assembled were discarded. Operational taxonomic units (OTUs) were clustered with a 97% similarity cutoff using UPARSE (version 7.1 http://drive5.com/uparse/) and chimeric sequences were identified and removed using UCHIME. Pick representative sequences for each OTU using Qiime2 (https://qiime2.org) Taxonomic assignment was performed using UCLUST (http://www.drive5.com/usearch/manual/uclust_algo.html) against the Silva 138.1 (https://www.arb-silva.de/) 16S rRNA database at a similarity threshold of 0.8. OTUs that did not belong to bacteria and were annotated as mitochondria and chloroplast were removed. The number of Pacbio reads assigned to each OTU and the OTU taxonomy details are stored in Supplementary Table 2 . To account for different sequencing depths, the OTU abundance matrix was rarefied to 4592 reads per sample.

The rarefaction analysis was performed using Mothur v.1.21.1 (Schloss et al., 2009) to reveal the Chao1, Simpson, and Shannon diversity indices. The principal coordinate analysis (PCoA) with a weighted-unifrac matrix (Lozupone and Knight, 2006) was performed using the community ecology R-forge package (Vegan 2.0 package was used to generate a PCoA figure). Mantel tests were carried out to examine the analysis of similarities (ANOSIM) correlation between the sponge taxonomy and the bacterial community similarity using weighted-unifrac distance matrices with 999 permutations, using the vegan package in R. Non-parametric multivariate analysis of variance (ADONIS) was conducted to further confirm the observed differences.

The PhyloSuite workflow desktop platform (Zhang et al., 2020) was used to construct phylogenetic trees of bacteria. The most abundant OTUs belonging to the clade SUP05 and the family Methylomonadaceae in each sponge species were used as queries to search against the NCBI NR database and close relatives were selected. The symbionts and their relatives were aligned together using MAFFT with the default parameters, the SUP05 dataset consisted of 1475 bp, while the Methylomonadaceae dataset comprised 1465 bp. Phylogenetic relationships were using both maximum likelihood (ML) and Bayesian inference (BI) methods. The most suitable nucleotide substitution model was determined by ModelFinder 2 using the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC). Maximum likelihood (ML) analysis was performed in PhyloSuite batch mode, employing the IQ-TREE plug-in with the “Auto” mode and 100,000 ultrafast bootstrap replicates. Bayesian analysis was conducted using MrBayes. Two runs were carried out with four Markov Chains for 2,000,000 generations (SUP05 dataset) and 10,000,000 generations (Methylomonadaceae dataset) started from a random tree, with sampling every 1,000 generations. The first 25% of trees were discarded as burn-in, and a 50% majority-rule consensus tree was obtained from the remaining trees. The average standard deviation of split frequencies fell below 0.01. The phylogenetic trees and node labels were visualized using FigTree.

In order to identify biomarkers for highly dimensional bacteria, linear discriminant analysis effect size (LEfSe) analysis was performed (Segata et al., 2011). Kruskal-Wallis sum-rank test was performed to examine the changes and dissimilarities among classes followed by LDA analysis to determine the size effect of each distinctively abundant taxa (Li et al., 2018).

Morphological identification and molecular analysis revealed that our sponge samples belong to six sponge species affiliated with six different families in two classes (Demospongiae and Homoscleromorpha) ( Figure 2 ). Most of the species were classified at the genus level, except for Petrosiidae sp., which could only be classified at the family level. Among the six species, only Plakina sp. belongs to Homoscleromorpha. As for the five demosponges, Euchelipluma sp. and Coelosphaera sp. belong to the same order (Poecilosclerida), while Janulum sp. and Petrosiidae sp. belong to the same order (Haplosclerida). Notably, Euchelipluma sp. is the only carnivorous sponge. Several sponges are new species, and their detailed identification will be reported in a separate article. In this study, we mainly provide general taxonomic information about the sponges ( Table 1 ).

