Introduction

Reproduction is essential for species propagation and potential evolutionary adaptation, both of which are critical for species survival amid the unprecedented challenges of a rapidly changing climate. Worldwide coral reefs are experiencing increased frequency and severity of conditions related to climate change and human activity, such as increased sea surface temperatures, ocean acidification, sedimentation, and pollution (reviewed in1). Continuation of successful reproduction under these conditions will likely require both extensive physiological plasticity (e.g.2) as well as high genetic diversity (e.g.3) to enable adaptation to novel environments. Despite the importance of reefs to coastal ecosystems and economies (e.g.4), the underlying physiological acclimatization needed for continued resilient and successful reproduction in corals experiencing environmental stress are not known.

Successful reproduction despite experiencing physiological stress may come at a cost to the reproducing adult. If gametogenesis is energetically prioritized in a resource-limited physiological state, such as after thermal bleaching, colony growth5 or even survival and susceptibility to disease6 may be at risk. This balance between the long-term benefits of reproduction versus the short-term tradeoffs in growth and health may also have a tipping point. In the coral Montipora digitata, a reduction in photosynthesis led to an energetic shift, prioritizing gametogenesis to a certain extent. Partially shaded colonies allocated a larger percentage of their available energy to gametogenesis and less to colony growth, whereas fully shaded colonies ceased gametogenesis completely7. This suggests that the phenotypic plasticity has limits and beyond these limits short-term colony survival is prioritized over reproduction.

Bleaching is a stress response associated with a variety of environmental stressors that occur when a coral colony no longer maintains its symbiotic relationship with algal endosymbionts and appears pale or white in color. In Hawaiʻi, thermal bleaching typically occurs in August or September when seawater temperature peaks8, coinciding with the beginning of a new gametogenic cycle for Montipora capitata9. Various aspects of bleaching recovery in M. capitata have been investigated (e.g.8,10,11), but specific mechanisms facilitating continued gametogenesis post-bleaching have not been investigated. Specifically, the question of how corals overcome the physiological tax of bleaching at essential points in the gametogenic process to realize full gametogenesis has never been addressed. Uncovering the tradeoffs necessary for gametogenesis post-bleaching could give insight into what is required for population persistence in a changing ocean.

M. capitata is a simultaneous hermaphrodite with gametogenesis spanning most of the year12. Oogenesis begins in September or October and progresses through a period of rapid oocyte size increase in March/April until spawning in June or July12. Spermatogenesis is a shorter process, lasting four to five months, from April to June/July12. M. capitata has previously shown resilience in its gametogenic process in the face of environmental stresses, such as sedimentation, pollution, and thermal bleaching12,13,14. Its long gametogenic cycle could make M. capitata more vulnerable to disruptive environmental impacts, but may also provide more time for coral recovery and spawning. The latter seems to be the case for M. capitata, as the species has previously demonstrated high phenotypic canalization in gametogenesis in response to a variety of environmental drivers, such as ammonium enrichment14, sedimentation12, and thermal stress13. Resilience in reproduction is essential to overall population survival in changing environments. The population of M. capitata in Kāneʻohe Bay has high levels of genetic diversity, mixing, and dispersal, suggesting a capacity for acclimatization and adaptation to novel environments15. The specific mechanisms underlying this reproductive resilience, and the associated physiological tradeoffs, have yet to be fully defined9.

In this study, we investigated the effects of experimental temperature-induced bleaching in M. capitata, focusing on the impact of bleaching on gametogenesis. Over a 10-month period, we monitored key physiological indicators of health, including mass spectrometry-based proteomics, to uncover the main molecular pathways underlying the phenotypic response to and recovery from temperature-induced bleaching. Proteins are essential functional molecules of cells that drive physiological and phenotypic changes in response to environmental drivers and are relevant for identifying physiological responses to changing environmental conditions. Our results uncover the diverse processes impacted by bleaching and the physiological tradeoffs of bleaching resilience.

Results

Histology: gamete development

Corals that were experimentally bleached in September (T1) 2017 had significantly smaller oocyte Feret diameter (linear mixed-effects p-value < 0.05) and a higher frequency of smaller oocytes than control coral in June 2018 (T5) (Fig. 1A). At T5, there was no difference in the distribution of oocyte stages (Fig. 1B) or spermatocyte stages (Fig. 1C) between previously bleached and control corals (p > 0.05). The results in Fig. 1 may represent a combination of at least three scenarios: oocytes that are growing slowly, oocytes that have ceased growth (and will eventually be resorbed), or oocytes that are shrinking. Since nuclei are often not visible in resorbed or shrinking oocytes, the data presented include only those oocytes that have reached the stage of spawning.

