The Red Sea is a natural laboratory for the ocean of tomorrow. Even hundreds of metres below the surface, its waters stay above 21°C and salinity approaches 40 PSU, conditions that would be considered extreme in almost any other deep basin. The corals living in its twilight and deep zones are therefore already tolerating some of the warmest deep waters on Earth. The question is how much further they can be pushed as the basin continues to warm.

Knowledge gap

Deep-sea corals build complex three-dimensional habitats that sustain biodiversity, retain resources and support ecosystem functioning far below the reach of sunlight. The global deep ocean is warming at around 0.1°C per decade, and organisms adapted to stable conditions and slow life histories are expected to be highly vulnerable to even small changes. Yet surveying the deep seafloor requires remotely operated vehicles and submersibles, which are costly and rarely available, leaving most deep-sea species data deficient.

In the Red Sea, habitat models had only described where corals live today. No study had mapped how their distributions may shift under future climate change, a gap that has prevented the design of conservation and management measures able to remain effective over the coming decades.

Main approach

We compiled data from six oceanographic expeditions conducted between 2020 and 2023 on board R/V OceanXplorer, R/V Aegaeo and R/V Meteor. From 325 benthic video transects recorded by ROVs and manned submersibles, we extracted 653 occurrence records for four coral taxa representing four distinct orders: the octocoral genus Acanthogorgia, the endemic black coral Bathypathes thermophila, the endemic hydrocoral Stylaster tritoni and the scleractinian Madracis interjecta, a mesophotic bioherm builder. These records were related to seafloor environmental conditions with ensemble ecological niche models combining Boosted Regression Trees and MaxEnt, and projected across the entire Red Sea for the present and for the end of the century (2090 to 2100) under five Shared Socioeconomic Pathways, from SSP1-1.9 to SSP5-8.5.

Technological challenge

Deep corals respond to fine-scale features such as ridges, slopes and seamounts, but available ocean climate layers resolve the Red Sea at roughly 5 km, too coarse to capture the steep topography where these corals settle. We tackled this through a dedicated data and modelling pipeline:

We merged 79,654 km² of high-resolution multibeam bathymetry with GEBCO data to build a seamless 400 m digital terrain model of the whole basin, from which slope and seafloor ruggedness were derived.

Using the Bio-ORACLE pipeline, we integrated 146 in situ CTD casts of temperature, salinity and oxygen with Copernicus data, and applied kriging-based trilinear interpolation across longitude, latitude and depth to downscale present and future seafloor conditions from 5 km to 400 m.

To overcome sampling bias towards the eastern Red Sea, pseudo-absences were drawn from k-means clusters of environmental space, models were validated with spatial block cross-validation, and monotonic constraints were imposed on thermal and current responses to improve transferability to future climates.

The resulting models performed strongly, with average AUC between 0.87 and 0.95 and TSS between 0.66 and 0.86.

Main finding

Every coral taxon is projected to lose part of its currently suitable habitat by the end of the century, with losses intensifying under more severe emission scenarios. Madracis interjecta is the most threatened, losing over 45% of its habitat under SSP5-8.5 (from 29,317 km² to 15,893 km²) with almost no new areas becoming available. In contrast, the black coral Bathypathes thermophila retains nearly all of its range, losing only around 2.3%, although it will need to withstand waters almost 2°C warmer. Acanthogorgia and Stylaster tritoni may gain new habitat in deeper waters, with the mean depth of suitable habitat for S. tritoni moving from 424 m to 675 m under SSP5-8.5. However, many of these new areas lie along the isolated central axis of the basin, and their colonisation will depend on dispersal capacity. Even where habitats deepen, corals will still face rising temperatures, of up to 2.3°C for M. interjecta, highlighting the magnitude of change expected in the deep sea.

Areas suitable for all four taxa covered 16,960 km² and pinpointed potential biodiversity hotspots in the northern Red Sea, around the Farasan Bank and north of the Dhalak Islands. These hotspots persist under future conditions, although their extent shrinks under the most severe scenarios.

Implications for conservation and management

These results provide the first basin-wide, climate-informed baseline for the spatial planning of mesophotic and deep-sea coral habitats in the Red Sea, in support of the UN 2030 Agenda and Sustainable Development Goal 14:

Areas that remain suitable now and in the future, particularly the northern Red Sea, the Farasan Bank and the region north of the Dhalak Islands, should be prioritised as climate refugia within a dynamic network of marine protected areas.

Model uncertainty is highest outside Saudi Arabian waters, calling for exploration of the western Red Sea and stronger data sharing among all Red Sea nations to support coordinated, transboundary management.

Physiological experiments and larval dispersal models are now needed to determine whether corals can tolerate warmer waters or reach newly suitable deep habitats, the key uncertainties for their long-term persistence.