One- and two-month matching gave no clear slope change for coral bleaching and later cover in four samples in 2003-2018

We tested whether wider calendar-month matching changed the link between bleaching and next-year hard-coral cover change. Reef Check records in the Atlantic and Pacific in 2003-2018, with follow-up through 2019, gave 1,286 exact-month pairs, 2,286 pairs within one month and 2,704 pairs within two months. Four region and period models controlled current cover, heat, depth, year and month. Paired country resampling gave zero-containing 95% intervals for all eight wider-minus-exact slope contrasts. All twelve main slope intervals included zero after correction for four regional tests within each window. Wider windows added records but changed sites and elapsed time. They did not establish a stable bleaching effect or unbiased follow-up. Few country clusters, sparse high-bleaching records, unknown visits and survey error limit inference. The package gives code, cells, pairs, source hashes, all results and independent checks.

coral bleaching · marine heat · observation bias · record linkage · matching sensitivity · coral cover