Persistent Legacy Effects of Marine Heatwaves on Coral Symbioses

ABSTRACT

Obligate endosymbioses between eukaryotes and their single‐celled inhabitants form the basis of many ecosystems, yet little is known about the long‐term impacts of climate change on them. On coral reefs, extensive studies have shown that climate change‐driven heatwaves and other environmental stressors can disrupt the obligate symbiosis between reef‐building corals and Symbiodiniaceae, with consequences for coral fitness and survival. However, despite coral symbioses playing a fundamental role in reef resilience to climate change, whether, and at what rate, they recover following heatwave disruption is largely unknown. We used ITS2 DNA metabarcoding to characterise symbiont assemblages in colonies (n = 237; 598 samples) of the brain coral Platygyra ryukyuensis over a decade (2014–2023), spanning from before to long after the 2015–2016 El Niño at its epicentre, Kiritimati, in the central equatorial Pacific. Although before the heatwave only P. ryukyuensis colonies exposed to high levels of chronic local disturbance were dominated by stress‐tolerant Durusdinium symbionts, surviving colonies around the atoll transitioned during the heatwave from Cladocopium dominance to Durusdinium dominance. Here, we show that nearly eight years after this transition, these symbiotic partnerships had not recovered, but rather Durusdinium remained entrenched in virtually all (92%) Platygyra colonies. Recovery of symbionts in the genus Cladocopium was severely limited and restricted to taxa distinct from their ‘C3’ and ‘C50a’ pre‐heatwave congenerics. Moreover, in the three immediate post‐heatwave years, many tracked corals, and especially those at low local disturbance, high in‐water visibility sites, hosted transient symbiont assemblages codominated by Durusdinium and the previously rare genus Symbiodinium. Our results demonstrate that heatwave‐driven symbiont transitions can persist for longer than the average heatwave return time, potentially impairing coral resilience to future extreme weather events.

The algal symbionts hosted by Platygyra ryukyuensis corals on Kiritimati Island were distributed according to their exposure to local human disturbance before an extreme marine heatwave (1). The symbionts hosted by corals underwent a major shift as a result of this acute heatwave disturbance, with transient symbionts detected in corals at less turbid (more light‐exposed) sites in the three years that followed (2). More than seven years post‐heatwave, the expected recovery of pre‐heatwave symbionts (3) had not occurred, with heatwave‐transformed symbionts still hosted by corals across the island (4).

1. Introduction

Symbiotic microbes support many of the world's most biodiverse ecosystems (Tylianakis et al. 2008). However, the increasing frequency and intensity of ecological perturbations, both acute (e.g., heatwaves, disease outbreaks) and chronic (e.g., pollution, climate change), threaten the persistence of ecosystem services provided by symbioses that co‐evolved with host species to specific environmental niches (Kiers et al. 2010). These stressors are known to be responsible for range shifts, local extinctions and changes to the population structure of eukaryotic hosts (Chen et al. 2011; Wiens 2016). Similar transformative changes are likely for microbial symbiotic species (DeAngelis et al. 2015; Zhou et al. 2016), but despite the disproportionate importance of symbioses for ecosystem function, multi‐year monitoring surveys of microbial symbionts are rare compared with those of their eukaryotic hosts. The rarity of examples from natural populations greatly constrains current understanding of symbiotic stability under climate change and the persistence of the ecosystems that these symbioses support.

In the ocean, climate change poses a major threat to coral reef ecosystems, with rising temperatures driving increasingly frequent and intense heatwaves (Smith et al. 2023). Marine heatwaves often trigger coral bleaching, during which corals lose many of their symbiotic dinoflagellates (family: Symbiodiniaceae), which provide the majority of the host's energy through photosynthesis (Glynn 1996). Bleaching can restructure symbiont assemblages and prolonged bleaching may result in host mortality (Buddemeier and Fautin 1993). Symbiont identity plays an important role in mediating coral responses to thermal stress, recovery and long‐term resilience (Silverstein et al. 2015). Symbionts in the genus Durusdinium are thermotolerant generalists and often increase in relative abundance in a range of coral host species following coral bleaching events (Baker et al. 2004; Berkelmans and van Oppen 2006). In contrast, the genus Cladocopium is comprised of a diversity of highly host‐specific symbiont species (LaJeunesse et al. 2018). The fitness trade‐offs for corals hosting different taxa within the Symbiodiniaceae are responsive to environmental conditions: Whilst Durusdinium symbionts may resist bleaching under elevated temperatures, their photosynthetic contribution can be comparatively lower under ambient temperatures, with coral traits such as growth and reproductive output suffering the consequences in some coral species (Jones and Berkelmans 2011; Matsuda et al. 2023). However, this trade‐off may be restricted to geographic regions other than the tropical Pacific, colonised more recently by Durusdinium (Kemp et al. 2023).

Heatwave‐transformed symbiont assemblages that are dominated by thermotolerant symbionts are generally considered to be ephemeral as the former host‐specific symbiont taxa are recovered. This expectation is based on the putative competitive inferiority and reduced benefits conferred to coral hosts by incumbent generalist symbionts under ambient, non‐stressful conditions (Buddemeier and Fautin 1993; Cunning et al. 2018). However, in natural populations, the few reports of post‐heatwave persistence of heat‐tolerant symbionts have been variable. Intrageneric symbiont changes have been reported to persist for only 2–8 months following heat stress (Sampayo et al. 2008; Thornhill et al. 2009; Lewis et al. 2019), whereas intergeneric replacement, most notably by Durusdinium, has been shown to endure for as little as 5 months in some coral species and in others for 2 to 4 years after heat stress (Thornhill et al. 2006; LaJeunesse et al. 2009; Elder et al. 2022). Despite the critical importance of coral symbioses for reef resilience in a warming ocean, there are few other studies that have tracked coral symbioses following an acute disturbance, leaving the trajectories and drivers of symbiont recovery largely unknown. Moreover, the few long‐term studies of coral symbioses in the absence of acute stress had contrasting results—with one 9‐year study reporting symbiont persistence over time (n = 55 tracked colonies; Lewis et al. 2024), a 16‐year study finding intracolonial shifts in symbiont species and genera relative abundances (n = 12 tracked colonies; Edmunds et al. 2014) and a 10‐year study of haphazardly‐selected corals finding symbiont dynamism but no directional changes (Baker et al. 2013), thus providing little insight into the long‐term stability of coral symbiont partnerships.

