Growth and Survival in Common Garden Before Bleaching.

In the common garden nursery, we measured growth by buoyant weight of each replicate nubbin from August 2014 to April 2015, when the first signs of bleaching occurred. Species differed in relative growth rate, with the branching Acropora growing by three- to fivefold during this period compared with two- to threefold growth in P. cylindrica (SI Appendix, Fig. S1A). The lowest net growth was recorded for P. damicornis but this was due to consistent predation of branches by resident corallivores.

Pool of origin was an important predictor of growth for two species. In P. cylindrica, nubbins from parent colonies grew more slowly when they were from the warmer HVP (88% vs. 114% growth, P < 0.0001, t test). The opposite trend was seen in A. hyacinthus, where nubbins from parent colonies grew more slowly when they were from the less variable MVP (144% vs. 188% growth, P = 0.007, t test). Origin was not a significant factor in growth of A. gemmifera or P. damicornis during this prebleaching period.

Symbiont genus played a significant role in growth for A. hyacinthus (ANOVA P = 0.004): colonies hosting primarily D. trenchii symbionts and switchers tended to grow faster than those hosting primarily genus Cladocopium. Colonies of A. hyacinthus with D. trenchii grew by 295% compared with 249% for those with Cladocopium. Though we saw the same relationship in A. gemmifera (274% growth with Durusdinium vs. 225% with Cladocopium) this difference was not significant, largely because of the small number of colonies hosting Cladocopium. By contrast, in P. damicornis, colony growth was higher for colonies with Cladocopium symbionts (188% vs. 156%, ANOVA P = 0.0007). Symbionts were not variable in P. cylindrica.

For all species but P. cylindrica, identity of the parent colony played a very strong role in determining overall growth rate. There was about a twofold difference in growth across the 20 colonies of each Acropora species, with relatively low variance among growth rates of the 10 replicates of each colony (average coefficient of variation ~25% of the mean, SI Appendix, Fig. S1A). As a result, a significant proportion of the variance in growth among clones in the nursery was explained by parent identity (ANOVA P < 7e-10).

We found that survival was most affected by conditions of the transplant site. In particular, predation by small benthic reef fish at the transplant site in Sili Reef caused the most mortality, particularly in P. damicornis and A. hyacinthus (SI Appendix, Fig. S1B). By April 2015, survival was 94% for A. gemmifera, 97% for P. cylindrica, 67% for A. hyacinthus, and 54% for P. damicornis.

Bleaching and Recovery in the 2015 and 2017 Events.

Eight months after the establishment of the nursery, the 2015 El Niño brought temperatures reaching above 35 °C to back reef areas of American Samoa, causing bleaching in both the parental populations and the nursery nubbins (Figs. 2 and ​and3).3). Bleaching occurred again in 2017, 31 mo after establishment (SI Appendix, Fig. S2). Monitoring 10 nubbins per parent allowed us to evaluate the way replicate nubbins from different parents reacted to a natural bleaching event. In both events, bleaching was variable within species, allowing us to test the value of our proxies of thermal tolerance, origin, biomarkers, and microclimate measured for parent colonies to predict nursery results.

Open in a separate window

Fig. 2.

(A) Average temperature in the HVP, the MVP, and the transplant site in Sili with observed natural bleaching marked in the red box. The gray line marks the expected threshold bleaching temperature. (B) Visual bleaching severity of nubbins in the nursery partitioned using the three simplest proxies for predicted resilience. Labels on the predicted resilient or vulnerable stocks axis are as follows: (A, in blue lettering) corals from the HVP; (B, in blue lettering) top 10 experimental stress performers; (C in blue lettering) hottest extreme microclimates and the corresponding predicted vulnerable stocks: (a) corals from the MVP; (b) bottom 10 experimental stress performers; and (c) coolest extreme microclimates.

Open in a separate window

Fig. 3.

Correlations between the average bleaching severity of replicate nursery nubbins in Sili versus the average bleaching severity of their parent genotypes in their original habitats. Photograph demonstrates severely bleached A. gemmifera next to nonbleached nubbins.

After the 2015 bleaching event, visual bleaching of nubbins was less severe for P. cylindrica (average bleaching was 12%), moderate for P. damicornis (32%), and severe for A. hyacinthus (39%) and A. gemmifera (43%). However, there was also a great deal of variation within species. Even after 8 mo of growth side by side in common garden settings, bleaching of nubbins was 1.9–2.5 times higher when they originated from parents in the MVP compared with the HVP (Figs. 2B and ​and4,4, two-way ANOVA F = 67.1, P = 10⁻¹²). This result shows that common garden growth of 200–500% and 8 mo of acclimatization did not erase the differences in heat tolerance among parent colonies.

Open in a separate window

Fig. 4.

Average percent bleaching across four species in the Sili nursery after the 2015 natural bleaching event. Bleaching of nubbins that came from parents from the HVP (red) was two- to threefold lower than for nubbins with parents from the MVP (orange).

