Evidence of habitat specificity in sponge microbiomes from Antarctica Authors: Maria F. Manrique-de-la-Cuba, Génesis Parada-Pozo, Susana Rodríguez-Marconi, Marileyxis R. López-Rodríguez, Sebastián Abades, Nicole Trefault Published: 2024 ABSTRACT BackgroundMarine sponges and their microbiomes are ecosystem engineers distributed across the globe. However, most research has focused on tropical and temperate sponges, while polar regions like Antarctica have been largely neglected. Despite its harsh conditions and geographical isolation, Antarctica is densely populated by sponges. In this study, we explored the extent of habitat specificity in the diversity, community composition, and microbial co-occurrence within Antarctic sponge microbiomes, in comparison to those from other marine environments. We used massive sequencing of 16S rRNA genes and integrated multiple databases to incorporate Antarctic sponges as a habitat in global microbiome analyses.ResultsOur study revealed significant differences in microbial diversity and community composition between Antarctic and non-Antarctic sponges. We found that most microorganisms present in Antarctic sponges are unique to the South Shetland Islands. Nitrosomonas oligotropha, Candidatus Nitrosopumilus, Polaribacter, SAR116 clade, and Low Salinity Nitrite-Oxidizing Bacteria (LS-NOB) are microbial members characterizing the Antarctic sponge microbiomes. Based on their exclusivity and presence across different sponges worldwide, we identified habitat-specific and habitat-generalist bacteria associated with each habitat. They are particularly abundant and connected within all the Antarctic sponges, suggesting that they may play a crucial role as keystone species within these sponge ecosystems.ConclusionsThis study provides significant insights into the microbial diversity and community composition of sponges in Antarctica and non-Antarctic ecoregions. Our findings provide evidence for habitat-specific patterns that differentiate the microbiomes of Antarctic sponges from elsewhere, indicating the strong influence of environmental selection and dispersal limitation wrapped into the Antarctic ecoregions to shape more similar microbial communities in distantly related sponges. This study contributes to understanding signatures of microbial community assembly in the Antarctic sponges and has important implications for the ecology and evolution of these unique marine environments.Supplementary InformationThe online version contains supplementary material available at 10.1186/s40793-024-00648-4. Background Marine sponges are sessile and filter-feeding macroinvertebrates that are widely distributed in oceans and seas. They play a critical role in ecosystem dynamics by acting as ecosystem engineers of benthic habitats [1–3]. They cycle nutrients [1–4], provide refuge for many micro and macroorganisms, and contribute significantly to the formation of reefs by stabilizing the substrate [5–7]. Sponges are considered holobionts, which are complex evolutive and functional units [1, 8] where microorganisms represent up to 38% of the host biomass and play pivotal roles in the health of the host [1, 5, 6, 9]. However, much is still unknown about how abiotic and biotic factors shape specific patterns of the microbial ecology in sponges of distinct environments and phylogeny. Sponges acquire microbial symbionts from their environments and their parents during their development stages [10–12]. The microbial community composition of sponges is primarily driven by the host species and is distinct from the surrounding seawater [10]. Sponges can be categorized into two types: High Microbial Abundance (HMA) sponges and Low Microbial Abundance (LMA) sponges, based on their bacterial abundance [2, 13]. However, most of the knowledge regarding sponge microbiomes comes from tropical and temperate environments, and there is limited research on whether the microbial patterns in sponges from other environments are similar. In Antarctica, sponge holobionts dominate the benthos and are the most significant influence on this ecosystem [14, 15]. Bacteria, Archaea, and microbial eukaryotes have distinct diversity and similarity patterns in Antarctic marine sponges, suggesting inherent differences in how organisms from the three domains of life form symbiotic relationships [16]. In addition, the bacterial communities exhibit variable host specificity, depending on each sponge species, but their inferred functions remain conserved [17, 18]. Antarctic sponge holobionts share common functions related to the symbiotic lifestyle (CRISPR system, Eukaryotic-Like Proteins, and transposases) and nutrient cycling (ammonia oxidation, nitrite oxidation, denitrification, sulfite and sulfur oxidations, synthesis of polyphosphates, and carbon fixation) [19]. Together, these functions help the sponge holobiont to survive in the Antarctic habitat, further contributing to the major nutrient cycles in this ecosystem. Habitat specificity plays a strong role in microbial community assembly and co-occurrence patterns in different habitats, including aquatic ecosystems [20, 21]. In symbiotic systems, genetic isolation and natural selection can enhance specialization for specific habitats [22, 23]. Habitat generalists are microorganisms distributed widely across diverse environments, maintaining similar abundances [24–29]. In contrast, habitat-specific microorganisms are adapted to particular environments, thriving best in their optimal habitat [24–29]. The Baas Becking hypothesis, “everything is everywhere, but the environment selects” [30, 31], aligns with the concept of habitat specificity illustrating how microorganisms distribute and adapt to various environments . Here, we compared the microbiomes of sponges found in Antarctica with those found in other environments. We identify microbial features that differ between Antarctic sponge habitats and other sponge habitats. We seek to address the extent of habitat specificity in the microbial community composition, diversity, and microbial co-occurrence of Antarctic sponge microbiomes compared to sponges of other environments. Our results indicate habitat specificity in sponge microbiomes