Post‐heatwave recovery of coral symbioses may also be influenced by environmental factors. Light intensity has been implicated in symbiont niche‐partitioning, as light‐tolerant symbionts in the genus Symbiodinium are detected in higher abundances at shallow depths, in low‐turbidity (high in‐water visibility) and even on the shallowest, least shaded surfaces of individual colonies (Rowan and Knowlton 1995; Kemp et al. 2015). In contrast, the proliferation of Durusdinium symbionts is seemingly favoured in corals subject to high turbidity and high inorganic nutrient pollution (nitrogen in particular), characteristic of coral reefs exposed to higher anthropogenic disturbance (Claar et al. 2020; Poquita‐Du et al. 2020). It remains unclear if local variation in human disturbance or other environmental factors influences symbiont competitive dynamics during the symbiotic instability that follows coral bleaching.

Here, we present one of the longest time‐series of Symbiodiniaceae assemblages in tracked coral colonies to date, 598 samples from 237 distinct coral colonies collected over a decade (2014–2023) at the epicentre of the 2015–2016 El Niño‐induced global coral bleaching event (Kiritimati, Republic of Kiribati, central equatorial Pacific) at sites spanning a strong gradient of chronic local disturbance (Baum et al. 2023). Platygyra brain corals are widely dispersed across Indo‐Pacific reefs and demonstrate the symbiotic flexibility typical of the Merulinidae family. Before the 2015–2016 El Niño, local human disturbance exposure strongly influenced symbiont assemblages hosted by Platygyra ryukyuensis on Kiritimati, with those exposed to higher human disturbance hosting predominantly Durusdinium (Claar et al. 2020). However, by the end of 2016, following this marine heatwave, the entirety of the Kiritimati P. ryukyuensis population had shifted to Durusdinium (Claar et al. 2020). In this study, we investigate recovery trajectories and rates for coral‐Symbiodiniaceae associations following heatwave‐induced disruption for coral colonies exposed to distinct levels of underlying chronic local anthropogenic disturbance. We hypothesise that corals will revert to their pre‐heatwave Symbiodiniaceae assemblages as the competitive superiority of these symbionts returns following the end of heat stress. We further hypothesise that the post‐heatwave trajectory of symbiont succession will be influenced by local environmental variables known to define symbiont niches, as they compete to recolonise corals recovering from bleaching.

2. Materials and Methods

2.1. Study Site

This study was conducted at Kiritimati (Christmas Island), Republic of Kiribati, a remote atoll situated in the central equatorial Pacific Ocean (01°52′ N, 157°24′ W). Kiritimati is the largest atoll in the world by landmass (388 km²), with a continuous fringing reef encircling the island. Kiritimati's population was estimated as 6500 in 2015, increasing by 14% to 7400 by 2020 (Kiribati National Statistics Office 2016, 2021), distributed mostly along the northeast lagoon face (Figure 1a), exposing adjacent reefs to disturbance from wastewater runoff, subsistence fishing, coastal construction and dredging (Figure 1a). A total of 12 sites were sampled for this study across this disturbance gradient, ranging from reefs adjacent to villages to minimally impacted reefs at the remote ends of the atoll. At each site, samples were collected along 60 m transects, laid along the 10–12 m isobath.

FIGURE 1.

Spatiotemporal patterns of chronic and acute stressors on Kiritimati reefs (a) Labelled points indicate locations of the 12 reef sites sampled on a map of Kiritimati. Sites are categorised as very low, low, medium, high or very high human disturbance. Permanent settlements are depicted by red circles scaled according to population size. A dashed line bisects the island to indicate shores designated as windward or leeward. (b) Mean (+/− standard error) measurements of environmental metrics at each site from 2014 to 2023. Satellite‐derived sea‐surface chlorophyll a concentration, surface water nitrate and nitrite concentration, and in‐water visibility are shown. Study sites are ordered along the x‐axis by location, moving clockwise windward to leeward. (c) Satellite‐derived sea‐surface temperature (°C) and accumulated heat stress (maximum degree‐heating‐weeks, DHW) are shown by blue and black line graphs, respectively. Colour fills under the DHW line represent periods over which thresholds of 4, 8, 12, 16 and 20 DHW were breached, corresponding to respective Coral Reef Watch bleaching alert levels 1–5. The dashed blue horizontal line indicates the average threshold for SST anomalies (1°C above maximum monthly mean for all sites), above which DHW begin to accumulate. Dotted vertical lines mark ten sampling expeditions carried out between August 2014 and September 2023. Grey boxes indicate categorical expedition classes used to visualise symbiont recovery over time.

Kiritimati's reefs were heavily impacted by the 2015–2016 El Niño, which caused widespread coral bleaching and mortality across the atoll (Baum et al. 2023). Sampling for this study spanned both pre‐ and post‐heatwave periods (2014–2023), with collections conducted two years before the marine heatwave and extending to more than seven years after temperatures returned to baseline conditions.

2.2. Quantifying Acute and Chronic Disturbance Exposure

To quantify acute heat stress exposure on Kiritimati throughout this study, satellite‐derived sea surface temperature (SST) at ~5 km² resolution was extracted from NOAA's ‘CoralTemp’ product and used to calculate daily average SST between 1st July 2014 and 1st October 2023 (NOAA Coral Reef Watch 2019) (Figure 1c). Degree‐Heating‐Week (DHW) data were extracted from NOAA's ‘Coral Reef Watch’ product to evaluate accumulated heat stress. The DHW metric represents the compounded duration and amplitude of SST anomalies in ~5 km² pixels over a 12‐week rolling window when temperatures exceed the pixel's maximum monthly mean (MMM) by more than 1°C. In situ temperature loggers were also deployed at the 12 reef sites in this study, and when compared for 2011 to 2016, revealed close agreement with NOAA's satellite‐derived temperature data and consistency in SST variance and SST anomalies (Claar et al. 2019). The 12 reef sites included in this study all experienced severe heat stress during an extreme pan‐Pacific marine heatwave in 2015–2016, with minimal variation in maximum accumulated heat stress between all 12 reef sites (24.26–25.28 DHW) when evaluated using in situ and satellite‐based data (Claar et al. 2019).