Because we have data on three bleached species, from 10 parental colonies from each of two pools, and 10 replicates each, we tested how much variation in bleaching is partitioned by parent, origin, and symbiont. Different colonies, even from the same pool, had nubbins that bleached very differently than those of other colonies (Fig. 4, two-way ANOVA, F = 21.9, P = 10⁻¹⁸). For example, A. hyacinthus colony SU03 had nubbins that bleached 80–100% (mean 92%, n = 10), whereas conspecific SU08 had nubbins that bleached only 20–40% (mean 23%, n = 8). These colonies were from the same pool but had different symbionts (SI Appendix, Fig. S1A).

To tease apart the relationship between pool of origin, symbiont, and nubbin bleaching, we used both general linear models (GLMs) and multiple linear regression. For both analyses (implemented in R), pool of origin was a significant factor in bleaching after accounting for symbiont (SI Appendix, Tables S1 and S2). Symbiont was separately significant in the multiple linear regression but only marginally so in the general linear model analysis. These results suggest that bleaching in the nursery was affected by both the pool of origin of the parent and to a slightly lesser degree by the parent’s symbiont type: the combination of these effects explains much of the complex variation of nubbins in their bleaching patterns (SI Appendix, Fig. S1A).

This variation was consistent with heat resistance we previously measured for these parents. Bleaching of nubbins in the common garden in 2015 was higher when they were derived from parents with higher heat tolerance. The correlation was significant for the two Acropora species (P = 0.006 and P = 0.032) and was in the same direction but nonsignificant for P. damicornis (P = 0.18) (SI Appendix, Fig. S3).

Similar results were recorded after the 2017 bleaching event. The two Acropora species bleached, whereas P. cylindrica did not (predation-based mortality between 2015 and 2017 was too high to score bleaching in P. damicornis). In both Acropora species, bleaching was less severe for parents from the HVP than the MVP, and for parents scored as more heat resistant in 2014 (SI Appendix, Fig. S2). This second event was less drastic in A. gemmifera than the first bleaching.

Parent Versus Nursery Bleaching.

In addition, by monitoring parents during the bleaching event, we could compare bleaching in nursery nubbins and parents. Even though they experienced high water temperatures on different reefs in different months, clones and corresponding nursery replicates showed a strong and significant correlation in the 2015 bleaching for the three species that bleached (Fig. 3; from Pearson’s correlation: P. damicornis P = 8.89e-5, A. gemmifera P = 8.82e-8, A. hyacinthus P = 4.14e-4; P. cylindrica colonies did not bleach in their original environments). For example, A. gemmifera clone GE_J bleached 80% in its native habitat and the nursery replicates from this colony bleached an average of 94%. By contrast, GE_C did not bleach in its native habitat and none of its replicates in the transplant bleached more than 20%. The strong correlation from clone to corresponding nursery replicates and low SD (<10%) across nubbins indicate stability in the bleaching response.

Acclimatization in Common Garden Settings.

Acclimatization to overall lower heat tolerance was also apparent in our study. The bleaching-inducing temperatures experienced in the nursery site were relatively mild compared with the environmental history experienced by parent colonies (SI Appendix, Fig. S4). Nevertheless, nursery replicates at Sili bleached more (average 11%) than their parents did in their original habitat. Despite this small loss of bleaching tolerance in a milder environment (SI Appendix, Fig. S3), the rank order of bleaching remained highly correlated between clones in their home environment and corresponding nursery replicates (Fig. 3).

Symbiont Switching During Bleaching.

For the two Acropora species, we observed symbiont shifts to occur in 11 of the 40 parent colonies, all shifting from Cladocopium-dominated to Durusdinium-dominated symbiont populations. For nubbins, lower bleaching was seen for parent colonies that switched to genus Durusdinium (52% vs. 88% for A. gemmifera P = 0.01, 16% vs. 80% for A. hyacinthus, P = 0.008) than for parents that retained Cladocopium.

Variation in symbionts allowed a limited test of the expected tradeoff between increased growth in corals with genus Cladocopium and increased heat tolerance in corals with genus Durusdinium (21). Transplant corals from the MVP with genus Durusdinium symbionts bleached less than the MVP transplants with genus Cladocopium symbionts. In addition, clones that switched from genus Cladocopium to genus Durusdinium showed much less bleaching than colonies that started with genus Cladocopium and did not switch (Fig. 4). However, the impact of symbiont type on growth was negligible. Within the MVP, parent colonies with genus Cladocopium did not produce nubbins that grew more quickly than those with genus Durusdinium, as might be expected if the tradeoff between growth and bleaching were strong (15).

Correlations Between Bleaching and Simple Predictive Proxies.

