### BIEA25 — Eagle Island

**Date:** 2015-03-17 (dive window 12:26–13:19, from frame EXIF — **the first dive of a new day**; BIEA22–BIEA24.5 all fall on the 16th, and BIEA26 follows later on the 17th from P3170439 onward)
**GPS:** 6.19072°S, 71.31839°E
**Frames reviewed:** 20 of 20 (100% of `BiEa25-finished`; 0 evicted)
**Coral cover:** 15.7% (formal transect surveys) · **% bleached: not recorded in the manifest** (the existing master-file entry's "17.7% of coral bleached" is unsourced — see flag 7)
**Reef type:** fringing, intermediate exposure · **Temperature:** 29.4 °C
**Live hard coral cover across methods** (n=20 photos): transect 15.7% · CoralNet points 21.0% · Coralscapes 17.0%. Deviation vs transect **+1.3 to +5.3 pp**. **Mean of 3: 17.9% ± 2.8.** **All 20 frames are CoralNet-annotated and all 20 matched to Coralscapes masks — the soundest CV basis of the five sites in this batch**, and the reference against which the thin ones (BIEA24, 11 of 15 frames; BIEA24.5, 5 of 23) must be read. Restricted to the **16 benthic frames (<40% water)** the photo methods rise to **25.6% and 20.7%**, i.e. **+9.9 and +5.0 pp above the transect**: the blue-water filter does **not** flip the sign here as it did at BIEA20 — **it widens an already-positive gap.** See the flags.
**Coral genera identified:** 6 scleractinian groups — ***Acropora*** in two guises (broad corymbose/plating thickets, P3170433–438; small digitate colonies and recruits, P3170411, P3170416); ***Porites***, smooth massive/domal and knobby-columnar olive forms (P3170390, P3170411); ***Pocillopora***-type small branching colonies (P3170411, P3170437, pale recruits in P3170415/P3170416); cerioid and meandroid brains compatible with ***Favites*/*Goniastrea*/*Platygyra*** (P3170411, P3170432); large partly-dead tabular plates compatible with tabular ***Acropora*** (P3170387, P3170388); thin horizontal discs on the algal ridge compatible with ***Turbinaria*** or ***Leptoseris*** (P3170397). Plus abundant crustose coralline algae, ***Halimeda*** (one annotated point, P3170397) and encrusting sponges. Growth-form calls from photographs, given as "compatible with"; **no species is asserted.**
**Representative photos:** P3170386.jpg, P3170411.jpg, P3170421.jpg, P3170434.jpg, P3170438.jpg

**BIEA25 is a site of two halves, and the split is almost exactly chronological.** For the first forty minutes (P3170386 at 12:26 through P3170432 at 13:08) the camera is over **a low-relief plain of pale consolidated carbonate, rubble and coarse sand** carrying small, well-spaced colonies — olive knobby and domal *Porites*-type heads, scattered small branching colonies, old tabular plates already given over to turf and coralline algae (P3170387, P3170388, P3170411, P3170432 all show the same ground). Then, from 13:15, the dive arrives at **broad corymbose *Acropora*-type thickets standing clear of white sand** — P3170433 through P3170438 — and frame-level live cover jumps from single figures to 60%. **The 15.7% transect number is the average of two worlds and describes neither: the last six frames, shot inside three minutes, carry roughly three times the coral of the first fourteen.**

**The opening sequence is a methodology record, not a benthic one.** P3170386 shows a diver kneeling on the pavement working a **black-and-white banded transect pole**, camera housing and strobe beside them; P3170387 a second diver with slate and housing; P3170388 **two divers with the pole strung between them**. **P3170389 is the fullest apparatus frame at the site** — five divers in open water, the banded pole, a camera tray, and at upper right a diver carrying **a distinct white rectangular open frame that reads as a photoquadrat**; **P3170437 has a diver in the background blue holding the same frame**. These six carry most of the site's blue-water load, which is why the un-filtered photo estimates sit closer to the transect than the benthic-only ones do.

**The macrofauna is dominated by one animal.** **A large octopus, compatible with *Octopus cyanea*,** occupies **three consecutive frames** — P3170417, P3170418, P3170421, shot between 13:01:53 and 13:02:19 — wedged against a maroon- and pink-encrusted ledge, mantle, papillae and red-rimmed eye filling the frame; P3170421 is the best. (The existing description calls this "four consecutive close-up shots" and then lists three; there are three.) Beyond it, **a blue sea star compatible with *Linckia laevigata*** appears on CCA-crusted pavement in P3170415 and on the sand under the thickets in P3170433 and P3170434 — the latter two five seconds apart and almost certainly one animal, so **two sightings, not three**. The thickets swarm with **blue-green chromis compatible with *Chromis viridis***, a **blue-and-yellow *Chrysiptera*-type damselfish** in the branches of P3170435 and a **humbug-type *Dascyllus*** in P3170437. **P3170438 is the site's best fish frame**: three **lined surgeonfish (*Acanthurus lineatus*)** and two **powder-blue surgeonfish (*Acanthurus leucosternon*)** over a broad mauve-pink corymbose table. **P3170397**, the deepest and most algal frame — an eroded ridge falling into blue under pink CCA, maroon crusts and yellow-green turf — carries a **stream of fusiliers (Caesionidae)** and a dark blotched mass that is probably an encrusting sponge but is not resolvable.

**Two frames matter for condition.** **P3170414** is a macro of a **cream-white knobby branching colony with prominent, well-separated circular corallites**, tissue apparently intact but pigment absent. It is the one frame where both methods flag bleaching in the same place — 2 human bleached points and **Coralscapes' highest bleached reading at BIEA25, 5.53% of image area** — but its corallite scale looks coarser than the *Porites* the annotation called it, so **genus and bleaching both want checking at full resolution**. **P3170416** is a downward macro of pavement carrying a **dense field of small dark boreholes or oscula** — bioerosion or a boring sponge, not resolvable — ringed by **pale coral recruits with blue-white polyp tips**, which are healthy and should not be read as bleaching.

