When researchers re-recorded the same 10 Great Barrier Reef sites after two cyclones and the 2016 mass bleaching, the damaged reefs were an average of 15 dB re 1 µPa quieter than before (Gordon et al., PNAS 2018). That silence matters because fish and coral larvae use reef sound to find a home: playing healthy-reef recordings on degraded patches doubled juvenile fish abundance in a 40-day field trial (Gordon et al., Nature Communications 2019). The backdrop is a reef system under record heat, with 84.4% of global reef area hit by bleaching-level heat stress from 2023 to 2025 (NOAA Coral Reef Watch 2026). And the tools that turn those recordings into a health score are still imperfect: acoustic indices correlate with biodiversity at only r = 0.33 on average (Alcocer et al., Biological Reviews 2022). We aggregated data from NOAA Coral Reef Watch, the Global Coral Reef Monitoring Network (GCRMN), NOAA’s SanctSound project, peer-reviewed studies in PNAS, Nature Communications, Royal Society Open Science and PLOS Computational Biology, and a dozen other primary sources. For the shipping, sonar and whale side of underwater sound, see our separate roundup of ocean noise and marine bioacoustics statistics.
TL;DR
- Damaged reefs recorded 15 dB re 1 µPa quieter than the same reefs before cyclone and bleaching damage (Gordon et al., PNAS 2018).
- Healthy-reef playback on degraded patch reefs doubled juvenile fish abundance and raised species richness by 50% (Gordon et al., Nature Communications 2019).
- Coral larvae settled at 1.7 times the control rate on average, up to 7 times, near a reef speaker (Aoki et al., Royal Society Open Science 2024).
- Larval fish counts were 4 to 14 fold higher at a sound-enriched site in Hawaii (Scripps Institution of Oceanography, Scientific Reports 2026).
- 989 fish species are documented to produce active sounds (Looby et al., Reviews in Fish Biology and Fisheries 2022).
- Acoustic indices correlate with biodiversity at r = 0.33 on average across 364 effect sizes (Alcocer et al., Biological Reviews 2022).
- The best single acoustic index classified only 47% of reef recordings correctly, versus 91.7% for 12 indices combined (Williams et al., Ecological Indicators 2022).
- The reef sound model SurfPerch scored a mean AUC-ROC of 0.933 on reef tasks (Philosophical Transactions of the Royal Society B 2025).
- NOAA’s SanctSound recorded at 30 sites and gathered close to 300 terabytes of sound (NOAA NCEI 2022).
- 84.4% of the world’s reef area faced bleaching-level heat stress from January 2023 to September 2025 (NOAA Coral Reef Watch 2026).
- Global coral cover fell by a relative 8.9% between 2023 and 2024, the largest one-year drop on record (GCRMN Status of Coral Reefs of the World 2025).
1. Healthy vs Degraded Reefs: What a Reef Loses When It Goes Quiet
A healthy reef is loud because it is crowded: snapping shrimp crackle, fish grunt, pop and chorus at dawn and dusk. When coral dies, the animals that make those sounds thin out, and the reef’s acoustic signature fades with them. The most cited test of that idea compared the same 10 reefs near Lizard Island in November 2012 and November 2016, before and after Cyclone Ita, Cyclone Nathan and the 2016 bleaching. The post-damage recordings were an average of 15 dB re 1 µPa quieter, and they also scored lower on acoustic complexity, acoustic richness and shrimp snap rate.
