Temporal stability of Orbicella annularis symbioses: a case study in The Bahamas

.— Orbicella annularis (Ellis and Solander, 1786), a key reef building species, is unusual among Caribbean corals in the flexibility it displays in its symbioses with dinoflagellates in the family Symbiodiniaceae. This variability has been documented at a range of spatial scales; from within and between colonies to scales spanning the entire species range. However, temporal variability in Symbiodiniaceae communities found within O. annularis colonies is not well understood. Evidence suggests that symbiont communities in this coral species fluctuate temporally in response to environmental stressors (sporadic changes in abundance and in community composition). In this study, we investigated temporal stability of symbiont communities in O. annularis at four sites in The Bahamas over a period spanning 6 yrs. While the dominant symbiont species, Breviolum minutum (LaJeunesse et al.) J.E.Parkinson & LaJeunesse (formerly ITS2-type B1), remained stable across four patch-reef study sites, finer resolution molecular techniques revealed inter-annual variability in the presence/ absence of cryptic species Durusdinium trenchii (LaJeunesse) LaJeunesse (formerly ITS2-type D1a). Durusdinium trenchii is known to play a role in resistance to environmental stress and may have a protective effect under warm conditions. These results suggest that, while it might take an extreme environmental perturbation to trigger a long-term shift in the dominant symbiont, at background levels, less prevalent symbiont taxa are likely to be continually shuffling their relative abundances as they change in response to seasonal or environmental changes.

Seahorse and Snapshot reefs (San Salvador Island) varied across a 6-yr sampling period. Secondly, we explored how detectable background occurrences of thermallytolerant Durusdinium trenchii (LaJeunesse) (referred to in previous literature as D1a) hosted by O. annularis at these sites varied with time. From this, we explore whether the observed partitioning of Symbiodiniaceae diversity represents a snapshot in time of a dynamic partnership, or whether it represents a suite of more stable symbioses; dependent on this, conclusions regarding symbiont traits can be derived, and betterinformed decisions regarding placement of marine reserves and other conservation actions can be formulated (Mumby et al. 2011). The consequences of symbiont shuffling are important for understanding the capacity of corals to respond to thermal stress events.

Materials and Methods
Sample Collection.-Field sites, established as part of a study on O. annularis (Foster 2007), were accessed using scuba during the third week in June in 2006, and again at the exact same time of year in June 2010, June 2011, and June 2012. Propeller and School House reefs are located approximately 6 km apart on the southwest of New Providence Island (Fig. 1). Propeller reef (25.0064°N, 77.5524°W) is located approximately 1 km from a power station waste-water outflow pipe, and was dominated by large colonies of O. annularis. School House reef (24.9734°N, 77.5051°W), a similar (but slightly smaller) patch reef east of Propeller reef, lies farther (approximately 3.5 km) offshore.
Snapshot and Seahorse reefs are located on San Salvador Island, lying southeast of New Providence on the eastern limit of The Bahamas archipelago, and more exposed to the open waters of the Atlantic Ocean. Seahorse reef (24.1582°N, 74.4839°W), an extensive patch reef with a number of large gorgonians and several Acropora spp. across the site, is located 4.5 km off the most northerly tip of San Salvador. Snapshot reef (24.0314°N, 74.5297°W), located on the western leeward side of San Salvador, is slightly more sheltered. Again, the site was dominated by O. annularis. All sites were between 3 and 4 m deep.
A 10-m wide circular sampling plot encompassing multiple O. annularis colonies was established at each location sensu Foster (Foster 2007). Orbicella annularis typical growth form is multiple disjunct columns (or "ramets"), often with senescent margins, emerging from a basal colony (Weil and Knowlton 1994). Each year, individual ramets were randomly sampled from every O. annularis colony within each plot (to a total of 30 samples per plot), with the target number of samples collected across the four sites and 4 yrs being 480. Each spatially independent ramet (i.e., with no connected tissue) was sampled only once. Seahorse Reef was not accessible in 2010, so the total number of collected samples was 450. The location of each sampled ramet was mapped by recording distance (to the nearest 5 cm) from the center of the sampling plot (marked with a fixed stake), and bearing (°). Individual ramets were not tagged, which means sampling between years did not target the exact same ramet through time. Molecular analysis of 2006 samples revealed ramets within each sampling plot were genetically diverse. While up to three clonemates were detected within a plot (e.g., Seahorse Reef had three sets of clones; independent colonies with identical genotypes), host genetic variability was proven to have little to no influence on symbiont species hosted (see Kennedy et al 2016 for details). Bahamian O.
