Changes with a common survey had no clear lower covariance in four coral samples in 1998-2019

We tested whether coral-cover changes with a common survey had lower covariance than changes with no common survey. HeatCRD version 1 gave 338 four-visit windows in Atlantic and Pacific records from 1998-2019. We held site, source, depth and calendar month fixed. Visit gaps were one to five years. Each site had equal total weight. Common-survey minus separate-survey covariance estimates were +3.16, -1.82, +25.41 and +1.83 in squared cover points per year. All four 98.75% site bootstrap intervals included zero. Annual-only and adjusted estimates changed some signs. These selected records did not establish a repeated covariance deficit from a common survey. The result cannot separate survey error from real recovery, bounded cover or disturbance timing. Small site counts, source dominance and missing visits limit inference. The package gives matched records, source hashes, code, a figure and numerical checks.

coral cover · HeatCRD · observation bias · recovery · shared survey · covariance · null result