Kiritimati's local human disturbance gradient has been previously quantified using two spatial data layers to calculate a combined metric of chronic human disturbance at each reef site (Table S1): a kernel density function of subsistence‐fishing pressure within 10 km based on community interview data (Watson et al. 2016) and estimates of point‐source waste‐water exposure based on a geographic buffer of human population size within 2 km. Anthropogenic nutrient pollution may directly mediate coral‐Symbiodiniaceae interactions (Allgeier et al. 2020), whereas fishing practices can shape these symbioses directly through increased sedimentation and physical damage to corals (Ying et al. 2021) or through changes to reef fish populations. Specifically, herbivorous fish can reduce reef turf algal cover with implications for coral–microbial associations (Zaneveld et al. 2016) and corallivorous fish shape the dispersal of Symbiodiniaceae (Grupstra et al. 2021). Natural breakpoints in this combined disturbance metric allow sites to be discretely categorised by local disturbance, which have been used to explain variability in many facets of coral reef health and function around Kiritimati (Magel et al. 2019; McDevitt‐Irwin et al. 2019; Claar et al. 2020; Baum et al. 2023).

Several additional environmental factors differ between reef sites (Baum et al. 2023) and were therefore also characterized: reef aspect (windward‐ or leeward‐facing), surface‐water productivity, inorganic nitrogen and in‐water visibility. The prevailing wind direction from 1979 to 2009 (Bosserelle et al. 2015) was used to categorise sites as windward‐ or leeward‐facing (Figure 1a). Persistent and concentrated upwelling off Kiritimati's western shore (Walsh 2010) was observed using 4 km² resolution daily chlorophyll a satellite data as a proxy for surface water phytoplankton content and productivity (NOAA CoastWatch Central 2012), and mean annual chlorophyll a concentration (μM/m³) was calculated for each reef site for 2014 to 2023 (Figure 1b). In‐water visibility (in metres) was estimated at each site by the same scientific diver each day the site was visited and averaged per expedition. Visibility is inversely proportional to turbidity and is a reliable indicator of the photosynthetically active radiation reaching the benthos on tropical reefs (Jones et al. 2015); on Kiritimati, visibility is correlated with local human disturbance, with more disturbed sites tending to have lower visibility (Baum et al. 2023; Figure S1). Three surface water samples of 30 mL were collected per site per expedition for inorganic nutrient content analysis and frozen on the day of collection before quantification at the Institute of Ocean Sciences (BC, Canada) (McDevitt‐Irwin et al. 2019). Combined nitrate and nitrite molar concentrations are taken as a proxy for seawater inorganic nitrogen content. Not all sites were visited on every expedition to collect visibility and inorganic nutrient data due to inclement weather or logistics (Table S2).

2.3. Sampling Coral‐Associated Symbiodiniaceae

Coral tissue samples were collected over the course of 10 expeditions to Kiritimati: August 2014, January 2015, May 2015, July 2015, March 2016, November 2016, July 2017, July 2018, July 2019 and September 2023 (Figure 1). Samples for this study were exclusively collected from the most common morphotype of the genus Platygyra at Kiritimati, described here as belonging to the species P. ryukyuensis based on key skeletal characteristics (meandroid colonies with short and occasionally monocentric valleys and irregular septo‐costae) (Veron 1995) (Figure S2). On each expedition, only P. ryukyuensis colonies that were at least 20 cm in diameter and at least 5 m apart were tagged and sampled. Coral colony size can be used as a reasonably reliable proxy for age (Madin, Hoogenboom, et al. 2016) and the congener Platygyra sinensis has been found to increase slowly in diameter by 6–12 mm annually (Madin, Anderson, et al. 2016). Hence the minimum size threshold of 20 cm diameter was used to select Platygyra ryukyuensis colonies that were alive before peak heatwave exposure. Previously tagged colonies were re‐sampled whenever possible on subsequent expeditions and are referred to as tracked colonies (Figure S3). However, substantial mortality of this species during the heatwave, especially at sites exposed to Very High disturbance (Claar et al. 2020; Baum et al. 2023), necessitated tagging new colonies on each post‐heatwave expedition. Tissue samples were taken from the central, uppermost surface of colonies using a woodworking chisel, and then stored in individual bags on ice until the end of each day, when ~1 cm³ of tissue from each sample was transferred to microcentrifuge tubes containing a guanidinium isothiocyanate buffer. Samples were stored at 4°C until DNA extraction was performed. Samples were collected from colonies' uppermost surface for consistency and to avoid sampling shaded microhabitats in a colony's morphology, which can contain distinct Symbiodiniaceae (Kemp et al. 2015; Lewis et al. 2022).

2.4. Symbiodiniaceae DNA Extraction and Sequencing

DNA was extracted from coral tissue samples preserved in guanidinium isothiocyanate buffer using a protocol designed for Symbiodiniaceae (Stat et al. 2009). 200 μL subsamples of homogenised tissue/buffer slurry were incubated overnight at −20°C with an equal volume of isopropanol. Precipitated DNA was then collected into a pellet by centrifugation and washed three times with 70% ethanol before resuspension in 50 μL Tris buffer. The internal transcribed spacer 2 (ITS2) ribosomal DNA was amplified using a two‐step PCR for DNA metabarcoding. The first round of PCR used forward primer ITS‐DINO (Pochon et al. 2001) and reverse primer ITS2Rev2 (Stat et al. 2009) with standard Illumina adapters to amplify a ~271‐303 bp amplicon. PCR amplicons were visualised with gel electrophoresis on a 2% agarose gel. Samples that failed to amplify were re‐amplified using template DNA that was cleaned using a Genomic DNA Clean & Concentrator kit (Zymo Research) according to the manufacturer's protocol. Negative controls from each DNA extraction and PCR performed were used to check for contamination using gel electrophoresis. Library preparation followed an adapted Illumina 16S Metagenomics protocol (Illumina 2015; Hume et al. 2020), with two AMPure bead (Beckman Coulter) clean‐ups and an indexing PCR using Nextera primers (Illumina). Indexed amplicons were normalised using a PicoGreen dsDNA assay (Invitrogen) before pooling for multiplex sequencing. Sequencing was performed across ten runs on the Illumina MiSeq platform (MiSeq Reagent Kit v3 600‐cycle), either at the Hawaii Institute of Marine Biology Genetics Core Lab or at the John G. Shedd Aquarium (Chicago) Molecular Ecology Laboratory, producing on average 28,140 raw contigs per sample.