from Antarctica. These microbiomes present lower alpha diversity, higher compositional similarity, and more habitat-specific microorganisms than sponges from tropical and temperate environments, reflecting the role of the unique environmental conditions and geographic isolation of Antarctica in shaping a distinct microbial community assembly in sponges. Methods Antarctic sponge sampling Adult samples from Antarctic sponges (n = 33) were sampled during three expeditions between 2018 and 2021 from Chile Bay, Greenwich Island, South Shetland Islands, Antarctica (Supplementary Table S1 and Supplementary Table S2). Sponge samples were collected by scuba diving and kept individually in plastic bags containing natural seawater until the initial processing at the laboratory in Captain Arturo Prat Base at Greenwich Island. Seawater (SW) was also sampled approximately 5 m from each sponge sampling site. Sample treatment, DNA extraction, and sequencing Sponge samples were rinsed three times with sterilized SW, cleaned under a stereomicroscope to remove dirt and ectoparasites, and stored at -80 °C until further processing in Santiago, Chile. For this, subsamples of about 1 cm3 of sponge tissue were obtained and rinsed again to remove loosely attached SW microorganisms. Sponge-associated microbial communities were separated from the host grounding the tissue with stainless steel beads in a TissueLyser II (QIAGEN) for five minutes at 30 Hz. The resulting pulp was resuspended in 10 mL of sterilized SW, filtered and centrifuged for 15 min at 100 g, then the supernatant was recovered and centrifugated for 20 min at 8,800 g, both at 4 °C. Following the manufacturer’s instructions, the metagenomic DNA was extracted from the resulting pellets using the DNeasy PowerSoil Pro Kit (QIAGEN). The quantity and integrity of the DNA were evaluated with a Qubit fluorometer and by electrophoresis, respectively. The extracted DNA was sequenced in the Integrated Microbiome Resource (IMR, Canada), targeting the V4-V5 hypervariable region of the 16S rRNA gene using primers 515 F-926R [32] and the Illumina Miseq platform (300 bp paired-end) (Dataset 1, Supplementary Table S1). The library preparation and sequencing followed the protocol initially described by Comeau et al. [33] and updated as in Comeau and Kwawukume [34]. Dataset collection of sponge microbiomes We retrieved raw sequencing datasets of the V4 hypervariable region from the 16S rRNA gene amplicons deriving from Antarctic sponge microbiomes (n = 56) from the SRA database (Supplementary Table S1). These datasets belong to sponge individuals of several species sampled from different locations around the South Shetland Islands (King George Island for Dataset 2 and Deception and Half Moon Islands for Dataset 3) and the Antarctic Peninsula (Rothera and Bernardo O’Higgins Research Stations for Dataset 3) during 2013–2016 (Supplementary Table S2). Amplicon datasets from samples of sponge microbiomes of tropical (n = 50) and temperate (n = 40) environments, sequenced within the Sponge Microbiome Project (SMP) [35], were retrieved from the SRA database (Dataset 4, Supplementary Table S1 and S2). The selection criterion of the non-Antarctic sponge datasets was to match the taxonomy of the available Antarctic sponges, at least at the family level (Supplementary Table S2). Taxonomy at the order level of non-Antarctic sponges was verified and corrected (if necessary) using the sponge species name reported in the SRA run tables of the SMP to search the respective sponge order within the World Porifera Database [36] and the NCBI Taxonomy Database [37] (in-house script). Other selection criteria were the sponge status (healthy individuals), sampling depth (< 50 m), life stage (adult or not provided), and number of species replicates per environment (n ≤ 5, selected randomly and constrained to resemble the Antarctic sponge replicates in number). Microbiome datasets were labeled by their habitat as ‘associated with Antarctic sponges’ (n = 89) and ‘associated with non-Antarctic sponges’ (n = 90). Tag sequence processing Raw reads were demultiplexed and quality checked using QIIME2 version 2021.4.0 [38]. Adapters were removed with Cutadapt [39]. We included only samples with > 10,000 reads. The amplicon processing was performed using DADA2 version 1.24.0 [40] in R Statistical Software version 4.2.1 (R Core Team 2022) [41]. Only forward reads matching the single-end sequencing data from the SMP were used. Raw datasets were processed on a per-run basis, as these belonged to 12 different sequencing runs. Samples generated from this study (Dataset 1) were all sequenced in a single run. The other samples correspond to 11 different runs. For quality control, filtering parameters were set to read quality > Q15 (truncQ = 2) and a minimum read length of 100 bp. Also, read truncation to 125 bp length was set for Antarctic sponge datasets (as these were 150–300 bp length and most of the SMP sequences were 125 bp). To avoid overestimating Amplicon Sequence Variants (ASVs) due to length variation, sequences were aligned and collapsed together if they had no mismatches or internal indels (function collapseNoMismatch). Sequence denoising and chimera removal followed the default parameters from the DADA2 pipeline. We verified each sample to recover more than 50% of sequences after chimera removal. For the taxonomic assignment of ASVs, the Silva NR99 v138.1 database [42] was used with a minimum of 80% bootstrapping support. We removed sequences assigned to chloroplasts, mitochondria, Eukaryota, non-assigned sequences at kingdom level, singletons, and doubletons. To report the taxonomy at the species/genus level, ASVs not assigned according to the Silva database were annotated with the highest score against the NCBI 16 S rRNA database (21-10-2023), using BLASTN 2.14.1+ [43], if they agreed with Silva taxonomy at the lowest level originally assigned. To link the sponge microbiome to an environment (i.e., tropical, temperate, or Antarctic), and a marine ecoregion (i.e., Antarctic Peninsula, South Shetland Islands, Adriatic Sea, Bahamian, Central and Southern Great Barrier Reef, Central New Zealand, Hawaii, South European Atlantic