A later-heat link to earlier coral change did not repeat across four Atlantic and Pacific samples in 1998-2018

We tested whether heat at a later annual survey was linked to an earlier coral-cover change. HeatCRD version 1 supplied 377 matched survey triples in four Atlantic and Pacific samples in 1998-2007 and 2009-2018. Models controlled two earlier heat fields, initial cover, year, month and source. The early Atlantic slope was +10.82 cover points per four future TSA degree-heating weeks, with a corrected interval of +2.64 to +18.99. It used 13 sites and only one high-heat future survey. Its Reef Check-only interval included zero. Two later slopes had intervals across zero. All early Pacific future heat values were zero, so that slope was not identified. The lead pattern did not repeat. Sparse heat, selected return sites and incomplete past exposure prevent causal or bias-free claims. The package gives code, aggregate results, numerical checks and an external-data retrieval recipe.

coral cover · marine heat · temporal negative control · observation bias · HeatCRD · regional replication