In all, 16 sponge samples yielded a total of 179,026 qualified reads, with an average length of 1456.52 bp (the length of 99.98% reads >1350 bp). The reads were clustered into 1,783 unique operational taxonomic units (OTUs) using a similarity threshold of 97%. Out of these OTUs, 1,123 were classified at the species level, accounting for 63.02% of the total, and 532 were exclusively classified at the genus level, representing 29.83% of the total, whereas the remaining 128 can be classified at other higher taxonomic levels, comprising 7.18% of the total.

The coverage rates ranged from 95.34% (Euchelipluma sp.-2) to 99.87% (Stylocordyla sp.-2) ( Figure 3 ), suggesting most of the bacteria have been covered by our sequencing efforts. The number of OTUs (reads can be seen in Table 2 ) and the Chao1 index indicated that Stylocordyla sp.-1 has the lowest species richness and diversity, while Euchelipluma sp.-2 has the highest species richness and diversity ( Table 2 ). Significant differences were found among sponge species in the Shannon index (ANOVA: df = 5, P < 0.001) and Simpson index (ANOVA: df = 5, P < 0.001), whereas no significant difference was found in the Chao1 index (ANOVA: df = 5, P > 0.05) ( Figure 3 ).

The taxonomic abundance of the sponge-associated microbial communities was summarized at both the phylum level and genus level ( Figure 4 ). In total, 25 bacterial phyla were detectable among all sponge samples (based on SILVA classification). The dominant bacterial phyla are Proteobacteria (0.907 ± 0.107), Bacteroidota (0.025 ± 0.001), Actinobacteriota (0.0165 ± 0.0028), Bdellovibrionota (0.0118 ± 0.0005), Planctomycetota, and Acidobacteriota. The predominant phylum Proteobacteria is mainly composed of the classes Gammaproteobacteria and Alphaproteobacteria, of which Gammaproteobacteria are found in all the sponge samples and represent the most abundant group in most of the samples. Especially, the relative abundance of Gammaproteobacteria in the Stylocordyla sp. samples was up to 99.78% (Stylocordyla sp.-1), 99.23% (Stylocordyla sp.-2), and 97.80% (Stylocordyla sp.-3), respectively, and that in the Coelosphaera sp. and Janulum sp. samples exceeded 73.43%. Alphaproteobacteria is relatively abundant in Petrosiidae sp. and Plakina sp. sponges, with an average relative abundance of 36.50% and 51.10%, respectively. Bacteroidota were mainly found in Euchelipluma sp., with an average relative abundance of 10.51%. Actinobacteriota accounts for 3.40% and 21.00% of the carnivorous sponge Euchelipluma sp. and 1.61% in Petrosiidae sp.-3. Bdellovibrionota is mainly found in Janulum sp. and Euchelipluma sp., with an average relative abundance of 10.51% and 6.15%, respectively. Planctomycetota were found in all samples, with up to 6.12% in Petrosiidae sp.-3.

At the genus level, the bacterial symbionts associated with sponges are characterized as specialists and generalists rather than opportunists. The SUP05 clade at the genus level is present in all sponges and appears to be a generalist in the sponge ground. It makes up more than half of the total bacterial reads and is the predominant bacterial species in the Stylocordyla sp., Coelosphaera sp., and Janulum sp. samples. An unclassified UBA10353 marine group in the class Gammaproteobacteria is more closely related to samples from the carnivorous sponge Euchelipluma sp. samples. An uncultured group in the family Rhodobacteraceae belonging to order Rhodobacterales is abundantly present in Plakina sp. but is nearly absent or found in trace amounts in the remaining five Demospongiae sponges. This suggests that the group potentially specializes in Homoscleromorpha sponges. Similarly, an unclassified group in the family Methylomonadaceae is exclusively found in the three Homoscleromorpha samples, suggesting another specialist associates with Homoscleromorpha sponges. Alternatively, an unclassified group in the family Kiloniellaceae, belonging to the class Alphaproteobacteria, is exclusively found in the Petrosiidae sp. and accounts for 27.05% to 47.43% of the total bacterial reads. In addition, bacterial reads assigned to genus-level taxa in the Sva0996 marine group, the S25−593 group, and the Marine Methylotrophic Group 2 also occupy a relatively high proportion in these cold seep sponges.

