The Palk Bay reef extends approximately 200–600 m in width and occurs at shallow depths of 1–5 m, located 1–4 km offshore (Figure 1). The reef system is not designated as a marine protected area and is subject to multiple anthropogenic stressors, including effluent discharge from coastal industries (e.g., fish processing units and ice factories) and intensive fishing activities such as trap fishing (Asir Ramesh et al., 1996; Karuppanapandian et al., 2007). Live coral cover in Palk Bay was estimated at 37.8% in 2008, with the remaining benthos largely comprising seagrass meadows and sandy substrates (Sukumaran et al., 2008). Major species found in this region were Favites, Favia, Platygyra, Astreopora, Turbinaria, Siderastrea, Hydnophora, Symphyllia, Montastrea and Syphastrea sp (Marimuthu et al., 2016). Subsequent surveys documented a pronounced decline in coral cover in the Mandapam North region, decreasing to 22.62% in 2013 and further to 3.41% by 2015 (Edward et al., 2015; Thinesh et al., 2017a, b), within longitudes 79°17′40″–79°08′ E and latitude ~9°17′ N. All biological samples were collected from the same shallow reef zone (reef flat to upper reef slope, 1–5 m depth) to minimize potential effects of depth-related zonation on microbial community composition.

Despite the decline in adult coral cover, recent field surveys at the study site recorded high densities of coral recruits on CCA, indicating ongoing recruitment activity (unpublished observations; Figure 1). CCA in the vicinity of the sampling locations predominantly exhibited a fruticose growth form and were identified as Lithophyllum kotschyanum following the identification guide of Perry (2005), consistent with the first regional record by Sreeraj et al. (2018). Published reports and field observations indicate that Lithophyllum spp. constitute the most abundant and spatially dominant CCA taxa in Palk Bay reefs, occurring at higher cover than other genera such as Porolithon spp. and Sporolithon spp.

Accordingly, L. kotschyanum was selected as the focal CCA species because it represents the dominant and ecologically representative CCA substrate in the region. Field observations suggest a frequent association between Lithophyllum-dominated CCA surfaces and coral juveniles, indicating a potential role in facilitating coral settlement and early recruitment. Restricting sampling to a single, dominant CCA species minimized taxonomic variability among CCA substrates and enabled clearer comparisons of microbiome structure across reef substrates, consistent with the aims of this study to assess substrate-driven patterns in reef-associated microbial communities.

During a coral reef survey conducted in Palk Bay in 2022, we observed a high density of coral recruits in close association with CCA at a shallow fringing reef (2–3 m depth). Based on this field observation, we speculated that CCA-associated microbial communities may differ from those inhabiting other reef substrates and could potentially influence early coral settlement processes. This observation formed the basis for the present study.

A representative reef-flat site (~200 m²) was selected where live corals, CCA, coral rubble, sediment, and seawater co-occurred, enabling comparative assessment of substrate-associated microbial communities under similar environmental conditions. Sampling was conducted in June 2023 across these reef niches.

At the time of sampling, basic water-column parameters were recorded in situ to characterize the ambient reef environment. Seawater pH was ~8.1, salinity ranged between 35 and 37 ppt, and the average sea surface temperature was 29.8°C. These measurements indicate typical warm, fully marine conditions for shallow fringing reefs in Palk Bay and provide environmental context for the microbial community analyses. No additional nutrient or dissolved oxygen measurements were undertaken as the study primarily focused on substrate-associated microbial assemblages rather than water-column biogeochemistry and also absolute microbial abundances (e.g., cell counts by microscopy) were not quantified. Hereafter, ‘n’ refers exclusively to the number of individual fragments collected per colony for coral and CCA samples, whereas ‘N’ denotes the number of independent biological replicates included in the sequencing-based microbial community analysis. For sediment, rubble, and seawater, samples were collected and processed directly at the replicate level and are therefore represented only by ‘N’.