Fig. 1
figure 1

The distribution of coral oocyte Feret diameters in June 2018 (T5) for previously bleached (blue) and non-bleached control (brown) coral colonies (A). Vertical lines represent the mean Feret diameter for each group, which were significantly different from each other. Oocyte stage distributions for bleached and control coral from June 2018 (T5) demonstrated no significant differences (B). Spermatocyte stage distributions for previously bleached and control coral were also not significantly different in June 2018 (T5) (C). The coral genets for the histology analyses are not the same as the ones used for proteomics.

Full size image

Chlorophyll, symbionts, and lipids

The amount of chlorophyll a and c2 in coral tissue significantly differed by bleaching status and time (p < 0.05; Supplementary Information 8; Supplementary Information 8). Chlorophyll content (both a and c2) generally decreased in bleached corals between T1 and T2 (October) (64% and 47% reduction, respectively), followed by an increase across both bleached and control corals by T3 (December) (Fig. 2A; Supplementary Information 1). Chlorophyll content was very similar between bleached and control corals by T4 (March 2018) for chl a and T5 for chl c2.

Fig. 2
figure 2

Chlorophyll a (A), symbiont counts (B), lipid biomass (C), and average egg diameter of M. capitata from Kāneʻohe Bay in 2009 adapted from12 (D) across time. For symbiont counts, chlorophyll a, and lipids, all measured in the current 2017/2018 study, values from bleached corals are in blue and non-bleached controls are in brown. Points represent average values with standard error bars. The gray point represents the value for T1 (before thermal stress) with the bleaching event marked with an orange dotted line. The 2009 egg diameter data (with standard deviation represented as gray vertical lines) serves to situate the progression of thermal bleaching recovery within the context of the gametogenic cycle.

Full size image

Symbiont densities also varied significantly by time and by treatment (Fig. 2B, Supplementary Information 8). Symbiont counts decreased at T2 for bleached corals. By T4, bleached and control symbiont counts were about equal and remained similar for the duration of the experiment. Generally, corals maintained the same genus of symbiont throughout the time series (Supplementary Information 2). Seven genets were dominated by Durusdinium at all time points measured; three genets had Cladocopium as the dominant symbiont for all time points. Two genets went from being Cladocopium- to Durusdinium-dominated between T1 and T2, regardless of bleaching treatment. One genet changed from Cladocopium-dominance at T1 and T2 to Durusdinium-dominance at T6 (July), regardless of bleaching status (Supplementary Information 2).

Lipid biomass was significantly impacted by time (Fig. 2C, Supplemental Table 8). Lipid biomass decreased dramatically in experimentally bleached coral between T1 and T2 but attained the same levels as paired controls by T4 and surpassed lipid biomass in control corals at T5 and T6. Lipid biomass decreased in control corals around the time of spawning, between T5 and T6.

Proteomics

We identified 4346 proteins in the M. capitata adult proteome across all time points (Supplemental Table 9). In the results that follow, findings from the hierarchical clustering (Supplementary Information 3) and the PLS-DA (Fig. 3) are presented together. Hierarchical clustering results are referred to by cluster ID (e.g., NB17 = cluster #17 from non-bleached controls); PLS-DA protein trends are referred to by the figure panel. Supplemental Tables 9 and 3 contain details on which specific proteins were important for each analysis.

Fig. 3
figure 3

Plots of the proteins identified with discriminant analysis of partial least squares (PLS-DA), including NMDS of the top 100 proteins that contribute to PLS-DA variance for non-bleached controls (A) and the top 100 proteins for bleached corals (F) across all time points; and line plots of the abundance trends over time for each cluster for control (B–E) and bleached corals (G–J). Details on these proteins can be found in Supplemental Table 10.

Full size image

Progression through gametogenesis of non-bleached control colonies

Each time point in this dataset represents the proteomic profile underlying an important step in the gametogenic process: September, October, January, March, June, and July. These processes are summarized in Table 1 for the control cohort. In M. capitata in Kāneʻohe Bay, the new gametogenic cycle begins between July and October12 (Fig. 2D), which is represented by T1 and T2 (September and October) in this study. PLS-DA trends B and C include proteins that increase and decrease, respectively, across early gametogenesis. This time period is marked by increasing abundance of proteins involved in lipid, protein, and carbohydrate metabolism. Many of these general metabolic themes continue through December (T3), which is still within early gametogenesis for M. capitata12.

