Within-region Reef Check models give a bleaching odds ratio of 0.91 per 0.1 °C of weekly temperature variation in 2002–2019

Does the link between weekly temperature variation and coral bleaching remain within the same ecoregion and year? We used Reef Check records from the Global Coral-Bleaching Database, BCO-DMO version 2. We removed repeated rows and fitted logistic models of at least 10% bleaching. The primary sample has 6,767 surveys at 2,693 sites in 41 ecoregions, from 2002 to 2019. It includes 761 positive outcomes in 186 ecoregion and year groups with both outcomes. At fixed measured heat stress, depth and latitude, the model gives an odds ratio of 0.911 per 0.1 °C higher standard deviation of weekly temperature anomalies. The 95% site-cluster interval is 0.828 to 1.003. A 1,000-sample site bootstrap gives 0.817 to 0.995. Spatial-cluster intervals include 1. All planned bleaching thresholds give negative point estimates, but most intervals include 1. Separate fits give different estimates across survey sources. Total SST variation gives an odds ratio of 0.989, unlike anomaly variation. The evidence favors a modest negative association in these surveyed reefs. It does not establish a precise common effect across settings or show that temperature variation causes resistance. The exposure is measured over the full reference record, and local daily temperature variation is not tested. The package includes source data, provenance, an analysis plan, all model results, figures, code and reproduction instructions.

coral bleaching · temperature variation · Reef Check · ecoregion · open biodiversity data · reproducibility