Given a large number of OTUs cannot be accurately classified at the species or genus level, we conducted a detailed analysis of the convergence and divergence of OTUs among sponge species ( Figure 5B ). The results indicate that around 85.73% of all OTUs were exclusive to a single sponge species, 4.05% of all OTUs were shared by two sponge species, and 10.23% of all OTUs were shared among three or more sponge species. Notably, only OTU55 was detected in all samples, making it the most widely distributed taxon. OTU55 was classified as unclassified Gammaproteobacteria and may have a correlation with environmental microbiota. Interestingly, most highly abundant OTUs are not shared by sponge species.

The Principal Coordinates Analysis (PCoA) clearly demonstrated the distinct separation of microbial communities among the six sponge species ( Figure 6A ). The analysis based on weighted-unifrac revealed that the host species diversity accounted for a substantial proportion of variation in community compositions (ANOSIM: test statistic=1.0, p<0.001; ADONIS: F=20.47, R² = 0.91099, p<0.001). However, ANOSIM analysis revealed no significant variation in community compositions at the host sub-order level (from order level to Plakina sp.) (test statistic = 0.2995, p = 0.011). The PCoA plots ( Figure 6B ) indicated that the differences among samples from different sub-orders were partially attributed to variations in dispersion between groups. Specifically, the sub-order groups exhibited greater dispersion compared to the species groups. On the host class level ( Figure 6C ), a highly significant difference in microbial community structure was observed based on ANOSIM (test statistic=0.3625, p =0.018).

LEfSe analyses were conducted to identify significantly different microbial abundance across sponge species and a bar plot was used to display the distribution of LDA scores above the default value of 2.0 for taxa deemed significant, identified as biomarkers ( Figure 5A ). The result indicated that all sponge species except for Coelosphaera sp. have unique biomarkers. Four species of sponge have their unique bacterial class, including Alphaproteobacteria in Plakina sp., Planctomycetia in Petrosiidae sp., Gammaproteobacteria in Stylocordyla sp., and Bdellovibrionales in Janulum sp. Biomarkers at lower taxonomic levels were also identified ( Figure 5A ).

The microbes in the SUP05 clade, particularly those symbiotic with macrofauna such as clams and sponges, are characteristic sulfur-oxidizing bacteria (SOB). The SUP05 clade is highly abundant in our cold-seep sponges. A Venn diagram illustrates that each sponge species has a specific SUP05 OTU ( Figure 7A ). The phylogenetic tree based on 16S rDNA of ML and BI analyses were highly congruent ( Figure 7B ). Furthermore, it also revealed that the predominant SUP05 OTUs in each sponge species do not cluster together or with any known species. The six predominant SUP05 OTUs in each sponge species were searched further against the 16S rRNA gene amplicons of sediment and seawater samples adjacent to them obtained during the same research expedition, and no closely related OTUs were found (the data has not been published yet). Additionally, we searched these OTUs against a large size of metagenomic reads produced from seawater and sediment samples in the Site F cold seep (NCBI accession numbers: SRX10272306, SRX10272305, SRX10272304, SRX10272316, and SRX10272320). Only a limited number of reads matched, which likely resulted from matches of conservative regions in 16S rRNA genes. Consequently, we hypothesize the uniqueness of our SUP05 OTUs to each sponge species.