The ecological cost shows up in behaviour. “Settlement” here means a drifting larva leaving open water to take up residence on a reef, and many reef fish use sound to choose where to do it. Juvenile settlement onto patch reefs playing post-damage recordings was 40% lower than onto reefs playing the pre-damage sound, and the damaged soundscape was statistically no more attractive than open-ocean ambient noise (Gordon et al., PNAS 2018). A degraded reef is therefore not just poorer: it is harder for the next generation to find.
| Metric | Value | Source |
|---|---|---|
| Sound pressure level of 10 matched reefs after damage (2016) compared with the same reefs before damage (2012) | 15 dB re 1 µPa quieter on average | Gordon et al., PNAS 2018 |
| Reduction in lagoon reef coral cover after Cyclone Ita (April 2014) | 40% | Gordon et al., PNAS 2018 |
| Attraction of pre-settlement fish larvae to post-damage recordings compared with pre-damage recordings (light traps) | 8% lower | Gordon et al., PNAS 2018 |
| Juvenile fish settlement on patch reefs playing post-damage recordings compared with pre-damage recordings | 40% lower | Gordon et al., PNAS 2018 |
| Correlation between night-time low-frequency (20 Hz to 2 kHz) Acoustic Complexity Index and number of fish species, 8 Moorea reef sites | Spearman ρ = 0.81, P = 0.01 | Bertucci et al., Scientific Reports 2016 |
| Correlation between night-time low-frequency sound pressure level and rubble cover, same 8 sites | Spearman ρ = -0.82, P = 0.01 | Bertucci et al., Scientific Reports 2016 |
| Fish species documented to produce active sounds worldwide | 989 species, from data extracted from 834 studies | Looby et al., Reviews in Fish Biology and Fisheries 2022 (University of Victoria release) |
Context note: restoration can bring the sound back. In Indonesia, reefs restored for 1 to 3 years on previously blast-fished rubble matched healthy reefs in phonic richness (the number of distinct biological sound types heard), and both were significantly higher than degraded reefs, while sound pressure level showed no significant difference in either frequency band (Lamont et al., Journal of Applied Ecology 2022, Dryad dataset). Most recent available data: the Moorea correlations come from Bertucci et al. 2016 and the fish sound inventory from the 2022 University of Victoria FishSounds release.
2. Acoustic Enrichment: Playing a Healthy Reef Back to a Damaged One
If damaged reefs lose fish because they go quiet, the obvious experiment is to make them loud again. “Acoustic enrichment” is exactly that: an underwater loudspeaker plays recordings of a healthy reef over a degraded one. “Recruitment” in these studies means young fish or corals arriving and staying. The landmark trial built 33 patch reefs of dead coral rubble on open sand in the Lizard Island lagoon and ran three treatments (healthy-reef playback, a silent dummy speaker, and no speaker) for 40 days in the November to December 2017 recruitment season. Enriched reefs ended with twice the total juvenile fish abundance and 50% greater species richness than the controls (Gordon et al., Nature Communications 2019).
The effect has since been reproduced with corals and in a second ocean. On St. John, U.S. Virgin Islands, larvae of the mustard hill coral settled at much higher rates near a solar-powered speaker on a degraded reef (WHOI 2024), and a 2023 to 2024 programme in Kāne’ohe Bay, Hawaii, used autonomous cameras to count larval fish at enriched and control sites (Scripps Institution of Oceanography, Scientific Reports 2026). The open question is no longer whether larvae hear the playback, it is whether they stay and survive.
