Artificial-reef evidence case v1.0 — 2026-10-21
Focused secondary case review, not a systematic review, field experiment, pooled meta-analysis or net production estimate. Two quantitative primary main papers; source supplements and raw records not acquired. Independent scientific review pending.

Streich et al.2017, DOI10.1080/19425120.2017.1347116: Corpus Christi Nearshore Reef, Texas, one complex plus one bare control,22m depth. Table1 p409 red-snapper arms transcribed and visually audited: vertical-line fish/set and trap fish/trap-hour kept separate, mean/SE/n retained. Vertical-line before about1year, after about2years; traps began only2months before construction. n means sets/traps, not independent reef sites. After sampling targets reef structure strata; temporal dependence and seasonal/monthly sampling limit simple independence assumptions. Authors’ model was hierarchical and used log-transformed CPUE. Our descriptive arithmetic (reef after-before)-(control after-before) is not a refit of that model. From rounded means: vertical3.424fish/set; trap(.455-.015)-(.002-.014)=.452fish/trap-hour. No new P or CI from n/SE; raw covariance/date structure missing. Zero baseline percentage undefined. Published SE whiskers are ±1SE, not95%CI. Zero sampled catch not fish absence. Juvenile condition/age progression and size-at-age regression support the authors’ nursery interpretation, but survival not directly measured and adult export/emigration vs mortality unresolved. Growth slope124.1mm/year is cross-sectional size-at-age, not tracked individual growth or a treatment difference.

Gardner et al.2022 DOI10.3389/fmars.2022.772292: Table3 p9 yearly/mean commercial landings fractions, Table4 p10 Louisiana state-specific extrapolation, visually audited. Initial habitat-linked allocation is itself inferred from VMS events and trip landings; not observed catch at each hook. Years2011,2012,2018,2019 only;1866linked trips424Louisiana.250m buffer to known habitat. Equal landing assignment across identified fishing events; unknown locations redistributed using identified habitat per trip. Reported means are preserved, not replaced with a mean-of-rounded-years or normalized Gulf percentages. Table3 mean natural22%,artificial25%,unknown53%; extrapolated natural54%,artificial46%. The reported mean artificial95%halfwidth14percentage points is copied, not recomputed, and excludes unquantifiable mapping/allocation assumptions. Louisiana Table4 reported mean artificial73%,annual66/74/74/75%; no present recreational catch probability. Four selected historical years are not a continuous annual record. Confidential vessel tracks not obtained or republished. No reef coordinates or spatial exploitation map reconstructed.

Czaja et al.2025 DOI10.1111/faf.12906 is a perspective/review used for research framing and design discussion. CC BY-NC: no adapted or redistributed table, rates, figures, text or dataset. This article’s own systematic search is not ours. No worldwide counts extracted or used as independent case trials.

Rights: quantitative primary papers CC BY; factual extraction attributed; original website prose/figures/calculations. Full source PDFs retained internally for audit, not public downloads. The code/data do not grant access, navigation clearance or permission to fish.
Reproduce: save evidence.json and reproduce.mjs together; Node24, node reproduce.mjs. Offline output: analysis.json plus red-snapper-cpue.csv,descriptive-changes.csv,commercial-landings.csv,evidence-map.csv. This reproduces our arithmetic/exports, not authors’ raw models. Repository charts: MPLCONFIGDIR=/tmp/ldsf-matplotlib /tmp/ldsf-noaa-plots/bin/python site/scripts/plot-reef-evidence.py (Matplotlib3.11.2); normal site builds use saved artifacts only.
Maintenance: recheck primary corrections and relevant new survival/movement/production evidence before production publication; change each endpoint claim separately. Current rules/access require responsible official sources. No new legal schedule asserted in this case.