2.5. Clustering and Classifying Symbiodiniaceae Sequence Data

ITS2 sequence data were analysed using two workflows to classify distinct taxa: (i) SymPortal, which uses co‐occurrence patterns of Symbiodiniaceae sequence variants to identify defining intragenomic variants (DIVs) and further clusters these into ITS2 profiles (Hume et al. 2019) and (ii) DADA2, which identifies amplicon sequence variants (ASVs) using a denoising algorithm that models sequencing errors (Callahan et al. 2016). These two approaches of distinguishing intra‐ and inter‐genomic sequence variation from the multi‐copy ITS2 region in Symbiodiniaceae were evaluated in parallel to provide more robust interpretation of sequencing results (Quigley et al. 2022). Sequence data were submitted to the remote SymPortal database (symportal.org). For the DADA2 approach, primer sequences were removed using the ‘cutadapt’ tool, then forward and reverse reads were trimmed by 100 bp at the 3′ and by an additional 31 bp at the 5′ end of reverse reads based on declines in read quality and to ensure equal amplicon sequence lengths with the SymPortal‐based approach. Trimmed reads were then denoised and merged using default parameters for read quality and error rates. Samples with fewer than 500 total post‐processing Symbiodiniaceae reads were removed from the analyses. Incidences of ITS2 profiles, DIVs or ASVs comprising < 1% of Symbiodiniaceae reads in a sample were also discarded. ASVs were assigned genus‐level taxonomy using a naïve Bayesian classifier algorithm, implemented in DADA2, with a minimum bootstrap confidence of 80 using a comprehensive Symbiodiniaceae reference database available within the online data repository (Nitschke 2020). Separate neighbour‐joining trees based on Tamura‐Nei distances were built in Geneious Prime (version 2023.2.1) for each symbiont genus, to calculate unifrac similarity between ASVs.

The impact of variation in ITS2 copy number on the quantitative accuracy of relative read abundance (RRA) of symbiont genera (Davies et al. 2023) was accounted for using two approaches. Changes in the numerical and RRA of the three most abundant symbiont genera were investigated in this study and include Cladocopium (C), Durusdinium (D) and Symbiodinium (A). Changes in Symbiodinium RRA were inferred only when samples were collected from the same coral colony. In all other cases, categorical assemblage designations were assigned based on the natural bimodality of symbiont RRA observed (Figure S4) and thresholds for dominant and background symbionts suggested in other studies (Silverstein et al. 2012; Quigley et al. 2014). Samples were designated dominant when only one genus was observed with > 90% RRA and codominant when more than one symbiont genus was detected with RRA between 10% and 90%.

2.6. Statistical Analyses

To investigate the probability of reversion to pre‐heatwave symbiotic associations in the P. ryukyuensis population, logistic regressions of the absence or presence of Durusdinium (hosted as the dominant or codominant genus) were fit against continuous human disturbance exposure and sample collection date as fixed effects, with coral colony included as a random effect. Unifrac distances, calculated from sequence similarities between ITS2 profile composite sequence variants and from ASV sequence alignments, were used to generate pairwise dissimilarities between Cladocopium or Durusdinium taxa detected in samples before and after the heatwave at the same site. Sites not represented in both 2014 and 2015 and in 2016–2023 were not included in pairwise analyses. Unifrac dissimilarities for the genus Symbiodinium were not reported because it was rarely detected before the heatwave and few corals hosted this genus. The ‘pairwiseAdonis’ wrapper was used to perform pairwise PERMANOVA with sites included as strata (Martinez Arbizu 2017), to test for significant unifrac dissimilarities within each symbiont genus. A subset of the 24 coral colonies that were sampled both before and after the heatwave was used to assess symbiotic reversion within tracked colonies (Sections 3.1 and 3.2; Figure S3).

Sequential sample pairs (363 samples in total) spanning 2016–2017 (either between expeditions 2016a and 2016b or 2016b and 2017), 2017–2018, 2018–2019 or 2019–2023, were used to track post‐heatwave changes in Symbiodiniaceae assemblages within 107 distinct tracked coral colonies (Section 3.3; Figure S3). Beta regression models were fit to post‐heatwave changes in Symbiodinium RRA in tracked corals against continuous human disturbance exposure. Corals were grouped by expedition class as described earlier and the Bonferroni approach was used to correct P values for multiple comparisons. The effects of additional environmental variables on changes in Symbiodinium RRA in these resamples were tested by fitting a beta generalised mixed‐effects model with ‘glmmTMB’ (Brooks 2024): site, coral colony and categorical human disturbance exposure were all included as random effects. Due to the low temporal resolution of these environmental data during this period (Table S2), site‐level annual measures of chlorophyll a, inorganic nitrogen and visibility were averaged from 2016 to 2023 and included as fixed effects along with the windward or leeward aspect of each site.

3. Results

3.1. Persistent Dominance of Durusdinium Across All Reefs

In 2023, more than seven years after the end of the heatwave, all but one P. ryukyuensis corals (n = 144/145) hosted the symbiont genus Durusdinium, in striking contrast to the reef‐wide associations observed before the event (25 of 54 colonies (46.3%) containing Durusdinium at some point) (Figure 2a). Almost all colonies sampled in 2023 (91.7%) were classified as Durusdinium dominant, while 7.6% of colonies were codominant with another symbiont genus (2.8% Cladocopium; 4.8% Symbiodinium; Figure 2a). Counter to our hypothesis of reversion to Cladocopium following a return to non‐stressful temperatures, time had no significant effect on the proportion of colonies hosting Durusdinium from 2016 onwards (z = 0.03, p > 0.98). The effect of human disturbance on the probability of P. ryukyuensis corals hosting Durusdinium, significant before the heatwave (z = 5.95, p < 0.001), was not apparent post‐heatwave (z = 0.20, p > 0.84; Figure 2b), reflecting a homogenisation of symbioses following exposure to chronic human disturbance and acute heatwave stressors.

FIGURE 2.

Variation in symbiont assemblages hosted by Platygyra ryukyuensis by expedition and disturbance. (a) 598 samples collected across expeditions grouped into four expedition classes. Total samples and coral colonies sampled in each expedition class are indicated on survey timeline. Symbiont assemblages are grouped by the incidence of genera with relative read abundance (RRA) > 10%. C = Cladocopium; D = Durusdinium; and A = Symbiodinium. One sample co‐dominated by all three genera (C + D + A), and two samples hosting codominant C + A not shown. Bars represent the proportion of samples within each period and are filled by human disturbance category. (b) Logistic regressions for each expedition class showing colonies with dominant or co‐dominant Durusdinium against site‐level human disturbance exposure (categorical designations shown by coloured bars above x‐axis). Points represent individual samples, with some random horizontal noise added to reduce overplotting. Peak marine heatwave (MHW) exposure is denoted with a red hashed vertical line between expeditions 2015c and 2016a and the hashed horizontal line marks a hiatus in sampling between 2019 and 2023.