| Metric | Value | Source |
|---|---|---|
| Experimental patch reefs built from dead coral rubble (Lizard Island lagoon, 2017) | 33 reefs, 3 treatments | Gordon et al., Nature Communications 2019 |
| Total juvenile fish abundance after 40 days, healthy-reef playback compared with control reefs | Twice as high | Gordon et al., Nature Communications 2019 |
| Species richness after 40 days, healthy-reef playback compared with control reefs | 50% greater | Gordon et al., Nature Communications 2019 |
| Nights the playback ran successfully, out of 40 | 34 to 36 nights | Gordon et al., Nature Communications 2019 |
| Porites astreoides larval settlement, degraded reef with playback compared with two control reefs (June 2022) | 39 ± 6% vs 13 ± 3% and 14 ± 3% | Aoki et al., Royal Society Open Science 2024 |
| Average settlement with playback compared with control sites, both trials combined | 1.7 times (maximum 7 times) | Aoki et al., Royal Society Open Science 2024 |
| Distance from the speaker at which coral settlement stayed above control levels | At least 30 m | Aoki et al., Royal Society Open Science 2024 |
| Larval fish counts at the sound-enriched site compared with the control site, Kāne’ohe Bay, Hawaii | 4 to 14 fold higher | Scripps Institution of Oceanography, Scientific Reports 2026 |
Outlier note: the coral result was not uniform. In the July 2022 repeat, settlement was 54 ± 4% with playback against 41 ± 6% and 39 ± 5% at the controls, a difference that was not significant (p = 0.0663) (Aoki et al., Royal Society Open Science 2024). In Hawaii, larvae dispersed from the enriched site in less than 48 hours, and the Lizard Island authors stress that attracting fish does not repair habitat on its own. Most recent available data for the fish-community result: the Lizard Island fieldwork ran in 2017. The Hawaii coral-larvae companion study (UC San Diego release, October 2026) placed 37 artificial structures 1 to 42 m from the speaker and found settlement highest closest to it, but reported no percentage increase.
3. Acoustic Indices: How Accurate Is a One-Number Reef Health Score?
An acoustic index compresses a recording into a single number meant to track biological activity. The best known is the Acoustic Complexity Index (ACI), which rewards sound intensity that changes rapidly over time, as animal calls do. Indices are attractive because they are cheap to compute over thousands of hours of audio. The catch: across 34 studies and 364 effect sizes, indices correlated with biodiversity at only r = 0.33 (Alcocer et al., Biological Reviews 2022), with no difference between terrestrial and aquatic settings and with effect sizes shrinking in later studies.
Coral reefs are an especially hard case because snapping shrimp produce a dense, broadband crackle that can swamp or distort an index. Off Kiritimati, shrimp snaps were negatively associated with ACI at every frequency resolution tested (Dimoff et al., Ecological Indicators 2021). The same indices are also used in parks and cities, where our roundup on urban green space soundscapes covers the human-perception side. On reefs, the lesson from the table below is consistent: single indices are weak, combined indices are far better.
| Metric | Value | Source |
|---|---|---|
| Studies and effect sizes in the acoustic index meta-analysis | 34 studies, 364 effect sizes | Alcocer et al., Biological Reviews 2022 |
| Average correlation between acoustic indices and diversity metrics | r = 0.33, CI [0.23, 0.43] | Alcocer et al., Biological Reviews 2022 |
| Share of tests affected by pseudoreplication (non-independent samples counted as independent) | 40% | Alcocer et al., Biological Reviews 2022 |
| Recordings correctly classified by the best single index, healthy (90 to 95% coral cover) vs degraded (0 to 20%) Indonesian reefs | 47% | Williams et al., Ecological Indicators 2022 |
| Recordings correctly classified when 12 indices across 3 frequency bands were combined (regularised discriminant analysis) | 91.7% | Williams et al., Ecological Indicators 2022 |
| Recordings classified unambiguously and correctly by individual indices: Indonesia, Australia, French Polynesia | 5.2%, 13.0%, 12.2% | Williams et al., PLOS Computational Biology 2025 |
| Acoustic indices compared with shrimp snap rates across five Polynesian islands | 13 indices; ACI, AEI and H changed behaviour with FFT settings | Raick et al., Bioacoustics 2025 |
| Frequency resolutions tested for ACI on Kiritimati reefs (2017 to 2018 recordings) | 3 (31.2 Hz, 15.6 Hz, 4 Hz); results depended on the choice | Dimoff et al., Ecological Indicators 2021 |
Context note: the combined Indonesian model was then applied to young restoration sites under 12 months old, where 25.6% of recordings were classified as healthy (Williams et al., Ecological Indicators 2022), consistent with a soundscape that recovers more slowly than the coral planted on it. The Polynesian study recommends avoiding indices whose results change with analysis settings.
