Hidden heatwaves and severe coral bleaching linked to mesoscale eddies and thermocline dynamics

Marine heatwaves (MHWs) have increased in frequency and severity and are projected to continue intensifying with global warming. Most assessments quantify MHWs using satellite-derived sea-surface temperature (SST) anomalies and accumulated heat metrics relative to climatologies. However, the extent to which heating penetrates below the surface and impacts ecosystems across depth—particularly on stratified fore-reef slopes—is poorly resolved due to limited long-term in-situ observations. Internal waves can intermittently deliver cooler water to reefs, potentially buffering subsurface habitats from surface warming. This study asks how the vertical structure of MHWs differs from SST-based assessments, what role mesoscale eddies and thermocline dynamics play in modulating internal-wave cooling (IWC), and how these processes influence coral bleaching across depths at Moorea, French Polynesia.

Conventional MHW definitions often rely on exceedance thresholds (e.g., 90th percentile) calculated seasonally from SST. In coral-reef contexts, heat stress is commonly quantified relative to a fixed bleaching threshold based on the maximum monthly mean (MMM) SST, using cumulative Degree Heating Weeks (DHW) or Degree Heating Months (DHM). While many El Niño–associated bleaching events have been linked to elevated SST and DHW, these surface metrics frequently under- or over-predict bleaching, even in shallow habitats, implicating incomplete environmental characterization, biological nuance in thresholds, and community shifts. High-frequency internal-wave processes and upwelling can cause substantial subsurface temperature variability, often cooling reefs and mitigating heat stress; however, such dynamics are rarely incorporated into MHW assessments. Spatial and temporal resolution strongly affect heat metrics; finer-scale products (e.g., CoralTemp 5 km) and shorter aggregation windows can better capture local events. Prior work at Moorea and elsewhere demonstrates that internal-wave climates are spatially variable and seasonally modulated by stratification and regional circulation, suggesting mesoscale eddies could alter thermocline depth, mixed-layer properties, and thus IWC, yet these links have seldom been integrated into bleaching risk assessments.

Study area and data: The fore reef of Moorea's north shore (17.5°S, 150°W) was instrumented from Dec 2004 to Aug 2019 with Seabird SBE39/56 temperature loggers (accuracy 0.002 °C; 2-min sampling) at 10, 20, 30, and 40 m depths (40 m logger sampled at 2-hr intervals Aug 2015–Aug 2016 due to setup error). Reef-level bottom pressure (SBE26plus) at 10 m provided sea level anomalies (SLAs) after removing atmospheric pressure and low-pass filtering (cutoff 1/20 day−1) and conversion using seawater density. SST and heat metrics: Daily SST (1985–2019) from NOAA Coral Reef Watch ‘CoralTemp’ (5 km) were analyzed at multiple spatial scales (2°×2°, 1°×1°, 0.1°×0.1° around Moorea). The Society Islands MMM SST was 28.8 °C; a bleaching threshold of MMM+1 = 29.8 °C was used. Heat accumulation was computed as Degree Heating Days (DHD, °C-days) over 12-day moving windows using 2-min in-situ data and daily SST, analogous to DHW but with finer temporal resolution. Local-scale events (0.1°×0.1° over Moorea’s north shore) were emphasized to capture heterogeneity relevant to reef conditions. Internal-wave filtering and metrics: To separate high-frequency internal-wave effects from lower-frequency background variability, observed in-situ temperatures were separated at the local inertial period (~40.0 h) into high-pass (HP) and low-pass (LP) components. To correct for the cold bias introduced by non-linear internal waves interacting with the reef slope, the root-mean-square of HP over a 40 h window (rms(HP)) was added to LP to estimate a non-internal-wave (NIW) temperature time series (NIW = LP + rms(HP)). Internal-wave cooling at 2-min resolution was then IWC = NIW − Obs. Average IWC was summarized over MHW periods; cumulative cooling was quantified as Degree Cooling Days due to internal waves (DCD_IW) analogous to DHD. Spectral analysis: Power spectral densities of temperature time series were computed (Welch method; 12-day windows, Hamming window, 50% overlap; 95% CIs) to assess variance at semi-diurnal frequencies indicative of internal waves across depths and years. Mesoscale context and hydrography: Satellite altimetry SLAs (Copernicus, 1993–2020) were averaged north of Moorea for comparison to reef-level SLAs. Regional SST maps and animations (10°×10° and 20°×20°) characterized surface heterogeneity. Offshore vertical thermal structure and stratification were derived from NOAA WOD CTD casts and Argo profiles (casts >200 m; 1977–2020; 302 casts since 2004). The depth of key isotherms (e.g., 26 °C, 29.8 °C) and buoyancy frequency (N) profiles were compared between 2016 and 2019. Ecological surveys: Coral bleaching prevalence was quantified from photoquadrats (0.5×0.5 m; 36–40 quadrats per site/depth; April–May annually) at two LTER sites (10 m and 17 m) using CoralNet with 200 random points per image. Metrics included percent live coral cover and proportion visibly bleached (all corals pooled, with additional summaries for Pocillopora).