Porifera in chemosynthetic environments, such as hydrothermal vents and cold seeps, have been reported to harbor substantial populations of bacteria capable of sulfide and methane oxidation (Rubin-Blum et al., 2019; Zhou, 2019). OTUs belonging to methane-oxidizing bacteria (MOX) in the family Methylomonadaceae were also present in our cold-seep sponges. Each sponge species possesses specific MOX OTUs in the family Methylomonadaceae, similar to the phylogenetically diverse SUP05 OTUs found in these sponges. Carnivorous sponges of Euchelipluma sp. exhibited the highest diversity of MOXs, while Plakina sp. had the highest abundance of MOXs. Phylogenetic trees based on the predominant MOX OTUs in each sponge species of ML and BI analyses were generally congruent ( Figure 8 ), revealing that these OTUs do not form a single clade or are clustered with any known species. The process of sponge animals acquiring MOX from the environment is hypothesized to have occurred multiple times throughout their evolutionary history.

Early studies were mainly focused on sponge-associated microbial communities in shallow-water marine habitats. These studies demonstrated that sponges maintain highly diverse and specific microbial communities in the face of a constant influx of seawater microorganisms due to their filter-feeding activities (Margalef, 1963). The Sponge Microbiome Project (SMP) employed high-throughput sequencing technologies to analyze 81 sponge species comprising 804 samples, primarily collected from shallow-water habitats (Moitinho-Silva et al., 2017; Busch et al., 2022). Limited sampling challenges research on sponges from cold-water and deep-sea environments. The Deep-sea Sponge Microbiome Project (D-SMP), conducted in 2022, aimed to characterize the microbial communities associated with 1,077 sponge samples, predominantly collected from the deep sea, representing 169 sponge species (Busch et al., 2022). Both studies utilized extensive sequencing data to demonstrate the microbial diversity and composition of global sponge microbiomes. The chemoautotrophic environment of the cold seeps in the SCS represents unexplored sponge habitats, with sponge grounds consisting of up to six sponge species rarely reported in deep-sea environments. In this study, we investigated the microbial communities of six cold-seep sponge species in the SCS using nearly full-length sequencing, thereby expanding the datasets of sponge-associated microbial communities.

At high taxonomic levels, such as the phylum, the bacterial compositions of the six sponge species showed similarities to shallow-water demosponges. The most abundant phyla included Proteobacteria (Gamma- and Alpha-), Bacteroidota, and Acidobacteriota. Generally, Gammaproteobacteria were the dominant bacterial communities associated with sponges, irrespective of geographical region or sampling depth (Thomas et al., 2016; Cárdenas et al., 2018; Steinert et al., 2019). Additionally, Plakina sp. and Petrosiidae sp. exhibit a high abundance of Alphaproteobacteria, which are commonly found in shallow-water and other deep-sea sponges. Notably, Cyanobacteria are generally absent or occur in low abundance in deep-sea sponges, attributed to the absence of sunlight in deep-sea environments (Thomas et al., 2016). In contrast, the phylum Poribacteria, which is highly specific to sponges, appears to be less prevalent in deep-sea and cold habitat sponges (Cárdenas et al., 2018; Steinert et al., 2020). Nevertheless, the district chemoautotrophic environment of the cold seep in the SCS could potentially influence the composition of sponge-associated microbial communities. Our analysis revealed that 63.02% of OTUs found in the six cold-seep sponges could be classified at the species level, while the remaining OTUs represent novel taxa. These findings highlight the novelty of sponge-associated microbes in deep-sea sponges that reside in unique environments.