4. Machine Learning: Teaching Computers to Hear a Reef
Neural networks avoid the hand-built formula problem by learning their own features from spectrograms. A three-ocean comparison using 20,387 one-minute recordings from Indonesia, Lizard Island and French Polynesia pitted a compound index (8 indices, 44 dimensions) against VGGish, a network pretrained on general audio, and against task-specific networks trained from scratch. The task-specific network classified Indonesian coral cover correctly 91% of the time, against 86% for the compound index (Williams et al., PLOS Computational Biology 2025). The margins were thin for fish diversity, and the task-specific network needed a datacentre GPU: the authors note a standard CPU could not finish one of 50 training epochs in 24 hours.
The bottleneck is labelled data. ReefSet, the annotated reef sound library behind the SurfPerch model, is about 2% the size of comparable bird libraries, so the team mixed bird, reef and unrelated audio during pretraining (SurfPerch study, Philosophical Transactions of the Royal Society B 2025). AUC-ROC in the table is a 0-to-1 score of how well a model separates the right sound class from the wrong ones, where 0.5 is chance. Similar pattern-recognition approaches power industrial listening systems, covered in our acoustic anomaly detection statistics.
| Metric | Value | Source |
|---|---|---|
| Accuracy classifying high vs low coral cover (Indonesia): compound index / pretrained network / task-specific network | 0.86 / 0.88 / 0.91 | Williams et al., PLOS Computational Biology 2025 |
| Accuracy classifying high vs low fish diversity (Lizard Island): compound index / pretrained network / task-specific network | 0.71 / 0.72 / 0.73 (chance = 0.50) | Williams et al., PLOS Computational Biology 2025 |
| Labelled reef sound clips in ReefSet | 57,074 clips of 1.88 seconds | SurfPerch study, Phil. Trans. R. Soc. B 2025 |
| Datasets and countries contributing to ReefSet | 16 datasets from 12 countries | SurfPerch study, Phil. Trans. R. Soc. B 2025 |
| ReefSet size compared with comparable bird sound libraries | 1.99% | SurfPerch study, Phil. Trans. R. Soc. B 2025 |
| Mean AUC-ROC on reef tasks: SurfPerch compared with VGGish | 0.933 vs 0.813 | SurfPerch study, Phil. Trans. R. Soc. B 2025 |
| Estimated cloud cost to train about 139 networks for the SurfPerch study | 26,668 USD (Google Cloud spot pricing) | SurfPerch study, Phil. Trans. R. Soc. B 2025 |
Outlier note: specialisation cuts both ways. On six non-reef test sets, the bird models Perch and BirdNET both beat SurfPerch, so a reef model is not a general-purpose ocean listener.
5. Passive Acoustic Monitoring Deployments: How Much Ocean Is Being Recorded
Passive acoustic monitoring (PAM) means leaving a recorder in the water to listen, without emitting any sound. The largest U.S. effort, SanctSound, was run by NOAA and the U.S. Navy and recorded at 30 sites across seven national marine sanctuaries and one marine national monument, from Hawai’i to both continental coasts (data.gov SanctSound dataset). It collected close to 300 terabytes of underwater sound (NOAA NCEI 2022), capturing fish choruses and snapping shrimp alongside vessel traffic. Ship noise is the main competitor for that acoustic space, a topic our propeller cavitation and vessel noise roundup covers in detail.
