Historically, massive coral bleaching in Hawaiian ecosystems was unusual, until 1996 [78]. The consecutive heatwaves of 2014 and 2015 in Kane‘ohe Bay allowed us to track temporal shifts in bleaching susceptible and resistant coral microbiomes in situ, during and after the bleaching peaks. Pcom_B corals recovered faster (after ~ 2.5 months) than Pacu_B (~ 3 months), and Mcap_B (~ 6 months), according to color scores [51], yet actual Symbiodineaceae densities could have been regained faster [38]. Prokaryotes, in turn, were expected to exhibit more rapid responses to stressors, due to their fast generation times [49, 79, 50].
Algal and prokaryotic communities in our corals followed a species-specific pattern, frequent in sympatric populations [29, 27], whereas intraspecific Symbiodiniaceae signatures were identified at the colony scale [80]. Mcap had the most variable ITS2 profiling, followed by Pcom and Pacu, whilst Symbiodiniaceae composition was influenced by bleaching susceptibility. Algal-genotypes conferring different bleaching resistance in conspecific hosts may appertain to the same genus, as in Pcom [4], or to different ones as in Mcap [22, 29]. But also, susceptibility can be independent from symbiont-type [81, 27].
Bacterial compositions were more diverse in Mcap and Pacu than in Pcom, and were practically constant between bleaching phenotypes. Microbial stability after natural thermal disturbance has been reported in corals undergoing sub-bleaching [44] and bleaching [29]. While, community shifts were documented after induced stress [45, 47, 46, 82]. In our corals, certain bacterial-ASVs/ Symbiodiniaceae-DIVs were differentially abundant across time and/or bleaching susceptibility, highlighting the potential of fine-scale microbiome changes in coral resilience [79, 83, 44]. Below we discuss corals’ microbiomes’ dynamics during bleaching and recovery by host species.
Pocillopora acuta
Pacu had the highest bleaching incidence, and was associated with eight fluctuating Symbiodiniaceae C1d-profiles. This agreed with the C1d-dominance described for this species in Hawai’i [84]. Predicted profiles dominated by C1d and C42.2 likely reflect the preponderance of mixed Cladocopium pacificum and C. latusorum [35]. Lack of acclimatization patterns agrees with stabilities of dominant symbionts in pocilloporids under thermal stress. Whilst, profile shifting driven by minor ITS2-sequences shifts, is presumably matching with background genotype variability reported in previous studies [85, 37, 44]. In other geographies, higher bleaching thresholds have been reported in populations harboring Durusdinium glynnii (previously D1) [86, 87, 88, 89, 90], or in chunky (versus fine) morphotypes, even when presenting C1d [81, 44]. Therefore, the bleaching incidence observed in Pacu could rely on a combination of having fine morphology and Cladocopium–profiles, both correlated to higher susceptibilities [81].
Bacterial communities were dominated by phylum Proteobacteria, followed by Bacteroidetes, Actinobacteria, Firmicutes and Cyanobacteria, similar to pocilloporids from other regions; whereas, Family Amoebophilaceae (mostly Candidatus amoebophylus) and genus Acinetobacter (largely A. calcoaceticus) were more preeminent, and Endozoicomonas less abundant in Pacu [45, 48, 88, 89, 90]; but see [44, 91]. Prokaryotic community did not show significant changes over time, as in other surveys involving coral bleaching [50, 29]. Nonetheless, balances of differentially abundant taxa revealed lower log-ratios in corals at the bleaching peak M0. Upon recovery (M1–M12) Pacu was correlated to Endozoicomonas, Cyanobium, Acinetobacter, Pseudomonas and Neisseriaceae, whereas bleached colonies in M0 were associated to Micrococcus, Lawsonella, Synechococcus, Bacillus and Staphylococcus. Likewise, cross co-occurrence networks showed an increase in node complexity and positive interconnections from M1. This implied that sparse interactions between bacteria and Symbiodiniaceae during thermal stress, increased in number as algal cells repopulated in the recovery process after M0, yielding larger networks.
Recovery in Pacu happened after 2–3 months [51]; probably thanks to heterotrophic feeding [92, 93] and, microbiome rearrangements in early recovery phases [94, 49].