Table 1 Summary of proteomic changes in control corals throughout gametogenesis.
Full size table

In the spring, M. capitata undergo a rapid increase in oocyte size, concomitant with an increase in solar radiation (Fig. 2)12. March (T4) also marks a distinctive proteome change in control corals (Fig. 3D,E). Changes in protein abundance span a diverse range of processes including extracellular matrix remodeling, transcriptional activation, and steroid synthesis.

Spawning in M. capitata occurs in June and July and T5 and T6 capture time periods just days before the first and second spawning dates. Similar to T4, T5 represents another inflection point in protein abundance, representing large changes in proteomic profile from T4 to T5 and from T5 to T6 (Fig. 3D,E).

Bleaching response and gametogenesis in experimentally bleached corals

Bleached proteomes had strikingly different temporal trends than those seen in non-bleached controls. Trends of proteomic divergence continued from bleaching through spawning with some results summarized in Table 2. In the hierarchical clusters at the September (T1) time point, there is limited overlap in the proteins that peak and are suppressed at later time points between bleached and control corals (Supplementary Information 3). October (T2) represents a large proteomic shift in bleached corals since it is the time point that immediately followed thermal bleaching (Fig. 3G–J). There is a dramatic change in all of the PLS-DA trends in bleached corals between T2 and T3. From the hierarchical clustering, almost twice as many proteins in the bleached corals increased in abundance at T2 (n = 248) compared to the controls (n = 130), suggesting that T2 is a critical time of recovery, supported by proteome remodeling in response to thermal stress (Supplementary Information 3). Elevated proteins at T2 include some that may be biomarkers of catabolism of stored lipids since the corals remained in tanks with filtered seawater and had limited access to food via heterotrophy.

Table 2 Summary of proteomic changes in bleached corals throughout bleaching recovery and gametogenesis.
Full size table

Unlike their paired controls, the bleached corals do not have a dominant proteome signal in March (T4) in the PLS-DA (Fig. 3), the time of rapid egg size increase (Fig. 2D). However, T4 represents an attainment of proteomic stasis that is maintained for many proteins through spawning (T5 and T6). In July (T6), of the 177 proteins with elevated abundance in previously bleached corals (Supplementary Information 3), only sixteen were also elevated in the controls.

Discussion

The full impact of thermal bleaching on population and ecosystem scales is in part determined by the reproductive success of the corals that survive but have still experienced sub-lethal impacts of the stress. Thermal bleaching responses can be variable even within a single species from the same reef16, suggesting that individual variability in responses can have reef-wide consequences. We followed a cohort of M. capitata through 10 months of recovery after a simulated thermal bleaching event and found that the corals surviving bleaching regain lipid reserves within two months post-bleaching and reacquire symbionts within six months, before spawning occurs. However, differences in oocyte size and protein biomarkers of certain metabolic pathways suggest that even 10 months after thermal bleaching stress, corals still express a phenotype at least partially dictated by prior thermal stress. Although natural bleaching susceptibility in M. capitata can be explained by differences in their algal symbionts10, symbiont type was not strongly associated with the ability of coral colonies to recover from bleaching in this experiment. The focus on the impacts of bleaching on the essential process of gametogenesis in this study provides a novel perspective on thermal bleaching recovery that has not been directly addressed in previous studies on long-term impacts of bleaching using molecular markers (e.g.17). Reproductive success is a complicated mosaic that is determined by parent physiology and environmental conditions and culminates in larval survival and settlement; this study provides a detailed depiction of the underpinnings of reproductive success after thermal bleaching.

Immediately following thermal bleaching in September, M. capitata colonies lost their symbionts and much of their lipid reserves (Fig. 2A,C). It is well established that thermal stress can disrupt the symbiotic relationship between the coral and its dinoflagellate symbionts (Symbiodiniceae) (e.g.18) and once the symbionts are expelled and/or digested, the host loses an important metabolic resource. Symbiont genus did not impact bleaching recovery in this study, as has been observed in natural, field-based bleaching studies (i.e.10). We hypothesize that symbiont genus had no impact on bleaching recovery because we intentionally induced complete bleaching in corals hosting varying symbiont community types to generate our bleached cohort, but additional work is needed to distinguish the relationship between thermal resistance and recovery dynamics.

There are proteomic signals detected in this study of symbiont rejection and reacquisition in the corals that underwent experimental bleaching. The GS/GOGAT (glutamine synthase/glutamine oxoglutarate aminotransferase) system is the dominant means of nitrogen acquisition in a healthy holobiont19; this system breaks down when the symbiotic relationship is disrupted and is in part replaced by host urease expression to acquire N16,20. In October (T2), there was a decrease in abundance of coral glutamine synthetase (part of the GS/GOGAT system) and an increase in urease enzymes relative to control corals, suggesting a shift in host N acquisition reflecting the loss of photosynthate (Fig. 4, Supp. Fig. 4)16.

