Eight to ten P. lobata colonies were sampled at each of the six locations in Fouha Bay, Guam (Figure 1) via hammer and chisel in April 2019. These locations were further clustered into sites (2–4) that reflect position along the sedimentation gradient as described by Rongo (2004; Figure 1). Site 1 was excluded from this study as it is virtually devoid of coral, with only a few colonies of Leptastrea purpurea remaining. The site closest to the river mouth (site 2) not only experiences higher sedimentation but also lower average light and more frequent decreases in salinity compared to the other two sites. All three sites experience similar temperatures (Supplementary Figure 1). The La Sa Fua River discharges into Fouha Bay through a small canyon cut through the reef flat. This canyon is about 20–30 m wide with depth varying between 0.5 m at the shore and about 7 m at the reef edge. Corals grow along the walls of the canyon with some additional coral outcrops scattered throughout the bay. All specimens collected for this study were growing on the edges of the canyon. All colonies (N = 52) were sampled in the center of the colony, half (N = 23) were also sampled at the edge and sediment (N = 18) was collected adjacent (<0.25 m) to the coral colony. Samples were placed in collection bags and flash frozen in liquid nitrogen upon collection and transferred to −80°C until DNA isolation.

Environmental data were sampled at 30 min intervals over 8 months (April 19, 2018–January 19, 2019) that spanned both dry and wet seasons. Temperature and conductivity were sampled using a HOBO U24-002-C (Onset Computer Corporation, Bourne, MA, United States) saltwater temperature/conductivity logger; conductivity was converted to salinity following calibration using a refractometer. Illuminance was recorded using HOBO UA-002-08 pendant temperature/light data logger.

To supplement the center versus edge intra-colony symbiont structure hypothesis addressed herein, contributions from fifteen P. lobata colonies collected during Klepac and Barshis (2020) were also included. P. lobata is a dominant species of the backreef lagoon pools on Ofu Island in American Samoa. Although these colonies are not influenced by sedimentation, three well-studied pools experience contrasting environmental regimes (high variability, moderate variability, and low variability), which could contribute to site-wide differences in Symbiodiniaceae composition. Five large colonies (6–10 m²) per site were sampled using a core punch (1.27 cm) and rock hammer. From each colony a core was taken from the top centroid and ∼0.5–1 m down (depending on the size of the colony; N = 30 samples) from the top sampling location to determine intra-colony differences in Symbiodiniaceae.

DNA was extracted using the DNeasy PowerSoil (Qiagen, Hilden, Germany) kit at the University of Guam Marine Laboratory. Tissue samples were first placed in a bead beater for 30 s (with C1 solution), followed by centrifugation and transfer of the resulting supernatant to a fresh 1.5 ml tube. DNA was purified using a QIAcube automated liquid handler (Qiagen, Hilden, Germany). Bacterial 16S (for center, edge, and sediment samples) and algal ITS2 libraries (for center and edge only) were prepared at Boston University. 16S preps also included a negative control prepped with nuclease-free water. ITS2 libraries were generated following Baumann et al. (2017) and 16S libraries were generated by amplifying the V4/V5 region using modified 515f (Parada et al., 2016) and modified 806r (Apprill et al., 2015). These two library sets were then pooled in 1:2 ITS2:16S and sequenced on an Illumina MiSeqV2 (paired-end 250 bp) at Tufts University Core Facility (TUCF). All samples were prepared for 16S but sediment samples were excluded from ITS2 analysis as P. lobata transmits symbionts vertically (Richmond and Hunter, 1990). Colonies from Ofu Island were prepped for ITS2 as described in Klepac and Barshis (2020).

