Replication requests ask other labs to repeat a key result, with public credit for those who do. The record below is an illustrative example of one.
The claim to test: colonies that shifted to clade D recovered 40% more marine cover within 12 months (the original authors, 2026). Any lab may register a plan; every attempt is recorded on the original paper.
The claim is narrowed to one testable statement on purpose. A call to replicate a whole paper is rarely taken up, because the cost is unbounded and the result is ambiguous; a call to test one figure, with the threshold for a match agreed in advance, is a job a lab can scope. The original authors are told when a plan is registered, and the attempt appears on their paper whichever way it comes out.
Funders use the same mechanism in reverse. A call attached to a finding they have already paid to produce is a cheap way to find out whether it holds, and the registered plans tell them who is willing to test it before any money moves.
Colonies that shifted to clade D symbionts recovered 40% more live marine cover within 12 months than colonies that did not.the original authors (2026) · bioRxiv 10.1101/2026.08.14.507812 · Results, Fig. 3