These enduring population‐scale changes in the symbiont assemblages of P. ryukyuensis are further supported by the persistence of Durusdinium in individually tracked P. ryukyuensis colonies that were sampled both before and after the heatwave (n = 24 colonies). All but one of these colonies hosted assemblages dominated or codominated by Durusdinium post‐heatwave, despite most exclusively hosting Cladocopium before peak heatwave exposure (Figure S6).

3.2. No Recovery of Pre‐Heatwave Cladocopium Taxa

In addition to the post‐heatwave entrenchment of Durusdinium, none of the 16 Cladocopium ITS2 profiles found in P. ryukyuensis before the heatwave (n = 107 samples) were observed after the heatwave. The most abundant pre‐heatwave Cladocopium ITS2 profiles, characterised by the ‘C3’ and ‘C50a’ sequence variants, were entirely replaced by phylogenetically distinct profiles comprised of mainly ‘C15’, ‘C15h’ or ‘C116’ variants, which had not been detected in the P. ryukyuensis population before the heatwave (Figure 3a). The disparity between pre‐ and post‐heatwave Cladocopium taxa is further indicated by significant differences in paired unifrac dissimilarities between pre‐ and post‐heatwave samples containing dominant, codominant or background Cladocopium hosted by corals at the same site (ASVs: F = 25.0, R² = 0.191, p < 0.001, ITS2 profiles: F = 74.7, R² = 0.430, p < 0.001). No analogous difference was detected in Durusdinium ASVs (F = 0.621, R² = 0.002, p > 0.24) or ITS2 profiles (F = 0.436, R² = 0.001, p > 0.931) hosted before and after the heatwave, with > 90% (n = 434 of 481) post‐heatwave samples containing a Durusdinium profile that had been detected pre‐heatwave. Furthermore, the ITS2 profiles in 530 of the 531 samples with dominant or codominant Durusdinium were all characterised by the ‘D1’ sequence variant (Figure S7).

FIGURE 3.

Shift in Cladocopium biodiversity before and after heatwave. (a) Bubble plot showing the number of colonies hosting Cladocopium as dominant or codominant symbionts (n = 88) at any timepoint. ITS2 profiles are ordered by their majority sequence variant(s) (b) Stacked bar plot showing Symbiodiniaceae assemblages in the n = 24 colonies sampled both before and after the heatwave. Cladocopium profiles coloured as depicted in (a). Vertical dashed red lines indicate peak marine heatwave (MHW) exposure.

Of the 24 individually tracked P. ryukyuensis colonies that were sampled both before and after the heatwave, most that hosted Cladocopium pre‐heatwave hosted Durusdinium or Durusdinium‐Symbiodinium codominant assemblages post‐heatwave (Figure 3b). The few colonies that hosted Cladocopium before and after peak heat stress underwent a shift in symbiont sequence variants matching the changes observed at the population scale (Figure 3b), indicating that the phylogenetic shift in Cladocopium lineages was not an artefact of sampling different colonies pre‐ and post‐heatwave.

3.3. Post‐Heatwave Symbiodinium Dynamism

Extensive dynamism in symbiotic assemblages was also observed after the heatwave, driven by the emergent genus Symbiodinium. Symbiodinium was observed in 43% of all samples (214 of 491, in 113 distinct colonies) collected after the heatwave, contrasting 9.3% of samples (10 of 107, in 8 distinct colonies) before the heatwave (Figure S6). The RRA of Symbiodinium post‐heatwave was highly variable within each expedition (0%–55.1%) and over subsequent expeditions (mean 8.14% ΔRRA; Figure 4a), from the 107 colonies sampled in at least two successive post‐heatwave expeditions. This contrasts with the overall consistent proportion of colonies codominant for both genera observed during expeditions from 2016 to 2019 (Figure 2a).

FIGURE 4.

Dynamism of Symbiodinium (measured as RRA) in Platygyra ryukyuensis colonies post‐heatwave. (a) Heatmap of Symbiodinium RRA by expedition for those colonies that were sampled at least twice post‐heatwave over successive expeditions, with dashed vertical red line showing the timing of the marine heatwave. Each row represents a distinct tracked colony ordered by human disturbance exposure. For a list of the 107 tracked colonies included in this figure, see Figure S3. (b) Beta regressions of the change in Symbiodinium RRA in corals sampled in successive expeditions against human disturbance exposure. Points represent tracked colonies with some random horizontal variation added to reduce point overlap. (c) Estimated effects on the change in Symbiodinium RRA in successively sampled colonies post‐heatwave attributed to variation in site environmental variables. Lines represent 95% confidence estimates of effect sizes (points) based on beta mixed‐effects model coefficients, shown on the model's log‐odds scale. Effects significantly different from zero are shown in black, those not significantly different from zero are grey.

Further comparison of colonies tracked across successive post‐heatwave expeditions (n = 107 corals, 363 samples) identified distinct trends in the change in Symbiodinium RRA. Although there was no significant trend in Symbiodinium RRA associated with human disturbance exposure over the 8–12‐month sampling periods between successive expeditions from March 2016 to July 2019 (z = −2.173, P _Bonferroni  = 0.119), a significant positive relationship between the change in Symbiodinium RRA and human disturbance was observed over the longer and later time period (due to COVID‐19) between July 2019 and September 2023 (z = 2.701, P _Bonferroni  = 0.028, Figure 4b). This largely reflects differences in the retention of Symbiodinium across the disturbance gradient, with high RRA enduring until 2019 at the least disturbed sites, when most colonies at medium and high disturbance sites hosted exclusively Durusdinium or Symbiodinium with low RRA. Subsequently, colonies under lower human disturbance underwent larger declines in Symbiodinium RRA from 2019 to 2023 as colonies became dominated by Durusdinium (Figure 4a).

The dynamism of Symbiodinium associations may be attributed to between‐site differences in in‐water visibility. Although the estimated effects of site‐level nitrate and nitrite, chlorophyll a and the windward or leeward status of each site on changes in Symbiodinium RRA were not significantly different from zero at any point post‐heatwave, post‐heatwave Symbiodinium RRA significantly increased with higher site‐level in‐water visibility until 2019 (z = 2.28, p < 0.023), and subsequently decreased with in‐water visibility between 2019 and 2023 (z = 4.00, p < 0.001) (Figure 4c).