ITS2 Metabarcoding data showed that samples were dominated by Cladocopium C15, but several distinct ITS2 profiles (commonly referred to as types; LaJeunesse, 2002) exist in this population (Figure 2). Individual coral colonies can harbor distinct C15 types between intra-colony positions, although intra-colony position does not predict the Symbiodiniaceae type (X-squared = 26.682, df = 22, p-value = 0.2235). Similarly, colonies from Ofu Island also showed distinct profiles within colonies but intra-colony position did not predict Symbiodiniaceae type (X-squared = 5.5824, df = 6, p-value = 0.4716; Supplementary Figure 3). Within Fouha Bay, site did not predict Symbiodiniaceae type within both center (X-squared = 26.682, df = 22, p-value = 0.2235) and edge samples (X-squared = 15.754, df = 16, p-value = 0.4703). Comparisons of 16S beta diversity between a center-edge pair for a colony that showed intra-colony position-specific differences in dominant ITS2 profiles were no different from the pairwise beta diversity of a center-edge pair for colonies that maintained the same ITS2 profile (t = −1.2871, df = 36, p-value = 0.2063; Supplementary Figure 4).

16S Alpha diversity and beta diversity comparisons for both center and edge communities revealed no significant differences along the sedimentation gradient (Figures 3A,B and Supplementary Figure 5). Alpha diversity for sediment communities was also not significantly different across sites (Supplementary Figure 5). However, beta diversity was significantly different between site 2 (the site with the highest sedimentation) and sites 3 and 4 (Figure 3C). Comparisons of diversity between intra-colony positions revealed striking differences, with edge, center and sediment communities all exhibiting significant differences in both alpha and beta diversity (Figure 4). Sediment bacterial communities were more similar to edge communities than they were to center communities (Supplementary Figure 6). Dominant bacterial families from center samples included Endozoicomonadaceae (mean 42.2% relative abundance), Xenococcaceae (6.3%) and Amoebophilaceae (5%) (Supplementary Figure 7). Dominant bacterial families from edge samples included Endozoicomonadaceae (11.7%) and Amoebophilaceae (6.1%) (Supplementary Figure 8). Dominant bacterial families from sediment samples included Cyclobacteriaceae (5.4%) and Rhodobacteraceae (5.3%) (Supplementary Figure 9).

For edge samples, several ASVs were more abundant at site 2 (the site with the highest sedimentation) or more abundant at sites 2 and 3 compared to site 4 (least sedimentation). These included ASVs belonging to genera Catenococcus, Cohaesibacter, Filomicrobium, Halioglobus, Peredibacter, Pirellula, Rhodopirellula, Woeseia, and Kiloniellaceae (Supplementary Figure 10). For samples collected from the center of the coral colony, several ASVs belonging to the genus Endozoicomonas were more abundant at sites 3 or 4 compared to 2 (i.e., associated with lower sedimentation). One ASV, of the genus Burkholderiaceae, was more abundant at sites 2 and 3 compared to 4 (Supplementary Figure 10). Only one ASV, which belonged to the genus Planococcaceae, was an indicator species for both edge and center colonies across sites. This ASV was most abundant for both intra-colony positions at site 2 (Figure 5A). Additionally, only 1 ASV, belonging to Peredibacter, was an indicator species for both coral (edge only) and sediment across sites. This ASV was also the most abundant for both sample types at site 2 (Figure 5B).

Functional pathways significantly differentially enriched when comparing colony edge and center, but not between colony edge and reef sediment, were examined to investigate differences between the center and edge of colonies that were likely mediated by bacteria available in nearby sediment. All pathways that were overrepresented at the edge and sediment compared to the center are involved in cobalamin and amino acid synthesis (Supplementary Figure 11).

Pathways that were significantly enriched when comparing both coral colony edge and center as well as colony edge and reef sediment communities were then examined to investigate differences between the center and the edge of the colony that were not necessarily due to uptake at the edge from local sediment communities. There were many more pathways enriched in this group compared to those that had no differences between edge and sediment. This comparison was further split into three groups. Pathways that were most highly represented in the center, then the edge, and then the sediment included carbohydrate and amino acid degradation and TCA cycle pathways (Supplementary Figure 12). Pathways that were most highly represented in the sediment, then the edge, and then the center involved denitrification and methanogenesis (Supplementary Figure 13). Finally, pathways that were most highly represented in the edge, then sediment, and then center consisted of amino acid metabolism pathways (Supplementary Figure 14).

