Priority recovery halved the calculated depth bound width with 149 reef values under equal cost in 2003-2019

We tested how to recover missing bleaching values for paired shallow-reef surveys. We used Reef Check records in BCO-DMO 773466 version 2. The frame has 1,965 site-date pairs at 1,127 sites, with depths of 0-30 m and gaps of at least 3 m. Each site has equal weight. There are 546 missing sample percentages. Their unrestricted support gives a 26.271-point bound width. Under equal cost and error-free recovery, ranking records by their absolute estimator coefficient gives the smallest remaining width for every fixed budget. Halving the width needs 149 priority values versus 273 for the random expected width. A width of at most one point needs 472 recovered values under this rule, versus 526 for the expected width under uniform random recovery. The method ranks source records; it does not recover any values. Unknown source availability, unequal retrieval costs, survey selection, and coral composition limit its use. The package gives source hashes, derived records, priority lists, exact width curves, code, and an independent arithmetic check.

coral bleaching · missing outcomes · source recovery · depth refuge · observation bias · marine heat stress