Coral samples were collected from four genera - Acropora, Favia, Favites, and Symphyllia, as these taxa represent the major growth forms and reef-building groups occurring across Palk Bay and frequently co-occur with CCA in reef-flat and shallow-slope habitats. Their broad spatial distribution and ecological connectivity with CCA made them appropriate model hosts for examining potential microbial community sharing between corals and coralline algae. For each genus, three independent colonies were sampled, and five small fragments (~200 mg each; n = 5) were collected per colony to minimize colony-level variability, yielding 15 samples per genus (total coral samples = 60). We selected the above-mentioned colonies as they are dominant coral genera CCA samples were collected following the same strategy, with three colonies sampled and five fragments (~200 mg each; n = 5) collected per colony (total CCA fragments = 15). Sediment samples (~1 g per replicate; N = 2) and coral rubble samples (N = 2) were collected directly as independent biological replicates. For seawater, approximately 1 L of surrounding reef water was collected per replicate and filtered through sterile 0.2 µm membrane filters, with two independent replicates (N = 2).

All coral and CCA fragments were gently rinsed onboard with 0.2 µm filtered seawater to remove loosely associated microbes, photographed, and preserved in sterile DMSO-saturated salt (DESS) buffer for DNA preservation. Samples were transported to the laboratory at 4°C. Upon arrival, samples were removed from DESS, rinsed with sterile seawater to remove excess salts, and stored at −20°C until further processing.

DNA was extracted from all samples using a uniform protocol using the DNeasy PowerSoil DNA Isolation Kit (Qiagen, Germany) following the manufacturer’s instructions. Prior to extraction, solid samples were crushed in liquid nitrogen using a sterile mortar and pestle with 2 mL of sterile water. For seawater samples, the 0.2 µm membrane filters (Millipore) were cut into small pieces prior to extraction. For coral and CCA samples, 100 µL of the supernatant obtained after crushing was used for DNA extraction. As this study aimed to characterize overall, community-level microbial diversity and variability across reef substrates, a fragment pooling strategy was applied to coral and CCA samples following Ray et al. (2019). DNA extracted from the five fragments collected per colony (n = 5) was pooled to generate a single composite sample per colony. This approach was implemented to capture representative colony-level microbial community structure, integrate within-colony heterogeneity, and reduce potential biases associated with fragment-level sample size. Whereas, sediment, rubble, and seawater samples were not pooled, and each replicate (N) was processed independently for DNA extraction and downstream analyses.

In total, 21 samples were processed for bacterial 16S rRNA gene amplicon sequencing: CCA (N = 3; one pooled sample per colony), four coral genera (Acropora, Favia, Favites, and Symphyllia; N = 3 per genus), coral rubble (N = 2), seawater (N = 2), and sediment (N = 2). Bacterial community profiling followed the protocol of Rodriguez-Lanetty et al. (2013). The V₃-V₄ hypervariable regions of the bacterial 16S rRNA gene (~460 bp) were amplified using primers 341F (5′-CCTACGGGNGGCWGCA-3′) and 805R (5′-GACTACHVGGGTATCTAATCC-3′), each containing Illumina overhang adapters. PCR amplification consisted of an initial denaturation step at 95°C for 3 min, followed by 30 cycles of denaturation at 95°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 30 s, with a final extension at 72°C for 5 min. PCR products were purified using AMPure XP magnetic beads (Beckman Coulter, Brea, CA, USA). Five microliters of purified amplicons were used for index PCR with DIY Illumina indices according to the manufacturer’s protocol. Indexed amplicons were again purified using AMPure XP beads and quantified using a FLUOstar Omega microplate reader (BMG Labtech, Germany) with the Quant-iT PicoGreen dsDNA assay kit (Invitrogen, USA). The quality and size distribution of the final pooled library were assessed using a 1% agarose gel and an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA).

Single-end reads were processed using the QIIME 2 pipeline (Bolyen et al., 2019) after the reverse reads were excluded due to poor quality. Quality filtering and primer trimming were performed, and overlapping reads were merged. DADA2 was used for denoising and ASV (Amplicon Sequence Variant) inference following the standard QIIME 2 workflow (Callahan et al., 2016). Taxonomic assignment of 16S rRNA ASVs was conducted using the SILVA database (release 138) in conjunction with the QIIME 2–implemented RDP classifier (Bolyen et al., 2019). The initial ASV table was pre-processed prior to downstream analyses. Potential contaminant ASVs were identified based on their prevalence in negative controls and removed manually. Additionally, sequences unclassified at the kingdom level and non-bacterial reads (Mitochondria, Chloroplasts, Archaea, and Eukarya) were filtered out using QIIME 2’s filtering tools. ASVs with fewer than 20 total reads were excluded to minimize potential sequencing artifacts (Callahan et al., 2016). The resulting curated dataset was used for all subsequent analyses unless otherwise specified. Alpha diversity indices, including Chao1 richness, Simpson’s evenness, Shannon diversity, and phylogenetic diversity, were computed within QIIME 2.