annularis hosts displayed a significant amount of diversity compared to other Wider Caribbean sites sampled (Kennedy et al 2016). An approximately 1-cm 2 sample was chipped from the edge of each ramet using a hammer and a small chisel. Samples were collected in separate labelled plastic ziplock bags, and on returning to the shore were preserved in 90% ethanol and stored at 4 °C. GPS coordinates, time, weather conditions, and reef position of each site were also recorded. Molecular Analyses.-Holobiont tissue was homogenized and both host and symbiont DNA extracted overnight using Qiagen DNEasy Blood and Tissue kits. Roughly 1 cm 2 of tissue was removed from the skeleton using a scalpel, and overnight lysis performed as per the manufacturer's instructions. Extracted DNA was amplified in a PCR reaction using Symbiodinium-specific rDNA primers "ITS2 Clamp" [modified "ITS-reverse" primer with an additional 39 bp GC clamp (underlined), 5´-CGC CCG CCG CGC CCC GCG CCC GTC CCG CCG CCC CCG CCC GGG ATC CAT ATG CTT AAG TTC AGC GGG T-3´] and "ITSintfor2" (5´-GAA TTG CAG AAC TCC GTG-3´) using a PCR protocol [95 °C × 5 min; followed by 30 cycles of 94 °C (45 s), 57 °C (45 s), 72 °C (60 s); with a final annealing step of 59 °C for 20 min], based on LaJeunesse (2002). The reaction mix comprised 1× PCR reaction buffer, 2.5 mM MgCl 2 , 0.2 mM dNTPs, 2 U Taq DNA Polymerase, and 0.6 µM primer in a 12.5 µl reaction. One microliter of the final eluate was used as a template for amplification. DNA quality in the final eluate was assessed using a nanodrop.
Imaged gels were examined carefully by eye and scored for types, with comparison to a database of other gels used to help identify haplotypes. Representatives of every discrete, prominent band were excised under a UV-transilluminator using 10 µl tips and stored at 4 °C overnight in 1.5 ml eppendorf tubes containing 30 µl RNAse free water. Reamplification was performed with 1 µl eluate, using ITSinfor2 and ITS2-reverse lacking the GC-clamp. Two microliters of the PCR product was then cleaned using Exo-sap (per 100 samples: 5 µl Exonuclease 1 (20U µl −1 ), 10 µl Exonuclease buffer, and 85 µl dH 2 0, along with 20 µl Antarctic phosphatise (5U µl −1 ), 10 µl buffer and 70 µl dH 2 0). Two microliters was added to each sample and put in the thermocycler at 37 °C for 15 min and a further 15 min at 80 °C. Following the determination of concentration of each sample on the nanodrop, samples were diluted to a suitable concentration (6-12 ng µl −1 ) for sequencing. The product was sequenced in both directions using both forward and reverse amplification primers separately (Macrogen). A sequence alignment was performed in Clustal X and checked by eye, prior to comparison against a database of all known Caribbean Symbiodinium types in GeoSymbio database (Franklin et al. 2012).
