1 Introduction

Octocorals (Phylum Cnidaria, Order Alcyonacea) are benthic marine organisms found in littoral waters to deep-sea abysses (McFadden et al. 2010). With recent declines in scleractinian coral cover, octocorals have become more prevalent on shallow Caribbean reefs and on some reefs are numerically dominant (Reyes-Bonilla and Jordán-Dahlgren 2017; Tsounis and Edmunds 2017; Rioja-Nieto and Álvarez-Filip 2019; Lasker et al. 2020). Scleractinian corals and octocorals form intricate relationships with associated microorganisms (Rohwer et al. 2002; Rosenberg et al. 2007; Blackall et al. 2015). These interactions play important roles in acquiring and cycling nutrients, defending against potential pathogens, and increasing host resilience to environmental stressors (Reshef et al. 2006; Raina et al. 2009; Lema et al. 2012; Rosado et al. 2019; Doering et al. 2021; Peixoto et al. 2021; Schul et al. 2022).

Because corals are keystone organisms in reef communities (Wilson et al. 2007, 2019; Kerry and Bellwood 2015) and their health and functions can be affected by their associated microbiota, investigations into factors that might influence the composition and nature of the interactions between members of the coral microbiota are needed. This is particularly true for Caribbean octocorals, which have received relatively little attention compared to scleractinian corals and Mediterranean octocorals (reviewed in van de Water et al. 2018a). Spatial and temporal analyses showed that gorgonian octocorals from the Mediterranean Sea supported highly conserved bacterial communities that were different from those in the surrounding seawater (van de Water et al. 2018b). In contrast, Haydon et al. (2022) identified seasonal changes in the composition of the bacterial assemblages associated with two of the four Pacific octocoral species studied from Australian waters. These octocoral-associated bacterial communities were often dominated by Proteobacteria within the order Oceanospirillales (in some cases, accounting for over 90% of the community), with Endozoicomonas being the most common bacterial genus detected across latitudes (Bayer et al. 2013; Correa et al. 2013; La Riviere et al. 2015). Thus, increasing knowledge regarding the composition, stability, and functions of octocoral-associated bacterial communities across time and space and their responses to local environmental conditions (e.g., increasing temperatures, health of the coral host) may provide important insights to explain the recent success of octocorals on Caribbean reefs.

Eunicea flexuosa and Antillogorgia americana, two broadcast-spawning octocoral species that support dinoflagellate endosymbionts within the family Symbiodiniaceae and are common inhabitants of shallow reefs in the Caribbean Sea and western North Atlantic Ocean, were selected for this study. Although both species belong to the order Malacalcyonacea, each is a member of a distinct family (Plexauridae for E. flexuosa and Gorgoniidae for A. americana) and exhibits different morphological and physiological features (Sánchez et al. 2003; McFadden et al. 2022). E. flexuosa colonies are characterized by dense stands of thick dichotomous branches and spawn after the full moon during the summer months (Beiring and Lasker 2000) while A. americana forms tall, plumose and flexible colonies that produce copious amounts of mucus and spawn after the full moon in November (Coelho and Lasker 2016).

It is known that host physiology and phylogeny can influence the structure and dynamics of the associated microbial communities and evidence for phylosymbiosis has been reported for several coral reef invertebrates, including octocorals from the Great Barrier Reef (O’Brien et al. 2019, 2020). However, comparisons of the spatial–temporal dynamics of the bacteriomes associated with the often dominant but morphologically and taxonomically different E. flexuosa and A. americana Caribbean octocorals are still lacking in the scientific literature, limiting the ability to investigate phylosymbiosis in these species. In response, this study characterized the bacterial communities associated with these octocorals, examining their stability across local and distant geographical locations as well as exploring potential temporal patterns. Microbial co-occurrence network analyses were employed to predict interactions between octocoral-associated bacteria and to identify genera that may play important roles in structuring the associated bacterial community. Overall, the results obtained in this study provide new insights into the structure and stability of the bacterial communities associated with E. flexuosa and A. americana over space and time and identify areas of focus for future research.

