### BIEA52 — Eagle Island (island label disputed — see Flag 2)
**CRD page:** [Three Brothers / Central Chagos Coral Surveys (25–28 Mar 2015)](https://coralreefdiagnostics.com/three-brotherscentral-chagos-20150325-to-0328)

**25 Mar 2015**
**GPS:** 6.19878°S, 71.49310°E
**Setting:** protected (leeward) submerged bank reef, inside the lagoon, 20.0 m
**Coral cover:** 34.1% (14 transects across 4 depth strata, D2–D5; **algal cover 24.1%**) · **published image bleaching:** 0.0% (`Mean(%bleaching)` column — the lowest value of any Chagos site with a figure; see Flag 5)
**Live hard coral cover across methods:** formal transects 34.1% only. **BIEA52 has no row in `chagos_multimethod_cover_by_site.csv`** — there is no CoralNet-points or Coralscapes figure to place beside the transect value, so the usual three-method comparison cannot be made here. A photo-derived cover of 38.13% over 22 frames was reported by an earlier reviewer from the Mayfield & Dempsey supplemental "formal vs. informal comparison" sheet; that figure is carried below but could not be re-verified in this pass. Transect and image products have different sampling footprints and denominators and must not be averaged.
**Temperature:** 29.9 °C · **Salinity:** 34.5
**Photo review:** all 26 of 26 frames examined, in the folder `BiTb52-meter site-finished` (the folder prefix differs from the site code — see Flags 1 and 2). All 26 carry a 25 Mar filename stamp (P325xxxx); this is a single dive, not a multi-date set.
**Representative photos:** P3250416.jpg, P3250413.jpg, P3250423.jpg, P3250420.jpg, P3250412.jpg

**Correction, 2026-08-26 — the photographs of this site are NOT missing.** The previous version of this shard stated that "no habitat photo folder exists anywhere in the Chagos tree and none of the frames can be located," and escalated the loss as the most consequential finding of its batch. **All 26 frames are on disk**, in `edited site photos for website/BiTb52-meter site-finished/` — the folder is filed under the phantom prefix `BiTb52` rather than `BiEa52`, the same misfiling that had already been found and corrected in the site's benthic workbook. The frame stems are exactly the P3250386–P3250424 sequence the analysis record listed. Everything the earlier reviewer inferred from the numbers is preserved below; the reef description that shard said could not be written now can be, and one substantial field observation was waiting in the imagery.

BIEA52 is a submerged bank reef inside the lagoon, surveyed on 25 Mar 2015 under protected, leeward conditions at 20 m. **Coral cover of 34.1% ranks 67th of 109 Chagos sites and algal cover of 24.1% ranks 102nd of 109** — low-to-middling cover, but among the eight least algae-dominated reefs in the archipelago. Neither of the usual algal categories is running away with this reef: crustose coralline algae hold 8.4% of the benthos (91st of 109) and fleshy turf just 3.0% (**105th of 109, near the archipelago minimum**), with sediment-laden turf 7.3%, erect algae 2.5%, macroalgae 1.6% and cyanobacteria 1.2% making up the rest. Bare substrate is 13.3%.

**What is unusual here is not the coral or the algae but the invertebrates. Non-coral invertebrates (PITS) hold 28.5% of the benthos — larger than the entire algal budget, and the biggest single non-coral, non-algal share at the site.** The published table does not resolve that category to taxon, and the earlier shard flagged it as unexplained. The photographs offer a candidate: **large, flat, dark maroon-to-purple encrusting mats spread across the pavement in P3250389, P3250411, P3250417, P3250418 and P3250424**, sometimes covering several square metres of a single frame and reading as a smooth, living, non-calcified veneer rather than as rock or algal turf. These are consistent with encrusting sponges, which a point-annotation protocol would score as non-coral invertebrates. They could equally be a cyanobacterial or encrusting-algal mat, and no frame resolves the surface texture well enough to settle it, so this is offered as a hypothesis worth testing rather than an identification. Coral, algae, bare substrate and invertebrates together account for exactly 100.0% of the benthos, so nothing is unallocated.