Shelf, Southern Norway, Southwestern Caribbean, Western Caribbean, and Western Mediterranean), the latitude coordinates of the sampling sites were interpolated into the Marine Ecoregions Of the World (MEOW) classification system metadata [44] using an in-house script. Diversity and community composition analyses Taxonomy composition and diversity analyses were performed using the phyloseq package version 1.40.0 [45] in R. To explore differences in the microbial community composition of Antarctic sponges vs. non-Antarctic sponges, Bray-Curtis dissimilarity and unweighted UniFrac phylogenetic distance were used in ordination analysis, and the former also in hierarchical clustering (Ward’s method, package pheatmap version 1.0.12). These measures were calculated over Hellinger-transformed (square-root) relative abundances at different taxonomy levels, from phylum to genus, and at the ASV level. The function tax_glom was used to agglomerate ASVs assigned to the same taxonomy at each level, enabling the inclusion of non-assignments (NAs). To determine the diversity of microbial communities, measurements of alpha diversity as observed richness, Shannon index, and Simpson index were calculated at the ASV level. Generalized linear mixed-effects models (GLMM) were applied to compare the calculated alpha diversity indexes between Antarctic sponges and non-Antarctic sponges (habitat, fixed effect, two levels), taking into account the sponge phylogeny (family, random effect, 16 levels) and the sequencing depth (run, random effect, 12 levels), using the packages lme4 version 1.1.33 [46], sjPlot version 2.8.14, and ggeffects version 1.2.3.1. The Akaike information criterion (AIC) was used to select the best model for the data analyzed per each ecological measure through the AICcmodavg package version 2.3.2 [47]. To determine significant differences in the community composition and diversity between habitats (associated with Antarctic sponges vs. associated with non-Antarctic sponge), non-parametric tests were calculated in R. Bray-Curtis dissimilarity and unweighted UniFrac distances were used to perform analysis of similarity (ANOSIM) and non-parametric multivariate analysis of variance (PERMANOVA). ANOSIM was carried out with the function anosim of the vegan package version 2.6-4 [48] to estimate significant similarity differences between and within habitats (Antarctic sponge vs. non-Antarctic sponge). To evaluate significant effects in the community composition from the predictor pair ‘habitat’ and ‘sponge family (host phylogeny)’ and the predictor pair ‘ecoregion’ and ‘sponge family’, PERMANOVA was carried out with the function adonis2 of the vegan package. The sequencing depth was treated as a random factor in each case. The number of permutations was set to 999 and the sed.seed to 1000 for these statistical tests. The analysis of Similarity Percentage (SIMPER) was conducted over ASV relative abundances using vegan to assess which ASVs primarily drive the observed difference between habitats. The Linear Discriminant Analysis (LDA) was performed over absolute abundances of the microbial genera to assess differentially abundant taxa using the package microbiomeMarker version 1.3.3. For this, we utilized the function run_lefse and set the LDA score > 3 and the p-value < 0.01 of the Wilcoxon test. Survey of habitat-specific and habitat-generalist microorganisms To determine Antarctic habitat-specific microorganisms at the ASV level, we determined shared and exclusive ASVs using the MicEco package version 0.9.15. Habitat-specific microorganisms from Antarctic sponges were defined as those ASVs present exclusively in Antarctic sponges, in at least 50% of the sponges. Habitat-specific microorganisms from non-Antarctic sponges were defined as those ASVs present exclusively in non-Antarctic sponges in at least 50% of the sponges. Habitat generalists were those ASVs shared between Antarctic and non-Antarctic sponges in at least 50% of the sponges, considering the total sponge samples (n = 179). Co-occurrence networks inference Absolute abundances of ASVs were rarefied to equal the sample with the minimum number of sequences (ID ERR1772080; 8,688 sequences) using phyloseq. Subsequently, the samples were divided by their habitat. The co-occurrence networks were performed for each habitat using the CoNet plugin version 1.1.1.beta [49] of Cytoscape version 3.9.1 [50]. Only the genera identified in at least 10% of the sponge samples were kept. Pairwise correlations between genera were carried out by combining the Pearson and Spearman measures and requesting the network union. The correlation threshold for both methods was set to 0.6 (for positive and negative correlations). Additionally, we avoided correlations between pairs of genera with double-zero values in more than half of the sponge samples. The p-value threshold was set to 0.05 and calculated from measure-specific permutation with 1,000 iterations. Simes method and Benjamini–Hochberg correction combined the networks from each measure. The resulting networks were visualized with the default Prefuse Force Directed Layout with default edge weight settings, implemented within Cytoscape. The network topological properties were obtained with the NetworkAnalizer plugin. Results Diversity and community composition of sponge microbiome differ between habitats We assessed the bacterial/archaeal microbiomes associated with 179 sponges collected from Antarctic, tropical, and temperate environments (Fig. 1A). These sponges comprised at least 59 species across 16 families within seven orders (Fig. 1B and Supplementary Table S2). Across all the microbiomes, we recovered 27,803 ASVs of Bacteria (11,631 in Antarctic sponges and 17,880 in non-Antarctic sponges), along with 270 ASVs of Archaea (44 in Antarctic sponges and 239 in non-Antarctic sponges). Antarctic sponges had fewer lineages and more assigned ASVs compared to non- Antarctic sponges (Supplementary Table S3). We tested whether the microbiomes of Antarctic sponges differed from non-Antarctic sponges in microbial richness and diversity (Fig. 1C and Supplementary Tables S4 and S5). Antarctic sponges exhibited a significantly lower microbial