### BIDI12 — GCB-West (Danger Island)
**Date:** 2015-03-13
**GPS:** 6.4534°S, 71.2337°E
**Frames reviewed:** 27 of 27 (100% of the site folder, `BiDi12-finished`)
**Coral cover:** 4.2% · **% bleached:** 6.8%
**Live hard coral cover across methods** (n=27 photos): transect 4.2% · CoralNet points 19.5% · Coralscapes 1.7%. Deviation vs transect -2.5 to **+15.3 pp**. **Mean of 3: 8.5% ± 9.6.** **The anomaly here is NOT that a photo method reads high — that is the expected archipelago pattern at low cover, and neighbouring BIDI13 shows it behaving normally (+4.5/+6.0 pp, the two photo methods agreeing within 1.5 pp). The anomaly is that the two photo methods disagree with EACH OTHER by nearly 18 pp, which no cover-dependent bias explains. The imagery says the CoralNet figure is the wrong one — see the assemblage note; do not use the mean for this site.**
**Coral genera identified:** 2 scleractinian — *Pocillopora* and *Acropora*, both only as small, scattered, low-relief colonies. A few encrusting faviids may be present but are not resolvable. **This near-absence is the site's defining benthic fact, not a review shortfall:** the dominant cover is fleshy green macroalgae compatible with *Caulerpa*, over eroded algal-turf pavement.
**Representative photos:** P3130330.jpg, P3130345.jpg, P3130302.jpg, P3130321.jpg, P3130333.jpg

**BIDI12 is not a coral reef in any functional sense, and that is the finding.** The dive begins over **flat, eroded, algal-turf pavement** — P3130298, P3130300, P3130301 and P3130302 show grey-white consolidated hardground with scattered rubble, pink coralline patches and essentially no living scleractinian framework. Then, from P3130330 onward, the substrate changes completely: **a continuous, dense meadow of fleshy green macroalgae — strap-bladed and compatible with *Caulerpa* — carpets the seafloor to the limit of visibility.** Eleven consecutive frames (P3130330, P3130331, P3130332, P3130334 through P3130340, P3130342 through P3130345) are wall-to-wall algal meadow, with P3130346 catching its margin where it abuts a clean sand plain scattered with purple coralline nodules and brittle-star arms. Live hard coral in this half of the dive is effectively zero. Only P3130322 and P3130323 hold a modest rubble ridge with small *Pocillopora* and *Acropora* colonies, orange anthias and pink CCA — the closest thing to reef structure at the site.

**The methods disagreement here is a data-quality flag, not an ecological result.** Coralscapes reads 1.7% live hard coral; the transect reads 4.2%; **CoralNet human point-counts read 19.5%.** Photo methods running above the transect at low cover is normal and expected here; what is not normal is the ~18 pp split between the two photo methods themselves. Across 27 frames reviewed at full resolution there is nowhere near 19.5% live coral — the honest visual estimate is low single digits, consistent with both the model and the transect. The most likely explanation is that 30-point annotation on frames dominated by algal turf, coralline-crusted pavement and *Caulerpa* meadow has assigned points to hard-coral classes that the imagery does not support. **This site should be pulled and re-checked before any archipelago-wide CoralNet summary is published**, and it is a caution against treating human point-counts as ground truth on low-relief, algal-dominated substrate. It is the mirror image of BIDI08, where the human and model agreed with each other and both fell far below a tabular-dominated transect. Neighbouring BIDI13 — same rim, next day, also low-cover and algal — is the control: there the human and model agree within 1.5 pp.

Bleaching cannot meaningfully be assessed from these frames — the 6.8% figure applies to a coral population that is barely present, and no bleached colony is identifiable in any frame. **The far more important condition signal is the macroalgal dominance itself.** Whether this reflects a phase shift following the older mortality seen at neighbouring BIDI10, a naturally algal-dominated lagoonal or back-reef habitat, or a seasonal *Caulerpa* bloom cannot be resolved from a single dive — but for a remote, unfished, unpolluted atoll it is striking and deserves Anderson's attention.

Faunally the site is dominated by one thing: **grey reef sharks compatible with *Carcharhinus amblyrhynchos*, cruising low over the algal meadow in nine separate frames** (P3130298, P3130299, P3130335, P3130336, P3130337, P3130339, P3130340, P3130342, P3130343, P3130345), with two animals in frame together in P3130340 and P3130345. **P3130299 and P3130300 also show a school of large parrotfishes** working the eroded pavement, one visibly expelling a sand plume — bioerosion photographed in the act, and directly relevant to the site's condition. A **green turtle** crosses the hazy blue in P3130303, a large jack compatible with *Caranx* patrols above the meadow in P3130338, and an emperor compatible with *Monotaxis grandoculis* forages over the rubble ridge in P3130322. **An octopus occupies a coralline-crusted crevice in P3130320 and P3130321**, the second octopus record of the expedition after BIEG05. **P3130333 shows staged dive gear — twin cylinders, a yellow mesh collecting bag and hardware resting on the pavement with a diver above** — a working-logistics frame rather than a survey-method one.

<b>Assemblage composition.</b> See the cover line and the paragraph above: the headline is that **the three methods span 1.7% to 19.5% on the same 27 frames**, and the imagery supports the low end. The 9.6 pp SD for this site encodes an apparent annotation error, not method bias and not precision, so **the mean-of-3 value (8.5%) is not defensible for BIDI12 and should be suppressed or footnoted in any table**. Ecologically, the honest description of this site's benthos is: fleshy green macroalgal meadow > algal-turf pavement > coralline-crusted rubble >> live hard coral.

**Flags:** (1) **DATA-QUALITY — CoralNet 19.5% is not supported by any of the 27 frames, and the tell is the ~18 pp split from Coralscapes, not the positive gap vs the transect** (a positive gap at low cover is normal; see BIDI13). Re-check this site's point annotations before publishing any archipelago CoralNet summary; suppress or footnote the mean-of-3. (2) **MACROALGAL DOMINANCE** — a continuous *Caulerpa*-type meadow covers roughly half the dive. Phase shift vs natural habitat vs seasonal bloom is unresolved and needs Anderson. (3) **Grey reef sharks in nine frames**, two together in P3130340/P3130345 — the densest shark record of the trip. (4) **Parrotfish bioerosion photographed** (P3130299, P3130300), including a visible sand plume. (5) Octopus in P3130320/P3130321 — second of the expedition. (6) Green turtle in P3130303, too distant to separate *Chelonia* from *Eretmochelys*. (7) P3130333 is staged dive gear, not survey instrumentation — distinct from the BIDI08 deployed logger. (8) The green alga is called as *Caulerpa*-compatible on blade form alone; *Halimeda* is excluded by the absence of calcified segments, but a definitive call needs a specimen. (9) No per-frame depth data.

**Benthic data:** [Download BIDI12 benthic survey data (CSV)](../benthic_data/BIDI12_benthic.csv) — scleractinian cover by genus with full names, plus substrate and algal categories (bare substrate, CCA, turf, macroalgae, invertebrates), from the LOF BIOT-1 survey (Mayfield & Dempsey 2026, *Oceans*, supplemental file S1).