The assemblage is genus-diverse and unusually evenly divided. **25 non-zero genera were recorded, 51st of 109 Chagos sites**, and the 25 genus percentages sum to 34.12%, an exact match to the published cover figure. All values below are **percent of total benthos, not percent of coral**. ***Porites* leads at 12.38% of the benthos — 36% of all coral recorded** — followed by *Acropora* 5.85% (17%), *Pocillopora* 3.98% (12%), *Favia* 2.45%, *Pachyseris* 1.50%, *Leptastrea* 1.33%, *Goniastrea* 1.06%, *Symphyllia* 0.82%, *Montipora* 0.77%, *Montastrea* 0.70%, *Stylophora* 0.62%, *Favites* 0.61%, *Platygyra* 0.56%, *Hydnophora* 0.47%, *Echinopora* 0.39% and *Millepora* 0.18%, then a nine-genus tail below 0.15% — *Cantharellus*, *Fungia*, *Galaxea*, *Pavona*, *Cyphastrea*, *Astreopora*, *Acanthastrea*, *Gardineroseris* and *Seriatopora*. **No genus exceeds 37% of the live coral present**, which makes this one of the more balanced *Porites*/*Acropora*/*Pocillopora* mixes in the dataset — a real site strength distinct from its mid-pack cover.

**The depth profile is the dominant structure at BIEA52, and it is a three-fold difference between the shallowest stratum and the mid-depths.** **D2 carries just 13.7% coral across eight genera** — the poorest band on the reef — against 25.3% algae and, remarkably, **55.3% non-coral invertebrates**, by far the highest invertebrate share of any stratum here; *Acropora* leads at 4.67%, then *Pocillopora* 2.67% and *Porites* 2.33%. **D3 jumps to 46.2% coral across 17 genera**, with *Porites* 16.00%, *Pocillopora* 9.17%, *Acropora* 7.33% and *Favia* 5.67% — this is the site's best band and the one that carries the headline number. **D4 holds 42.7% coral and is the most genus-diverse stratum on the reef, 24 genera**, led by *Porites* 15.20% and *Acropora* 11.40% with *Montastrea*, *Hydnophora*, *Favites* and a long tail all present. **D5 falls back to 34.0% coral across only seven genera on 29.0% bare substrate**, and the composition changes character: *Porites* 16.00%, then ***Pachyseris* 6.00% — 18% of the coral on that transect, where at site level *Pachyseris* is only 1.5%** — plus *Goniastrea*, *Leptastrea* and *Symphyllia* at 3.00% each and *Acropora* absent from the leading ranks entirely. The site-level genus list therefore describes no single stratum: it averages an invertebrate-dominated, coral-poor shallow band, two coral-abundant and genus-diverse mid-depth bands, and a deep *Porites*/*Pachyseris* transect with *Acropora* gone.

**The headline number is moderately weighting-sensitive.** Site cover is the unweighted mean of the four per-stratum means, 34.12%; the same 14 transects (D2 n = 3, D3 n = 6, D4 n = 4, **D5 n = 1**) re-computed as a transect-weighted mean give **37.33%, a 3.2 pp difference** — because the two best-sampled strata are the two coral-abundant mid-depth bands, while the poorest band (D2) and the thin deep band each get a full quarter of the published figure. **D5 rests on a single transect** and is the stratum carrying the *Pachyseris* signal, so the deep end of the depth story above is unreplicated. The earlier shard reported the same 37.33% transect mean and attributed the gap to a mixture of 500-point and 100-point transects; the stratum-weighting arithmetic accounts for it directly.

The 26 photographs show a low-to-moderate-relief bank reef of coral patches and thickets standing on broad expanses of pale pavement and sand — exactly the patchy structure a 34% site figure with 13.3% bare substrate implies. **P3250413 and P3250414 are the highest-structure frames**: large golden-brown staghorn *Acropora* thickets a metre or more across, standing clear of the substrate with shaded understoreys, backed by further thickets receding into the blue. **P3250419** is the highest-cover view, a near-continuous branching *Acropora* garden with damselfish, a Moorish idol and a powder-blue surgeonfish working through it. Tabular and corymbose *Acropora* recur throughout (P3250391, P3250393, P3250417, P3250421, P3250422), massive and lobate colonies consistent with *Porites* anchor the foreground of many frames, and large foliose plates — a bright yellow-green whorled colony in **P3250409**, a dark brown vase-shaped colony in **P3250424** — punctuate the pavement. Against those, **P3250388, P3250400 and P3250412 are the open frames**, wide pale plains of sand and coralline-nodule pavement carrying only scattered small colonies, and they are the photographic face of the 13.7%-cover shallow stratum. Pink and mauve crustose coralline algae encrust the pavement throughout; olive-green turf appears locally in P3250410 but never dominates a frame, consistent with turf at 3.0% and 105th of 109.