Although the median sequencing depth across samples was 64,770 reads, rarefaction to 3,000 reads per sample was applied only for alpha diversity (Faith’s PD and Pielou’s evenness) (Faith, 1992) and UniFrac-based beta diversity metrics to standardize sampling effort for phylogeny-based analyses in QIIME 2 (Pielou, 1969; Bolyen et al., 2019). This depth was selected as the minimum that retained the majority of samples and captured near-saturation in rarefaction curves (Supplementary Figure S1). Rarefaction curves indicated that ASV richness approached saturation for most samples despite heterogeneous sequencing depths. All other analyses of community composition were performed on relative abundance data without rarefaction, preserving the full sequencing information. Seawater, sediment, and rubble samples (N = 2 each) were combined as a single Environmental group, while corals and CCA were treated as independent host-associated groups. This grouping enabled meaningful statistical comparisons between host-associated and environmental microbiome.

For R-based analyses, raw counts were converted to per-sample relative abundances, and Bray–Curtis dissimilarities were calculated for Principal Coordinates Analysis (PCoA) using phyloseq and ape. Differences among groups were tested with PERMANOVA (vegan::adonis2; 999 permutations; formula: Bray–Curtis ~ group), and homogeneity of multivariate dispersion was assessed with betadisper followed by Tukey’s HSD. Taxa contributing to between-group dissimilarity were identified using SIMPER at the genus level (999 permutations), retaining taxa explaining up to 70% of cumulative dissimilarity.

Shared and unique microbial assemblages were visualized with UpSet analysis, where a genus was considered present in a sample at ≥0.05% relative abundance and in a group if detected in ≥20% of its samples. Core microbial sets (e.g., all coral samples, all non-coral host-associated samples, or all samples combined) were defined using the same criteria. Unless otherwise stated, statistical tests were two-sided with α = 0.05 and Benjamini–Hochberg FDR correction. Analyses were conducted in R v4.2.1 using phyloseq, vegan, UpSetR, and ggplot2 (R Core Team, 2025).

Co-occurrence patterns were inferred using all 21 samples pooled across substrates. ASVs were agglomerated to the genus level (SILVA-138), converted to relative abundances, and only fully classified genera were retained after removing placeholder taxa (e.g., “uncultured/unknown/norank”, “group/clade/bacterium”, NB1-j/SAR324/PAUC34f, and non-Latin tokens). Genera present in at least 10% of samples and among the top 20 by total abundance were selected for network analysis.

As described above, seawater, sediment, and rubble samples (N = 2 each) were grouped as the Environmental group, while corals and CCA were treated as independent host-associated groups. Pairwise Spearman correlations were calculated on a pseudocount-adjusted matrix, with significance corrected using the Benjamini–Hochberg FDR. Network edges were drawn for |ρ| ≥ 0.5 and q < 0.05 (undirected), with edge width proportional to correlation strength and sign indicating positive or negative associations. Node fill represents the substrate group in which a genus is most abundant (dominant group), node size scales with its mean abundance within that group, and labels are italicized. Co-occurrence reflects cross-sample co-variation rather than direct ecological interactions. Given the limited sample size of the study groups, these networks are treated as exploratory and interpreted cautiously.

