Pooled depth equivalence did not repeat across heat-exposed coral survey samples in 1998-2019

We tested the precision of paired coral bleaching contrasts under heat exposure. We used Reef Check population records from BCO-DMO 773466 version 2. Pairs had the same site and date, depths from 0 to 20 m, and a depth gap of at least 3 m. Only 96 of 1,665 pairs with valid heat data in 1998-2019 had CoRTAD TSA_DHW of at least four. Their mean deep minus shallow contrast was -0.072 percentage points. Its nominal 90% country-cluster interval was -2.934 to +2.790 points. This interval was within an example equivalence range of plus or minus five points. Neither time period met that range. Three of six Atlantic, Indian and Pacific period cells had fewer than three countries or no data. Positive Atlantic contrasts offset negative Pacific contrasts in the pooled sum. The result shows why a small pooled contrast cannot establish a common depth response during heat stress. The margin is illustrative and was chosen after related published results were read. Survey selection, coral composition, heat timing, and few spatial clusters limit inference. The data do not establish causal depth protection, species loss, or an air pollution effect. The package gives code, survey inputs, paired records, source hashes, precision tables, a figure, and an independent arithmetic check.

coral bleaching · heat stress · equivalence test · observation bias · regional replication · precision audit