**One frame is a genuine recovery observation.** **P3250423 shows a large dead tabular colony — its upper surface now bare, pink-CCA-encrusted skeleton — carrying a dozen or more small live corymbose and digitate *Acropora* colonies growing directly on top of it.** The old table is functioning as a settlement platform, and the recruits are well established rather than newly settled. It is a single frame and not a quantitative recruitment measure, but it is the clearest evidence in this set that the reef is regenerating on its own dead framework.

**The survey itself is visible in the imagery, and the folder name "meter site" is explained by it.** **P3250386** shows a diver working over the substrate with a yellow-framed instrument and a grey cylindrical housing laid on the reef, and **P3250412** shows a deployment on open pavement — a dark cylindrical logger housing, a yellow instrument or bag, a black frame, and a vertical line running up out of frame toward the surface. These are field documentation of an instrument deployment, not a data concern.

**Condition: the photographic record and the published bleaching figure do not fully agree, and this is the most important caveat at the site.** The published image bleaching value is **0.0%, which the central rank file places 98th of 98 — the lowest bleaching figure of any Chagos site with a value**, with both CoralNet training runs agreeing at 0.0. Most of the imagery supports a healthy reef: colonies are normally pigmented brown, olive, gold and mauve, and the pale ground in the open frames is plainly sand, coralline pavement and old eroded skeleton rather than bleached tissue. **But P3250420 contains a small, entirely bone-white branching colony** sitting on the pavement among fully pigmented neighbours under identical lighting, with a second, smaller whitish colony beside it; the branch structure is intact rather than eroded, which makes recent bleaching or very recent tissue loss the likelier reading over old dead skeleton. **A published figure of exactly 0.0% and a visibly white colony in the reviewed frames are not strictly contradictory** — a point-annotation product returns 0.0% whenever no annotation point happens to land on bleached tissue, and a single small colony can easily fall between points — but the site should not be described as having had *no* bleaching on 25 Mar 2015. **A separate condition observation: P3250415 shows a large golden staghorn *Acropora* colony whose entire centre is dark brown and apparently dead while its periphery is brightly pigmented and alive** — classic partial mortality with die-back from the colony interior, the sort of thing that follows sediment accumulation, overgrowth or an older stress event, and distinct from acute bleaching.

**FAUNA — CROWN-OF-THORNS RECORD. P3250416 shows a crown-of-thorns starfish (*Acanthaster planci*) on the reef, alongside a large stainless-steel bone-cutting forceps laid on the substrate beside it.** The animal is small — roughly the span of the instrument's jaws — with the diagnostic dense covering of long spines over a mottled olive-grey disc and short arms, on coralline-encrusted pavement next to an orange encrusting sponge. The presence of the tool in the same frame indicates a deliberate collection, removal or measurement operation rather than a chance encounter. **This is a positive COTS record for BIEA52 on 25 Mar 2015 and belongs in the running COTS/bleaching/disease log.** It is one individual in one frame; no aggregation, and no fresh white feeding scar was visible on any colony in the surrounding frames, so this does not evidence an outbreak. Other fauna is modest and typical: blue-green chromis in numbers over the coral patches (P3250408), damselfish sheltering in the staghorn thickets, powder-blue surgeonfish (P3250418, P3250419), butterflyfish, a Moorish idol, and small black surgeonfish throughout.