The host-species specificity is a major characteristic of the global sponge microbiomes. The six sponge species in our study exhibit substantial variations in both bacterial diversity and compositions ( Figures 3 , 4 , 6 ). These results align with previous studies on shallow-water and deep-sea sponges (Moitinho-Silva et al., 2017; Busch et al., 2022), underscoring the existence of distinct microbial communities within each sponge species. This rule also applies to cold-seep sponges in narrow geographical sponge grounds. Analyses of the global sponge microbiome have revealed that the majority of OTUs are specialists (that is, found in only one or a few sponge species), while only a few are truly core (that is, found across many sponge species) (Busch et al., 2022). In our result, 85.73% of all OTUs were found only in one sponge species, indicating their specialization. This divergence is evident in the varying species composition biomarker quantities among the five sponge species ( Figure 5 ). These findings indicate significant differences in bacterial communities among sponges inhabiting the same restricted geographic area. However, due to the inherent randomness and limitations of collecting samples from deep-sea sponge grounds, only a single sponge species from class Homoscleromorpha was collected. Further analyses will be conducted upon supplementing the sample with additional sponge samples from sponge grounds.

Sponges can be broadly classified into two groups: “high microbial abundance (HMA)” and “low microbial abundance (LMA)” (Hentschel et al., 2003). Generally, LMA sponges have fewer microbial species and lower alpha diversity. In our study, the microbial diversity of Plakina sp., Stylocordyla sp., and Coelosphaera sp. was significantly lower than the other three sponges, referring to both diversity indices ( Figure 3 ) and the number of unique OTUs ( Figure 5B ). However, other direct evidence, such as observations using transmission electron microscopy to detect the presence of microorganisms in the mesohyl matrix, quantitative comparative analyses between fresh samples, and in situ seawater assessments, still need to be comprehensively considered to further determine the HMA and LMA types of sponges.

The SUP05 clade (class: Gammaproteobacteria) is mainly composed of unculturable chemoautotrophic SOBs in marine environments (Anantharaman et al., 2013; Anantharaman et al., 2014). Additionally, these SOBs establish symbiotic relationships with invertebrate animals in deep-sea environments (Kuwahara et al., 2007; Zhou, 2019). As an example, Candidatus Ruthia magnifica, bacteria affiliated with the SUP05 clade, is an obligate chemolithoautotrophic bacterium with a fully sequenced genome (Nelson and Fisher, 1995). Throughout the extended process of symbiosis, SUP05 symbionts have undergone significant genome reduction (Kuwahara et al., 2007), while retaining the complete set of genes necessary for sulfur oxidation, which allows them to obtain energy from environments abundant in reduced sulfur compounds (Ikuta et al., 2015). Existing evidence of strict maternal co-transmission suggests that certain endosymbiotic SUP05 bacteria may have lost their capability to live independently in the marine environment and re-infect invertebrates (Hurtado et al., 2003). Previous studies have reported the presence of sulfur-oxidizing SUP05 strains with high abundance in deep-sea sponges inhabiting asphalt seeps and hydrothermal vents (Tong et al., 2011; Rubin-Blum et al., 2019), but the association of SUP05 symbionts with sponges remains unclear (Kuwahara et al., 2007; Zhou, 2019; Georgieva et al., 2020). The genome of sponge-associated SUP05 was found to contain a variety of genes related to sulfur oxidization, including sat, aprAB, dsrAB, soxA, soxB, soxX, soxY, and soxZ (Zhou, 2019). The genomes possess a complete set of genes necessary for sulfur oxidation, allowing the bacteria to use surrounding reduced sulfur compounds as an energy source (Kuwahara et al., 2007; Newton et al., 2007; Ikuta et al., 2015). The generated energy benefits various symbiotic processes, such as the synthesis of essential amino acids and cofactors, which drives carbon dioxide fixation through the Calvin cycle to fulfill the sponges’ nutritional needs, as well as the uptake and utilization of nitrate produced by the sponge during nitrate respiration(Zhou, 2019).

Our study demonstrates that the microbial communities of cold-seep sponges in the SCS are dominated by sulfur-oxidizing bacteria belonging to the SUP05 group. The predominant SUP05 16S rRNA genes in six sponges exhibited genetic divergence from each other and from other symbiotic and free-living SUP05 members of the SUP05 group ( Figure 7 ), indicating their novelty. In the cold seep environment that we investigated, dark sulfide oxidation appears to serve as the primary source for the sponge microbiome, thus highlighting the potentially significant roles of the SUP05 symbionts.