The scale problem is analysis, not recording. Three hundred terabytes cannot be listened to by people, which is why the index and machine learning results in sections 3 and 4 decide how useful these archives become. New sensor types are arriving too: in the Red Sea, a fibre-optic cable laid across a reef was used as a continuous line of sound sensors, a technique called distributed acoustic sensing (DAS).
| Metric | Value | Source |
|---|---|---|
| SanctSound recording sites (2018 to 2021) | 30 sites | NOAA and U.S. Navy, SanctSound dataset (data.gov) |
| Protected areas monitored by SanctSound | 7 national marine sanctuaries and 1 marine national monument | NOAA NCEI 2022 |
| Sound data collected by SanctSound | Close to 300 terabytes | NOAA NCEI 2022 |
| Researchers involved in SanctSound | Over 50 scientists from 20 institutions | U.S. IOOS 2022 |
| Coral reef sites with recorders around Singapore, noisy urban waters | 10 sites over two years | Vishnu et al., arXiv 2025 (revised 2026) |
| Reef cable section analysed with fibre-optic DAS, central Red Sea | 50 m, spatial sampling about 0.5 m | Scientific Reports 2025 (KAUST, Thuwal) |
| Frequency range of snapping shrimp compared with fish sounds in the Red Sea DAS study | 1.5 to 20 kHz vs 100 to 1,000 Hz | Scientific Reports 2025 (KAUST, Thuwal) |
| Raw reef audio shared from the three-ocean machine learning study | More than 340 hours | Williams et al., PLOS Computational Biology 2025 |
Context note: in noisy Singapore waters, the authors needed a neural-network denoiser before acoustic indices correlated with diver surveys, while shrimp snap rate from the high-frequency band was robustly correlated with reef parameters (Vishnu et al., arXiv). In the Red Sea study, shrimp snapping became more prominent after onshore lights were switched off (Scientific Reports 2025), a reminder that light pollution can shape a soundscape too. SanctSound’s public portal launched in April 2022 (U.S. IOOS).
6. Bleaching Context: Why Reef Listening Became Urgent
The fourth global coral bleaching event, confirmed by NOAA on April 15, 2024, is now judged to have ended in mid-2025. Between January 1, 2023 and September 30, 2025, bleaching-level heat stress reached 84.4% of the world’s coral reef area, against 68.2% in the third event of 2014 to 2017 (NOAA Coral Reef Watch). “We are now in the era where reefs will bleach on a near-annual basis,” said Coral Reef Watch coordinator Derek Manzello (NOAA NESDIS 2026). Note that heat stress is not the same as confirmed bleaching: not every reef exposed to it bleached.
Surveys confirm the damage. GCRMN’s seventh global assessment, built on 87,299 surveys at 36,886 sites, finds average hard coral cover of 27.3% in 2020 to 2024 against 30.2% in 1980 to 2009 (GCRMN Status of Coral Reefs of the World 2025). Diver surveys are expensive and mostly daytime, which is where acoustics adds value: at Moorea, recorders picked up a night-time rise in reef sound between 2015 and 2021 and higher invertebrate activity at protected sites after the 2016 and 2019 bleaching (Raick et al., Royal Society Open Science 2025).
| Metric | Value | Source |
|---|---|---|
| Share of global coral reef area under bleaching-level heat stress, January 2023 to September 2025 | 84.4% | NOAA Coral Reef Watch 2026 |
| Share of global reef area affected in the third global bleaching event (2014 to 2017) | 68.2% | NOAA Coral Reef Watch 2026 |
| Countries and territories with documented mass bleaching in the fourth event | At least 83 | NOAA NESDIS 2026 |
| Global average hard coral cover, 1980 to 2009 compared with 2020 to 2024 | 30.2% vs 27.3% (relative decline of about 9.5%) | GCRMN Status of Coral Reefs of the World 2025 |
| Relative fall in global coral cover between 2023 and 2024 | 8.9%, the largest single-year decline on record | GCRMN 2025, as reported by Reef Check |
| Coral cover in 2021 inside compared with outside Moorea marine protected areas | 21.7% vs 14.9% | Raick et al., Royal Society Open Science 2025 |
| Average change in night-time low-frequency (below 2 kHz) sound level at 8 Moorea reefs, 2015 to 2021 | +4.07 dB (protected +4.55 dB, unprotected +3.58 dB) | Raick et al., Royal Society Open Science 2025 |
| Outlook probability of bleaching-level heat stress, central and eastern equatorial Pacific, May to August 2026 | 90 to 100% | NOAA NESDIS 2026 |
Context note: GCRMN reports that global bleaching events now recur every 5 to 6 years, and that cover rose by a relative ~6% between 2017 and 2019 before the fourth event reversed much of that gain. The 2024 point value of 25.8% coral cover is reported by Reef Check from the GCRMN report but does not appear on GCRMN’s own summary page. The Moorea lead author’s own caveat: “It’s a story on one island and needs to be repeated in other parts of the world.”
