### BITB47 — Three Brothers
**CRD page:** [Three Brothers/Central Chagos Coral Surveys BITB38-BITB49 (Mar 2015)](https://coralreefdiagnostics.com/three-brotherscentral-chagos-20150321-to-0324)

**24 Mar 2015**
**GPS:** 6.02659°S, 71.61644°E
**Setting:** protected lagoonal submerged patch reef, 29 m (inside the lagoon; `exposure-original` "leeward", `exposure-revised` "protected")
**Coral cover:** 53.7% (17 formal transects, D2–D4; **algal cover 35.5%**) · **published image bleaching (`Mean(%bleaching)` column):** 5.56%
**Live hard coral cover across methods:** formal transects 53.7% · CoralNet points 30.0% · Coralscapes masks 25.7% (**2 frames only** — see Flags). These products have different sampling footprints and denominators and should not be averaged; at this site the CV products rest on two photographs and are illustrative, not representative.
**Temperature:** 29.8 °C · **Salinity:** 34.5
**Photo review:** both of the 2 jpg stills examined — the entire photographic record for the site. The folder is named "few site shots (camera fogging)"; the dive's imagery was curtailed by a flooding/fogging camera housing, and both surviving frames (timestamps 09:20:55 and 09:21:59, 64 seconds apart) show visible blue haze.
**Representative photos:** P3240477.jpg, P3240478.jpg

BITB47 is one of the best reefs in this dataset. A protected lagoonal patch reef at 29 m, it carries **53.7% coral against 35.5% algae — 20th of 109 Chagos sites for coral cover, inside the top fifth of the archipelago** — and it does so with **29 nonzero genera, 19th of 109 for genus count**, so it is coral-abundant and genus-rich at the same time. The remaining benthos is remarkably tidy: only 6.6% bare substrate and 4.1% non-coral invertebrates, the lowest invertebrate share of the four sites in this rebuild. The algal budget is middling and mixed — crustose coralline algae 14.0% of the benthos (54th of 109), turf 7.4% (75th of 109, lower quartile), macroalgae 5.7%, sediment-laden turf 4.5%, erect algae 3.7% and cyanobacteria 0.19%. Nothing about the algal community is unusual; the coral is what distinguishes this site.

All genus values below are **percent of total benthos, not percent of coral**, and the 29 genus percentages sum to 53.748% — an exact match to the published cover figure. ***Acropora* leads at 15.95% of the benthos (30% of all coral present)**, with *Porites* close behind at 12.73% (24%), then *Pocillopora* 5.04%, *Favia* 4.27%, *Montipora* 2.85% and *Stylophora* 2.48% — these six genera hold 80% of the coral. Behind them: *Echinopora* 1.71%, *Pavona* 1.67%, *Platygyra* 1.41%, *Favites* 1.16%, *Leptastrea* 0.80%, *Astreopora* 0.49%, *Montastrea* 0.48%, *Goniastrea* 0.44%, ***Heliopora* 0.43% — the blue coral, and a genuine record here**, *Gardineroseris* 0.43%, *Millepora* 0.39% and *Cyphastrea* 0.34%, then a tail of 11 genera under 0.25% each (*Acanthastrea*, *Fungia*, *Hydnophora*, *Cantharellus*, *Alveopora*, *Ctenella*, *Oxypora*, *Galaxea*, *Leptoseris*, *Goniopora*, *Seriatopora*). Diversity indices computed across the 29 codes give Shannon H′ = 2.23, evenness J′ = 0.66 and an effective diversity of 9.31 genera — the most even assemblage of the four sites rebuilt here.

**Cover rises monotonically with depth while algae falls just as steadily, and the dominance handover between *Acropora* and *Porites* happens at the deepest stratum.** D2 (n = 6 transects) gives 48.2% coral against 43.8% algae, and is unambiguously *Acropora* ground — *Acropora* 19.57%, which is 41% of all coral on the stratum, ahead of *Porites* 7.13%, *Pocillopora* 6.07%, *Montipora* 3.53%, *Favia* 3.37% and *Stylophora* 3.27%, across 19 genera. D3 (n = 7) climbs to 55.14% coral against 38.17% algae and is **the diversity peak of the profile at 29 genera — every genus recorded at the site is present on this one stratum**; *Acropora* still leads at 14.69% but its share of the coral drops to 27%, and *Favia* doubles to 7.46% to take second place ahead of *Porites* 5.97%, with *Pocillopora* 5.20%, *Stylophora* 4.11% and *Montipora* 3.37% behind. **The blue coral *Heliopora* (1.29%) occurs at D3 and nowhere else**, and it contributes the site's entire 0.43%. D4 (n = 4) reaches **57.9% coral against just 24.4% algae — the highest cover and lowest algal cover of the profile — and the leadership flips: *Porites* jumps to 25.10%, 43% of all coral on that stratum and more than four times its D3 value**, pushing *Acropora* (13.60%) into second for the first time. *Pavona* also peaks here at 3.95%, but genus count falls back to 20. Bare substrate climbs with depth (2.7% → 4.4% → 12.9%) and non-coral invertebrates stay low throughout (5.3% → 2.3% → 4.8%). **The site-level figures of 15.95% *Acropora* and 12.73% *Porites* therefore describe no single transect**: they average an *Acropora*-dominated upper reef, a maximally diverse and *Favia*-rich middle, and a deep stratum that is *Porites* ground.