Bacterial 16S rRNA gene amplicon sequencing was performed on samples representing coral tissue (Acropora, Favia, Favites, Symphyllia), Coralline algae, Rubbles, Sediment and Water (N = 21). This generated a total of 22,410 ASVs across all samples. Sequencing depth per sample ranged from 30,865 to 81,920 reads, with a median of 64,770 reads, ensuring adequate coverage of the microbial communities. Analyses used forward reads only. Sequence length ranged from 232 to 251 bp (median = 251 bp). After quality filtering, including the removal of unclassified taxa and low-abundance ASVs, a final dataset suitable for downstream ecological analysis was obtained. Taxonomic assignment successfully classified sequences across seven hierarchical levels (Kingdom to Species), providing robust resolution for community analysis. The bacterial community across all substrates was dominated by Proteobacteria (56.6%) and Bacteroidota (10.9%). Other phyla were present at moderate levels, including Chloroflexi (4.8%), Desulfobacterota (3.7%), Acidobacteriota (3.4%), Actinobacteriota (2.8%), Gemmatimonadota (2.4%), Firmicutes (2.4%), Cyanobacteria (1.9%), Myxococcota (1.5%), Planctomycetota (1.4%), and Nitrospirota (1.3%) (Figure 2). Collectively, several rare phyla each contributed less than 1% to the overall community, together accounting for only a minor fraction of the total bacterial assemblage. Notably, minor phyla together accounted for less than 1% of the community, including Spirochaetota, Dadabacteria, NB1-j, Bdellovibrionota, Entotheonellaeota, Campilobacterota and Nitrospinota, each occurring in small proportions (0.2–0.8%), thereby reflecting a broad spectrum of microbial diversity within the samples. After filtering unclassified taxa, a total of 562 distinct bacterial genera were identified. The most abundant genera were Roseospira (3.6%), Ralstonia (2.1%), and Pelagibius (1.5%), followed by Nitrospira (1.2%), Ruegeria (1.1%), Paenibacillus (0.9%), Woeseia (0.8%), and Pseudomonas (0.8%). Additional genera, such as Labrenzia, Chlorobium, Stenotrophomonas, and Constrictibacter, each contributed between 0.7% and 0.9% of the total community. Several other genera, including Albidovulum, Vibrio, Desulfovibrio, Desulfocella, Algisphaera, Marinifilum, and Bacteroides, were also represented at lower relative abundances (0.3–0.4%), highlighting a broad spectrum of genus-level diversity across the samples (Figure 3).

Alpha diversity, measured using the Shannon index, was broadly comparable across the six substrate groups comprising four coral genera (Acropora, Favia, Favites, and Symphyllia), crustose coralline algae (CCA), and pooled Environmental samples (sediment, rubble, and seawater) (Figure 4A). A Kruskal–Wallis test did not detect strong overall differences in Shannon diversity among substrates, and pairwise Wilcoxon rank-sum tests with Benjamini–Hochberg correction similarly showed no significant differences between individual substrate pairs (all adjusted p-values > 0.05). Consistent with these results, Shannon diversity distributions showed substantial overlap among coral-associated, CCA, and Environmental microbiomes, indicating comparable levels of microbial richness and evenness across substrates.

Beta diversity analyses based on Bray–Curtis dissimilarities indicated substrate-associated structuring of microbial community composition. PERMANOVA indicated that substrate type was associated with differences in bacterial assemblages (F = 5.35, R² = 0.742, p = 0.001), with substrate identity explaining approximately 74% of the observed variation. Tests of multivariate dispersion confirmed that within-group variability did not differ significantly among substrates (betadisper: F = 0.61, p = 0.74; Tukey’s HSD: all adjusted p-values > 0.73), indicating that the PERMANOVA result was not driven by heterogeneity of dispersion. Principal Coordinate Analysis (PCoA) based on Bray–Curtis distances further supported these findings, showing partial separation of samples according to substrate type, with some overlap among groups (Figure 4B). Pairwise PERMANOVA comparisons suggested that microbial communities associated with crustose coralline algae differed from those of other substrates, with relatively high effect sizes (R² = 0.56–0.66); however, these pairwise differences did not remain statistically significant after Benjamini–Hochberg correction (all adjusted p-values > 0.05). This likely reflects limited statistical power within individual host-associated groups (N = 3 for each coral genus and CCA) and the inherent heterogeneity of the pooled Environmental category, which comprises distinct substrate types (rubble, sediment, and seawater; each N = 2).