Fig. 4
figure 4

A conceptual timeline of protein abundance changes in bleached (bottom timeline) and non-bleached control (upper timeline) corals. Physiological processes that are elevated for bleached or control corals are listed for each time point along the timeline (blue and brown lines, respectively). For some of the processes, representative plots of protein abundance are indicated below. Additional proteins are shown in similar plots in Supplementary Information 4.

Full size image

Coral nitrogen acquisition via symbiont pathways is still suppressed in December (T3), reinforcing that symbionts are not yet providing all the resources the host needs (Fig. 4, Supp. Fig. 4). Isocitrate dehydrogenase is another enzyme associated with GS-GOGAT and may be a biomarker of symbiotic dysfunction; some homologs of this enzyme are elevated in bleached compared to control corals in December (T3) and March (T4) (Supp. Fig. 4). Thus, during this critical period of early gametogenesis, the previously bleached corals may be relying predominantly on heterotrophic feeding to rebuild lipid reserves.

Symbiont reacquisition was almost complete by March (T4) in previously bleached coral and symbiont abundances are about equal with their non-bleached paired controls. In the proteome, there was an almost eight-fold decrease in Rab11a abundance in March (Fig. 4), a protein that is suppressed during healthy symbiosis21. Healthy, photosynthesizing symbionts would provide the coral host metabolic by-products, such as simple carbohydrates, which the host then metabolizes22. Two metabolic enzymes (pyranose oxidase and succinate-semialdehyde dehydrogenase), which may be involved in processing of symbiont-derived resources, decrease post-bleaching and rebound through symbiont reacquisition, perhaps signaling a reestablishment of symbiont byproduct-derived carbohydrate metabolism (Supp. Fig. 4).

During physiological stress, organisms may reallocate energetic resources to respond to the stress while maintaining essential physiological functions. In corals, skeletogenesis decreases immediately following bleaching stress or reduction in photosynthate (e.g.7). In the M. capitata proteome of experimentally bleached corals, we see a suppression in the abundance of all isoforms of carbonic anhydrase, a skeletogenesis protein, suggesting a decrease in skeletal growth relative to non-bleached controls (Fig. 4, Supp. Fig. 4), which has been previously observed5. This trend in carbonic anhydrase abundance is recapitulated in the M. capitata transcriptome post-thermal stress23. Another aspect of the coral response to bleaching is an increase in reactive oxygen species (ROS) and a concomitant increase in ROS scavengers and repair proteins to mitigate and respond to cellular damage23,24. As in24, homologs of glutathione-S-transferase and caspase were elevated in bleached corals compared to non-bleached controls in October, suggestive of a ROS response (Fig. 4, Supp. Fig. 4). ROS response proteins elevated in October (T2) also include peroxidasin (a known inhibitor of ROS), as well as proteins that are likely responding to the consequences of cellular damage: protein disulfide isomerase A6 (inhibits aggregation of misfolded proteins) and leukocyte elastase inhibitor (protects cells from proteases released into the cytoplasm during stress). These proteins all potentially represent a cellular effort to repair and mitigate damage to DNA, proteins, and other cellular structures in the aftermath of thermal stress.

The timing of symbiont loss, which corresponds to when most natural bleaching events occur, overlaps with the onset of gametogenesis in M. capitata12. Gametogenesis is an energy-intensive process, and the combined effect of lipid store catabolism with the loss of symbiont resources could conceivably deplete the energy store necessary to form viable gametes. October (T2) also represents the start of a months-long recovery period for bleached corals. During these months post-bleaching, we show evidence that bleached coral survive bleaching by first catabolizing lipid reserves (Figs. 2C, 4, Supp. Fig. 4) and then switching to heterotrophic feeding (Fig. 4, Supp. Fig. 4) to rapidly restore lipid levels (Fig. 2C) while reacquiring symbionts (Fig. 2A) and acquire enough autotrophically derived carbon before spawning (Figs. 2A, 4, Supp. Fig. 4) to successfully undergo gametogenesis, with potential tradeoffs in egg size (Fig. 1A).

Lipid and proteomic data in the present study suggest that bleached corals began to catabolize their lipid reserves post-bleaching to make up for the loss of metabolic products from their symbionts. The rapid mobilization of stored lipids to compensate for reduced photosynthate after bleaching has been well documented in M. capitata9,17,25 and other coral species18,22. Many protein biomarkers detected at this time point also confirm a probable up-regulation of lipid catabolism immediately post-bleaching (Fig. 4 and Supplementary Information 4).