Real-time PCR in conjunction with high resolution melt (HRM) analysis were used to screen each O. annularis ramet specifically for the presence or absence of low abundance Durusdinium (following Kennedy et al. 2015a). A 312 base pair target region specific to Symbiodinium clade D, located in domain 2 of the LSU gene, was amplified using published qPCR primers (Correa et al. 2009). A 10-µl reaction mix containing 1 mM of both forward and reverse primers-1 µl DNA template, 2× Absolute qPCR SYBR Green Flourescein Mix (Thermo Scientific), and made up with dH 2 0-was amplified in qPCR reactions (CFX96 real-time PCR detection system, Bio-Rad Laboratories, Inc.) using the FAM filter. Reaction conditions were an initial denaturing step of 95 °C for 10 min, followed by 50 PCR cycles of 95, 61, and 72 °C for 30 s each (Correa et al. 2009). A final high-resolution melting (HRM) step entailed a 55 to 95 °C temperature ramp, of 0.2 °C every 2 s. Fluorescence data were collected during each PCR annealing step, and each temperature step of the HRM melt cycle. Each DNA sample was run in duplicate for the clade D primer set, and positive (standard) and negative controls were included on every plate.
While RT-PCR primers lack the specificity to distinguish among Durusdinium species, D. trenchii (ITS2 type D1a, also known as and D1-4) is (1) the species detected by DGGE and (2) understood to be the only representative of Durusdinium in the Caribbean Sea (Kennedy et al. 2015b). For these two reasons, we interpreted RT-PCR detection of Durusdinium as presence of D. trenchii. Data Analysis.-Symbiodiniaceae communities and occurrences of low-abundance D. trenchii were compared within and between sites, and between years. An extended Fisher exact test of independence was used to determine shuffling between dominant symbiont species at each of the four locations, as well as for the region overall (Freeman and Halton 1951).

Results
Temporal Stability of Dominant Symbionts.-The symbiont communities sampled from 248 spatially independent O. annularis ramets were categorized using a combination of PCR-DGGE fingerprinting and sequencing. Every individual generated one bright farthest-migrating band on the gel, corresponding to the B. minutum symbiont (Fig. 2). In many lanes, additional bands were present. Sequencing revealed these either to be B. minutum heteroduplexes, or other Breviolum or Cladocopium species.
Profiles generated from Snapshot reef samples were homogenous and comprised one bright, low-molecular weight B. minutum band on a gel, and two or more slightly larger bands a short way above (Fig. 2). Sequencing revealed all three of these bands to be B. minutum heteroduplexes, in addition to confirming the dominant band as B. minutum (GenBank Accession no. AF333511). There was no observable difference in the dominant symbiont hosted by the Snapshot reef population across the four time points, with B. minutum dominating every sample tested (Fig. 3).
Propeller reef (P) displayed more variable DGGE profiles, although every sample had a dominant B. minutum band, and many showed heteroduplexes similar to those generated from Snapshot reef symbiont samples (Profile A). An additional slowermigrating band was present in several samples from 2006 (e.g., P03 and P06, Fig. 2, Profile A+). Although these bands appeared in a Cladocopium sp. C7 position (an O. annularis specialist), sequencing of this band revealed a different endosymbiont Figure 2. Denaturing gel gradient electrophoresis images revealing the most common banding patterns found at each site. Breviolum minutum (B1) was found in every single coral. Cladocopium ITS2 type C1 (C1) was also commonly dominant. Other species identified included a species similar to ITS2 type B8 (Breviolum sp.) and similar to ITS2 type C62 or maybe C7 (Cladocopium sp.) type, which aligned best with Cladocopium sp. C62 in alignments with all known types (EMBL Accession no. LR215825 GeoSymbio; Franklin et al. 2012). This type has not been previously reported as inhabiting O. annularis. Propeller reef samples shared the faster migrating heteroduplexes present in Snapshot samples, but several samples also showed a slower-migrating band in the approximate D. trenchii position (GenBank Accession no. AF499802). Symbiont communities sampled in 2010 were all dominated by the B. minutum band, but again were quite varied, some communities displaying an uncommon lower band that was unable to be identified, although these were not dominant. The 2011 and 2012 samples appeared more uniform with little difference between symbiont communities hosted by each ramet.
