Intrapopulation adaptive variance supports thermal tolerance in a reef-building coral

Abstract

Coral holobionts are multi-species assemblages, which adds significant complexity to genotype-phenotype connections underlying ecologically important traits like coral bleaching. Small scale heterogeneity in bleaching is ubiquitous in the absence of strong environmental gradients, which provides adaptive variance needed for the long-term persistence of coral reefs. We used RAD-seq, qPCR and LC-MS/MS metabolomics to characterize host genomic variation, symbiont community and biochemical correlates in two bleaching phenotypes of the vertically transmitting coral Montipora capitata. Phenotype was driven by symbiosis state and host genetic variance. We documented 5 gene ontologies that were significantly associated with both the binary bleaching phenotype and symbiont composition, representing functions that confer a phenotype via host-symbiont interactions. We bred these corals and show that symbiont communities were broadly conserved in bulk-crosses, resulting in significantly higher survivorship under temperature stress in juveniles, but not larvae, from tolerant parents. Using a select and re-sequence approach, we document numerous gene ontologies selected by heat stress, some of which (cell signaling, antioxidant activity, pH regulation) have unique selection dynamics in larvae from thermally tolerant parents. These data show that vertically transmitting corals may have an adaptive advantage under climate change if host and symbiont variance interact to influence bleaching phenotype.

Fig. 1. Genomic and symbiotic dynamics underly bleaching phenotype in Montipora capitata.

a Hierarchical clustering of genetic distances calculated from genotype likelihoods, with one sequenced pair of biological replicates to confirm the absence of clonality and 3 M. flabellata. b PCA of 20,407 loci, shaded by historical bleaching phenotype. c Symbiont communities in three haphazard replicates of each colony used in this study, shaded by proportion of Cladocopium or Durusdinium measured via qPCR.

Fig. 2. Comparison of genetic associations with binary phenotype and symbiont state.

Points represent LRT values from two association tests: case-control association by allele frequencies using phenotypes as the dependent variable and continuous association by gene dosage using proportion Durusdinium as the dependent variable. Points (n = 10,803) represent loci shared between the two analyses, excluding those which were filtered to complete continuous association. Point color represents absolute effect size (beta) from symbiont association. Gray lines represent independent LRT value thresholds at p = 0.0001 determined by 10 k permutations. Venn diagram contains counts of these significant loci for phenotype, symbionts and shared between the two. Black line is x = y.

Fig. 3. Functional ontologies distinguish bleaching phenotype and relate to biochemical patterns.

a The 28 gene ontologies that were significantly enriched in loci highly associated with bleaching phenotype (FDR p < 0.0001). Points represent each phenotype in the study (mean ± 1 SE) along the first principal component that describes all loci associated with the ontology of interest. Number of loci corresponding to each ontology are in parentheses, red ontologies are significantly enriched in loci with high associations with symbiont state and thus impact phenotype via symbiosis. b WGCNA modules of the corals in this study, shaded by correlation with loading in (a). Each column represents a metabolite module resolved by WGCNA and each row corresponds to gene ontologies to the left. No correlations were significant after FDR correction.

Fig. 4. Fitness and Symbiont Dynamics in Larval and Juvenile corals.

a Final survivorship 213 h post fertilization for larvae at ambient and high temperatures. b Final survivorship 34 days post settlement for juveniles at ambient and high temperatures. Boxplots are mean 1± IQR, red fill denotes high temperature treatments. c Proportion Durusdinium of the three phenotype larval pools from fertilization to 84hpf in ambient and high temperatures. Line is best-fit with 95% confidence interval. Letters represent post hoc significance values for each life-history stage.

Fig. 5. Selection Dynamics in Larval heat stress experiment.

GO delta rank derived from PBS statistics of cross and nonbleached larval pools. High values represent gene ontologies enriched in large PBS statistics. Tan points represent the 4 gene ontologies that were significant (FDR p < 0.1) in both cross and nonbleached phenotypes, with lm best fit line. Black points represent the ontologies that were not significant in both treatments, with gray lm best fit line. Blue points represent gene ontologies with enrichment in Nonbleached phenotype and delta rank <100 (not significant) In the cross phenotype, isolating the genomic effect of selective breeding. Venn diagram shows shared significant ontologies between phenotypes.