diversity than non-Antarctic sponges as measured by the Simpson and the Shannon index (LMM, pSimpson=0.014, pShannon=0.036). However, the microbial richness was not significantly different between the two groups (GLMM, p = 0.103), although the variability in the microbial richness was higher in Antarctic sponges. Fig. 1Geographic and taxonomic distribution of sponges used in this study and diversity of the sponge-associated microbiomes. (A) Geographical distribution of the sponge samples used according to the environment. The pie chart represents the sponge sample size by environment. (B) Distribution of sponges in ecoregions according to the Marine Ecosystems of the World (MEOW), a biogeographic classification of coastal and continental shelf waters of the world, and taxonomic classification at the family level according to the Porifera Database and the NCBI taxonomy Database. (C) Ecological indexes in Antarctic and non-Antarctic sponge microbiomes, represented by observed richness, Simpson, and Shannon Geographic and taxonomic distribution of sponges used in this study and diversity of the sponge-associated microbiomes. (A) Geographical distribution of the sponge samples used according to the environment. The pie chart represents the sponge sample size by environment. (B) Distribution of sponges in ecoregions according to the Marine Ecosystems of the World (MEOW), a biogeographic classification of coastal and continental shelf waters of the world, and taxonomic classification at the family level according to the Porifera Database and the NCBI taxonomy Database. (C) Ecological indexes in Antarctic and non-Antarctic sponge microbiomes, represented by observed richness, Simpson, and Shannon Furthermore, we searched for compositional differences between the microbiomes of Antarctic vs. non-Antarctic sponges at the phylum level. We identified a total of 61 phyla, among which Abditibacteriota was exclusively associated with the Antarctic sponges Haliclona (Gellius) sp. (sample ID E11), Myxilla (Burtonanchora) sp. (sample ID E4), and Kirkpatrickia variolosa (sample ID E8) (Supplementary Table S6). Interestingly, this phylum was also absent from the whole SMP dataset. The sponge microbiomes were primarily made up of Proteobacteria and Bacteroidota, independently of the habitat (Fig. 2). However, Antarctic sponges had significantly lower relative abundances of unassigned bacterial/archaeal phyla than non-Antarctic sponges (t-test, p = 4.898e-08). Chloroflexi and Cyanobacteria were only present in non-Antarctic sponges with relative abundances > 5%. Moreover, we observed differences in the relative abundance of particular phyla. For instance, Crenarchaeota (dominated by the class Nitrososphaeria) and Nitrospinota (dominated by the class Nitrospinia) were present with abundances > 5% in more Antarctic sponges than sponges from the other environments (Supplementary Table S7). Fig. 2Microbial composition at the phylum level contrasting Antarctic and non-Antarctic sponges. Proteobacteria was split at the class level. Relative abundances were calculated including Hellinger transformation. <5% abundance: phyla with less than 5% of abundance in each sponge; therefore, a phylum with more than 5% of abundance in some sponges could have less than this percentage in others. Sponge samples within each category are ordered by species Microbial composition at the phylum level contrasting Antarctic and non-Antarctic sponges. Proteobacteria was split at the class level. Relative abundances were calculated including Hellinger transformation. <5% abundance: phyla with less than 5% of abundance in each sponge; therefore, a phylum with more than 5% of abundance in some sponges could have less than this percentage in others. Sponge samples within each category are ordered by species The microbiomes of Antarctic sponges showed remarkably distinct composition compared to those of non-Antarctic sponges based on Bray-Curtis dissimilarities (ANOSIM, p = 0.001, R = 0.7068) and Unweighted UniFrac distances (ANOSIM, p = 0.001, R = 0.4761; PERMANOVA, Pseudo-F = 17.507 p = 0.001, r = 0.078) at the ASV level (Fig. 3A, Supplementary Table S8, and Supplementary Figure S1). We observed a significant effect of the sponge phylogeny at the family level in the microbial community composition, accounting for 22.9% of the variation (PERMANOVA, Pseudo-F = 4.158, p = 0.001, Supplementary Table S9 and S10). The habitat was responsible for 11% of the variation in the dissimilarities of the community composition (PERMANOVA, Pseudo-F = 30.079, p = 0.001, Supplementary Table S9). The interaction between the habitat and the sponge family contributed an additional 11.5% (PERMANOVA, Pseudo-F = 2.418, p = 0.001). The difference in dispersion between Antarctic and non-Antarctic sponge microbiomes also accounted for at least part of the variation in the community composition with both distances (PERMDISP, FBray−Curtis=86.147, FUnweighted UniFrac=45.591, p = 0.001). No clear compositional clustering pattern separating tropical and temperate sponges was observed at this level of resolution, and dispersion between these two sponge groups was similar. We identified the ASVs responsible for the observed differences in composition between habitats. Among the top 10 abundant ASVs, seven ASVs belong to Gammaproteobacteria, while the rest are Nitrospinia (LS-NOB only), Alphaproteobacteria (SAR116 clade only), and Nitrososphaeria (Candidatus Nitrosopumilus only) (Supplementary Table S11). A single ASV, annotated as Nitrosomonas oligotropha (order Burkholderiales), accounted for 7.7% of these differences. The remaining ASVs contributed up to 2%. Apart from the top 10, an ASV assigned to Synechococcus CC9902 (Cyanobiaceae) and two Polaribacter (Flavobacteriaceae) ASVs appeared in the 12th, 15th and 18th positions, respectively, with a 1% average contribution to the differences. Fig. 3Comparison between microbial community composition of Antarctic and non-Antarctic sponge microbiomes. (A) nMDS of community composition based on Bray Curtis dissimilarities at the ASV level. Ordination was performed using Hellinger-transformed abundances. (B) Linear discriminant analysis (LDA) of microbial genera