We sampled a total of 52 P. lobata corals along a steep sedimentation gradient in Fouha Bay, Guam, and characterized their Symbiodiniaceae algae (ITS2) and bacterial (16S) microbiome communities across the center and edge of coral colonies (ITS2, 16S) and immediately adjacent to sediment (16S only). We observed that P. lobata in Fouha Bay almost exclusively associated with algae in the genus Cladocopium and harbors different putative C15 genotypes even within a single colony. Colonies of P. lobata from Ofu Island also harbored different genotypes of Cladocopium C15 genotypes within a single colony. Massive Porites corals were historically considered to have fixed and non-mixed Symbiodiniaceae associations (Fay and Weber, 2012) with most populations of this coral being associated with a single Cladocopium species, commonly referred to as the C15 lineage (Lajeunesse et al., 2004; Thornhill et al., 2006). These strict associations with a single algal lineage were previously thought to be a result of the vertical transmission during reproduction (Fay and Weber, 2012). However, it has since been discovered that massive Porites corals can harbor mixed Cladocopium and Durusdinium communities (Terraneo et al., 2019; Tan et al., 2020) and application of the within-sample informative intragenomic sequences used by SymPortal (Hume et al., 2019) revealed different colonies within a population of P. lobata can harbor different genotypes of Cladocopium (Gardner et al., 2019; Ziegler et al., 2019; Camp et al., 2020). This study adds to the literature providing evidence against the fixed and non-mixed infection theories by demonstrating intra-colony variation of dominant Cladocopium C15 genotypes within a single colony of P. lobata across small (Fouha Bay) and large (Guam and American Samoa) geographic locations. There are two possible scenarios for explaining mixed genotypes here and they are not mutually exclusive: (1) multiple genotypes are inherited vertically and symbionts “shuffle” (Baker et al., 2004; Berkelmans and van Oppen, 2006); (2) cryptic horizontal transmission of Symbiodiniaceae (Quigley et al., 2018). Further work sequencing ITS2 profiles of the water column, nearby sediment, and P. lobata larvae is needed in order to test these hypotheses.

There was neither a clear pattern of ITS2 profile structuring according to intra-colony position nor a discernable pattern of ITS2 profiles along the sedimentation gradient, suggesting that the ITS2 community composition in P. lobata is not directly driven by sedimentation. Massive Porites can harbor different dominant ITS2 profiles across latitudinal gradients (Terraneo et al., 2019), seasons (Ziegler et al., 2015), and habitats [e.g., higher diversity of ITS2 profiles among colonies in the open ocean compared to those in mangrove habitats; (Camp et al., 2019)]. These previous studies suggest that ITS2 profile structuring for this genus of corals is determined by the local environment. However, we do not observe such structuring here, implying either differences in profiles are random or due to differences in the microenvironment not captured by this study. It is possible resampling following the flooding events might reveal shifts in dominant profile-types. Interestingly, there was a trend where 7/10 colonies that showed differing profiles between the colony center and the colony edge hosted C15.C15bn.C15 on the edge. Furthermore, of these seven, six hosted C15.C15ev at the center. This suggests colonies hosting certain Symbiodiniaceae taxa might be more likely to host divergent populations between the center and edge. However, a greater sample size will be needed to properly investigate this pattern.

For P. lobata-associated bacterial communities, we observed largely homogenous bacterial communities along the sedimentation gradient. The majority of coral tissue samples from the center of the colony show high homogeneity and were dominated by ASVs belonging to the bacterial genus Endozoicomonas (Proteobacteria: Oceanospirillales). The ecological role of Endozoicomonas in the coral microbiome remains unknown, albeit there are numerous theories, most of which reflect a positive role in coral health [reviewed in Neave et al. (2016)]. Previous work found that Endozoicomonas increased in abundance across several species of coral in response to heat stress and persisted 10 months after heat stress events (Maher et al., 2020). It is therefore possible that the dominance of Endozoicomonas observed here may be a result of the sediment-induced stress associated with the riverine environment of Fouha Bay. As Endozoicomonas abundance did not change across sites within the bay, we theorize that “stress events”–in this case acute large-scale flooding and associated sedimentation events that occur several times a year affecting the entire bay (Rongo, 2004)–play a stronger structuring role than the chronic sedimentation gradient observed for most of the year.

