Macroalgal data selection raised coral cover means in Atlantic and Pacific heat records in 2009-2019

We tested how a macroalgal data rule changes the coral-cover sample in HeatCRD version 1. Four Atlantic and Pacific samples in 1998-2019 had 21,497 heat-eligible sample IDs. In 2009-2019, the rule raised mean coral cover by 5.57 and 3.11 percentage points. Early shifts were 0.12 and 0.11 points. Changes in source shares explain nearly all late shifts. Removing NOAA NCRMP reduced them to -0.052 and -0.003 points. No archive site ID had both reporting states within any tested sample. A within-site reporting comparison therefore lacks support. Late shifts persisted under site and site-year weights and a depth filter. These are fixed-archive descriptions, not causal heat effects or population estimates. Shared sources, selected sites, unknown visits and coarse exposure timing limit inference. The package gives code, aggregate evidence, exact source hashes, independent checks and an external-data retrieval script.

coral cover · macroalgae · marine heat · observation bias · missing data · source selection · HeatCRD