SIMPER analysis was used to identify the genera driving compositional differences among substrates. In the comparison between coralline algae and Acropora, Roseospira was the dominant contributor, accounting for approximately 24% of the dissimilarity (p = 0.016), followed by Pelagibacter (~9.5%), Ruegeria (~7.5%), and Labrenzia (~7.1%). This highlights a strong and substrate-specific association of Roseospira with coralline algae. In contrast, differences between Acropora and sediment were primarily driven by Woeseia (~6.3%, p = 0.003), along with contributions from Vibrio and Bacteroides. Comparisons among coral taxa (e.g., Acropora vs. Favia or Symphyllia) were characterized by more evenly distributed contributions across several genera, with no single taxon dominating. These findings show that while overall microbial communities differed significantly among substrates (PERMANOVA), a small number of genera — particularly Roseospira in coralline algae and Woeseia in sediment — were major contributors to these dissimilarities.

Differential abundance analysis using LEfSe highlighted genus-level bacterial biomarkers distinguishing substrate types (Figure 5). Coralline algae-associated communities were enriched in Labrenzia, Poribacter, Oceanicaulis, Ruegeria, Roseibacillus, and Desulfovibrio, consistent with SIMPER results indicating Rhodobacteraceae-related taxa are distinctive to coralline surfaces. Coral tissues were enriched in Dadabacteriales, Desulfobacter, Albidovulum, Aquimarina, and Collinsella, whereas environmental samples (sediment, rubble, and seawater) were dominated by opportunistic and generalist taxa such as Paenibacillus, Vibrio, and Terasakiella. These findings reinforce that while overall bacterial diversity was broadly similar across substrates, community composition is substrate-specific, with coralline algae hosting a particularly unique bacterial assemblage.

UpSet analysis revealed distinct patterns of shared and unique bacterial genera among substrates (Figure 6). Environmental samples harbored the largest number of unique genera (182), while Acropora corals contained fewer (25). Favia (60) and CCA (79) exhibited relatively higher numbers of unique genera, whereas Symphyllia and Favites had none. Core microbiomes were also evident: 39 genera were shared across all coral hosts, and three were common to non-coral substrates (Haliangium, Pelagibius, Vibrio). Pelagibius was the only genus detected in every substrate type. Notably, Roseospira, although not identified as a unique genus in the UpSet analysis, was consistently present in all coralline algae and rubble samples (100% prevalence) but absent from coral tissues, sediment, and water. This pattern suggests a strong ecological association with hard benthic substrates, in contrast to genera such as Pelagibacter and Woeseia, which were broadly distributed across multiple habitats.

We inferred a genus-level co-occurrence network across all 21 samples (pooled data; Figure 7), with edges representing significant cross-sample correlations (Spearman |ρ| ≥ 0.5, BH-FDR q < 0.05). The network is dominated by positive associations, forming a compact core, while negative correlations are relatively sparse. CCA biomarkers identified by LefSe - Rhodobacteraceae genera Roseospira, Labrenzia, and Ruegeria - occupy a well-connected region and exhibit multiple positive co-variations with benthic-associated lineages characteristic of the Environmental group (rubble, sediment, and water; e.g., Nitrospira, Woeseia, Desulfovibrio, Bacteroides) as well as broader generalists (e.g., Acinetobacter, Pseudomonas). Negative links are concentrated among opportunistic taxa, notably Vibrio and occasionally Shewanella, which show inverse relationships with several CCA-linked and Environmental genera, consistent with competitive replacement or stress-related shifts.

These network patterns align with the PCoA and LEfSe results: beta diversity analyses indicated substrate-associated structuring of bacterial communities, with coralline algae partially separated from other substrates. LEfSe enrichment highlighted distinctive CCA-associated taxa such as Labrenzia, Ruegeria, Roseibacillus, and Desulfovibrio. Coral tissues were enriched in genera including Dadabacteriales, Desulfobacter, and Aquimarina, while the Environmental group was dominated by generalist and opportunistic taxa such as Paenibacillus, Vibrio, and Terasakiella. Overall, these observations indicate that CCA microbiomes are compositionally distinct yet connected within a broader benthic microbial network. Co-occurrence patterns reflect statistical co-variation among taxa rather than direct interactions.