Lipids are essential macromolecules in the development of gametes in marine organisms (e.g.26) and catabolism of lipids for other physiological purposes may have lasting impacts on the gametogenic process. The experimentally bleached corals depleted lipid reserves through the first two months post-bleaching (Fig. 2C), with proteomics revealing some of the molecular mechanisms of lipid catabolism during this time. Phospholipase B is likely involved in this catabolism and is one of the proteins that follows a trend of elevated abundance in bleached corals compared to non-bleached controls in October (T2)24 (Fig. 4; Supplementary Information 4). Even if symbionts remain in the bleached corals, thermally stressed symbionts may provide fewer lipids to their host24. Therefore, the lipid metabolism enzymes that are elevated at this time point are either acting on 1) the few lipids provided by the remaining symbionts, 2) heterotrophically-derived lipids, or 3) storage lipids from the coral tissue. M. capitata use stored lipid reserves when photosynthate and heterotrophy are unavailable25, while as much as 70% of M. capitata lipids can be heterotrophically derived immediately post-bleaching17. In October, the corals had not yet been outplanted to racks on the reef and therefore had access to only limited suspended plankton in tanks for heterotrophy. Given the low number of symbionts and rapid decrease in lipid reserves, the corals likely derived energy from their stored lipids at T2. Lipids are the main energy source that corals allocate to their gametes27 and the depletion of these stores in early gametogenesis impacts gamete production.

In December, bleached corals had not yet reacquired all their symbionts, yet lipid content was almost on par with their non-bleached control counterparts. This rapid reacquisition of lipids has been previously observed in M. capitata and may be in part attributable to more efficient resource conservation due to a lower respiration rate than other coral species25, as well as heterotrophic feeding. In December, many lipid hydrolysis enzymes, potentially linked to food acquisition and lipid digestion, were at relatively high abundance in bleached corals compared to non-bleached controls (Fig. 4, Supp. Fig. 4). Some proteins involved in peptide and protein breakdown were also elevated, and two proteins involved in ATP synthesis increased in abundance from October (T2) through March (T4, Fig. 4, Supp. Fig. 4). Combining all the molecular evidence, it suggests that M. capitata engages in external food acquisition and breakdown, as well as cellular energy production despite low levels of symbionts. M. capitata is known for its trophic plasticity and can acquire the food necessary to fulfill its daily metabolic needs with heterotrophy alone25. This flexibility in the nutrient acquisition is not present across all coral species25 and may be part of the phenotype of increased bleaching resilience observed in M. capitata25. Between October (T2) and December (T3), bleached M. capitata are likely relying on heterotrophy almost exclusively to rebuild lipid reserves, which could provide energy to support spawning and/or survival during a future bleaching event.

March (T4) represents a peak in lipid biomass for these corals and is when lipids are allocated to oocytes, followed by adult tissue lipid level decline post-spawning28,29. This period of lipid accumulation leading up to allocation to oocytes is essential for embryo and larval development and survival in their planktonic phase30. Some lipid metabolism enzymes were relatively lower in bleached corals in March (e.g., lipase, phospholipase D3, saposin domain-containing protein), while others were relatively elevated (e.g., phospholipase A2 and B, prosaposin). Two hypotheses for these trends could be that 1) the bleached coral have “overcompensated” for the depletion of lipid reserves during bleaching and have upregulated their lipid metabolism and storage pathways via increased heterotrophy and/or the return of symbiosis coinciding with increased light and temperature during spring months. Alternatively, 2) bleached coral may be dedicating relatively less lipid to their oocytes and thus maintaining a higher reserve in their somatic tissue than controls. The elevated level of lipids in adult coral is likely due to M. capitata’s ability to store heterotrophically derived carbon, as suggested by31, and as evidenced by the trends seen in early lipid reacquisition in this study.

The June (T5) and July (T6) time points represent the time periods just before spawning events and are marked in control coral by a decrease in lipid biomass (Fig. 2C) and another inflection point in trends of protein abundance in the PLS-DA trends (Fig. 3D,E). The dip in lipid biomass is not observed in bleached corals (Fig. 2C); maintenance of higher lipid levels may help these corals survive future bleaching events18. The dip in lipid levels in adult tissues is a consistent trend across broadcast-spawning corals that marks the allocation of lipids to oocytes7,27. It has been hypothesized that the drop in tissue lipid content with spawning is directly correlated with energy allocation to gametes7. Even though bleached M. capitata prioritize energy transfer to oocytes9, the lack of early access to autotrophic carbon may have resulted in less energy allocation to gametes in bleached corals, compared to controls. In the proteome, homologs of vitellogenin, a major egg yolk protein, are all at lower abundance in previously bleached coral compared to non-bleached controls in June, suggesting less lipid allocation to oocytes (Supp. Fig. 4).