2 Methods

2.1 Sample collection and DNA sequencing

Samples of Eunicea flexuosa and Antillogorgia americana, were retrieved from visually healthy colonies (i.e., no visible signs of stress, disease, discoloration or necrosis) at two locations, the Florida Keys (Florida, USA) and Roatán (Honduras), which are approximately 1,170 km apart. In the Florida Keys, two shallow reefs (2 – 4 m) with different temperature profiles but similar benthic community assemblages were selected for this study. Little Grecian (25.119°N, 80.302°W) is an offshore reef located in the upper Keys characterized by low turbidity and an annual temperature range of 18 – 30 °C whereas Jaap reef (24.585°N, 81.583°W) is a nearshore reef in the lower Keys with high turbidity and a temperature range of 15 – 33 °C. On both reefs, a plastic numbered cattle ear tag was nailed into the non-living substrate immediately adjacent to four octocorals of each species that were located at least 10 m apart to reduce the likelihood of obtaining clonal propagates. Samples (2–3 cm) were taken in October 2019 and 2021 from a branch of the same individual colonies (4 samples for each of 2 octocoral species on 2 reefs for 2 years; N = 32 octocoral samples). One seawater sample (200 ml) was also collected from each reef. Water temperatures on the dates of sampling were 28.3 °C and 28.7 °C in 2019 and 28.5 °C and 28.8 °C in 2021 for Little Grecian and Jaap reef, respectively. The SARS-CoV-2 pandemic and resulting travel restrictions limited our ability to sample in 2020.

In Roatán, four E. flexuosa and four A. americana colonies at least 10 m apart, as well as three seawater samples, were obtained in June 2019 from three shallow, low turbidity reefs (CocoView Wall [16.358°N, 86.432°W], Menagerhea [16.352°N, 86.437°W] and Newman’s Wall [16.356°N, 86.431°W]; 2 – 6 m) located within the nearshore waters of the southern coast with sea-surface temperatures ranging from 26 – 30 °C (4 samples for each of 2 octocoral species on 3 reefs for 1 year; N = 24 octocoral samples). On the dates of sampling, water temperatures ranged from 27.8 °C at Newman’s Wall to 28.2 °C at Menagerhea.

All samples were transported on ice until processing (≤ 3 h) in the lab. There, each octocoral fragment was gently rinsed with filter-sterilized (0.22 μm) artificial seawater (ASW; 26.46 g NaCl, 0.75 g KCl, 11.97 g MgCl2·6H2O, 1.68 g CaCl2·2H2O, 4.44 g Na2SO4, 1.08 g NaHCO3, 0.00090 g Na2H2PO4, 1000 ml deionized water) and placed into a sterile cryovial filled with 1.8 ml of RNAlater® (Life Technologies) and frozen. Seawater samples were filtered through 0.22 mm filters and filters were placed into cryovials containing RNAlater® and frozen.

DNA was extracted from 250–300 mg of thawed octocoral samples and seawater filters using the MP Biomedicals FastDNA® Spin Kit for Soil following the manufacturer’s protocol, except that samples were incubated at 55° C overnight in kit buffers prior to continuing the protocol. Samples were purified using the New England Biolabs Monarch® PCR & DNA Clean Up Kit, and the quality and quantity of DNA were evaluated by electrophoresis on 1% agarose gels and via a Nanodrop ND2000 spectrophotometer (Thermo Fisher), respectively. DNA was sent to the University of Alabama at Birmingham Microbiome Institutional Research Core (Birmingham, Alabama, USA) for amplification of the V4 region of the 16S rRNA gene using the primer pair 515F – 806R (Caporaso et al. 2011) and subsequent 250 bp paired-end read sequencing on an Illumina MiSeq (San Diego, California, USA).

2.2 Bioinformatics and statistical analyses

The raw sequences were processed and analyzed using the microbiome analysis package QIIME 2 version 2021.8 (Bolyen et al. 2019, version released in September 2021). Briefly, the forward and reverse demultiplexed sequence reads were imported into QIIME 2, and primers were trimmed using the Cutadapt program (Martin 2011). DADA2 microbiome pipelines were implemented to quality filter, merge paired end sequences, dereplicate and remove chimeras from the amplicon sequence variants (ASVs). Taxonomy was assigned to ASVs at 99% sequence identity using the SILVA-132–99-515–806 Naïve Bayes classifier (Quast et al. 2013, released in December 2017). Reads identified as mitochondria, chloroplast and/or host contaminants were removed before the ASV table was rarefied to 12,500 reads (read depth determined by a rarefaction curve; Supplementary Fig. 1), which ultimately led to the loss of four samples of A. americana from Roatán. Subsequently, the ‘q2-diversity’ plugin was employed to compute alpha and beta diversity metrics.