4. Discussion

Our decade‐long survey of coral symbioses reveals enduring transformations of coral‐Symbiodiniaceae associations in the wake of an extreme marine heatwave. The heat‐tolerant genus Durusdinium remained dominant in P. ryukyuensis corals across Kiritimati more than seven years after heat‐stress exposure, revealing an alternative pathway to long‐term symbiotic transformation beyond the cumulative impacts of multiple heatwaves (Quigley et al. 2022; Palacio‐Castro et al. 2023). We observed the decimation of a previously diverse community of Cladocopium symbionts. Expanding upon studies reporting heatwave‐driven turnovers and losses of Cladocopium richness (Quigley et al. 2022; Starko et al. 2023), our findings reveal that this transformation can persist, leading to the disappearance of several Cladocopium species from within P. ryukyuensis assemblages. Corals also experienced short‐lived increases in Symbiodinium symbionts, revealing a transient role of local environmental drivers in post‐heatwave symbiotic changes and providing support for ex situ experimental evidence of dynamic Symbiodinium‐Durusdinium competition dynamics (McIlroy et al. 2019). Taken together, these results reveal a path of post‐heatwave symbiont succession distinct from trajectories observed following milder acute disturbance exposure. As the frequency and intensity of marine heatwaves increase, such persistence of heatwave‐transformed symbioses and homogenisation across gradients of human disturbance may come to characterise Symbiodiniaceae associations in many symbiotically flexible coral species.

4.1. Extended Heatwave Duration Disrupts Symbiotic Reversion: Durusdinium

Despite a 7‐year period free from additional heat stress exposure following the 2015–2016 heatwave, Durusdinium remained overwhelmingly dominant in Kiritimati's P. ryukyuensis population. This result extends the duration that heat stress‐induced Durusdinium is found to persist in natural systems by several years, with previous reports ranging between 5 months and 4 years post‐heatwave (Thornhill et al. 2006; LaJeunesse et al. 2009; Keshavmurthy et al. 2020). Reef‐wide replacement of Cladocopium by heat‐tolerant Durusdinium was previously hypothesised to occur by repeated severe heat stress exposure without intervening symbiotic reversion, with step‐wise cumulative increase in the abundance of heatwave‐transformed assemblages (sensu Baker et al. 2017). This is an increasingly likely trajectory under current projections of marine heatwaves frequency, duration and severity (Laufkötter et al. 2020). Field evidence of such an increase in heatwave‐transformed assemblages has been reported for Durusdinium‐dominated colonies of Pocillopora spp. on a Panamanian reef‐tract between the 1997–1998 and 2015–2016 mass bleaching events (Palacio‐Castro et al. 2023). The reefs of Kiritimati had accumulated heat stress above 8 DHW four times in recorded history before the 2015–2016 event, most recently in 2002 (NOAA Coral Reef Watch 2019), although these were dwarfed by the accumulation of over 24 DHW in 2015–2016. The dominance of the genus Cladocopium symbionts in 2014–2015 in most Platygyra ryukyuensis corals demonstrates that any prior heat‐induced shifts towards Durusdinium had not persisted on Kiritimati. We therefore suggest that long‐term shifts in coral‐Symbiodiniaceae assemblages, as theorised following multiple successive heatwaves, can also be brought about by a single extreme heat stress event.

The unprecedented duration and intensity of the 2015–2016 marine heatwave may explain the enduring success of Durusdinium in Kiritimati's Platygyra population. Priority effects are thought to provide a competitive advantage to Durusdinium in early bleaching recovery (Buzzoni et al. 2023). However, as symbionts recover, the photic environment within bleached coral tissue is ameliorated and the competition for nitrogen between symbionts intensifies (Enríquez et al. 2005; Krueger et al. 2020). This could facilitate the recovery of stress‐sensitive symbionts, such as Cladocopium, which can have a competitive advantage over Durusdinium in nitrogen assimilation and population growth in hospite (McIlroy et al. 2020). Extended heatwaves, such as the 2015–2016 heatwave on Kiritimati, may provide opportunity for Durusdinium assemblages to become established and remain resistant to competition. Indeed, tagged Platygyra ryukyuensis colonies that survived the heatwave recovered Durusdinium at densities comparable to pre‐heatwave symbiont densities by March 2016, whilst still experiencing heat stress, following an average decrease in symbiont density of 78% between May and July 2015 (Claar et al. 2020). The dominance of stress‐tolerant generalists, such as Durusdinium, is typically restricted to communities of early‐ or mid‐successional taxa exposed to chronic disturbance (Huston and Smith 1987), as observed on Kiritimati in 2014–2015, before the heatwave. However, the enduring persistence of Durusdinium has obscured this spatial variation and homogenised symbiont assemblages across the atoll.

The ecological consequences of the long‐term entrenchment of thermotolerant symbionts, for coral individuals, populations and the broader reef ecosystem are not clear, but may be detrimental. Whilst these shifts might confer an advantage to corals in a warming ocean, providing increased bleaching resistance during subsequent heatwaves of short duration (Palacio‐Castro et al. 2023), this may be outweighed by any putative physiological trade‐offs of hosting these symbionts outside of heatwave periods (Ortiz et al. 2013; Matsuda et al. 2023; Turnham et al. 2023). Indeed, whilst some Platygyra ryukyuensis corals on Kiritimati survived the 2015–2016 heatwave by shifting to host Durusdinium after bleaching, those already hosting these symbionts on reefs exposed to high human disturbance suffered much higher mortality (Claar et al. 2020), reflecting either the accumulated costs of hosting Durusdinium, the impact of chronic anthropogenic stress or their interaction. Understanding the true cost of symbiont community transformations on the resilience of coral reefs will only be possible when the persistence of these changes is better understood, which our study suggests may be more enduring than previously thought.

4.2. Loss of Pre‐Heatwave Species Disrupts Symbiotic Reversion: Cladocopium

Previously common Cladocopium lineages were not detected post‐heatwave in the P. ryukyuensis population, reflecting a concomitant shift in Cladocopium species with the enduring dominance of Durusdinium. Symbiont ITS2 profiles comprising mainly C50a and/or C3 sequence variants, found in P. ryukyuensis pre‐heatwave, are commonly detected in Indo‐Pacific corals within the Acroporidae or Merulinidae families (Carballo‐Bolaños et al. 2019; Lewis et al. 2022; Marzonie et al. 2024). In contrast, the taxa that replaced pre‐heatwave lineages, defined by C15, C15h and/or C116 variants, are most often found in Indo‐Pacific Porites spp. (Hoadley et al. 2021; Starko et al. 2023; Chen et al. 2024). The survival of Porites sp. on Kiritimati following the 2015–2016 heatwave was higher than most Acroporidae and Merulinidae (including P. ryukyuensis ) species (Baum et al. 2023). Given the propensity for symbiont environmental availability to shape coral symbiont assemblages via horizontal acquisition (Ziegler et al. 2018; Williamson et al. 2021), we postulate that the switch in P. ryukyuensis ‐associated Cladocopium species arose as environmental or host reservoirs of C50a and C3 symbionts were severely depleted due to the severity and extended duration of the 2015–2016 heatwave. This would suggest that extreme heatwaves may reshape the Symbiodiniaceae metacommunity through the differential mortality of Cnidarian hosts, analogous to the population co‐extinctions projected for many bacterial symbionts and their insect hosts under climate change (Renoz et al. 2019; Doña and Johnson 2020).