In our study, substrate identity accounted for 74% of the variation in bacterial community composition, highlighting microhabitat as a primary driver of microbial assembly on coral reefs. Bacterial communities differed among corals, CCA, and the pooled environmental group (rubble, sediment, and seawater), underscoring the influence of substrate type in shaping reef microbiomes. This pattern aligns with global evidence that distinct reef habitats harbor unique microbial assemblages (Glasl et al., 2016; Glasl et al., 2019; Hernandez-Agreda et al., 2017; Bourne et al., 2016). For example, in the Red Sea, microbial communities vary significantly among coral, sponge, sediment, and water column habitats (Ziegler et al., 2016), while elsewhere, substrate-specific signatures distinguish corals, macroalgae, and turf assemblages (Kelly et al., 2014).

Ordination analyses revealed clear substrate-linked clustering. Coral tissues exhibited host-specific associations: Favia was enriched in Ruegeria, and Favites in Albidovulum. These profiles diverge from many Indo-Pacific systems, such as Palau and the Great Barrier Reef, where Endozoicomonas typically dominates coral microbiomes (Ainsworth et al., 2015; Dunphy et al., 2019; Gardner et al., 2019; Sanchez-Quinto and Falcon, 2021; Sanna et al., 2023). Within the Environmental group, sediments were enriched in Woeseia and Desulfovibrio, differing from the Deltaproteobacteria, Acidobacteriota, and Chloroflexi that dominate sediments in other tropical reefs (Zinger et al., 2011; Dong et al., 2019; Ziegler et al., 2017). Seawater communities within this group were dominated by pelagic lineages such as Pelagibacterales and Synechococcus, consistent with oligotrophic assemblages described in the Tara Oceans survey (Sunagawa et al., 2015; Torda et al., 2017).

The substrate-specific microbial communities observed in this study align with the concept of reef “microbialization,” where benthic organisms strongly influence surrounding microbial assemblages and ecosystem processes (Haas et al., 2016). Despite these compositional differences, Shannon diversity was broadly comparable across corals, CCA, and the pooled Environmental group, with no statistically significant pairwise contrasts. This indicates that within-sample diversity is maintained across microhabitats, even though community composition differs substantially. Together, these results highlight the role of substrate filtering in structuring microbial communities, with host identity and local environmental conditions shaping composition within the reef system (Lozupone et al., 2011). Cross-regional comparisons further emphasize that microbial assemblages are influenced by geography and habitat type, with community profiles diverging between the Great Barrier Reef (Tout et al., 2014), Daya Bay (Wu et al., 2023), and the central Red Sea (Ziegler et al., 2016). Establishing such baselines in understudied systems like Palk Bay is critical for understanding microbial biogeography and monitoring potential ecological responses to environmental change.

CCA-associated samples in this study supported microbial assemblages enriched in genera such as Roseospira, Labrenzia, and Ruegeria, all belonging to the family Rhodobacteraceae. These taxa have been repeatedly reported from CCA surfaces linked to coral larval settlement in other reef systems (Webster et al., 2004; Tebben et al., 2015; Jorissen et al., 2021). In our dataset, both SIMPER and LEfSe analyses consistently identified Rhodobacteraceae members as characteristic of CCA relative to coral substrates, with Roseospira contributing substantially to between-substrate dissimilarity and Ruegeria and Roseibacillus emerging as CCA-enriched biomarkers. The concordance between SIMPER and LEfSe strengthens confidence in the observed compositional patterns, with the fragment pooling strategy, in which each CCA sample represented a single colony composed of five pooled fragments, facilitating detection of consistent community-level signatures across CCA colonies despite the limited number of colonies analyzed (N = 3) (Ray et al., 2019).

While this study did not assess exometabolite production, predict functional pathways, or conduct larval settlement assays, the observed enrichment of Rhodobacteraceae on CCA in Palk Bay is consistent with patterns reported in other reef systems, where these taxa are frequently detected on settlement-inducing CCA species. For example, on the Great Barrier Reef, Titanoderma prototypum and Porolithon gardineri host Rhodobacteraceae ASVs that correlate with coral recruitment intensity (Siboni et al., 2020; Jorissen et al., 2021), and Porolithon onkodes retains Ruegeria and related taxa even under bleaching stress (Yang et al., 2021). Similarly, studies from the Northwestern Mediterranean report Rhodobacteraceae-dominated microbiomes on Lithophyllum stictiforme and Macroblastum dendrospermum, both of which enhance coral larval settlement under ambient and climate-stressed conditions (Manea et al., 2025a, 2025b). These observations provide contextual relevance for interpreting CCA-associated microbiomes in Palk Bay, while acknowledging that functional inferences cannot be drawn from 16S data alone (Tran and Hadfield, 2011; Gómez-Lemos et al., 2018).