Wilcoxon–Mann–Whitney tests were performed in R (v. 3.4.3; R Core Team 2017) with a significance level of 0.05 to evaluate richness, evenness, Shannon diversity, and Faith’s phylogenetic diversity among samples after verifying the normality of the data with a Shapiro–Wilk test at a significance level of 0.05. The unweighted and weighted UniFrac distance matrices were imported into R with the package ‘qiimer’ (Bittinger 2015) and used as metrics for beta diversity. The ‘betadisper’ command in the ‘vegan’ package was used to check for differences in homogeneity of variances (Anderson 2006; Oksanen 2013). Principal coordinates analyses (PCoA) were used to visualize patterns in beta diversity and permutational multivariate analyses of variance (PERMANOVA) were applied to both unweighted and weighted Unifrac distance matrixes to test for differences in the octocoral prokaryotic communities between years, geographical locations, and local reefs. All raw sequences were deposited in the NCBI Sequence Read Archive database under BioProject number PRJNA879770.

To identify the bacterial genera responsible for the observed differences in octocoral-associated bacterial community composition between years and geographical locations, linear discriminant effect size (LEfSe) calculated on prokaryotic reads at the genus level was used on the galaxy platform (https://huttenhower.sph.harvard.edu/galaxy/). Sample communities were considered significantly different at p ≤ 0.05 and a LDA score (log10) > 3, one order of magnitude greater than the default value of the LEfSe methodology (Segata et al. 2011).

Octocoral microbial co-occurrence network analyses for both E. flexuosa and A. americana were conducted in R using package ‘Hmisc’ and function ‘rcorr’ (Harrell and Dupont 2006) at the genus level for all bacterial genera with more than 10 reads. A correlation between two genera was considered statistically significant if the Spearman’s correlation coefficient (ρ) was > 0.6 (or < -0.6) and the p-value was ≤ 0.05 (Barberán et al. 2012). P-values were adjusted with the Benjamini–Hochberg method (Benjamini and Hochberg 1995) to account for multiple comparisons. Networks were visualized using the Frunchterman-Reingold layout in the package ‘igraph’ (Csardi and Nepusz 2006). Density, average shortest path length, network diameter and average degree were calculated to characterize the overall interconnectedness of the networks, while average clustering coefficient and modularity described the degree of aggregation within networks. For every node, betweenness centrality values were computed to understand which bacterial genera may play an important role in structuring host bacterial communities.

3 Results

3.1 Microbial community profiles and spatial/temporal stability analyses

A total of 7,026,201 demultiplexed raw sequences were obtained from the samples retrieved for the current study (Supplementary Table 1). After quality-filtering and removal of sequences identified as chloroplasts, mitochondria and host contaminants, reads were assigned to 1110, 1400 and 413 ASVs for Eunicea flexuosa, Antillogorgia americana and seawater samples, respectively.

The bacterial communities associated with both octocoral species differed significantly when compared to those within the surrounding seawater from both the Florida Keys and Roatán (PERMANOVA: Df = 54, F = 3.530, R2 = 0.119, unweighted UniFrac, p = 0.001; weighted UniFrac, Df = 54, F = 8.167, R2 = 0.239, p = 0.001). Alpha diversity metrics for E. flexuosa revealed that individuals from the Florida Keys hosted significantly richer (ASVs number: Wilcoxon test, W = 83.00, p = 0.01; Table 1) and more diverse bacterial assemblages than those from Roatán. These results were supported by the Shannon and Faith’s Phylogenetic Diversity indices, which reported a significantly higher bacterial diversity within the E. flexuosa individuals from the Florida Keys than those from Roatán sampled in 2019 (Shannon index: Wilcoxon test, W = 83.00, p = 0.01; Faith’s Phylogenetic Diversity index: Wilcoxon test, W = 84.00, p < 0.01; Table 1). Conversely, no significant differences in alpha diversity were found between octocorals across years (2019 vs. 2021) in the Florida Keys (Wilcoxon tests: ASVs number W = 36.50, p > 0.05; Shannon index W = 33.00, p > 0.05; Evenness W = 32.00, p > 0.05; Faith’s Phylogenetic Diversity index W = 27; p > 0.05; Table 1). For A. americana, although more bacterial ASVs were detected in individuals from the Florida Keys, there was no significant difference in the alpha diversity metrics across sites (Wilcoxon tests: ASVs number W = 34.00, p > 0.05; Shannon index W = 32.00, p > 0.05; Evenness W = 34.00, p > 0.05; Faith’s Phylogenetic Diversity index W = 35; p > 0.05) or years (Wilcoxon tests: ASVs number W = 28.00, p > 0.05; Shannon index W = 30.00, p > 0.05; Evenness W = 36.00, p > 0.05; Faith’s Phylogenetic Diversity index W = 16; p > 0.05; Table 1).