The Cladocopium symbionts hosted post‐heatwave, in addition to being taxonomically distinct, were hosted mostly at low relative abundances. Results from experimental inoculations of sea anemones with novel symbionts reach lower symbiont densities compared with normal symbiont densities, attributed to increased rejection by the host immune response (Matthews et al. 2017; Medrano et al. 2019). Hence, the C15‐, C15h‐ or C116‐dominated Cladocopium species detected in the post‐heatwave P. ryukyuensis population may lack the coevolved immune‐recognition or immune‐evasion traits that facilitated host colonisation of pre‐heatwave Cladocopium species to higher densities (Voolstra et al. 2009; Johnston et al. 2022). Reduced host recognition of recovered Cladocopium lineages may explain the observed lack of post‐heatwave reversion towards Cladocopium, as these distinct Cladocopium taxa are inferior to Durusdinium at colonising recovering host corals. Heatwave‐induced local extirpations of Cladocopium lineages have been reported in other reef coral populations, including Kiritimati's Porites lobata corals (Starko et al. 2023). The apparent loss of host‐adapted Cladocopium symbionts jeopardises the persistence of coral reefs, given the importance of Symbiodiniaceae functional diversity for coral reef ecosystem resilience (Suggett et al. 2017).

4.3. Post‐Heatwave Rise of a Rare Opportunist: Symbiodinium

The rapid post‐heatwave colonisation success of Symbiodinium in Kiritimati may be attributed to its advantage within the harsh photic environment that is bleached coral tissue (Enríquez et al. 2005). The genus Symbiodinium has been characterised as competitively dominant under higher light intensities, occupying spatial niches in shallower reefs or in the upper surfaces of colonies (Rowan and Knowlton 1995; LaJeunesse 2002). The influx of Symbiodinium contributed to three years of highly dynamic symbiont assemblages within coral colonies, with the RRA of Durusdinium and Symbiodinium oscillating year‐to‐year. This is consistent with prior studies reporting transient detections of Symbiodinium as it increases in relative abundance during or shortly after a bleaching event (Toller et al. 2001; Kemp et al. 2014; Leinbach et al. 2022). The effects of symbiont post‐heatwave dynamism on coral host fitness are uncertain but likely significant, considering the central role symbionts play in host metabolism and the host stress response (Jones and Berkelmans 2011; Silverstein et al. 2015), and any resource partitioning towards competition by cohabiting symbiont taxa may reduce energetic translocation to the coral host (Kenkel and Bay 2018). Moreover, energetic trade‐offs for corals hosting Symbiodinium at ambient temperatures have been reported. (Stat et al. 2008), resembling those incurred by hosting Durusdinium.

In addition to its increase from pre‐ to post‐heatwave, Symbiodinium RRA varied spatially in corals across Kiritimati. The post‐heatwave increase in Symbiodinium was greatest at sites experiencing lower local human disturbance and with higher in‐water visibility, exposed to higher benthic solar radiation compared with more turbid sites. The eventual competitive exclusion of Symbiodinium by Durusdinium as a function of in‐water visibility is consistent with previous experimental evidence of Symbiodinium persistence under high light and competitive exclusion by Durusdinium under low light conditions (McIlroy et al. 2019). This supports a role of local anthropogenic impacts on the recovery trajectories of coral symbioses, as heat‐tolerant symbionts are recovered following an intermediate symbiotic state under minimal local disturbance. However, the nuances of the symbiotic dynamism reported here were only revealed by time‐series sampling within tracked coral colonies, which would otherwise have been obscured without these longitudinal data.

4.4. Limitations and Future Directions

The limitations of our study design largely reflect trade‐offs incurred by prioritising long‐term sampling of a coral population. Most notably, because a primary aim was to take a new tissue sample from each tagged colony on each expedition, we wanted to minimise the amount of sampling (i.e., tissue scraped away on the colony surface) at each time point to minimise the surface area that the colony would need to heal before the next sampling occurred. From a logistical perspective, this also meant that we needed to be highly efficient with our dive time to relocate and sample each tagged coral, and do so at each site, thus prohibiting spending additional time to take multiple samples of each colony. We acknowledge that collecting only one sample from each coral per expedition may not have captured the entire symbiont diversity hosted by each colony. However, previous studies reporting partitioning of Symbiodiniaceae taxa between colony microenvironments have mostly noted differences in light exposure across the nodules of sub‐massive or lobed massive morphologies (Rowan et al. 1997; Kemp et al. 2015; Lewis et al. 2022), whereas more regular hemispherical morphologies, such as Platygyra ryukyuensis , are less liable to self‐shading. We were also unable to measure in situ environmental variables more than once a year during our long‐term monitoring of heatwave recovery. Although we did observe some temporal variation in these variables over the course of this decade long study (Table S2), this did not coincide with major shifts in Symbiodiniaceae assemblages or periods of Symbiodinium dynamism (Figure S5). However, responses occurring intraannually may have been overlooked, reflecting a common challenge for surveys conducted in remote locations of choosing between more accurate in situ data and higher temporal resolution, though potentially less accurate, satellite data (Claar et al. 2019). Future studies with more frequent environmental monitoring may help further resolve the role of parameters such as in‐water visibility on Symbiodiniaceae recovery (Doropoulos et al. 2022; Zweifler et al. 2024).