School House reef (N) DGGE fingerprints revealed fairly uniform communities, with most samples (e.g., N07 and N08, Fig. 2) exhibiting profiles similar to Snapshot reef, with a band below B. minutum (identified as Breviolum sp. B10, GenBank Accession no. AF499787), and a few samples showing slightly different banding, with a slower-migrating band around the D. trenchii position. However, re-running some School House reef samples adjacent on the same gel to some samples from Snapshot indicated the paler bands from School House were consistently in a different fingerprint position-slightly slower migrating than in the Snapshot profile. Sequencing of the dominant band and heteroduplexes revealed all to be B. minutum or heteroduplexes. The unusual band in N05 and N06 also had a different sequence (Fig. 2, EMBL Accession no. LR215821); however, the method was unable to confidently identify this symbiont-a BLAST search revealed Breviolum sp. B8 to be the closest match. The 2010, 2011, and 2012 samples all generated a mix of profiles, with some co-dominant for B. minutum and Cladocopium sp. C1 (GenBank Accession no. AF333515), but most being exclusively dominant for B. minutum. Seahorse reef (K) showed relatively simple profiles dominated by a single band in the B. minutum position. A much slower-migrating band (Cladocopium sp. C1) was often co-dominant with B. minutum.
An extended Fisher exact test confirmed that the relative proportion of dominant Cladocopium and Breviolum clades observed remained independent of the sampling time point (P = 0.088, Table 1). This result was also confirmed at site level, except for at Seahorse reef, where a significant change in the proportion of Cladocopium symbionts hosted between 2011 and 2012 meant that community composition was not independent of time (P = 0.042, Table 1). Overall, these results suggest temporal stability in the dominant types hosted by O. annularis across The Bahamas, and at all reef sites included in this study, with the exception of Seahorse Reef.
Temporal Stability of Cryptic Durusdinium trenchii.-Durusdinium trenchii, or D1a, was detected in O. annularis colonies at all four sites, and also at all four time points (Fig. 4). Where D. trenchii was present at a site, it was harbored by 5%-42% of colonies. Over time, there appeared to be a decline in the number of colonies hosting low abundances of D. trenchii, from an average 23% of colonies (across all sites) in 2006, to 12% in 2010, to 9% in 2011, and just 1% in 2012. By 2012, D. trenchii was only detected at one site (School House reef) and in just one sample (Fig. 4). Table 1. Results of Fisher's exact test comparing temporal changes in the dominant symbiont hosted at each location, as determined by denaturing gel gradient electrophoresis analysis, demonstrates that relative proportions of dominant symbionts did not change from year to year at the 1% significance level (significant differences highlighted with an asterisk). Bold indicates total for all sites. a Seahorse reef was excluded from The Bahamas total calculation, as data were missing from 2010. Extended Fisher exact tests suggest temporal differences in the number of colonies hosting D. trenchii at two of the sites-Seahorse reef and School House reef-but not at Snapshot or Propeller reefs (Table 2).

Discussion
The DGGE data show broad temporal stability of dominant symbionts in O. annularis in The Bahamas over a 6-yr time period, suggesting that spatial patterns in symbiont biogeography are likely to be fairly robust over time periods of 5-10 yrs. This stability appears to be a feature of not just O. annularis, but other Caribbean species (Thornhill et al. 2006a, Warner et al. 2006. The finer scale RT-PCR technique revealed interannual variability in background abundances of cryptic species D. trenchii. Together, these results suggest that, while it might take an extreme environmental perturbation to trigger a shift in the dominant symbiont or symbionts, symbiont species within the community are likely to be experiencing continual shuffling in their relative abundances as they jostle in response to seasonal or environmental changes. Temporal Stability of Dominant Symbiont Types.-At all four study sites, at all sampling time points, ITS2-type B1 (B. minutum) was found to be the dominant species in every community, and was hosted by 100% of colonies. Snapshot reef showed a stable community of B. minutum, in all samples, over the 6-year sampling period. At Propeller, Seahorse, and School House reefs, B. minutum was shown to share dominance with Cladocopium sp. C1 in a proportion of colonies, while in 2006, B. minutum from four colonies at Propeller and two colonies at School House reef was co-dominant with Cladocopium and Breviolum symbiont types that could not be confidently identified (possibly C62 and B8, respectively). However, annual variation in the absolute number of ramets hosting a dominant pair of symbionts (as opposed to exclusively B. minutum) were not deemed significant by Fisher's exact tests (P > 0.01). These findings imply substantial temporal stability over a 6-yr time period.