<b>Assemblage composition.</b> Across 20 analysed frames the Coralscapes masks put live hard coral at 17.0% of image area, of which <i>Acropora</i> (non-tabular) is by far the largest component at 60.1%; the rest breaks down as other hard coral 19.8% and massive and meandroid 15.5% of live cover. The independent 30-point human annotation of the same frames (600 points) gives live hard coral as 21.0% of all points and apportions it <i>Acropora</i> 49.2%, <i>Porites</i> 40.5%, other hard coral 5.6%, <i>Pocillopora</i> 4.8% of live coral. Non-coral cover is dominated by crustose coralline algae 4.3%, turf 23.0% of all points. Structural loss registers as dead coral 1.4% of image area and rubble 5.2%. Bleaching stands at 2.5% of annotated points — **15 points in total** — concentrated in <i>Acropora</i> 1.7%, other hard coral 0.8% of points. Frame-to-frame live cover runs 0% to 60.0% by point count (0% to 72.5% by mask area). Conspicuous fauna and sessile invertebrates appear as sponges in 7 frames, giant clams in 3 frames, sea cucumbers in 3 frames, anemones in 3 frames, debris in 3 frames. No COTS were recorded by either method or seen in review.

**Flags:** (1) **THE MANIFEST GPS FIELD FOR BIEA25 IS MALFORMED.** `chagos_work_manifest.json` stores `gps` as `"6.19072°S, 71.31839°E · **In-situ temperature:** 29.4°C"` — a temperature string with markdown bold concatenated onto the coordinate field. Both values are correct and used above, but **any parser splitting `gps` on a comma returns a garbage longitude.** Check the other 108 Chagos entries for the same concatenation. (2) **THE BLUE-WATER TEST WIDENS THE GAP HERE; IT DOES NOT FLIP IT.** Four frames are ≥40% water (P3170387, P3170388, P3170389, P3170397). Dropping them moves CoralNet 21.0→**25.6%** and Coralscapes 17.0→**20.7%** against a 15.7% transect (**+9.9 and +5.0 pp**). **BIEA20's flag (1) found a sign flip; BIEA25 shows the other outcome — both methods already read high, and de-diluting makes them read higher.** (3) **THE LARGEST BLEACHING CONCENTRATION AT THIS SITE IS ALMOST CERTAINLY WRONG.** **P3170386 carries 5 of the site's 15 bleached-*Acropora* points (33%) on a frame the same annotator scored 0.0% live coral** — a methodology shot of bare pale pavement. **Nothing in it supports a bleaching call.** P3170387 and P3170388 (1 each) repeat the error: **7 of 15 bleached points fall in the first three frames, all methodology shots.** (4) **THE REMAINING BLEACHED POINTS SIT ON WHITE *ACROPORA* BRANCH TIPS** — P3170435 (3), P3170434, P3170436, P3170437 (1 each), all inside well-pigmented tan thickets. **Growth-tip pallor, not bleaching; the same trap will recur at every *Acropora*-dominated Chagos site.** After (3) and (4), **one defensible bleaching observation remains: P3170414.** (5) **ONE OCTOPUS BREAKS BOTH METHODS.** Coralscapes scores **38.65% live hard coral and 7.77% dead** (the site's highest) on **P3170421**, a close-up that is mostly octopus, giving it only 1.94% "other animal". The same animal drives the site's two highest human Unknown fractions, **50.0% at P3170417 and 36.7% at P3170421**. **Exclude P3170421 from per-frame Coralscapes cover and screen large-animal close-ups generally**; BIEA25's Unknown load is not an image-quality problem. (6) **P3170414 — GENUS DISAGREEMENT PLUS THE ONE REAL BLEACHING CANDIDATE.** Human annotation calls it *Porites* (14 of 30 points); the corallites are large, circular and well separated, coarser than *Porites* and closer to *Pocillopora*/*Stylophora*. **Both the identification and bleached-vs-naturally-pale need the full-resolution original.** (7) **THE MASTER-FILE ENTRY'S "17.7% of coral bleached" HAS NO SOURCE.** `pct_bleached` is `null`, and 17.7% is not reproducible from the CoralNet composition (15 bleached against 126 live-coral points = 10.6% of coral). **Do not carry it forward.** (8) **METHODOLOGY IMAGERY, SIX FRAMES** — P3170386–389, P3170437, plus equipment in P3170390. **P3170389 and P3170437 both appear to show a photoquadrat; verify at full resolution.** Second-densest after BIEA20's eight; belongs with BIEA20/P3150311, BIDI08/P3120298, BIEG04/P3100401, BIEA18/P3150341 and BIEA19/P3150369. (9) **A GREEN WOOLLY HAT ON A DIVER IN P3170386 AND P3170389** — the dive is 17 March and BIEA26 later the same day has a green hat on a coral head, so **a St Patrick's Day prop running across the day**. Trivial, but it links the two sites independently of the log; not equipment. (10) **THE COMPOSITION JSON REPORTS GIANT CLAMS IN 3 FRAMES, SEA CUCUMBERS IN 3, ANEMONES IN 3 — I CONFIRMED NONE VISUALLY.** Sponges (7) and debris (3) are plausible. **Audit those three counts before quoting them.** (11) No per-frame depth data; **no COTS, no disease and no predation scars** in any frame, and Coralscapes' `cots_frames` is empty.

**Benthic data:** [Download BIEA25 benthic survey data (CSV)](../benthic_data/BIEA25_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).