Spatial grids kept positive fitted heat slopes but changed their size in four coral samples from 2003 to 2019

We tested whether spatial cells can replace recorded sites in a coral heat model. We used 5,336 monthly Reef Check records in four Atlantic and Pacific samples from 2003 to 2019. We kept all records fixed and replaced site effects with country-by-cell effects at three grid widths and sixteen origins. All 192 grid slopes were positive. At 0.1 degrees, slopes fell by 10-14% in the early Pacific sample and 9-22% in the late Pacific sample. They fell by 0.2-5.2% in the early Atlantic sample and rose by 4.7-13.6% in the late Atlantic sample. A nested-projection identity accounts for these changes through the extra heat contrast admitted by pooling sites. These are finite sensitivity ranges, not confidence intervals. Selected reports, one monitoring system, fixed grid choices and unknown field errors limit wider inference. The study does not establish a causal heat effect or test air pollution. Code, aggregate results, source hashes, independent checks and a figure are included.

coral bleaching · thermal stress · spatial aggregation · grid origin · observation bias · fixed effects · sensitivity analysis