DR06 / Red snapper descending versus venting / v1.0.0 / 2026-10-21
Focused secondary research review, not original fishing records or a systematic review.

Sources: full main papers by Stallings et al.2023 (Fisheries Research264:106717), Curtis et al.2015 (Marine and Coastal Fisheries7:434-449), Runde et al.2021 (same journal13:478-495). Citations, URLs and locators in evidence.json; retained PDF hashes and rights in source-hashes.json. No separate supplements or fish-level raw files acquired. Three different endpoints/settings are not pooled.

Stallings: Table1 p.4 sample sizes; Results p.5 rounded return percentages, odds ratios, exact-test CIs/P values. Florida eastern Gulf, July-Oct2014/2016, captured30-50m, lip-hooked without hook injury. Initial random treatment followed by sequential assignment with equipment-availability overrides; two descent tools pooled by source. 40m arm n=3 not compared. Surface releases inside a bottomless cage. No-barotrauma untreated group not a randomized treatment control. Red grouper not included in our snapper dataset. Tag reporting, recapture effort and movement affect returns; an unreported tag is not death.
IMPORTANT: table n=202 does not yield16.5% with any integer return count at one decimal; n=175 similarly fails to yield15.5%. Published summaries retained separately and labeled. No back-calculated counts, binomial CI, new hypothesis test or claimed model reproduction. Descriptive percentage-point differences and ratios use published rounded percentages only; those ratios are not the published odds ratios. Published intervals/P values attributed, not independently reproduced or adjusted for multiple comparisons.

Curtis: Table1 p.440 all15 existing rows transcribed (winter vent absent, not zero). Southern Texas platforms, winter/summer2010 and spring2011. Bottom descent to30/50m. All method-depth-season counts remain separate. n=tagged-unknown; survival=survivors/n. Known deaths=surface+delayed. Unclassifiable telemetry excluded by source:25/111 overall including controls. Controls acclimated to surface in tanks, not freshly caught untreated fish. Winter vent absent; summer vent added during trial. Open-bottom cage protected surface releases during immediate observation. Long-term survivor here means >3days in source definition, not lifetime survival. Lower/upper all-tagged descriptive bounds assume every unknown died/survived: survivors/tagged and (survivors+unknown)/tagged. These are extreme missing-fate bounds, not confidence intervals, and do not repair nonrandom unknown fates, season imbalance or tagging effects. No pooled odds/hazard ratio or causal test.

Runde: ~37m North Carolina2019. 44 live-tagged,36 eligible for Kaplan-Meier baseline: successfully descended once and no deep hooking. Excludes3 self-descending,3 multiple-descended and2 deep-hooked fish. Reported mortality0.08 CI0.00-0.17 is source time-to-event estimate with censoring; not raw3/36 and not all44. Separate fishery simulation median0.13 (2.5-97.5percentiles0.10-0.17) uses hooking evidence plus assumptions of universal descending and no device failure. No venting control. Not Louisiana or Gulf mortality, not a direct comparative method effect. We do not rerun telemetry, Kaplan-Meier or Monte Carlo models.

Reproduce: save evidence.json and reproduce.mjs together; Node24: node reproduce.mjs. No packages/network required. Regenerates analysis.json and4CSVs. Plot script in repo site/scripts/plot-release-evidence.py uses matplotlib; chart typography outlined as SVG paths. No paper figures republished.

Maintenance: verify source corrections, rule changes and release guidance before production and every6months; original numeric extracts change only with a versioned audit. Independent fisheries/statistical review pending, especially source rate/n discrepancy. Current legal applicability must be checked for trip/jurisdiction: NOAA July30,2026 action page records expirationJan2026 and proposed continuation, not a blanket current-law statement.