Table 1 Alpha diversity indices for the octocoral species analyzed in the Florida Keys and Roatán in 2019 and 2021
Full size table

The beta diversity metrics, used to evaluate the spatial stability of the bacterial communities associated with the two octocorals (Fig. 1), showed that E. flexuosa-associated communities differed between the Florida Keys and Roatán in terms of ASVs presence/absence (unweighted UniFrac PERMANOVA, Df = 18, F = 4.574, R2 = 0.216, p = 0.001) but not in terms of ASVs overall relative abundance (weighted UniFrac PERMANOVA, Df = 18, F = 2.547, R2 = 0.131, p = 0.108). No significant differences were observed between the bacterial communities of E. flexuosa on the reefs within locations (Florida Keys PERMANOVA: unweighted UniFrac, Df = 7, F = 1.053, R2 = 0.149, p = 0.401; weighted UniFrac, Df = 7, F = 3.868, R2 = 0.392, p = 0.185; Roatán PERMANOVA: unweighted UniFrac, Df = 10, F = 1.076, R2 = 0.212, p = 0.334; weighted UniFrac, Df = 10, F = 1.933, R2 = 0.325, p = 0.278). In contrast, A. americana-associated communities did not show significant differences at any geographical scale (Florida Keys vs. Roatán PERMANOVA: unweighted UniFrac, Df = 13, F = 1.062, R2 = 0.081, p = 0.329; weighted UniFrac, Df = 13, F = 4.031, R2 = 0.252, p = 0.089; Florida Keys PERMANOVA: unweighted UniFrac, Df = 7, F = 1.983, R2 = 0.248, p = 0.154; weighted UniFrac, Df = 7, F = 1.345, R2 = 0.183, p = 0.259; Roatán PERMANOVA: unweighted UniFrac, Df = 5, F = 0.898, R2 = 0.184, p = 0.667; weighted UniFrac, Df = 5, F = 1.439, R2 = 0.265, p = 0.400).

Fig. 1
figure 1

Spatial and temporal stability of the bacterial communities associated with Eunicea flexuosa (left) and Antillogorgia americana (right). Barplots and tables show means of relative abundance for bacterial phyla (mean [± standard error] %), while PCoA graphs of unweighted UniFrac distance matrixes show separation of the samples into clusters based on geographical location, individual reefs, and year. Small circles reflect communities within individual octocorals while large circles represent the centroids of the groups. FK = Florida Keys; RT = Roatán

Full size image

Similar results were obtained for temporal differences in the octocoral-associated bacterial communities (Fig. 1). In particular, beta diversity metrics showed significant temporal differences between the bacterial communities associated with E. flexuosa individuals from the Florida Keys in 2019 and 2021 for the presence/absence of ASVs (unweighted UniFrac PERMANOVA, Df = 17, F = 2.651, R2 = 0.141, p = 0.02) but not for the overall relative abundance of ASVs (weighted UniFrac PERMANOVA, Df = 17, F = 2.176, R2 = 0.155, p = 0.11). No significant differences in the composition of the bacterial communities associated with A. americana between 2019 and 2021 were found (PERMANOVAs: unweighted UniFrac, Df = 17, F = 2.319, R2 = 0.112, p = 0.08; weighted UniFrac, Df = 17, F = 1.983, R2 = 0.182, p = 0.113).