<b>Assemblage composition.</b> Formal transects give 34.1% coral against 24.1% algae, 13.3% bare substrate and an exceptional 28.5% non-coral invertebrates, across 25 non-zero genera in which no taxon exceeds 37% of the live coral — *Porites* 12.38%, *Acropora* 5.85%, *Pocillopora* 3.98% of the benthos. The four-stratum depth profile runs from an invertebrate-dominated, 13.7%-cover, eight-genus shallow band through two coral-abundant mid-depth bands (46.2% and 42.7%, 17 and 24 genera) to a single deep transect where cover falls to 34.0%, *Acropora* drops out and *Pachyseris* takes 18% of the coral. All 26 photographs corroborate that patchy structure — large staghorn *Acropora* thickets and foliose plates standing on wide pale pavement — and add four things the transects do not carry: **a crown-of-thorns starfish (P3250416)**, **a small bleached colony at a site published as 0.0% bleached (P3250420)**, **partial die-back in a large staghorn colony (P3250415)**, and **a dead table colonised by a dozen live *Acropora* recruits (P3250423)**.

**Flags:** (1) **THE PHOTOGRAPHS ARE NOT MISSING — the earlier shard's central claim is withdrawn.** All 26 frames are in `edited site photos for website/BiTb52-meter site-finished/`, filed under the phantom prefix `BiTb52` — the same misfiling already corrected in the site's benthic workbook, whose internal header also read BITB52. Any downstream text, popup or table describing BIEA52 as photographically lost needs the same correction, and the folder should be renamed to `BiEa52-...` only after Flag 2 is resolved, since the prefix may not be an error at all. (2) **THE ISLAND LABEL IS THE REAL PROBLEM, AND IT NEEDS ANDERSON'S JUDGEMENT.** Four independent pieces of evidence say this site is not at Eagle Island. **(a) Longitude:** BIEA52 sits at 71.4931°E, while the nine other Eagle Island sites span 71.276–71.404°E and the Three Brothers cluster spans 71.498–71.593°E — BIEA52 is east of every Eagle Island site and inside the Three Brothers range. **(b) Date:** Eagle Island work ran 14–17 Mar 2015 and ended at BIEA26 on 17 Mar; BIEA52 is dated 25 Mar, by which time the survey had been working Three Brothers and Central Chagos for a week, and the day's other two dives (BICC50, BICC51) are both Central Chagos. **(c) The photo folder prefix is `BiTb52`.** **(d) `chagos_site_to_page.json` maps BIEA52 to the Three Brothers / Central Chagos CRD page, not to an Eagle Island page.** If the island label is wrong, then any "highest-cover Eagle Island site" statement — including the one in the previous version of this shard — is void, and BIEA52's neighbours for comparison are the BITB/BICC sites, not the BIEA ones. **The archipelago-wide rank of 67th of 109 for cover is unaffected either way.** (3) **The frame-count discrepancy now reconciles exactly, which supports the earlier reviewer's site-split concern.** The folder holds **26** frames; the analysis record scored **22**. The earlier shard recorded four of this folder's stems — P3250389, P3250408, P3250409 and P3250419 — as assigned to **BICC50** in the source sheet, and **26 − 4 = 22**. That is an exact arithmetic match, and it means the folder on disk and the scored set are not the same set of images: four frames sitting in the BIEA52 folder are attributed to a different site in a different region on the same continuous camera counter (P3250386–P3250431, split across the two sites with each site's runs interleaving the other's). **Treat the reported photo-derived 38.13% cover and 0.0% bleaching as provisional until the split is confirmed**, and note that the 26-frame review above necessarily includes the four disputed frames. (4) **BIEA52 has no multi-method row.** `chagos_multimethod_cover_by_site.csv` contains no entry for this site, so no CoralNet-points or Coralscapes figure exists to compare against the 34.1% transect value. The 38.13% photo-derived figure quoted above comes from the earlier reviewer's reading of the supplemental "formal vs. informal comparison" sheet and **was not re-verified in this pass**. Do not average it with the transect figure. (5) **The published bleaching figure of 0.0% is the lowest of 98 Chagos sites, and one reviewed frame shows a bleached colony.** P3250420 contains a small, intact, bone-white branching colony among pigmented neighbours. A point-count product legitimately returns 0.0% when no point lands on bleached tissue, so this is not necessarily an error — but BIEA52 should not be described as having had zero bleaching. **Worth a targeted re-check of the scored frames.