in Antarctic and non-Antarctic sponges. Only genera that obtained an LDA score > 3 and a Wilcoxon-test p-value > 0.01 are displayed. (C) Occurrence of microorganisms in different numbers of sponges. The circle numbers represent the number of microorganisms found/shared in at least the percentage of sponges indicated beside each circle Comparison between microbial community composition of Antarctic and non-Antarctic sponge microbiomes. (A) nMDS of community composition based on Bray Curtis dissimilarities at the ASV level. Ordination was performed using Hellinger-transformed abundances. (B) Linear discriminant analysis (LDA) of microbial genera in Antarctic and non-Antarctic sponges. Only genera that obtained an LDA score > 3 and a Wilcoxon-test p-value > 0.01 are displayed. (C) Occurrence of microorganisms in different numbers of sponges. The circle numbers represent the number of microorganisms found/shared in at least the percentage of sponges indicated beside each circle We performed an LDA to identify the differences in the abundance of genera between Antarctic and non-Antarctic sponges (Fig. 3B). This analysis identified that 21 genera were differentially abundant between the two habitats. Among the Antarctic sponges, the most enriched genera correspond to Polaribacter, Candidatus Nitrosopumilus, Rubritalea, and two undetermined genera, one of the order Burkholderiales and the other of the class Gammaproteobacteria. In contrast, the number of enriched genera was lower in non-Antarctic sponges, and the most prominent were Synechococcus and an unidentified Bacteria, even at the phylum level. We focus on the number of ASVs that occur exclusively and consistently in Antarctic and non-Antarctic sponges (Fig. 3C). We did not identify any ASV present across all sponges. We observed a few ASVs in at least 70% of the sponges. Seven ASVs were exclusive of Antarctic sponges (corresponding to Polaribacter irgensii, Cochleicola gelatinilyticus, Owenweeksia hongkongensis, the SAR92 clade, the SAR116 clade, the NS9 marine group, and an undetermined Nitrincolaceae), one ASV was exclusive of non-Antarctic sponges (Synechococcus CC9902), and only one was shared between habitats (Candidatus Pelagibacter ubique). Most microorganisms from sponges are unique to each ecoregion Our sponge microbiome dataset comprised 12 marine ecoregions, two ecoregions of Antarctica, including the Antarctic Peninsula and the South Shetland Islands, and ten non-Antarctic ecoregions. The latter include the Adriatic Sea, Bahamian, Central and Southern Great Barrier Reef, Central New Zealand, Hawaii, South European Atlantic Shelf, Southern Norway, Southwestern Caribbean, Western Caribbean, and Western Mediterranean. We observed that the ecoregion exerted 24% of the variation in the microbial community composition based on compositional dissimilarities (Pseudo-F = 7.2384, p = 0.001). The ecoregion factor is as crucial as the phylogeny of the host in explaining the variation in the microbiome composition of sponges (Supplementary Table S12 and S13). Most bacterial/archaeal ASVs identified were exclusive to a particular marine ecoregion (Fig. 4), with a higher number of exclusive ASVs from the South Shetland Islands. The ASVs with the highest presence and abundance were those shared between sponges of the South Shetland Islands and the Antarctic Peninsula, rather than those exclusive to each Antarctic ecoregion. Exclusive ASVs of sponges of the Antarctic Peninsula presented only low abundances and presence across sponges. Surprisingly, the sponges of the South Shetland Islands shared more ASVs with Hawaiian sponges than with others, accounting for over three-quarters of the ASVs shared between Antarctic and tropical sponges, though being rare taxa (< 0.05% relative abundance). Moreover, ASVs shared between sponges of the South Shetland Islands and the South European Atlantic Shelf accounted for almost half of the ASVs shared between Antarctic and temperate sponges (Fig. 4). Fig. 4Exclusive and shared microbial ASVs between sponges of different ecoregions. Microbial ASVs that are unique to sponges of specific ecoregions and those common among sponges from diverse ecoregions using a criterion of shared ASVs if they were present in at least one sponge belonging to a different ecoregion. The horizontal bars indicate the sum of ASVs in sponges for each ecoregion, and the colors represent tropical, temperate, and Antarctic sponges from top to bottom. The number next to each ecoregion name indicates the sponge sample size. Only the 25 first intersects are displayed to favor visualization. The upper panel represents the abundance average and occurrence of ASVs in sponges and highlights in colors those ASVs that are unique and shared between each Antarctic ecoregion Exclusive and shared microbial ASVs between sponges of different ecoregions. Microbial ASVs that are unique to sponges of specific ecoregions and those common among sponges from diverse ecoregions using a criterion of shared ASVs if they were present in at least one sponge belonging to a different ecoregion. The horizontal bars indicate the sum of ASVs in sponges for each ecoregion, and the colors represent tropical, temperate, and Antarctic sponges from top to bottom. The number next to each ecoregion name indicates the sponge sample size. Only the 25 first intersects are displayed to favor visualization. The upper panel represents the abundance average and occurrence of ASVs in sponges and highlights in colors those ASVs that are unique and shared between each Antarctic ecoregion Antarctic sponge microbiomes exhibit habitat-specificity Habitat-specific microorganisms from Antarctic sponges were those ASVs found exclusively in this habitat in at least 50% of the Antarctic sponges. We found 13 bacterial ASVs fitting into these criteria, classified into Burkholderiales, Puniceispirillales, Pseudomonadales, Flavobacteriales, Thiotrichales, and Rhodospirillales (Fig. 5 and Supplementary Table S14). These ASVs reached between 0.01% and 97% of the total microbiome abundances, where the ASVs of N. oligotropha and SAR116 clade presented inflated abundances in Mycale spp. and Dendrilla antarctica, respectively (Supplementary Table S15). Habitat-specific microorganisms of the non-Antarctic sponges (ASVs found exclusively in temperate and tropical sponge microbiomes in at least 50% of the sponges) correspond to five bacterial ASVs which reached between 0.005% and 31% of the total microbiome abundances (Fig. 5 and Supplementary Table S14). The main difference of this habitat-specific group compared with Antarctic sponges is the presence of two ASVs of Cyanobacteriia assigned to Cyanobium gracile and Synechococcus strain CC9902. Fig. 5Abundance and taxonomy patterns of habitat-specific and habitat-generalist bacteria in Antarctic and non-Antarctic sponges. The dots in boxplots represent the sum of abundances of habitat-specific or habitat-generalist ASVs in each sponge. The pie charts display the average relative abundance of these microorganisms. All ASVs were assigned to the lowest taxonomy level (i.e., species, genus, or family) according to the Silva and NCBI databases. Asterisks indicate the highest-scored species hit according to the NCBI 16S rRNA database (unassigned species when using the Silva database) Abundance and taxonomy patterns of habitat-specific and habitat-generalist bacteria in Antarctic and non-Antarctic sponges. The dots in boxplots represent the sum of abundances of habitat-specific or habitat-generalist ASVs in each sponge. The pie charts display the average relative abundance of these microorganisms. All ASVs were assigned to the lowest taxonomy level (i.e., species, genus, or family) according to the Silva and NCBI databases. Asterisks indicate the highest-scored species hit according to the NCBI 16S rRNA database (unassigned species when using the Silva database) We identified 11 bacterial ASVs as habitat-generalist microorganisms, classified into Rhodobacterales, Flavobacteriales, Pseudomonadales, the SAR11 clade, and Burkholderiales (Fig. 5). These microorganisms reached between 0.3% and 64% of the total abundances in Antarctic sponges, compared to 0.01–20% in non-Antarctic sponges. Two ASVs belonging to Sulfitobacter undariae (family Rhodobacteraceae, Alphaproteobacteria) and Polaribacter sp. showed the highest relative abundances across Antarctic sponges. In contrast, in non-Antarctic sponges, the highest corresponded to an ASV of the methylotrophic OM43 clade (Gammaproteobacteria). Potential keystone microorganisms in sponges are both habitat-specific and habitat-generalist bacteria. Analysis of microbial co-occurrences within sponge microbiomes from the two habitats indicated that the number of nodes (microbial ASVs), edges (abundance correlations among them), and network properties varied for each habitat (Fig. 6 and Supplementary Table S16). Interestingly, the Antarctic sponge network showed a lower clustering coefficient and the highest number of microorganisms and connections, most of which were positive. Mutual exclusions between some microorganisms (i.e., inverse correlations in the abundance of pairs of ASVs) were only 12 out of the 503 edges. Fig. 6Microbial co-occurrence networks for Antarctic and non-Antarctic sponges. Nodes represent microbial ASVs, and edges represent connections between ASVs abundances based on significant correlations (r > 0.6). Node colors indicate the taxonomy assignments at the phylum level. Node size is equivalent to the node degree (number of connections of each node). Node shapes indicate the type of microbe according to their exclusiveness and occurrence across sponges. Edge colors represent direct or inverse correlations between pairs of nodes Microbial co-occurrence networks for Antarctic and non-Antarctic sponges. Nodes represent microbial ASVs, and edges represent connections between ASVs abundances based on significant correlations (r > 0.6). Node colors indicate the taxonomy assignments at the phylum level. Node size is equivalent to the node degree (number of connections of each node). Node shapes indicate the type of microbe according to their exclusiveness and occurrence across sponges. Edge colors represent direct or inverse correlations between pairs of nodes All the habitat-specific and habitat-generalist microorganisms appeared in the Antarctic sponge network. In contrast, in the non-Antarctic network, four of the five habitat-specific microorganisms and only two of the eleven habitat generalists appeared (Fig. 6). In each network, a subgroup of habitat-specific and habitat-generalist microorganisms are within the top 10 abundant ASVs with the highest number of connections, i.e., with the highest degree (Supplementary Table S17). These subgroups of ASVs are also superabundant, likely representing keystone microorganisms in sponges of the different habitats. Discussion Diversity and community composition of Antarctic sponge microbiomes Studies on the diversity and composition of microbial communities associated with sponges have been restricted to tropical and temperate zones. Here, we performed a comprehensive analysis of the microbiome of 179 sponges using a 16S rRNA gene-based approach, to elucidate specific patterns in the ecology of microorganisms associated with Antarctic sponges that differentiate them from tropical and temperate counterparts. The microbiomes of Antarctic sponges differed from tropical and temperate sponges in diversity. These differences accounted for rare and abundant microorganisms, as we used the Shannon and Simpson indexes to represent diversity better. The Simpson index is more sensitive to abundant species, accounting better for relative abundance and microbial evenness [51]. Meanwhile, the Shannon index is equally sensitive to rare and abundant species, accounting better for microbial richness [51]. Our results showed that uneven microbial relative abundances in Antarctic sponges, instead of the high variability in their microbial richness, substantially impact observing lower microbial diversity against non-Antarctic sponges. We suggest that the extreme cold and seasonal changes in Antarctica may limit the diversity of microbial communities in sponges, leading to unevenness and specialization. We detected 61 different microbial phyla across all microbiomes, encompassing almost the total richness of phyla (63 