Our results are particularly interesting in light of those reported in Sweet et al. (2019), which found low abundance of Endozoicomonas in massive P. lobata less than 30 km from Fouha Bay. Given that the Guam site from Sweet et al. (2019) experiences little to no sedimentation, it is tempting to speculate that severely sedimented environments are associated with the dominance of Endozoicomonas. Alternatively, given massive Porites is a cryptic species complex (Forsman et al., 2009, 2020), these differences might reflect microbial associations correlating with discrete host lineages. The Fouha Bay P. lobata population might show an absence of community changes along the sedimentation gradient because it belongs to a lineage with an inflexible bacterial microbiome, for which there is also precedent in Endozoicomonas coral dominated colonies (Pogoreutz et al., 2018). Regardless, future work investigating host genetic structure as well as common garden experiments exposing corals to different sediment loads would be the next steps to understanding the forces that structure these microbial communities.

Only the rarer bacterial taxa demonstrated abundance patterning according to the site along the sedimentation gradient. The single ASV that was differentially abundant across sites for both center and edge samples belongs to the family Planococcaceae and was most abundant closest to the river mouth, regardless of where on the colony we sampled. This family contains many genera that exhibit diverse functions (Shivaji et al., 2014) making it difficult to disentangle the potential role these microbes are playing. However, this family is well represented in coral samples worldwide (Rodríguez-Gómez et al., 2021) and has also previously been reported to change abundance based on proximity to different pollution sources (Qin et al., 2020). Another ASV which belongs to the genus Peredibacter was differentially abundant across the sedimentation gradient for both edge and sediment samples. This ASV was also more abundant closest to the river mouth and is part of a genus classified as a predatory, Gram-negative, bacteriovorus group ubiquitous in soil and sewage (Davidov and Jurkevitch, 2004) that has previously been found in association with Pocillopora coral hosts (Doering et al., 2021). In edge samples, two other bacteriovorus ASVs–Pirellula and Rhodopirellula–were also in their greatest abundance at sites by the river mouth. The presence of these three bacteriovores in greater abundance near the river mouth at the edge of colonies might be significant as micropredators can drive microbiome changes in corals even at low abundances (Doering et al., 2021).

Two pathogenic bacteria, Catenococcus and Cohaesibacter were also present in highest abundances at the edge of coral colonies closer to the river mouth. Catenococcus is a member of the Vibrionaceae family and has been described as a pathogen in the seaweed Kappaphycus alvarezii in which infection by Catenococcus thiocyli causes bleaching (Zheng et al., 2016), and showed toxic activity in both sponges (Yoghiapiscessa et al., 2016) and clams (Guibert et al., 2020), suggesting that edges of coral colonies may be more susceptible to infection closer to the river mouth. Cohaesibacter has been shown to be enriched in corals exhibiting stony coral tissue loss disease (Rosales et al., 2020; Becker et al., 2021) as well as in white plague-affected corals (Sunagawa et al., 2009; Roder et al., 2014), providing further evidence that colonies at the high sedimentation site might be more at risk for coral disease. Two ASVs from putative denitrifiers Filomicrobium (Asakura et al., 2014) and Kiloniellaceae (Wiese et al., 2020) were also abundant closest to the river mouth. These findings suggest the functions of the bacterial communities are largely the same across sites, consistent with the observed lack of pathway enrichment, but differences in some of the rarer bacteria reflects the potential for differential functional capabilities. Additionally, these site differences in rare bacteria are only seen in edge samples suggesting the edges of coral colonies are more vulnerable to infection by potentially harmful bacteria compared to the center of the coral colony.