Montipora capitata
Mcap colonies were associated with Cladocopium and Durusdinium symbionts. At the DIV level C31, C17 and C21 were predominant genotypes in both B and NB corals, while D4, D1, D6, D1ab and D3h characterized NB colonies, in agreement with recent studies [95]. Bleaching resistant Mcap_NB colonies contained either pure C or mixed D/C profiles (50% of the times), and were different from susceptible Mcap_B, which contained basically C-profiles. Adjacent colonies never shared the same ITS2-profile. In both bleaching phenotypes, six colonies (66%) maintained their corresponding dominant profiles, the remaining (three) experienced temporal shifts, in agreement with Cunning et al., [38]. C31-C17d-C31.1-C31a-C21-C31f-C17e-C31l-C21ac might represent a thermosensitive ITS2-profile, as 8 out of 9 Mcap_B bleached colonies in M0 contained this profile, whilst its presence in Mcap_NB (1–2 colonies) was always in combination with D-profiles. In purity or mixed, D-genotypes provide thermal resistance in M. capitata, but colonies with C-profiling also demonstrated stress-tolerance [38]. Our analyses based on ITS2-types [30] identified different Cladocopium profiles, in comparison to previous surveys reporting solely C31-genotype [84, 85, 38], which could resolve the disparate stress-resistance of Mcap_NB vs Mcap_B. In one exception though, two colonies containing the same profile (C31/C17d-C21-C31.9-C21ac-C17e-C31h-C31i) at M0, one underwent bleaching and the other one not, suggesting multiple factors other than symbiont type regulating thermal tolerance. Mcap_B and Mcap_NB maintained different Symbiodiniaceae compositions, based on profiles and underlying ITS2-sequences, while colony heterogeneity in bleached Mcap_B increased with time, with no clear stabilization pattern. Actually, only one colony acquired a partial Durusdinium profile at M6, supporting the low prevalence of symbiont shuffling described in this species [38].
Prokaryotic communities were dominated by Proteobacteria (Family P3OB-24, Order Myxococcales), and by genera Acinetobacter (largely A. calcoaceticus) and Endozoicomonas. In general, they matched with M. capitata microbiomes, characterized by the presence of Cyanobacteria and Deinococcus-Thermus, and low abundance of Vibrio [54, 96]. Even if non statistically significant, increased alpha diversities observed in Mcap_B at M1 and M3 may suggest microbial rearrangements after thermal-stress [47, 48, 97], or seasonal fluctuations in Mcap_NB at M6 [38]. Log-ratio rankings of differentially abundant taxa were higher in Mcap_NB with respect to Mcap_B at M0 and M6. At these two time points of symbiont depletion: bleaching peak (M0) and seasonal algal downturn (M6; as in [38]), Mcap_NB was ranked to numerator taxa –Endozoicomonas, Acinetobacter and Pseudomonas; whereas bleached Mcap_B were correlated to denominator taxa –Myxococcales, Lawsonella, Micrococcus, Synechococcus, Bacillus and Staphylococcus. Cross networks became more complex in Mcap_B from M1 to M6, as algal densities recovered (M1–M3), and bacteria established interactions with Symbiodiniaceae. Instead, Mcap_NB showed higher network complexity in M0 compared to bleached Mcap_B colonies, reflecting stress response rearrangements between thermo-tolerant algal and prokaryotic symbionts during the heatwave.
M. capitata was found to rely on heterotrophy to compensate for energy losses when experimentally bleached [16]. Mcap did not evidence such trophic plasticity, and would have regained symbiont populations at expense of biomass resources by January 2015 [18, 51], in agreement with the microbial outcomes.
Porites compressa
ITS2-profiling in Pcom revealed C15-dominance, in accordance with older surveys on Porites compressa [84]. Pcom_NB and Pcom_B corals held distinct Symbiodiniaceae patterns 70–90% of the times, across M0–M12. While, other characteristics in the holobiont should explain why 20% adjacent Pcom_B and Pcom_NB colonies sharing the same profiles had different susceptibilities in M0. During the peak of the heatwave in Oct-2014 (M0) Pcom_NB associated to DIVs C15cc and D6, and more often to the ITS2 profile C15-C5ci-C15cc-C15cl-C15n-C15cj-C15l, which could represent a thermotolerant symbiont-type found in 7 out of 10 resistant colonies, and in only one susceptible Pcom_B. Accordingly, this profile was less prevailing in M6 (May 2015), coinciding with a period of minor thermal disturbances and lower symbiont abundances [98, 36]. C15-genotypes with higher temperature tolerance were already described associating to Porites spp. from the Great Barrier Reef [23]. Dissimilarities in ITS2-sequences between Pcom_B and Pcom_NB tended to vanish after M1, reflecting algal rearrangements linked to recovery from this time point. This concurs with coral photo-physiology data supporting intense symbiont repopulation (elevated cell mitosis and photopigment synthesis) from Nov 2014 [18, 52, 51] (.
Bacterial communities in Pcom were less diverse than in the other hosts, accounting for many low abundance taxa, and ~ 90% predominance of a single Endozoicomonas microbe. The bacterial community structures were relatively constant, across bleaching phenotypes and time. Salerno et al., [53] also found stable microbiomes in P. compressa under mild thermal treatments; whereas Vega Thurber et al., [47] observed switches from healthy to pathogenic microbiota after intense high temperature exposures. Both of these thermal stresses were administered in an experimental setting. In our field data the prevalent ASV (694df3c7f8b6b66c922ed51a965d75d0a) matched with a symbiont (Oceanospirillaceae-OTU C7-A01: FJ930289.1; Supplementary Material S2) broadly documented in Porites spp. (including P. compressa from Maui) and other hermatypic corals from Australia, Hawai‘i, and Bermuda [99], suggesting a conserved large-scale partnership with corals [100] (. Coral-microbiomes dominated by one or few Endozoicomonas phylotypes were described to have microbial inflexibility in stress responses [50]. In our susceptible Pcom_B corals dominated by one Endozoicomonas strain though, the microbial balance composed by two Endozoicomonas (the predominant ASV above and another congeneric strain), Candidatus Amoebophilus, Acinetobacter calcoaceticus, Pseudosmonas stutzeri, Synechococcus and Roseitalea features; over five antagonistic Endozoicomonas strains, Micrococcus, Staphylococcus and Neisseriaceae taxa, pinpointed a longitudinal discontinuity of increased log-ratios in Pcom_B at M1. Microbial communities of bleaching resistant Pcom_NB phenotypes, in contrast, remained stable and dominated by Endozoicomonas.