Repeated bleaching events can lead to increased mortality in some species, perhaps because energetic reserves used to survive prior stressful events are depleted32. M. capitata in the dynamic Kāneʻohe Bay may have evolved physiological mechanisms to survive repeated environmental stresses (e.g.33). The tradeoff, however, could have implications for larval viability if there are less maternal lipids to sustain the non-feeding stages of development and to maintain positive buoyancy29.

While the dominant protein abundance trends in bleached corals were in response to bleaching, a significant proteomic shift occurred in non-bleached controls during March (T4), coinciding with the period of most rapid oocyte size increase and onset of spermatogenesis (Figs. 3D, 4E,F). A wide range of functions are represented by the proteins of control corals that peak in abundance in March. The proteins encompass functions in DNA and cellular replication, protein turnover, and signaling via neurons and neuropeptides (Fig. 4, Supp. Fig. 4). Signals of extracellular matrix (ECM) reorganization, a basal process essential to gametogenesis34, were detected in the control corals: two proteins that negatively regulate or degrade the ECM were at decreased abundance in March relative to previous time points (Fig. 4, Supp. Fig. 4). These proteins represent some of the complex physiological processes necessary to support gametogenesis.

As in other animals, sex hormones regulate reproductive maturation in corals (e.g.35) and we identified four proteins with likely sex hormone regulation functions that are impacted by thermal bleaching. The hierarchical clusters that peak in control corals in March (Supp. Fig. 3) include 1) steroid 17-alpha-hydroxylase (also known as CYP17A1), a protein important in regulation of sex hormones during gametogenesis36 (Fig. 4, Supp. Fig. 4). CYP17A1 abundance is not impacted by bleaching recovery at the time points investigated. 2) Steroidogenic acute regulatory protein (StAR) abundance is depressed in previously bleached corals at T2 and T3 and then increases to levels higher than those in controls at T5. StAR is an essential early step in steroid synthesis and has been identified in other corals genomes35 (Supp. Fig. 4). 3) Protein flightless-1 and 4) 17B hydroxysteroid dehydrogenase are both at relatively low abundance in bleached corals at T4. The former may regulate transcription downstream of estrogen and androgen receptors, and the latter is known to regulate estrogen and androgen levels in mammals and has known estrogen activity in corals37 (Supp. Fig. 4). In the weeks leading up to spawning, sex hormones increase in M. capitata23 and the proteins regulating these hormones detected in this study give further insight into patterns of sex hormone production during gametogenesis.

In June (T5), oocyte and spermatocyte stages were similar between bleached and control coral, but oocyte Feret diameter was significantly smaller in previously bleached corals (Fig. 1). Oocyte diameter reduction is a common effect of bleaching across coral species38. Even though bleached coral had attained the same lipid levels as controls by December, these lipid stores are likely derived mostly from heterotrophically derived carbon since symbiont levels were still low. M. capitata colonies do not allocate heterotrophically derived carbon to oocytes, only autotrophic carbon9,29. Previously bleached corals likely had less autotrophically derived carbon for their oocytes than controls. M. capitata and other corals provision eggs with lipids, proteins, symbionts, and photoprotectant molecules29,30; adult bleaching can reduce the amounts of these molecules allocated to oocytes28, which may be reflected in the smaller size of bleached corals’ oocytes in this study.

Bleaching impacts on coral gametogenesis vary both across species and populations within a species. If a coral population is highly tolerant to bleaching stress, they may still be able to execute full gametogenesis as seen in this study and others (e.g.38). Bleaching tolerance may be species-specific or genetically determined but may also be a result of acclimatization over multiple bleaching events (e.g.39). The severity of the bleaching event and/or a history of thermal bleaching exposure can dictate whether and to what degree a coral can achieve gametogenesis38. M. capitata in Kāneʻohe Bay experienced thermal stress and may have bleached in 2014 and 2015 (two and three years before this study, respectively8) and seem to have evolved resilience to disturbances33. Their combined genetic and ecological histories may give them the physiological capacity to accomplish gametogenesis despite bleaching, while other species or populations may fail. Some of the functional protein biomarkers identified here may help identify essential traits and mechanisms of thermal bleaching-resilient phenotypes.