Rubble, sediments, and seawater represented environmental, non-coral substrates that harbored distinct microbial assemblages with high numbers of unique genera (rubble: 39, sediments: 34), reflecting substantial compositional diversity (Supplementary Figure S2). These habitats contained taxa frequently reported from marine sediments and benthic environments, including Woeseia, Desulfovibrio, and Vibrio (Carlos et al., 2013; Ziegler et al., 2019). While these taxa have known metabolic capabilities in other systems, functional roles were not assessed here; thus, these environmental substrates are best interpreted as microbial diversity reservoirs rather than demonstrated biogeochemical engines. Seawater communities were dominated by pelagic lineages such as Pelagibacterales and Synechococcus, consistent with oligotrophic oceanic microbiomes (Sunagawa et al., 2015).

Together, CCA and environmental substrates provide a substrate-resolved baseline of reef-associated microbial diversity in Palk Bay. CCA samples capture taxa previously associated with settlement cues, while rubble, sediments, and seawater harbor diverse microbial pools that may support microbial redistribution and compositional resilience across the reef. It can be speculated that by providing functional redundancy and an “insurance effect” (Berry and Widder, 2014; Nyström, 2006), these non-coral microbial communities may help to maintain compositional heterogeneity and reef resilience in degraded ecosystems, contrasting with the more host-specific microbiomes observed in corals (Ainsworth et al., 2015). Consistent with findings from Sri Lankan reefs (Fairoz et al., 2023), we observed both host-associated microbial signatures in corals and strong environmental structuring in non-coral substrates, indicating multi-scale drivers of reef microbiome composition. This framework establishes reference points for future studies integrating functional predictions, metabolomics, and experimental larval settlement assays to determine whether the compositional signatures documented here translate into ecological function.

Microbial overlap across substrates in Palk Bay was minimal, with only Pelagibius detected across all habitats. Coral microbiomes were largely host-specific, with taxa such as Endozoicomonas restricted to coral tissues (Ainsworth et al., 2015; Ziegler et al., 2017). Settlement-associated taxa, including Roseospira, were present on CCA and rubble but absent from coral samples, indicating strong substrate specificity. These patterns suggest that most microbial lineages are tightly coupled to particular substrates, while a few generalist taxa provide limited compositional connectivity across the reef. Whether this minimal cross-substrate core represents a common feature of reef microbiomes or reflects environmental pressures in Palk Bay remains uncertain, highlighting the need for comparative baselines from less-disturbed reefs.

The ecological significance of substrate-specific microbial patterns lies primarily in compositional connectivity rather than directly demonstrated function. Despite >90% coral loss in Palk Bay (Thinesh et al., 2017a; Edward et al., 2015), high densities of recruits on CCA indicate potential for reef recovery, though our study does not provide functional or biochemical data to confirm microbial contributions. The enrichment of Rhodobacteraceae on CCA and the diversity of taxa in rubble, sediments, and seawater highlight substrates that harbor distinct microbial assemblages, which could support connectivity across the reef. While these patterns are consistent with ideas of ecological redundancy and cross-habitat connectivity, we refrain from inferring direct functional roles. This study establishes the first cross-substrate microbiome baseline for Palk Bay and provides a framework for future work. Seasonal and spatial replication will help test the stability of substrate-specific assemblages, while manipulative experiments integrating microbiome transplants, metabolomic profiling, and larval settlement assays will be necessary to resolve functional contributions (Vizon et al., 2025) of candidate taxa such as Roseospira and Labrenzia. Comparative studies across the Indo-Pacific will clarify whether the minimal cross-substrate core observed here reflects a stressed reef state or a general feature of reef microbiomes. Linking compositional patterns with functional assays in future studies will be critical to understanding how microbial networks influence reef resilience under accelerating global change.