phyla) reported in the SMP dataset [35]. Interestingly, we found that various taxonomic groups were unique to Antarctic sponges. For instance, members of the underexplored phylum Abditibacteriota, first isolated from ice-free Antarctic soil and formerly known as candidate phylum FBP [52], were identified in association with Antarctic sponges and absent from tropical and temperate sponges sampled in the SMP. Abditibacterium is the unique genus within Abditibacteriota, characterized as an oligotrophic chemoheterotrophic bacterium thriving in extreme ecosystems beyond cold ones [52–54]. Our results evidenced that Antarctic and non-Antarctic sponges also differed in their microbial community composition. Previous studies reported that Antarctic sponge microbiomes are highly similar, but with less similarity to tropical and temperate sponges [16]. On the other hand, 13% of the bacterial and archaeal OTUs of the sponges from the Antarctic Peninsula lacked a closest relative in the SMP database [17], suggesting a potential effect of the environment in shaping the Antarctic sponge microbiome. A recent study also found distinct microbiome diversity and taxonomic features between Antarctic and South American sponges [55]. Drivers shaping microbial community assembly in Antarctic sponges HMA and LMA status influence the microbial diversity and composition of sponge microbiomes [2, 13, 56–58]. The dominance of Proteobacteria and Bacteroidota, and the low abundance of Chloroflexi, Poribacteria, Acidobacteriota, and Gemmatimonadota in all the Antarctic sponges analyzed here suggests they belong to the LMA sponges [2, 13, 56–58], however quantitative microbial abundance measurements are needed to confirm this hint. Moreover, most of the temperate and tropical sponges analyzed here correspond to LMA. Usually, LMA sponges present lower richness and evenness than HMA sponges and higher variation in the microbial composition between host species [13]. Our results showed higher richness in Antarctic sponges but lower evenness and less variation in the microbial composition between distantly related species, compared to the LMA sponges from tropical and temperate environments. Although some microbial patterns define LMA sponges overall, we evidence here that LMA Antarctic sponges display a more distinctive microbiome. The sponge phylogeny also significantly influenced the variation in the sponge microbiome composition, as previously reported [6, 59, 60]. Patterns of co-evolution between the sponge and its microbiome have been characterized for some host species before [61]. Nonetheless, the evolution of sponge microbiomes is shaped by a complex blend of transmission of microorganisms across generations and acquisition of microorganisms from the marine environment, explaining community patterns that are variable even within the same host species (depending on the sponge species) [4, 10, 12, 18, 62]. The influence of the Antarctic environment (i.e., their extreme conditions and geographical isolation) could select microorganisms with specific characteristics and strongly contribute to shaping more similar microbiomes in the Antarctic-sponge habitat [20]. These microbiome similarities in sponges of different phylogeny may be sustained through time either by the vertical transmission of common microbial communities acquired at some point or by the host selection of common microorganisms from the Antarctic environment in each new sponge generation. Marine ecoregions delimit areas defined by biogeographic forcing agents, including isolation, upwelling, nutrient inputs, freshwater influx, temperature regimes, ice regimes, exposure, sediments, currents, and bathymetric or coastal complexity [44]. In our results we did not observe a temperature gradient distinguishing the microbiome composition from tropical and temperate sponges. Instead, comparing sponge microbiomes at distinct marine ecoregions, differences in the microbiome composition may be more related to geographic isolation, dispersal constraints, and environmental conditions other than temperature, along with the effect of the sponge phylogeny. Another study also reported a significant effect of the geographic region on microbial community composition of some tropical and temperate sponges [2]. Therefore, the differences observed in the composition between sponge microbiomes are related not only to the local environment, as postulated by the Baas Becking hypothesis [30, 31], but also to geographic patterns of the sponge-habitat. Notably, the sponge species are known to exert the most relevant effect on the diversity and community composition of sponge microbiomes [1, 10], leading to host specificity. The genetic differentiation and population structure of sponge species also influence the composition of their microbiomes, with host-specificity being detected at the genotypic level [4]. We suspect a greater amount of the residual variation in microbial community composition that our model could not explain might correspond to the sponge species specificity. Thereby, we suggest that a combination of host-related, environmental, and biogeographic factors all play a role in shaping the microbial community composition in Antarctic sponges, highlighting the complex nature of symbiotic relationships in these holobionts. Besides these deterministic drivers, we do not discard that stochastic drivers may also contribute to this complexity. We recognize a limitation of this study is a potential primer bias; however, including the primer set as a factor in the statistical analyses showed no significant effect on the microbial community composition, and differences observed between habitats in microbial diversity were almost the same (data not shown). Microbial signatures of antarctic sponges Nitrosomonas oligotropha, Candidatus Nitrosopumilus, SAR 116 clade, and Polaribacter influenced the compositional variation between Antarctic and non-Antarctic sponges. Nitrosomonas oligotropha, as an obligate chemolithoautotroph, uses ammonia as an energy source and carbon dioxide as a carbon source in the presence of oxygen [63]. A single N. oligotropha ASV became particularly dominant in the Antarctic Mycale