It is difficult to explain the difference between Seahorse reef, which had more ramets hosting mixed symbiont assemblages, and the other sites; this offshore reef site was considerably more exposed than the others, and differences between inner and outer reefs (generally driven by water quality factors) are consistently identified as important in driving symbiont community partitioning (Garren et al. 2006, LaJeunesse et al. 2010, Cooper et al. 2011. Overall, the result appears to be consistent with the other studies that have demonstrated broad temporal stability of O. annularis symbiont communities-in the absence of severe thermal stress events-over comparable time periods (e.g., 2 yrs in Warner et al. 2006; 5 yrs in Thornhill et al. 2006b). Researchers sampling 12 tagged O. annularis colonies in The Bahamas 15 times between August 2000 and August 2004 similarly found all colonies were dominated by B. minutum throughout the duration of sampling, although a few deeper colonies (12 m) had several instances of mixed ITS2 B1 and C12 (Thornhill et al. 2006b). In another study, the same researchers found that none of the genetic variation within the B. minutum populations was attributable to sampling time points from O. annularis [or Orbicella faveolata (Ellis and Solander, 1786)] across The Bahamas (or Florida) (Thornhill et al. 2009), with reef location being more important in explaining diversity (indeed, in O. annularis all symbiont genotypes were site specific).
However, two thermal stress events, one in November 2005, just 8 mo prior to the first sampling point, and a second in October 2010, between the second and third time points, occurred in the region (Fig. 5). These sea surface temperature (SST) anomalies might have been expected to influence symbiont community composition, especially if bleaching was triggered in the colonies studied here. Bleaching events near the study sites were reported by local scientists and dive operators, although no quantitative estimates exist for these periods. On San Salvador during November 2005, the Bahamian Reef Survey team (Earthwatch Institute) noted that "bleaching was evident primarily on Agaricia sp., Favia fragum, and Porites sp. ... limited bleaching of Orbicella annularis was also observed; however, many coral heads of this species displayed significant blanching" (Rollino 2005 . "All sites exhibited some bleaching" (Rollino 2010), although more detailed information on the bleaching severity and impact on different species was not provided. Thermal stress events clearly affected the reefs of San Salvador and New Providence over the period of our study. However, the lack of a severe bleaching response in O. annularis may explain why dominant symbionts hosted were maintained throughout the study period. Whether comparable temporal stability in symbiont dominance would have been maintained throughout a more severe bleaching event-such as observed in 1998-is unknown. Breviolum minutum has clearly maintained its dominance over several warmer-than-usual periods (Fig. 5) and demonstrated a high degree of temporal stability in the holobiont of O. annularis in this region.
Temporal Variability of Cryptic Durusdinium trenchii.-The second part of the present study focused on a thermally tolerant symbiont known to be present in relatively low densities in O. annularis colonies (Kennedy et al. 2015a). Durusdinium trenchii, or D1a, was detected in a proportion of colonies at all sites (at least one time point), but occurrence diminished over the study, with just one colony (out of 69 sampled in 2012) observed to contain D. trenchii in 2012, compared to 22 colonies (out of 95) sampled in 2006. At Seahorse and School House reefs, observed differences between sampling time points were shown to be significant by Fisher's exact tests (P < 0.01).
