Ocean deoxygenation is predicted to threaten marine ecosystems globally. However, current and future oxygen concentrations and the occurrence of hypoxic events on coral reefs remain underexplored. Here, using autonomous sensor data to explore oxygen variability and hypoxia exposure at 32 representative reef sites, we reveal that hypoxia is already pervasive on many reefs. 84% of reefs experienced weak to moderate (≤153 to ≤92 μmol O₂ kg^-1) hypoxia and 13% experienced severe (≤61 μmol O₂ kg^-1) hypoxia. Under different climate change scenarios based on 4 Shared Socioeconomic Pathways (SSPs), we show that projected ocean warming and deoxygenation will increase the duration, intensity, and severity of hypoxia, with more than 94% and 31% of reefs experiencing weak to moderate and severe hypoxia, respectively, by 2100 under SSP5-8.5. This projected oxygen loss could have negative consequences for coral reef taxa due to the key role of oxygen in organism functioning and fitness.

As described in the Methods section of the accompanying manuscript and the Supplementary Information by Pezner et al.:

Sensor and deployment information

The majority of dissolved oxygen data presented in the current study (25 of 32 sites) were recorded by SeapHOx or Sea-Bird Scientific CTD sensor packages with Aanderaa oxygen optodes (Table S1). The remaining oxygen datasets were recorded by Idronaut CTD and oxygen sensor packages (Dongsha 1 and Dongsha 2), Sea-Bird SBE19 Plus CTD and SBE 43 oxygen sensor packages (Baker, Jarvis, Palmyra 3, and Tutuila), or Sea-Bird Scientific CTD and PME miniDOT sensor packages (Taiping 1). Detailed site and deployment information can be found in the Supplementary Information Extended Methods section. For each location, different instrument deployment sites are represented by numbers (e.g., Dongsha 1 and Dongsha 2), or a combination of letters and numbers where letters represent either different depths at the same site (e.g., Bocas 1a, 1b, 1c, and 1d) or different deployments at the same site over time (e.g., Crocker 1a, 1b, and 1c). 

Calibrations and conversions of datasets

All sensors were calibrated by the manufacturer or according to the manufacturer’s instructions by the user as noted in previous publications. If applicable, salinity corrections were implemented according to the manufacturer’s specifications and analog measurements were converted from voltages to concentrations. Data were assessed for quality and erroneous data points (defined as missing values or values that exceeded the measurement range of the instrument), which were flagged and excluded from any subsequent analyses. Deployment data were restricted to measurements in seawater based on salinity values to exclude extraneous data points collected during instrument transport, initial deployment, or recovery (exposure to air). Density calculations using the Gibbs Seawater Toolbox functions in RStudio were used to convert all oxygen concentration units to μmol O₂ kg^-1. See Supplementary Information Extended Methods (and Table S8) for an assessment and discussion on the potential errors and uncertainty of the oxygen measurements. All data and relevant code (https://github.com/apezner/GlobalReefOxygen) files are available for download online and may be used with proper attribution.

The primary dataset is provided as an R data file and in CSV format for ease of access. All secondary datasets are provided in CSV format.