The sequences from the octocorals represented 29 bacterial phyla, with 22 phyla detected at less than 1% relative abundance from Florida Keys or Roatán individuals. The most dominant phylum for both octocorals, regardless of collection date or location, was Proteobacteria, followed by the phyla Tenericutes, Firmicutes, Bacteroidetes, Actinobacteria, Cyanobacteria and Verrucomicrobia (Fig. 1). Some of the phyla found in low relative abundances were exclusively detected in only one octocoral host. For example, members of Dadabacteria, Kiritimatiellaeota, Nitrospinae and PAUC34f were only found in E. flexuosa individuals while Atribacteria, Deinococcus-Thermus, Latescibacteria, Lentisphaerae, Synergistetes and WPS-2 were retrieved only from A. americana samples. ASVs assigned to the genera Endozoicomonas (Proteobacteria, Gammaproteobacteria, Oceanospirillales, Hahelaceae), Lactobacillus (Firmicutes, Bacilli, Lactobacillales, Lactobacillaceae), Dubosiella (Firmicutes, Erysipelotrichia, Erysipelotrichales, Erysipelotrichaceae) and Akkermansia (Verrucomicrobia, Verrucomicrobiae, Verrucomicrobiales, Akkermansiaceae) were found in at least 90% of the samples across species, collection date, and locations.

In the E. flexuosa colonies sampled in the Florida Keys, 16 bacterial genera had an average relative abundance ≥ 1% in 2019 whereas 14 genera met this criterion in 2021. From samples collected in Roatán, only 4 bacterial genera had an average relative abundance ≥ 1% (Table 2). For E. flexuosa, Endozoicomonas and Mycoplasma (Tenericutes) were the dominant genera regardless of sampling location and time, representing 23.77–76.27% and 12.57–26.49%, respectively, of the total bacterial reads (Table 2). A similar result was found for the A. americana colonies, with 12 and 14 bacterial genera with an average relative abundance ≥ 1% recovered from colonies in the Florida Keys in 2019 and 2021, respectively, and 12 genera from Roatán corals. Endozoicomonas and Mycoplasma also represented the dominant genera associated with A. americana (Table 2).

Table 2 Bacterial genera with an average relative abundance ≥ 1% found in A. americana and E. flexuosa in the Florida Keys and Roatán in 2019 and 2021
Full size table

Spatial differential abundance analyses (LEfSe) on E. flexuosa samples indicated the presence of 36 significantly enriched bacterial genera associated with the Florida Keys samples compared to those from Roatán, with half being placed within the Firmicutes (Fig. 2).

Fig. 2
figure 2

LEfSe analysis of the bacterial genera associated with Eunicea flexuosa (left) and Antillogorgia americana (right) that were significantly different between locations (top) and sampling years (bottom). Communities were considered significantly different at p ≤ 0.05 and a LDA score (log10) > 3

Full size image

Among these, Lactobacillus had the highest LDA score (p = 0.001; LDA = 4.78), differing in relative abundance from 0.86 ± 0.31% (mean ± SE) in the Roatán samples to 12.63 ± 4.14% in the Florida Keys samples. Endozoicomonas (Gammaproteobacteria) was the only genus in E. flexuosa that was significantly enriched in Roatàn samples (p = 0.002; LDA = 5.42; relative abundance: Roatán 76.27 ± 1.19%; Florida Keys 23.77 ± 0.77%). For A. americana, Mycoplasma was the only genus significantly enriched in Florida Keys samples (p = 0.03;

LDA = 4.65; relative abundance: Florida Keys 28.22 ± 2.31%; Roatán 0.00 ± 0.00%).

Temporal LEfSE analyses identified fewer significantly enriched taxa than spatial analyses. Eighteen bacterial genera were differentially abundant between E. flexuosa octocorals sampled in 2019 and 2021 in the Florida Keys and only Herbaspirillum (Betaproteobacterium) changed significantly in A. americana (Fig. 2).