In cold seep environments, it has been observed that nutritionally symbiotic MOXs constitute a significant proportion (50% to 70%) of the microbiota associated with sponges, consisting of nutritionally symbiotic MOXs (Rubin-Blum et al., 2019). These MOXs utilize methane from seeps to provide the chemosynthetic supply of organic matter to support the host, thereby sustaining the symbiotic system. In our sponge samples, we also found the presence of MOX. Phylogenetic analysis of the 16S rRNA genes indicated that the MOXs associated with the sponges largely belong to the family Methylomonadaceae in the order Methylococcales with some being classified as members of the Marine Methylotrophic Group (MMG) 2 clade (also referred as deep-sea clade 2) (Parks et al., 2018). Pure cultures from the MMG2 clade have not yet been obtained, although three MMG2 genomes have been reported previously (Vekeman et al., 2016; Rubin-Blum et al., 2019). The MMG2 clade is not exclusively associated with sponges, as its 16S rRNA gene clones have also been found in ciliates and the squat lobster Shinkaia crosnieri ( Rubin-Blum et al., 2019 ). MMG2 sequences were also found in certain sponge species. For instance, they were discovered in an unidentified poecilosclerida sponge from seeps in the Gulf of Mexico (Nishijima et al., 2010) and in Cladorhiza methanophila from mud volcanoes off Barbados (Hestetun et al., 2016). These findings suggest that the presence of MMG2 clade bacteria is not restricted to a specific symbiotic relationship but might be more widely distributed in various deep-sea environments and associated with different host organisms. The symbiotic MOXs in sponges assimilate methane through a limited number of essential enzymes, and they exclusively employ the highly efficient Embden-Meyerhof-Parnas (EMP) variant of the ribulose monophosphate (RuMP) cycle to assimilate single carbon compounds (Kalyuzhnaya et al., 2013; Rubin-Blum et al., 2019). Importantly, the symbiotic MOXs in sponges can provide nutrition to their hosts even at lower concentrations of methane, unlike the symbionts found in mussels (Rubin-Blum et al., 2019). Free-living MMG2 bacteria are also commonly found in methane-rich environments, including the sediment and water column of vents, hydrothermal fields, and cold seeps, in addition to their symbiotic relationships with animals (Ruff et al., 2013).

In our study, we observed lower OTU-sharing ratios of MOXs among the six different sponge species. This phenomenon may be attributed to the vertical transmission of symbiotic MOXs from one generation to the next (Rubin-Blum et al., 2019). The six dominant OTUs do not cluster in a single clade and are not clustered together with other known symbionts, indicating their genetic diversity. In future work, more attention should be given to understanding how these symbiotic MOXs adapt to different sponge hosts. In Figure 4B , the analysis of microbial community composition in the 16 samples collected from the six sponge species revealed that Plakina sp. has a higher predominance of MOXs from Methylomonadaceae compared to other sponge species. Plakina sp. stands out among the host sponges as it belongs to the class Homoscleromorpha, distinguishing it from other species. The distribution of MOXs among these sponge species appears to reflect the host-specificity of sponge symbiotic microorganisms. Furthermore, within the three samples of Plakina sp., Plakina sp.-1 exhibits significantly higher MOX enrichment compared to the other two samples. Plakina sp.-1 was collected from the area of gas flares in this cold-seep site, where abundant methane bubbles emerge, while the other two samples were collected from areas farther away from active seepage. Therefore, the distribution of MOXs in Plakina sp., a specific sponge species, is presumed to be influenced by the surrounding methane concentrations. However, in order to understand the impact of class Homoscleromorpha and methane concentration on the content of MOXs in symbiotic microorganisms within sponges, further comparisons should be conducted with more biological samples and environmental parameters collected from the cold-seep site. Furthermore, more attention is needed to investigate this potential pattern.