spp., similar to a previous report of a single dominant OTU but classified as EC94 family [64]. On the other side, using metagenomic approaches, Candidatus Nitrosopumilus was reported to be abundant in the Antarctic sponges Myxilla sp., Megaciella aff. annectens and Leucetta antarctica and involved in ammonia oxidation, nitrite oxidation, and denitrification [19, 65]. SAR116, a ubiquitous clade in ocean waters, became here particularly abundant in Dendrilla antarctica, agreeing that this sponge species seems to acquire symbionts horizontally from the surrounding seawater [4]. Polaribacter is a gas vacuolate bacteria first isolated from polar marine waters and ice sea [66]. However, 16S rRNA representative sequences of Polaribacter have been found in other environments/habitats elsewhere (e.g., seawater, soil, sediment, groundwater, freshwater, hydrocarbon, and coral) registered in MetaMetaDB [67]. Cold shock and ice-binding proteins have been reported and characterized in Polaribacter irgensii [68–71]. Some authors refer to this species as a psychrophilic bacterium. However, in this study Polaribacter is also present in tropical and temperate sponges suggesting a psychrotolerant lifestyle. The enrichment of these microbial members in Antarctic sponges may indicate a leading role in the sponge holobiont metabolism. Future studies aiming to reveal their functional roles within microbiomes are still needed. Several studies have proposed that Antarctic habitats, such as soil and lake water, comprise an assemblage of habitat-generalist and habitat-specific microorganisms [72, 73]. In concordance, we identified thirteen habitat-specific and eleven habitat-generalist bacteria with a high abundance across the Antarctic sponges analyzed. Habitat generalists showed a higher abundance in the Antarctic sponges, while this pattern was inverse in non-Antarctic sponges. The dominance of habitat-generalist bacteria in Antarctic sponges over habitat-specific bacteria, which were more abundant in non-Antarctic sponges, likely reflects a combination of the specific ecological roles and adaptations and the evolutionary history of the host-microbe interactions in these environments. The unpredictable and remote Antarctic environment may favor generalist bacteria that can survive in a wide range of conditions over habitat-specific bacteria that thrive in more stable environments [17, 74, 75]. Habitat generalist bacteria might be better adapted to the oligotrophic Antarctic waters, as they often possess a wide array of metabolic capabilities that allow them to utilize diverse energy sources, exploit a broader range of resources and niches, and survive in variable conditions [17, 74, 75]. This versatility of the habitat-generalists could be more advantageous in Antarctic sponges. Meanwhile, the specialized adaptations and often vulnerability of the habitat-specific bacteria under adverse conditions could lead to a higher risk of extinction [24], which is alarming considering the fast warming rates in the Antarctic Peninsula [15]. Thus, habitat-specific bacteria could serve as health indicators of Antarctic sponge holobionts, and habitat-generalist bacteria could aim to achieve ecosystem recovery and stability in future scenarios. Implications of microbial co-occurrence in antarctic sponges Microbial co-occurrence in sponges differed between habitats, mainly in the number of microorganisms whose abundance is significantly associated with others, which is related to the complexity of microbiome structures. Microbial co-occurrences (connections between microbes in networks) partially explain ecological interactions between microbes (i.e., mutualism, competition, commensalism, and amensalism) [10] and sometimes involve similar responses to common environmental conditions [76, 77]. Therefore, the differences in the microbial co-occurrence between Antarctic and non-Antarctic sponges reflect different structuring dynamics related to the environment and, likely, different microbial interactions [10]. A mixture of positive and negative interactions, mainly amensalism and commensalism, distinguishes the core microbiome of tropical and temperate sponge species [10]. When we analyzed microbiomes independently of the sponge species, positive microbial connections prevailed in Antarctic sponges with only some negative connections, while only positive connections occurred in non-Antarctic sponges. This fact may reflect positive microbial interactions or niche partitioning within sponge microbiomes. The highly “connected” microbes (sharing similar co-occurrence patterns with other microbes) have been suggested to perform an essential role within microbial communities [78]. Therefore, the habitat-specific and generalist bacteria recognized here for maintaining associations with many other microbiome members might represent keystone microorganisms performing relevant roles within Antarctic and non-Antarctic sponges. Conclusions Here, we compared the diversity, community composition, and microbial co-occurrence of the sponge microbiomes from Antarctica and other environments. Our results evidenced habitat specificity in the Antarctic sponge microbiomes where habitat-specific microorganisms are more numerous but less abundant than habitat generalists, while both are highly connected. Antarctic sponges of several species represent a reservoir of unique microbial assemblies different from tropical and temperate sponges. The particular environment of Antarctica and its geographic isolation contribute to these microbiome structural signatures in sponges. However, other deterministic drivers also shape the microbiome structure in sponges, such as the sponge phylogeny and interactions between microorganisms. Sponge microbiome research must be expanded and specialized to the Antarctic environment, especially concerning future environmental change scenarios. Further research on the functional profiles of Antarctic sponge microbiomes will brighten the habitat specificity patterns evidenced here and the adaptation mechanisms that likely contribute to the success of sponge holobionts in the extreme Antarctic environment. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2