3.2 Co-occurrence networks

In order to evaluate potential interactions within host bacterial communities, co-occurrence networks based on significant pairwise correlations among bacterial genera were employed. The co-occurrence network topology varied between E. flexuosa and A. americana, with A. americana supporting a more dense and interconnected bacterial network with a higher number of nodes (bacterial genera = 214 vs. 157), edges (correlations = 1594 vs. 557), average degree (14.89 vs. 7.09), density (0.07 vs. 0.04), and average clustering coefficient (0.87 vs. 0.78), and a lower modularity (0.68 vs. 0.78), diameter (8 vs. 13), and average shortest path (2.59 vs. 3.34) than E. flexuosa (Fig. 3). Of the 214 and 157 nodes in the networks of A. americana and E. flexuosa, respectively, only 30 bacterial genera with betweenness centrality values > 0 were common to the two octocorals. In both networks, the genera Actinomyces, Delftia, Peptoniphilus, Veillonella, and Mobiluncus were found in the 20 highest betweenness centrality values (Supplementary Table 2). There were a number of shared phyla, including Proteobacteria (4 and 1 genera for E. flexuosa and A. americana, respectively), Actinobacteria (4 and 5), Firmicutes (9 and 10), and Bacteroidetes (2 and 4) within the 20 genera found to have the highest betweenness centrality values. Only 1 genus within the phylum Verrucomicrobia in E. flexuosa was distinct.

Fig. 3
figure 3

Correlation-based bacterial networks obtained using Spearman correlations (coefficient ρ > 0.6 and < -0.6; p < 0.05). Colored dots indicate nodes (bacterial genera), and lines indicate edges. Nodes are color-coded by phylum (or class for Proteobacteria). Edge lengths are not biologically meaningful as they are a function of the layout. Relevant network metrics for E. flexuosa and A. americana are included at the bottom of the figure. Nodes = number of bacterial genera in the networks; Edges = number of connections between nodes in the networks; Average clustering coefficient = average number of edges between neighbors of a node based on total potential number of edges between neighboring nodes; Average shortest path length = average shortest distance in number of edges between all node pairs; Modularity = tendency of nodes to cluster in modules, with values ranging between 0, if all nodes interact equally, and 1 if nodes interact only within modules; Graph density = number of edges out of the total possible number of edges; Network diameter = shortest distance between the two most distant nodes in the network; Average degree = average number of edges per individual node

Full size image

4 Discussion

As coral reefs in the western North Atlantic Ocean and Caribbean Sea are transitioning to octocoral dominance (Ruzicka et al. 2013; Lenz et al. 2015; Kupfner Johnson and Hallock 2020), this study investigated the temporal and spatial stability of the bacterial communities associated with two common yet taxonomically and morphologically distinct Caribbean octocorals. The bacterial assemblages of A. americana were more stable across space and time than those associated with E. flexuosa, suggesting that general inferences on the stability of octocoral-associated bacterial communities should not be made. Although sampling of E. flexuosa on Roatán reefs occurred in June, which may have corresponded with spawning and potentially contributed to differences observed in the composition of the associated bacterial communities compared to octocorals collected in the Florida Keys in October, the separation of the associated communities from the same individual corals over two sampling periods suggests that this is unlikely. The discordant results observed for the two octocoral species reinforced some previous studies that revealed discrepancies among different coral hosts. The significant spatial heterogeneity we observed in E. flexuosa-associated bacterial communities across large geographical scales is in accordance with results from the tropical octocoral Erythropodium caribaeorum sampled in Florida and in the Bahamas (McCauley et al. 2016), as well as for Mediterranean gorgonians within the genus Eunicella (van de Water et al. 2018b). Conversely, the overall stability in space and time for the A. americana-associated bacterial communities supports results obtained for Corallium rubrum and other temperate octocorals (van de Water et al. 2016, 2018b). Our observations provide further evidence that the spatial and temporal stability of bacterial communities associated with octocorals is species-specific (van de Water et al. 2018a, b).