M. capitata demonstrated a potential for resilience to thermal bleaching stress as well as an ability to rapidly recover symbionts and lipid levels post-bleaching. However, the bleached colonies did not have access to symbiont-provided autotrophic carbon during the critical months of early gametogenesis, which likely resulted in a time lag in upregulating some of the physiological processes necessary for gametogenesis (Fig. 4). One result of this delay in resource access was significantly smaller oocytes in bleached corals. Corals in Kāneʻohe Bay have experienced increasingly frequent thermal bleaching events in recent years, the effects of which may be seen in smaller oocytes across the population when compared with oocyte sizes from previous years (e.g.14,40). Additionally, previous work has suggested that the impacts of bleaching can be long-lasting and more detrimental for coral sperm than for eggs41. A longer-term study that follows gametogenesis, fecundity42, and spawning synchrony29 in these corals across years and multiple bleaching events would clarify whether these strategies facilitate faster recovery from bleaching events. In corals, the number of oocytes can decrease following bleaching, while the remaining oocytes in the colony can grow and mature properly42. Gametes could also be resorbed after stress43. Additionally, it is important to better understand if there are any short- or long-term population-level consequences in allocating less of the adult lipid biomass to gametes in terms of the number of viable embryos and successfully settled juveniles. Since both adult coral survival and juvenile settlement success are integral to continued reef survival, it is not enough that we understand if corals survive thermal bleaching; we also must understand the cost of that survival.

Methods

Coral selection and bleaching

Details for coral collection and historical bleaching regime can be found in11. Seventy-four M. capitata coral (approximately 30 cm in diameter) were collected from patch reefs located in Kāneʻohe Bay, Oʻahu, Hawaiʻi around the Hawaiʻi Institute of Marine Biology (HIMB, 21.428°N, 157.792°W) in August 2017. Each coral was further divided in half so that genetically identical halves (i.e. ramets of a genet) were replicated in the control and treatment conditions (Fig. 5). Experimental corals were acclimated in flow-through outdoor tanks (three tanks per treatment group) for 7–10 days. The experimental bleaching occurred in September (T1), the period when bleaching events typically occur in Hawaiʻi8. To reach the 30 °C temperature goal for the bleaching treatment, experimental tank temperatures were increased by 0.6 °C per day over four days (Supplementary Information 5). After about one week, the corals in the bleaching treatment showed no signs of thermal bleaching and so the temperature was increased to 31 °C using the same rate of increase. Control tanks were maintained at the ambient temperature of 28 °C. The corals were rotated once a week between tanks of the same treatment to minimize tank effects. M. capitata in the bleaching treatment were kept at elevated temperatures for three weeks to induce complete bleaching. After complete bleaching occurred, the tank temperature was lowered to ambient temperature, following the previously described rate (Supplementary Information 5). Experimental corals were then placed on racks off HIMB after the sampling time point on October 1, 2017 (T2) to monitor their recovery and physiological response to the in situ environment through July (T6) (Supplementary Information 6). Mortality and bleaching assessments were conducted weekly using the Coral Watch Coral Health Chart44.

Fig. 5
figure 5

Experimental design and bleaching status of M. capitata exposed to thermal stress. September (T1; pre-bleaching) through March (T4; visual recovery from bleaching, based on Coral Watch Coral Health Chart categories). Corals were also sampled in June (T5) and July (T6) during the spawning season and their colors remained unchanged during this time period. The period of gametogenesis begins in October (T2) and progresses through a period of rapid increase in oocyte size (March, T4) until spawning in June or July (T5 and T6).

Full size image

The time points when corals were sampled for proteomics, chlorophyll, lipids, symbiont density, and symbiont clades are: 1) at the end of August (August 30, 2017) after corals acclimated in the tanks but before they were bleached (referred to as “September” throughout, T1); 2) in October, 24 h after bleached corals were returned to ambient temperature (T2; October 1, 2017); 3) in December (T3; December 20, 2017); 4) in March, during the period of the rapid increase in egg size (T4; March 29, 2018)12; 5) in June, at the beginning of the spawning season, when symbionts are vertically transferred to gamete bundles (T5; June 8, 2018)29; 6) and in July, later in the spawning season (T6; July 9, 2018). Samples for histology were taken only in June (T5). Additional details are provided in Supplemental Methods.

Histology

Histology slides were made from 5 to 7 µm thick serial sections of M. capitata tissue samples collected in June 2018 (T5). We cut enough sections to measure around 50 eggs in both bleached and control fragments per genet. Slides were stained with hematoxylin and eosin before observation under a compound microscope, photographed using a camera attachment (model DS-Fi3, Nikon Instruments Inc.), and analyzed with NIS-Elements imaging software. Oocyte measurements were made in Image-J software version 1.52 using Feret’s Statistical Diameter to estimate size. Only oocytes with visible nuclei were measured to ensure standardization of measurements along the widest axis of the oocyte. Developmental stage of gametes was assessed using morphological guidelines and Feret diameter size ranges12.