All of our simple environmental and physiological proxies allowed the identification of parental corals producing nubbins that bleached substantially less in both the 2015 and 2017 bleaching events (Fig. 2 and SI Appendix, Fig. S2). In the 2015 event, all three environmental and physiological proxies significantly explained the bleaching response (linear model P values <0.01). Origin (HVP vs. MVP) and symbiont were the proxies that performed best, explaining 21% of the response. Microhabitat and stress performance explained 11% and 8%, respectively after removing the effect of origin. Biomarkers proved effective at predicting tolerance for the species in which they were characterized (A. hyacinthus) (SI Appendix, Fig. S5), but were not effective in other species.

Limited Tradeoffs in Growth in a Resilient Nursery.

Host-driven bleaching tolerance is predicted to come with potential tradeoffs (22). For P. cylindrica, the parents that bleached less tended to grow at slower rates under prebleaching conditions (August 2014–December 2014), as expected by tradeoff predictions [Pearson’s P = 0.005, correlation coefficient (cor) = 0.6]. In the period after bleaching (April 2015–August 2015), this tradeoff persisted for P. cylindrica (Pearson’s P = 0.008, cor = 0.6). For A. gemmifera, a similar but weaker trend was also seen (Pearson’s P = 0.05, cor = 0.4), and was largely explained by high growth and high bleaching for colonies from the MVP. However, in the period after bleaching, the trend reversed and coral from more tolerant parents grew faster than coral from less tolerant parents, dominated by low growth in those that were bleached (Pearson’s P = 0.02, cor = −0.5). By contrast, we did not see significant trends in A. hyacinthus or P. damicornis between growth and bleaching.

Role of Holobiont in Bleaching and Recovery.

Physiological response of corals to environment is a complex feature of the coral holobiont, including species (26), host genetics (27), host acclimatization (10, 20, 23), symbiont genetics (8, 19), symbiont acclimatization, and an extensive microbiome. In our experiments, we tested corals from a common garden setting that were different species, came from different locations, and had different symbionts. There is a good correlation of nubbin bleaching in the nursery with both symbiont type and origin of the parent colony when analyzed separately (SI Appendix, Fig. S8). However, there is a tight association between symbiont type and pool of origin in this location. The HV pool is dominated by symbiont genus Durusdinium, whereas the MV pool has a more even mix of Durusdinium and Cladocopium symbionts (SI Appendix, Fig. S8; see also ref. 28). Both GLM and multiple linear regression analyses (SI Appendix, Tables S1 and S2) show that both origin and symbiont genus play a significant role in determining the holobiont’s response.

Our goal in these experiments was to assess whether the differences we see in parental colonies are also seen in the nurseries after 8–17 mo of growth, and our results show this to strongly be the case (e.g., see correlations in Fig. 3). However, three other features stand out in these analyses. First, colonies from the MV pool or with genus Cladocopium symbionts have a wider range of bleaching than do colonies from the HV pool or Durusdinium symbionts (SI Appendix, Fig. S8, ANOVA F = 51.3, P = 10⁻²⁹), reflecting a larger range of average bleaching scores in MVP than HVP. Because these data were from nubbins exposed to common garden conditions, this variation cannot be ascribed to local variation in environment. Several alternative sources of variation could be: genetic differences among the hosts, hidden variation among Cladocopium symbionts, and long-term impact of environmental history on nursery traits. Genetic differences among host colonies are well known in A. hyacinthus from this population (11, 17), and parental genotypes at 114 loci correlate well with nubbin bleaching (SI Appendix, Fig. S5). The parental genetics of the other species are not as well known. Analysis of transcriptome SNPs from symbionts among our parent colonies does indeed show variation within both symbiont genera (SI Appendix, Fig. S6, PC axis 2), even within the genus Durusdinium, that might relate to bleaching. Thus, even in this well-known system, the genetic basis for host or symbiont effects on variation in bleaching is just starting to be discovered.

Our bleaching data also show significant symbiont switching from genus Cladocopium to genus Durusdinium during the bleaching event, both in parental colonies and also in the common garden nursery clones. This is in contrast to strong persistence of symbiont types after transplantation in previous studies (20), but is in line with a host of studies in other systems on symbiont dynamics (16). Our previous transplants never experienced bleaching conditions, and so the differences in our two sets of results may hinge on the impact of bleaching on different symbionts.

Third, host genotype also played an important role in recovery. Recovery from bleaching has been shown to be dependent on both biotic and abiotic conditions (29, 30). In our system, biotic factors of genotype and species continued to play important roles in rapid recovery. Despite widespread bleaching in 2015, 65% of bleached replicates regained baseline levels of pigmentation within 5 mo. For A. hyacinthus, recovery depended solely on the severity of bleaching (GLM P < 0.01). However, for A. gemmifera and P. damicornis, recovery was also dependent on the genotype (GLM P < 0.01). For example, both genotype GE_M and GE_P had an average bleaching severity across replicates of >90%, but GE_M had 0% recovery while GE_P had 80% recovery.