This result is consistent with previous studies that have reported an eventual post-bleaching reversion to original symbiont community types, which involves a reduction in the presence of D. trenchii (thought to incur trade-offs in terms of calcification efficiency) following increases in density during (or in the buildup to) bleaching (Thornhill et al. 2006b, LaJeunesse et al. 2009). It has been suggested that D. trenchii may naturally persist at trace background levels within symbiont communities, and that stress events, such as elevated SSTs, may trigger a shift in relative abundance to detectable levels (Kennedy et al. 2015a). It is possible that the "blanching" observed by John Rollino in Bahamian O. annularis in November 2005 facilitated shuffling of low D. trenchii abundances to detectable levels 7 mo later, as observed in our 2006 data. Other studies show high amounts of D. trenchii prior to, during, and immediately after bleaching (Thornhill et al. 2006b, LaJeunesse et al. 2009). In a study of symbiont assemblages in Barbados, levels of D. trenchii remained high for at least 7 mo following a bleaching event in October 2005, and were still comparable with bleaching levels in April/May 2006, only dropping off after 24 mo (LaJeunesse et al. 2009). In a Florida study, D. trenchii was detected in the tops of O. annularis colonies prior to the 1998 bleaching event, and remained detectable (by DGGE) in a community of six coral colonies until May 2002, after which it was not again detected (the study ended in August 2004) (Thornhill et al. 2006b). Without sampling prior to 2005, it cannot be known whether these levels (23% of colonies) are typical for Bahamanian O. annularis, although evidence suggests at least 30% of colonies across the Caribbean region contain cryptic D. trenchii (Kennedy et al. 2015a), and 28% of healthy O. annularis colonies contained cryptic D. trenchii in Barbados (LaJeunesse et al. 2009).
Lack of survey data from 2007 to 2009 creates uncertainty as to whether prevalence of D. trenchii declined, or was maintained, between our first two sampling time points. If thermal stress is important in influencing D. trenchii, the 2008 accumulated Degree Heating Weeks (DHWs) (Fig. 5) may have affected the number of colonies hosting D. trenchii, although no bleaching reports were reported for our sites in 2008. The colonies from 2010 were sampled 6 mo prior to reported reef bleaching, although incidences of low abundance D1a have been shown to accumulate in O. annularis in the buildup to bleaching events (LaJeunesse et al 2009). The 12% prevalence of D. trenchii we recorded in 2010 may be a legacy from the 2005 (or possible 2008) thermal stress events, or as an immediate response to the elevated temperatures observed in June 2010 (Fig. 5). The latter seems more likely, as if the declining trends we observe are genuine, then D. trenchii densities are clearly able to drop to undetectable levels within the space of a year (e.g., 2011 to 2012) if environmental conditions are favorable.
In 2011, D. trenchii was detected at three of the four sites studied: again, 6 mo after a reported bleaching event (albeit one in which bleaching was not specifically reported in O. annularis). Durusdinium trenchii was barely detected in 2012; if abundances are linked to thermal stress events, this may be due to more stable temperature conditions enabling other symbionts to gradually re-establish full dominance, thereby supporting previous studies suggesting post-bleaching reversion in O. annularis (Thornhill et al. 2006b).
Two of our study sites, Snapshot and Propeller reef, did not show significant changes in the number of colonies hosting D. trenchii, while Seahorse and School House reef clearly did. Propeller reef-situated relatively near to a warm water outflowmay have maintained a high level of background D. trenchii for this reason. It is impossible to reliably attribute the decline in the number of ramets hosting detectable D. trenchii to thermal stress events with this data set; however, it seems likely that thermal stress may be providing a mechanism by which low level densities of different symbiont taxa are in a continual state of flux. Orbicella annularis is documented as being one of the first species to bleach and has been shown to be more susceptible to water temperature increases than other corals (Fitt et al. 2001). Perhaps a fluctuating presence of D. trenchii reported in the present study may explain why, in 2005 and 2010, O. annularis in The Bahamas appeared to avoid bleaching.
In summary, we revealed temporal variability in the presence of a low-abundance, but potentially ecologically-important symbiont, in key reef-building corals in The Bahamas, despite broader apparent stability in the dominant community members over a 6-yr period. While the results support other studies that show long-term temporal stability of associations between O. annualris and B. minutum, it also highlights the importance of the nuances in symbiont community stability over time, and the importance of detection limits.
With coral reefs facing an unprecedented global crisis (Hughes et al. 2017), understanding these nuances in the coral microbiome will be important in interpreting bleaching responses and determining outcomes for reefs. Underlying mechanisms, such as shuffling of symbiont communities at a micro-scale, have consequences for individual coral colonies, and for O. annularis-an abundant species and dominant reef builder-these small effects may contribute to a coral reefs wider ability to resist or rebound from thermal stress in the face of a rapidly changing climate (Guest et al. 2018).