We acknowledge that there are limitations with this study in regard to the association between abundance of these rare ASVs and sediment stress. There are other environmental variables that we are unable to decouple across the sediment gradient. For one, we note that the site closest to the river experiences more extreme drops in salinity throughout the year compared to the other two sites, likely due to the freshwater influx during the rainy season. Previous research on the impact of reduced salinity on coral microbiomes is limited, but short-term salinity stress has been shown to increase Vibrio infection of the coral Montipora capitata (Shore-Maggio et al., 2018), suggesting hyposalinity can restructure bacterial communities. The site closest to the river also experiences lower light. While this is likely due to the presence of sediment in the water column, it demonstrates the difficulty of disentangling the exact mechanisms by which higher sedimentation might be impacting microbial communities. Finally, we acknowledge the possibility that the overlap between sediment and coral bacterial communities for these few rare taxa could be a result of coral expulsion of these taxa into the sediment instead of pickup from the sediment into the coral microbiome. Future work should include sampling along a sediment gradient in areas without coral colonies as well as ex situ experiments designed to parse apart the relative impacts of sediment and salinity on the structuring of coral bacterial communities.

Contrary to the minimal differences in coral-associated bacterial communities observed across the sites along the sedimentation gradient, we find significant differences in bacterial diversity between samples collected from the center and edge of colonies. Similar variation in bacterial communities has previously been shown for massive Porites across the skeleton (Marcelino et al., 2018), but we describe discrete bacterial communities based on intra-colony position within the tissue of massive Porites. High diversity differences between positions of the colony are likely due to proximity of the edge samples to surrounding sediments. This is reflected in the greater similarity between edge and sediment communities versus center and sediment communities. Additionally, many ASVs that were observed in greater abundance at the edge compared to the center were not differentially abundant between the edge and the sediment. One might have also expected to observe higher overlap between edge and sediment communities at sites with greater sedimentation, but this was not a pattern observed in our dataset. However, this study only sampled a single time point during the dry season, making it possible that edge communities exhibit increased similarity with sediment communities during large scale sedimentation events during the rainy season. Additionally, we found that colonies that hosted different Cladocopium genotypes between intra-colony positions did not show any greater differences in their bacterial communities, suggesting the forces structuring microbial communities are different between the algal and bacterial partners.

These differences in bacterial composition between colony edge and center samples also corresponded to substantial differences in functional profiles of the bacterial communities. Edge and sediment communities both showed functional enrichment of pathways relating to cobalamin and amino acid synthesis compared to center communities. Both Symbiodiniaceae and coral cells lack the ability to synthesize certain vitamins, including vitamin B12 (cobalamin), which is important for nucleic acid biosynthesis (Wagle et al., 1958) and is provided to them by bacterial symbionts (Hopkinson and Morel, 2009; Agostini et al., 2012). Greater availability of these important compounds at the edge might sustain increased coral tissue growth rates at the leading edge compared to the center of a coral colony. Edge communities were also enriched for methanogenesis and denitrification pathways compared to the center. As high sedimentation can produce low oxygen environments (Li et al., 2014), enrichment of methanogenesis at the edge might compensate for carbon buildup by providing alternative mechanisms for carbohydrate breakdown at the colony edge that do not require oxygen. Higher denitrification at the edge might be a response to a disruption in the N to P ratio as bacteria that mediate N availability through denitrifying processes can strengthen host tolerance to nutrient replete conditions and help maintain a favorable N to P ratio (Tilstra et al., 2019). However, given Fouha Bay’s proximity to a riverine environment, it seems unlikely that these corals are living under nitrogen limited conditions. Regardless, nitrogen availability impacts the coral-algal symbiont relationship as this association is largely maintained through provision and limitation of N from host to symbiont (Falkowski et al., 1993; Beraud et al., 2013; Tilstra et al., 2019). Intra-colony spatial differences in nitrogen availability suggest possible differences in the stability of this relationship across the colony. At the center, bacterial communities showed greater enrichment in functions relating to carbohydrate metabolism. This same pattern is exhibited by bacterial communities in coral colonies that are more resilient to heat stress (Ziegler et al., 2017). As heat stress can modulate the relative abundance of different sugar compounds in coral mucus (Lee et al., 2016), it is possible sediment stress is acting similarly, resulting in niche differentiation between the edge and center and ultimately discrete bacterial communities that have different metabolic profiles. Further research should investigate the concentration of these relevant compounds across the colony and in coral tissue experiencing different levels of sediment exposure to corroborate these interpretations.