Climate change threatens the survival of scleractinian coral from exposure to concurrent ocean warming, acidification and deoxygenation; how corals can potentially adapt to this trio of stressors is currently unknown. This study investigates three coral species (Acropora muricata, Acropora pulchra and Porites lutea) dominant in an extreme mangrove lagoon (Bouraké, New Caledonia) where abiotic conditions exceed those predicted for many reef sites over the next 100 years under climate change and compared them to conspecifics from an environmentally more benign reef habitat. We studied holobiont physiology as well as plasticity in coral-associated microorganisms (Symbiodiniaceae and bacteria) through ITS2 and 16S rRNA sequencing, respectively. We hypothesised that differences in coral-associated microorganisms (Symbiodiniaceae and bacteria) between the lagoonal and adjacent reef habitats may support coral host productivity and ultimately the ability of corals to live in extreme environments. In the lagoon, all coral species exhibited a metabolic adjustment of reduced photosynthesis-to-respiration ratios (P/R), but this was accompanied by highly divergent coral host-specific microbial associations. This was substantiated by the absence of shared ITS2-type profiles (proxies for Symbiodiniaceae genotypes). We observed that ITS2 profiles originating from Durusdinium taxa made up < 3% and a novel Symbiodinium ITS2 profile A1-A1v associated with A. pulchra. Bacterial community profiles were also highly divergent in corals from the lagoonal environment, whereas corals from the reef site were consistently dominated by Hahellaceae, Endozoicomonas. As such, differences in host–microorganism associations aligned with different physiologies and habitats. Our results argue that a multitude of host–microorganism associations are required to fulfill the changing nutritional demands of corals persisting into environments that parallel climate change scenarios.

In order to allow full comparability with other ocean acidification data sets, the R package seacarb (Gattuso et al, 2020) was used to compute a complete and consistent set of carbonate system variables, as described by Nisumaa et al. (2010). In this dataset the original values were archived in addition with the recalculated parameters (see related PI). The date of carbonate chemistry calculation by seacarb is 2020-11-20.