The significant spatio-temporal heterogeneity in the bacterial communities associated with E. flexuosa was mostly driven by changes in the presence or absence of ASVs, suggesting profound differences in the overall bacterial community structure across large geographical scales and time. This appears to be in contrast to what was previously observed in the tropical Pacific octocoral Sinularia sp., where temporal changes were largely dictated by differences in the relative abundances of associated bacteria (Haydon et al. 2022). Interestingly, LeFSE analyses revealed that several bacterial taxa within the common human gut-associated phyla Firmicutes and Bacteroidetes (Arumugam et al. 2011; Rinninella et al. 2019) contributed to the large-scale spatial differences observed in the bacteria associated with E. flexuosa. These taxa were significantly enriched in Florida Keys octocorals compared to those from Roatán, with genera such Anaerococcus, Blautia, Ezakiella, Faecalibaculum and Veillonella also being detected at higher relative abundances in the Florida Keys seawater samples. Coral reef communities exist adjacent to highly populated urban areas in the Florida Keys where human wastewater disposal systems were historically on-site and largely ineffective (Paul et al. 2000). Previous studies recorded the presence of fecal bacterial indicators and enteric viruses from animal and human sources in the Florida Keys surface seawater and the mucus of scleractinian corals (Griffin et al. 1999; Paul et al. 1995; Lipp et al. 2002; Wetz et al. 2004). Although efforts to build effective wastewater treatment plants in the Florida Keys have been largely completed, testing for Enterococcus still regularly shows elevated levels at a number of locations (Florida Healthy Beaches Program (n.d.); https://www.floridahealth.gov/environmental-health/beach-water-quality/index.html). Lipp et al. (2002) demonstrated that coral mucus may serve to accumulate human enteric microorganisms in the reef environment despite low indicator levels in the surrounding seawaters. The disparity in the detection of gut-associated bacteria between sampling locations is especially interesting as Roatán also has limited wastewater treatment options for the human communities located in close proximity to coastal waters. This suggests that host physiological features, such as the amount of mucus produced, functioning independently or in combination with biotic and/or abiotic environmental conditions (e.g., temperature, current flow, nutrient concentrations) may be important factors influencing the stability of bacterial communities across space and time for E. flexuosa and A. americana.

The surface mucus layer (SML) of corals is a very dynamic environment (Brown and Bythell 2005) that is periodically sloughed off into the surrounding seawater (Nelson et al. 2013). A. americana, the slimy sea plume, produces copious quantities of mucus (Lasker 1981) that form a dense, thick layer on the coral surface. In contrast, E. flexuosa has a thin SML (Shirur et al. 2014). As mucus aging and shedding events were shown to significantly influence the composition and stability of scleractinian prokaryotic communities (Glasl et al. 2016), the same factors may influence the bacterial communities associated with A. americana and E. flexuosa. Further studies are warranted to explore the potential role of mucus aging and shedding in the retention or turnover of octocoral-associated bacterial taxa.

Although the bacterial communities associated with E. flexuosa and A. americana showed different trends over space and time, members of the genus Endozoicomonas (Gammaproteobacteria) were most commonly associated with both species. Endozoicomonas spp. form stable associations with various marine invertebrates and were recently shown to dominate bacterial communities in both scleractinian corals and octocorals (e.g., La Riviere et al. 2015; Apprill 2017; Camp et al. 2020; McCauley et al. 2020; Kellogg and Pratte 2021). Roles in nutrient cycling and acquisition and host health regulation have been proposed for this genus (Bourne et al. 2013; Correa et al. 2013; Ransome et al. 2014; Neave et al. 2016), but additional studies are needed to further clarify the functions of Endozoicomonas spp. to understand their consistent association with octocorals.

Members of the genus Mycoplasma (Tenericutes) were also found at high relative abundances in both octocorals, with reads accounting for 16 to 36% of the total bacteria across space and time in E. flexuosa and 16 to 30% in A. americana samples from the Florida Keys although this genus was not detected in any specimens of this species from Roatán. Previous studies found Mycoplasma spp. to be one of the dominant components of the microbiota of E. flexuosa and the tropical octocorals Eunicea forti, Pseudoplexaura crucis, P. flagellosa, and P. porosa (Shirur et al. 2016; McCauley et al. 2020). Once again, the reason for this association in various octocoral hosts has not been determined (van de Water et al. 2018a, b; McCauley et al. 2020) although members of this bacterial genus are thought to be commensal or mutualistic symbionts and connected to host feeding in a scleractinian coral (Neulinger et al. 2009).

Over the 2019 and 2021 sampling dates in the Florida Keys, the relative abundances of Endozoicomonas and Mycoplasma fluctuated differently across the two octocoral species. In A. americana, Endozoicomonas increased ~ 3% in relative abundance from 2019 to 2021 while Mycoplasma decreased ~ 13%. In E. flexuosa, Endozoicomonas decreased in relative abundance ~ 7% with Mycoplasma increasing ~ 14%. Although McCauley et al. (2020) showed that Proteobacteria significantly decreased in abundance while Tenericutes increased in abundance from summer to winter in P. crucis, they also reported a shift in dominance between Endozoicomonas and Mycoplasma in P. porosa across years. When all of these observations are combined, it suggests that octocoral-associated bacterial communities are flexible and potentially able to adapt to changes occurring in the surrounding environment and/or respond to specific physiological needs. Moreover, given that the two octocorals analyzed in our study displayed opposite trends in the relative abundance of Proteobacteria and Tenericutes across years, the dominance of specific members of the octocoral bacterial communities may be strongly driven by the differences noted in host phylogeny and/or specific characteristics, as already observed in both scleractinian corals and octocorals (Ziegler et al. 2017; Pollock et al. 2018; O’Brien et al. 2020).

