Dataset: Stable oxygen (δ18O) and carbon (δ13C) isotope measurements from a Curaçaoan Massive Starlet Coral core collected from Tugboat Beach, Caracas Bay containing data from 1937 to 2022

During two separate field campaigns in February 2023 and March 2024, we collected a single skeletal core (TB-01) from a living ~ 1 m tall colony of Massive Starlet coral (Siderastrea siderea (urn:lsid:marinespecies.org:taxname:207516)) from Tugboat Beach (12.0686°N, 68.8619°W), a tourist beach located in a semi-enclosed bay on the southeastern end of the island in the town of Jan Thiel. The second field campaign extended the length of the initial core from the first field campaign. This species was targeted due to its hardiness and resilience, slow (3 – 5 mm yr^-1) linear extension rate, and consistently robust geochemical proxy calibrations throughout the Caribbean (DeLong et al., 2014; 2023; Moses & Swart, 2006; Saenger et al., 2009). Curaçao also sports relatively high S. siderea benthic abundance amongst the island’s southeastern reefs (The Waitt Institute, 2017). Although the immediate surrounding bathymetry of the core location is shallow (~5 m), the coral head is located within 50 m of a steep terrace which drops quickly to 30 m depth. Within 200 meters of the sampled coral colony, the seafloor reaches 100 m depth. Skeletal cores were collected vertically from the top of the colony via a 1.5-inch diameter pneumatic air drill. The total core length retrieved was ~50 cm, which was split into 4 sections. These splits were clean, with minimal material loss, minimizing hiatuses in the final coral record. The core was slabbed and x-rayed to identify density banding associated with seasonal growth and to establish a preliminary age model based on the average growth rate. X-ray images showed a coral growth rate of about 5.5mm/yr, allowing for sub-seasonal temporal resolution. On average, each year contained ~5.36 samples equating to roughly a bimonthly resolution. Slabs were sonicated in ultrapure 18.2 MΩ-cm water for 15-30 minutes and dried in an 80°C oven prior to drilling. Thecal walls of slabs were sub-sampled (target mass = 300 mg) at a continuous 1-mm resolution along transects perpendicular to the growth axis using a Sherline Model 4510 drill. Actual powder masses were weighed using a microbalance before being transferred into 12 mL Exetainer borosilicate vials for stable isotope (δ¹⁸O and δ¹³C) analyses at Vanderbilt University using a ThermoScientific Delta V Isotope Ratio Mass Spectrometer (IRMS) equipped with a GasBench II online preparation and introduction system (e.g. Belanger et al., 2025; de Wet et al., 2024).

The age model for the TB-01 core was constructed using a seasonal tie-point approach, which is standard for high-resolution coral reconstructions. Annual seasonal extrema were identified in the δ¹⁸O record. The δ¹⁸O minima (warmest temperatures) were assigned to October, and the δ¹⁸O maxima (coolest temperatures) were assigned to February. These assignments were corroborated by annual density banding observed in X-ray images, which showed a mean linear extension rate of ~5.5 mm yr^-1. Between these fixed seasonal tie-points, ages were assigned to each 1-mm depth increment using linear interpolation. The "Decimal Date" column in the main data file represents the calendar year plus the fraction of the year elapsed (e.g., a value of 2022.79 corresponds to the 288th day of the year, or mid-October). This format allows for precise temporal alignment with instrumental climate indices and satellite SST datasets.