- Surface similarities but subsurface contrasts: Local SST maxima were similar (30.1 °C on 8 Apr 2016; 30.2 °C on 4 Apr 2019), yet 2019 had longer durations above the bleaching threshold within the local 0.1°×0.1° box (18 days vs 4 days in April) and stronger, prolonged subsurface heating. - Subsurface heat accumulation: In 2019, DHD peaked at ~17 °C-days across the reef and extended with magnitude 16.6–8.3 °C-days from 10–40 m, persisting days to weeks. In 2016, heat accumulation was weaker and largely shallow: ~3.2 °C-days at 10 m and <0.8 °C-days at 20–40 m. - Internal-wave cooling (IWC): Internal-wave temperature variance (semi-diurnal band) and IWC were high during 2016 and low at the peak of the 2019 MHW. Average IWC during 2019 was ~0.070 °C at 10 m and 0.31 °C at 40 m, with DCD_IW reduced by 36–60% relative to 2016. High-IWC events (2012, 2015, 2016) showed average IWC 0.14–0.60 °C; low-IWC events (2007, 2017, 2019) aligned with elevated SLAs and reduced cooling. - Thermocline deepening and SLAs: Elevated SLAs in early 2019 (reef-level Mar–Apr mean 8.3±0.2 cm vs 3.0±0.3 cm in 2016; peak 11 cm) coincided with deeper isotherms offshore. The 29.8 °C isotherm depth averaged 14.9 m in 2016 vs 27.7 m in 2019; the 26 °C isotherm in April averaged 89.0±1.2 m (2016) vs 130±1.2 m (2019). Peak N (stratification) depth averaged 62.0±9.9 m (2016) vs 99.9±51 m (2019), indicating deeper, weaker stratification. - Mesoscale eddies: Cyclonic/low SLA conditions (2012, 2015, 2016) enhanced IWC and limited subsurface heating; anticyclonic/high SLA eddies (2007, 2017, 2019) suppressed IWC, deepened thermocline, and intensified subsurface MHWs. - Ecological impacts: Minimal bleaching in 2016 at 10 m (11.5–17.4% of live cover bleached; no detectable mortality by Aug 2016) contrasted with severe 2019 bleaching (54±8.6% of live cover bleached at 10 m on 1 May; 71–72% Pocillopora colonies bleached, 55–65% severely). Post-2019, coral cover declined by ~56% at 10 m and ~10% at 17 m, offsetting nearly a decade of recovery since 2011.

Findings demonstrate that SST-based metrics alone can mischaracterize MHW severity experienced by reef organisms across depth. In 2016, despite similar SST maxima and moderate surface heat accumulation, robust internal-wave cooling coincided with peak surface heating, limiting subsurface thermal stress and ecological impact. In 2019, anticyclonic eddies elevated SLAs, deepened the thermocline and mixed layer, and reduced internal-wave energy reaching the fore reef, allowing heat to penetrate and persist across depths, producing a severe subsurface MHW and extensive bleaching and mortality. The mechanistic linkage between mesoscale eddies, SLA, and thermocline structure explains the paradox of similar SST maxima producing divergent ecological outcomes. These processes challenge assumptions about depth refuges: when thermoclines deepen and IWC weakens, deeper reefs can lose their thermal protection. Given projections of increasing stratification and mesoscale variability, subsurface MHWs may become more frequent and severe, with implications for reef resilience and management. Integrating in-situ temperature records, SLA, and hydrographic profiles with high-resolution SST offers a more accurate assessment of heat stress risk than coarse-resolution, surface-only metrics (e.g., DHW).

This study reveals hidden subsurface MHWs driven by mesoscale eddy-induced thermocline deepening that suppress internal-wave cooling, leading to unexpectedly severe bleaching and mortality across reef depths despite moderate or similar SST maxima. By combining long-term in-situ temperatures, reef and satellite SLAs, and regional hydrography, the work establishes a mechanistic framework linking eddies, stratification, and internal-wave climates to subsurface heat exposure and ecological outcomes. Management and forecasting should incorporate mesoscale dynamics, thermocline structure, and in-situ temperature variability alongside high-resolution SST. Future research should: (1) generalize and validate NIW/IWC methodologies across reef systems; (2) operationalize SLA- and stratification-informed subsurface heat stress forecasts; (3) resolve species-specific responses and depth-dependent vulnerability; and (4) assess potential mitigating roles of internal waves in delivering food resources (e.g., deep chlorophyll maxima) during heat stress.

- Spatial-temporal coverage: In-situ observations are limited to Moorea’s north shore and depths to 40 m; offshore hydrography (Argo/WOD) is opportunistic and sparse near peak events in some years. - Methodological caveat: The NIW estimation (adding rms(HP) to LP) is designed for non-linear internal waves interacting with reef slopes; it may not be appropriate where internal-wave signals are symmetric around a mean (e.g., linear waves near the pycnocline in deep water). - Proxy relationships: SLA–thermocline depth relationships are weaker away from the equator; inference of thermocline depth from SLA has uncertainty at Moorea’s latitude. - Instrument issue: The 40 m logger sampled at 2-hr intervals during Aug 2015–Aug 2016, potentially under-resolving high-frequency variability. - Ecological inference: Bleaching identification from photoquadrats may misclassify some white areas, though confounding factors (e.g., Acanthaster, disease) were considered rare in 2016 and 2019; other stressors (e.g., irradiance, calm conditions) may also contribute to bleaching. - Generalizability: Results may vary at sites with different bathymetry, stratification, internal-wave generation, and eddy fields.