Summary: Coral Reef Soundscapes by the Numbers
| Metric | Value | Source |
|---|---|---|
| Damaged reefs compared with the same reefs before damage | 15 dB re 1 µPa quieter | Gordon et al., PNAS 2018 |
| Juvenile fish settlement to damaged-reef recordings compared with pre-damage recordings | 40% lower | Gordon et al., PNAS 2018 |
| Fish species known to produce active sounds | 989 | Looby et al., Reviews in Fish Biology and Fisheries 2022 |
| Juvenile fish abundance with healthy-reef playback compared with controls | Twice as high | Gordon et al., Nature Communications 2019 |
| Species richness with healthy-reef playback compared with controls | 50% greater | Gordon et al., Nature Communications 2019 |
| Coral larval settlement with playback compared with controls | 1.7 times on average (up to 7 times) | Aoki et al., Royal Society Open Science 2024 |
| Larval fish counts at enriched compared with control site, Hawaii | 4 to 14 fold higher | Scripps Institution of Oceanography, Scientific Reports 2026 |
| Average correlation of acoustic indices with biodiversity | r = 0.33 | Alcocer et al., Biological Reviews 2022 |
| Best single index compared with 12 combined indices, healthy vs degraded reef classification | 47% vs 91.7% | Williams et al., Ecological Indicators 2022 |
| Coral cover classification accuracy, compound index compared with task-specific network | 0.86 vs 0.91 | Williams et al., PLOS Computational Biology 2025 |
| SurfPerch mean AUC-ROC on reef tasks | 0.933 | SurfPerch study, Phil. Trans. R. Soc. B 2025 |
| ReefSet size compared with comparable bird libraries | 1.99% | SurfPerch study, Phil. Trans. R. Soc. B 2025 |
| SanctSound recording sites | 30 | NOAA and U.S. Navy, SanctSound dataset |
| SanctSound data volume | Close to 300 terabytes | NOAA NCEI 2022 |
| Global reef area under bleaching-level heat stress, 2023 to 2025 | 84.4% | NOAA Coral Reef Watch 2026 |
| Countries and territories with mass bleaching | At least 83 | NOAA NESDIS 2026 |
| Relative fall in global coral cover, 2023 to 2024 | 8.9% | GCRMN Status of Coral Reefs of the World 2025 |
| Coral cover inside compared with outside Moorea protected areas, 2021 | 21.7% vs 14.9% | Raick et al., Royal Society Open Science 2025 |
Methodology and Sources
- Gordon, T.A.C. et al. (2018). Habitat degradation negatively affects auditory settlement behavior of coral reef fishes. PNAS 115(20). PubMed Central
- Gordon, T.A.C. et al. (2019). Acoustic enrichment can enhance fish community development on degraded coral reef habitat. Nature Communications 10, 5414. Nature Communications
- Aoki, N. et al. (2024). Soundscape enrichment increases larval settlement rates for the brooding coral Porites astreoides. Royal Society Open Science 11, 231514. PubMed Central and WHOI press release
- Scripps Institution of Oceanography (2026). Autonomous cameras reveal larval reef fish responses to acoustic enrichment and lunar phase. Scientific Reports. Scientific Reports and UC San Diego release
- Bertucci, F. et al. (2016). Acoustic indices provide information on the status of coral reefs: an example from Moorea Island. Scientific Reports. PubMed Central
- Lamont, T.A.C. et al. (2022). The sound of recovery: coral reef restoration success is detectable in the soundscape. Journal of Applied Ecology 59(3). Dryad dataset
- Looby, A. et al. (2022). A Quantitative Inventory of Global Soniferous Fish Diversity. Reviews in Fish Biology and Fisheries. University of Victoria release