** (6) **CROWN-OF-THORNS STARFISH RECORDED, P3250416** — one individual on coralline pavement with a stainless-steel bone-cutting forceps beside it, indicating a deliberate collection or removal. **Add to the running COTS/bleaching/disease log for 25 Mar 2015.** Single animal; no aggregation and no fresh feeding scars observed in the surrounding frames, so no outbreak is implied. (7) **Partial colony mortality, P3250415** — a large staghorn *Acropora* with a dead brown centre and a live golden periphery; die-back from the colony interior, not acute bleaching. (8) **Coral recruitment on dead framework, P3250423** — a dead tabular colony carrying a dozen or more established live *Acropora* recruits. A positive recovery observation, qualitative only. (9) **PITS = 28.5% is exceptionally high and still unresolved to taxon**, rising to **55.3% at D2**. The large dark maroon encrusting mats visible in P3250389, P3250411, P3250417, P3250418 and P3250424 are a plausible candidate (encrusting sponge), but the frames cannot distinguish sponge from a cyanobacterial or encrusting-algal mat. Offered as a hypothesis to test, not an identification. (10) **A large brown vase/funnel-shaped foliose colony in P3250424** is consistent with *Turbinaria*, which does **not** appear in this site's transect genus list; it is equally consistent with a foliose *Montipora* or *Pachyseris*, both of which do. Recorded as a growth-form observation only. (11) **D5 rests on a single transect** and is the stratum carrying the *Pachyseris* signal; the site figure moves from 34.12% (unweighted stratum mean, the published value) to 37.33% (transect-weighted) — a 3.2 pp swing. (12) **Integrity checks pass:** the 25 genus values sum to 34.12% against a published 34.12%, and the algal categories sum to 24.06% against a published 24.05%. (13) **The published bleaching figure is a COMPUTER-VISION product, not an in-situ observation.** It comes from CoralNet annotation of the informal survey images (Mayfield & Dempsey 2026, *Oceans*). Per-image human-vs-model agreement is substantially weaker for bleaching than for cover: r = 0.40 for bleached against r = 0.72 for live hard coral (`CHAGOS_MULTIMETHOD_README.md`). Note also that the `Mean(%bleaching)` column is **not** a mean — it equals the first CoralNet training run, while the dive-history map publishes the second; at BIEA52 both runs read 0.0, so they do not diverge, but the column carries at least one known bad value elsewhere (BITB58 listed at 22.22% where both runs give ~0.77%), so any rank taken off it is approximate. (14) **NON-CANONICAL BLEACHING RANK — added 2026-08-26.** Any "of 98" bleaching rank above was computed off the site table's `Mean(%bleaching)` column, which equals the **1st** CoralNet training run. Anderson's decision of 2026-08-26 (HOW_TO rule 4e) makes **`% bleaching (2nd training)` canonical**, The file carries TWO pools for it, and they are not interchangeable: a **110-row pool** ranking every dive (revisit dives included, so BIPB70B/72B/69C compete) and a **99-site pool** ranking one value per site. The rank quoted below is on the 110-row dive pool. For this site the canonical value is **0.00%, rank 109 of 110** (1st training: 0.00%, rank 109 of 110). The two runs agree to <0.05 pp on 100 of 110 rows and diverge only on the bleaching-onset dives, so for most sites the distinction changes the rank but not the number.

**Benthic data:** [Download BIEA52 benthic survey data (CSV)](../benthic_data/BIEA52_benthic.csv) — scleractinian cover by genus with full names, plus substrate and algal categories (bare substrate, CCA, turf, sediment-laden turf, macroalgae, erect algae, cyanobacteria, non-coral invertebrates), from the LOF BIOT-1 survey (Mayfield & Dempsey 2026, *Oceans*, supplemental file S1). **Note, carried forward from the previous version of this shard:** this site's individual workbook was originally misfiled under the phantom code `BITB52` — its internal header also read BITB52 — while its lat/lon, date, temperature, salinity and 34.12% cover match BIEA52 exactly. `BITB52` does not appear in the published 109-site table. The file was renamed, the header corrected, and the stale BITB52 copy removed from S3. The photo folder carries the same `BiTb52` prefix and has **not** been renamed, pending Flag 2.