Although the response variable, Feret diameter, was not normally distributed (Fig. 2A), the sample size was sufficiently large (n = 858) to fulfill assumptions of the Central Limit Theorem that the sampling distribution is approximately normally distributed. Means, standard deviations, and ranges of oocyte Feret diameter were calculated for both control and experimentally bleached coral. Feret differences and oocyte and sperm stages between bleached and control corals were evaluated following45. For this study, we did not examine polyp fecundity (oocytes per polyp). Previous studies have shown polyp fecundity does not differ between bleached and control corals13. Additional details are provided in Supplemental Methods.

Chlorophyll, symbiont counts, symbiont clades, and lipids

Chlorophyll a (Chl a), chlorophyll c2, and dinoflagellate symbiont (Symbiodiniceae) counts were analyzed as reported by11. Briefly, Chl a was extracted on ground whole coral (tissue and skeleton) samples using 100% acetone46, determining absorbance at 630, 663, and 750 nm, and standardized to total dry tissue weight.

For symbiont counts, dinoflagellate symbionts were separated from ground whole coral samples; isolated pellets were resuspended in filtered seawater with 1% formalin and 2–3 drops of Lugol’s iodine. Three subsamples were counted using a hemocytometer, and the mean count was reported.

Relative symbiont abundances (genera Cladocopium and Durusdinium) were measured with qPCR for tissue samples taken in September, October, and July (T1, T2, and T6). DNA was extracted from tissue using a CTAB-chloroform protocol (http://dx.doi.org/https://doi.org/10.17504/protocols.io.dyq7vv)). Proportions of Cladocopium and Durusdinium symbiont cells in each sample were quantified with qPCR using actin assays47 on an Agilent AriaMX system with two technical replicates run for 40 cycles. Additional details are provided in Supplemental Methods.

Lipids were extracted from ground whole coral samples with a 2:1 chloroform: methanol solution, 0.88% KCl and 100% chloroform washes48. Extracted lipid samples were dried to a constant weight under grade 5.0 N2 gas and standardized to total dry tissue weight.

Linear mixed effects models (lmer in the lme4 package49) were applied to determine significant differences in chlorophyll a and c2 concentration, symbiont counts, and total lipids with bleaching status and time as fixed effects and coral genet as a random effect. Significant effects (p < 0.05) were determined in lmerTest50. There was not enough data for the symbiont genera to fit a generalized linear model, but qualitative results are described.

Proteomics

Protein digestions were conducted with 100 µg of protein per coral sample as described in11. M. capitata samples were analyzed using liquid chromatography coupled to tandem mass spectrometry (LC‒MS/MS) on a Q‒Exactive‒HF (Thermo) in Data Dependent Acquisition (DDA) mode.

From each mass spectrometry experiment, M. capitata peptides were identified and proteins were inferred using a proteome derived from the M. capitata genome (NCBI Bioproject Accession no. PRJNA509219)51. The M. capitata proteome was concatenated with a predicted proteome for Symbiodiniaceae genus Cladocopium (http://symbs.reefgenomics.org/download/) and 50 common contaminants (cRAPome52); the final proteome database contained 99,221 sequences. The raw MS data (PRIDE accession PXD021262) were searched against the protein database using Comet v 2019.01 rev.553. Concatenated target–decoy database searches were followed by PeptideProphet and ProteinProphet54 with a probability cut-off of 0 to allow for FDR cut-off downstream in Abacus. Resulting data files across all samples were analyzed with Abacus55 to generate consistent protein inferences across replicates and to calculate normalized spectral abundance factors (NSAF) with an FDR cut-off of 0.01 (protein probability of 0.91). Proteins were included in downstream analyses if two unique peptides were identified across all mass spectrometry experiments and if they were not flagged as outliers56.

Proteins important to experimental bleaching response and/or to gametogenesis were identified with two main methods : 1) hierarchical clustering and 2) discriminant analysis by partial least squares (PLS-DA). These two methods are complementary and were used to identify which molecular processes change over time and how those changes may impact bleaching recovery and the timeline of gametogenesis. Additional proteomics details are provided in Supplemental Methods.

All analyses described above were accomplished in R57. All data files and R code used for analyses are available on Dryad: https://datadryad.org/stash/share/TH40Lri9NsCmVVGpYe92iJ2jTfVsvJT0ryORkPSba70