- Alcocer, I. et al. (2022). Acoustic indices as proxies for biodiversity: a meta-analysis. Biological Reviews 97(6). PubMed Central
- Williams, B. et al. (2022). Enhancing automated analysis of marine soundscapes using ecoacoustic indices and machine learning. Ecological Indicators 140, 108986. DOI
- Williams, B. et al. (2025). Coral reef soundscape classification with acoustic indices, pretrained networks and unsupervised learning. PLOS Computational Biology (DOI 10.1371/journal.pcbi.1013029). PubMed Central
- SurfPerch study (2025). Using tropical reef, bird and unrelated sounds for superior transfer learning in marine bioacoustics. Philosophical Transactions of the Royal Society B 380, 20240280. PubMed Central
- Raick, X. et al. (2025). Limitations of α-acoustic diversity indices in assessing invertebrate sounds in coral reefs. Bioacoustics 34(3), 207-220.
- Dimoff, S.A. et al. (2021). The utility of different acoustic indicators to describe biological sounds of a coral reef soundscape. Ecological Indicators 124, 107435.
- Raick, X. et al. (2025). Highlighting the resilience potential of marine protected areas in the face of coral bleaching with passive acoustic monitoring. Royal Society Open Science 12(7), 241938. PubMed Central
- NOAA and U.S. Navy, SanctSound Raw Passive Acoustic Data. data.gov; NOAA NCEI, SanctSound: Studying the Underwater World of Sound (2022). NCEI; U.S. IOOS, Live for Listening (2022). IOOS
- Vishnu, H. et al. Passive Acoustic-based Composite Indices for Reef Health Monitoring in Noisy Tropical Waters (arXiv preprint, 2025, revised 2026). arXiv
- Marine reef soundscape monitoring with fiber-optic distributed acoustic sensing (2025). Scientific Reports. Scientific Reports
- NOAA Coral Reef Watch, 4th Global Coral Bleaching Event report page (updated June 2, 2026). Coral Reef Watch; NOAA NESDIS, World’s Fourth Mass Coral Bleaching Event Likely Ended in 2025 (June 2, 2026). NESDIS
- GCRMN, Status of Coral Reefs of the World: 2025 (released August 31, 2026). GCRMN; 2024 point value via Reef Check
- Data watch: (1) Heat stress is not bleaching: NOAA’s 84.4% is reef area exposed to bleaching-level heat stress, and its NESDIS release rounds it to 84%. (2) SanctSound’s time span differs by source: the data.gov record says 2018 to 2021, NCEI says fall 2018 through spring 2022. (3) The 25.8% coral cover value for 2024 comes from Reef Check’s summary of the GCRMN report and is not on GCRMN’s summary page. (4) The PLOS Computational Biology paper gives 20,437 recordings in its discussion while its three dataset counts sum to 20,387; we use the sum. (5) The WHOI release on the coral playback study gives two different figures for historical reef loss, so we did not use either. (6) Several core soundscape studies (Gordon 2018 and 2019, Bertucci 2016, Dimoff 2021) are more than three years old: most recent available data for those field results. (7) The Hawaii fish-larvae multiplier comes from a 2026 paper whose per-site totals we could not read; the Singapore study is a preprint and not yet peer reviewed. (8) NOAA warned of a possible El Niño in 2026, so a new bleaching wave could change section 6 quickly.
- Last updated: October 24, 2026. We update this roundup quarterly, and the next refresh is expected when NOAA Coral Reef Watch publishes its assessment of 2026 El Niño heat stress.