Pcom was characterized by small cross networks with mild fluctuations between heatwaves, reflecting a much simpler microbial community. Increased edge complexity at M1 in Pcom_B again suggests a rapid recovery response, with reliance on few bacterial ASVs; as compared to Mcap_B and Pacu_B, reflecting larger bacterial consortia participating in the recovery. Reduced trophic plasticity, and intense loss of Symbiodiniaceae and photosynthetic pigments might obligate Pcom_B to regain symbionts faster, at high biomass investment with respect to the other species [18, 52]. Furthermore, intense algal repopulation in Pcom_B from October–November 2014 was correlated with low symbiont ∂15N [18], and assimilation of 15N depleted sources, possibly derived from diazotroph bacteria via N2 fixation [101, 102, 103]. Indeed, differentially abundant taxa ranked to recovering Pcom_B included various diazotroph taxa (see below).
Differentially abundant bacterial taxa defining temporal shifts in bleaching recovery
Coral microbiomes in the present study revealed minor community disruption in response to heatwaves. Similar outcomes were reported previously, together with increases in potentially beneficial taxa [94, 44]. One bacterial group widely associated to corals and documented to display diversified tolerances and/or functional traits to stress conditions is Endozoicomonas [45, 47, 46, 100, 97, 50, 49, 82, 83, 44]. In our corals, saving an initial decline in Mcap_B at M0, this genus displayed preponderance throughout bleaching stress, in agreement with other studies [82, 50, 44]. Endozoicomonas symbionts are proposed to play three kinds of functions: 1) nutrient acquisition/provision –carbon, nitrogen, sulphur, methane recycling, amino acid production, dimethylsulfoniopropionate (DMSP) metabolism; 2) microbiome modulation –quorum-sensing; and 3) promotion of host health –antimicrobial activity, pathogens exclusion [100, 104]. DMSP produced by Symbiodiniaceae and sulfur-derivatives from certain prokaryotes (Endozoicomonas, Acinetobacter, Pseudomonas, Vibrio) provide a selective environment structuring bacterial populations [105]. Hence, upon the downturn of DMSP production throughout bleaching/stress episodes, high abundances of Endozoicomonas might modulate microbiomes steadiness [82, 50, 44]. Further, diazotroph bacteria contribute to homeostasis during bleaching and sub-bleaching recovery after thermal stress [94, 44], by suppling limiting nitrogen to Symbiodiniaceae [101, 106, 102, 103]. Indeed, many differentially abundant taxa positively ranked to our recovering corals included diazotrophs and/or DMSP-metabolizing bacteria: e.g., Endozoicomonas, Acinetobacter calcoaceticus, Pseudomonas stutzeri, Cyanobium [107, 101, 106, 105]. High occurrence of Acinetobacter spp. and Endozoicomonas spp. is frequently documented in healthy and bleached Scleractinia, implying synergistic roles in fitness [108]. Another recently described symbiont in coral holobionts is Candidatus Amoebophilus, an intracellular associate of unicellular eukaryotes, like Symbiodiniaceae or amoebae, with undefined role [109, 44]. Differential features in this genus were correlated to algal repopulation, particularly in Pcom_B. Bleaching entails loss of major nourishment inputs and photoprotection, and corals therefore implement compensatory strategies [7]. For instance, bleached corals have been observed to reinforce feeding on planktonic diazotrophs and preferentially on nitrogen-rich Synechococcus cyanobacteria [110]. Accordingly, bleached colonies and incipient recovery stages in this study were associated to Synechococcus; but interestingly also to differential features with potential UV-absorbing properties, like Bacillus, Staphylococcus [111], Micrococcus [112], and the already mentioned Cyanobacteria –Cyanobium, Synechococcus [113]. Notably, Bacillus and Staphylococcus strains within the coral mucus have demonstrated to increase their UV-absorbance range in response to elevated temperatures, likely protecting bleached colonies from excessive irradiation prior to recovery [111]. Lawsonella was another genus frequently associated with bleached corals here. Despite little information exists on marine representatives, it could involve opportunistic/transient microbes, as those described in certain human abscesses [114]. Differentially abundant taxa were broadly shared between Mcap and Pacu, and partially matching with Pcom –this last chiefly influenced by Endozoicomonas spp. This outcome is appealing, and suggests that locally the same players may modulate stress responses in different coral species. Thus, understanding the dynamics of differentially abundant microbial consortia in correlation with bleaching and recovery, could provide regional indicators to forecast the fate of sympatric corals to upcoming heatwaves [79]. Furthermore, certain strains could be proposed as “probiotics” to improve coral resistance [115].