Symbiotic changes within adult coral colonies and populations are reported here, yet whether heatwave‐transformed symbioses impact reef corals at transgenerational scales remains unresolved. New recruits may be instrumental in repopulating reefs with symbiont lineages lost during major disturbance events in coral species that vertically transmit symbionts (Van Nynatten et al. 2025). For species that acquire their symbionts horizontally from the environment, including Platygyra ryukyuensis , dominance of bleaching‐resistant symbionts, such as Durusdinium, may increase the likelihood that new recruits acquire these taxa, and may also reduce bleaching‐derived contributions to the environmental symbiont pool, potentially jeopardising larval recruitment (Scott et al. 2024). Future research investigating heatwave legacies in the Symbiodiniaceae assemblages of coral juveniles that recruited post‐heatwave may further resolve the long‐term ecological impacts of marine heatwaves on reef‐building corals. Similarly, future studies could employ quantitative PCR (qPCR) to quantify the abundances of cohabiting Symbiodiniaceae and clarify their competitive interactions (McIlroy et al. 2019), disentangling the mechanisms of post‐disturbance symbiont succession that we have alluded to here.

5. Conclusion

Our findings stand in contrast to previous studies that have observed reversion to pre‐disturbance symbiont assemblages in coral hosts within a few years. Instead, following a major marine heatwave, we document long‐term persistence of heat‐tolerant Durusdinium coupled with diminished Cladocopium prevalence and an influx of the previously rare genus Symbiodinium during a successional period after the heatwave. These transformations, driven by both chronic and acute stressors, mirror rapid global increases in species turnover as these interacting stressors increase in frequency and intensity (Pinsky et al. 2025). Results from coral reefs may also foreshadow the extended persistence of disturbance‐transformed states in other microbial symbiotic systems (Hernandez et al. 2021; Radujković et al. 2025), threatening ecosystem function (Cavicchioli et al. 2019). This study offers unique insights into the endurance of disturbance‐transformed microbial communities and underscores the value of these critically vulnerable coral reefs as a globally relevant model for understanding how climate extremes impact the resilience of symbiotic species.

Author Contributions

Conceptualisation: D.B., J.K.B. Data collection: D.B., J.K.B. Data processing: D.B., A.V.N., R.C. Data analysis: D.B., A.V.N. Writing (original draft): D.B., A.V.N., J.K.B. Writing (review and editing): J.K.B., D.B., R.C., A.V.N.

Funding

This work was supported by The Leverhulme Trust, Study Abroad Studentship (SAS‐2021‐047), National Science Foundation (NSF) RAPID grant (OCE‐1446402), Rufford Maurice Laing Foundation, Natural Sciences and Engineering Research Council of Canada, Canada Foundation for Innovation (CFI) Leaders Opportunity Fund, University of Victoria, Government of Canada's New Frontiers in Research Fund (NFRF), BIOSCAN (NFRFT‐2020‐00073), National Geographic Society, Committee for Research and Exploration Grants (NGS‐146R‐18, NGS‐63112R‐19), The Pew Charitable Trusts, Pew Fellowship in Marine Conservation, and David and Lucile Packard Foundation.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Site‐level metrics of local human disturbance.

Table S2: Site‐level seawater metrics by expedition.

Figure S1: Comparison of in‐water visibility of sampling sites exposed to different levels of human disturbance. Mean values for visibility at each site were calculated from values collected across expeditions that in‐water visibility values were taken by a consistent experienced scientific diver each time. Human disturbance categories for each site are indicated in figure legend.

Figure S2: Morphological basis of taxonomic classification. Examples of the macro‐ and micro‐structure of the Platygyra ryukyuensis morphotype sampled are shown (photographs: D. Buzzoni).

Figure S3: Samples used in population‐level and tracked colony analyses. A representation of sampling across ten expeditions, with individual colonies represented by grid rows, separated by human disturbance category. Each filled grid cell represents one of the 598 samples from 237 distinct colonies that passed post‐sequencing quality control. Red vertical lines represent exposure to peak heat stress. For population‐level analyses, 188, 41, 229, 44 and 96 individual samples were collected from very low, low, medium, high and very high human disturbance sites, respectively. For post‐heatwave tracked coral resampling, 363 samples from 107 distinct coral colonies were used, ranging from 2 to 6 sequential samples per colony. The 24 colonies sampled both before and after the heatwave are indicated by star icons.

Figure S4: Symbiont per‐sample relative abundances inform ‘background’ and ‘dominant’ designations. (a) Histograms of the per‐sample relative abundances of Symbiodiniaceae genera Cladocopium, Durusdinium and Symbiodinium after sequence variants comprising < 1% of samples have been removed, binned by increments of 0.02, calculated from minimum sequence alignment results (amplicon sequence variant proportions). Dashed vertical lines indicate symmetrical threshold values of 0.1 and 0.9 used to designate the dominant symbiont genus in a sample (≥ 0.9), co‐dominant genera (≥ 0.1, < 0.9) and background genera (≥ 0.01, < 0.1). Thresholds were selected to best capture the bimodality of proportions of all three genera of Symbiodiniaceae.

Figure S6: Symbiont composition of all samples according to different read‐clustering methods used. Plots show the symbiont composition of each of the 598 samples that passed post‐sequencing quality control. Each stacked column represents a single sample sequenced from a single colony, with the height of coloured stacks representing the relative abundance (proportion of Symbiodiniaceae reads) of the taxon, ordered from highest (bottom) to lowest (top). Taxa comprising < 1% of a sample's symbiont assemblage are not shown. Blue colours represent taxa in genus Cladocopium, red represents genus Durusdinium and yellow represents genus Symbiodinium. Samples are ordered along the x‐axis by site from low to high human disturbance. Samples are ordered within each site by coral colony, such that samples collected from the same colony during different expeditions are all shown at the same x‐axis position. (a–c) ITS2 profiles detected by SymPortal's intragenomic clustering algorithm, following minimum entropy decomposition (MED) clustering. (d–f) Defining Intragenomic Variants (DIVs) detected by MED‐clustering. (g–i) Amplicon Sequence Variants (ASVs) detected by DADA2's minimum sequence alignment clustering.

Figure S7: Shift in Durusdinium and Symbiodinium biodiversity before and after heatwave. Bubble plot showing the number of colonies hosting Durusdinium or Symbiodinium as dominant or codominant symbionts at any timepoint. ITS2 profiles are ordered by their majority sequence variant(s).

Data Availability Statement

Raw ITS2 DNA metabarcoding sequence data are available on the NCBI Sequence Read Archive under the BioProject accession 1294313.

(https://dataview.ncbi.nlm.nih.gov/object/PRJNA1294313?reviewer=1n1r867604hqf7hmk9bpmub40b). All other data and code are available on GitHub (http://github.com/baumlab/Buzzoni_et_al_2026_Platygyra) and archived at 10.5281/zenodo.18792037.