Interestingly, members of the phylum Spirochaetes were consistently found at low relative abundances (< 1%) in both octocoral hosts across sampling locations and collection dates. This contrasts with observations of prokaryotic communities associated with two deep-sea Anthothela spp. and Corallium rubrum from the Mediterranean Sea (Lawler et al. 2016; van de Water et al. 2016) where this bacterial phylum was found to be dominant or co-dominant and possibly involved in carbon and nitrogen fixation (Lilburn et al. 2001; van de Water et al. 2016).

Host identity also appears to be an important factor in network analyses, as E. flexuosa and A. americana bacterial communities displayed different network topologies. Networks have been successfully used to model the co-occurrence of host associated bacteria, explore relationships, and indicate taxa that may play important roles for structuring the entire microbial community (Layeghifard et al. 2017). The modularity of networks is a key feature (Muff et al. 2005) that can be used as a proxy to identify the presence of host-associated niche partitioning (Faust and Raes 2012). In both octocoral species studied, network modularity revealed bacterial genera clustered into distinct modules that may be mediated by competitive and/or mutualistic interactions as was shown for different systems (González et al. 2010; Baldassano and Bassett 2016). The network of E. flexuosa had a higher modularity score compared to that of A. americana, which may be representative of host physiological characteristics (e.g., different SML production across octocoral species) and/or may influence network functionality by promoting bacterial functional redundancy in different modules. This redundancy would potentially diminish the effects of losing bacterial modules in unstable environments (Papin et al. 2004; Lurgi et al. 2019). Conversely, the bacterial network of A. americana was more complex and interconnected, with a higher number of nodes and edges, and double the density and average number of connections per bacterial genera. These results suggest a more stable and organized community structure in A. americana individuals where bacteria establish more complex interactions between each other than in E. flexuosa octocorals.

The greater complexity of the interaction network for A. americana was also reflected by the number of nodes characterized by high betweenness centrality values. Betweenness centrality indicates the number of times a node (in this case bacterial genera) is located on the shortest path between other nodes (Anthonisse 1971; Freeman 1977). This score has been considered one of the key attributes to identify ‘hub’ (i.e., keystone) taxa which may play important roles in structuring and stabilizing the network by bridging various parts, and whose removal from the network may cause disproportionately deleterious effects (Borgatti and Everett 2006; Martins et al. 2022). Interestingly, several bacterial genera with the highest betweenness centrality values in the networks for both octocorals are known to produce bioactive secondary metabolites. For example, Kocuria spp. previously isolated from marine sponges and octocorals demonstrated bioactivity against both coral and human pathogens and were shown to possess biosynthetic pathways (e.g., non-ribosomal peptide synthetases) involved in the production of natural products (Palomo et al. 2012; Monti et al. 2022). Other bacterial genera within the phyla Actinobacteria and Proteobacteria exhibiting high betweenness centrality values in both E. flexuosa and A. americana networks are known to synthesize antimicrobial compounds and/or molecules which can interfere with quorum sensing signals, thereby influencing the abundance of other host-associated prokaryotes and/or opportunistic pathogens (Bérdy 2005; Behie et al. 2017; Raimundo et al. 2018; Rajasabapathy et al. 2020; Siro et al. 2022).

To the best of our knowledge, the current study is the first to apply network analyses to octocoral bacterial communities, providing insights into co-occurrence patterns, predicting potential interactions between genera, and identifying genera that may play fundamental roles in mediating ecological interactions and structuring the octocoral-associated community. However, the disparity in composition and stability of the bacterial communities associated with the two Caribbean octocoral species studied suggests that broad generalizations are not appropriate and that additional studies are warranted to uncover the functional roles of the bacterial assemblages, which may be important for explaining the success of these cnidarians on modern coral reefs.