The two surviving photographs are consistent with that reading but cannot test it. **P3240477 shows the heterogeneous, high-cover face of the reef: massive and submassive domes consistent with *Porites*, olive-green digitate and corymbose colonies consistent with *Acropora*, pale grey blunt-branched colonies, meandroid brain colonies consistent with *Platygyra*, encrusting and plating forms, and small tabular colonies — with only narrow sand patches between them**, a frame that would sit comfortably at the 50–60% end of the transect range. Damselfish and dark surgeonfish work the water column above, and the blue haze characteristic of the fogged housing is visible across the mid-field. **P3240478 is dominated by a large yellow-gold tabular *Acropora* filling the upper left, and it carries the site's one condition observation: a sharply bounded pale, bare crescent on the plate's inner surface, already colonized by pink-mauve crustose coralline algae** — the CCA colonization dates the lesion, which is far more consistent with older partial mortality or breakage than with recent bleaching. Below it the foreground is a dense field of cream-grey blunt-branched colonies, **several with distinctly pale to white branch tips**; purple-brown foliose plates, a bright red encrusting sponge and green filamentous/*Halimeda*-type algal tufts fill the mid-ground.

**Condition cannot be assessed properly at this site, and the shard should not pretend otherwise.** `Mean(%bleaching)` is **5.56%, 65th of the 98 Chagos sites carrying a value — below the archipelago midpoint for bleaching severity** — and the two CoralNet training runs agree here (5.56 vs 5.5556), so this is not one of the ten sites where the runs diverge. The bleaching estimate depends entirely on which denominator is used: the published product gives 5.56%; the 60-point human audit across both frames found 18 live-coral points and 1 bleached point, so bleaching is **1.67% of all 60 points or 5.26% of the 19 coral points**, and the same audit gives 30.00% live coral; the Coralscapes masks give 25.66% live coral and 0.02% bleached area. **Two frames, sixty points, one bleached point — that is the entire image-based evidence for this site.** What the frames do show is real but limited: pale-to-white branch tips on several branching colonies in P3240478, and the older algal-colonized partial-mortality lesion on the large table. Neither frame shows a broad bleaching field, and the great majority of visible colonies carry normal olive, gold, tan and purple pigmentation. **No COTS individuals or feeding scars, no disease front, no *Drupella* aggregation and no clear corallivore scarring are visible in either frame** — a genuine but very weakly powered absence-of-evidence finding.

<b>Assemblage composition.</b> Formal transects give 53.7% coral against 35.5% algae across **29 nonzero genera — 20th of 109 Chagos sites for cover and 19th of 109 for genus count**, an unusual combination of abundance and breadth — with *Acropora* (15.95% of benthos) and *Porites* (12.73%) together holding 53% of all coral. Cover climbs steadily with depth (48.2 → 55.1 → 57.9%) as algae falls (43.8 → 38.2 → 24.4%), and the leading genus hands over from *Acropora* at D2–D3 to *Porites* at D4, where *Porites* reaches 25.10%; *Heliopora* occurs at D3 only. The photographic record is two frames curtailed by camera fogging: one shows a dense mixed-growth-form, high-cover mosaic; the other a large gold tabular *Acropora* with an older algal-colonized partial-mortality lesion above a field of branching colonies with pale tips. Cover, composition and condition at this site rest on the transects, not the imagery.

**Flags:** (1) **The CV cover products at this site rest on TWO photographs.** `n_photos=2` in `chagos_multimethod_cover_by_site.csv`, and the resulting spread across methods is the largest of any site in this rebuild — transects 53.7%, CoralNet points 30.0%, Coralscapes masks 25.7%, a deviation range of −28.0 to −23.7 pp and an across-method SD of 15.1. **Those CV figures should be treated as illustrative of two frames, not as an estimate of the site**, and should never be averaged with the transect value or with each other. (2) **The photographic record was curtailed by camera fogging** (the folder is literally named "few site shots (camera fogging)"), both surviving frames were taken 64 seconds apart at what is effectively one spot, and both carry visible blue haze. Photographic condition assessment at this site is close to unpowered, and the imagery cannot test the depth structure at all. (3) **The two frames have no depth assignment**, so they cannot be attributed to D2, D3 or D4. (4) **The transects mix 100-point and 500-point totals** and are compared through percentage fields; the site value is the unweighted mean of the three stratum means (48.2 / 55.1 / 57.9%) rather than an equal weighting of all 17 transects. D4 rests on 4 transects and carries a third of the headline number. Individual transects range from 35.6% to 77.0% coral, so within-stratum patchiness is roughly twice the difference between strata. (5) ***Heliopora* is a single-stratum record.** The blue coral appears at D3 only, at 1.29%, and contributes the site's whole 0.43%. The same caution applies to the *Porites* spike: its site-level 12.73% is driven hard by D4's 25.10%, where it is more than four times its D3 value. (6) **Bleaching is quoted four ways with four denominators** — 5.56% published, 1.67% of all 60 audited points, 5.26% of the 19 coral points, and 0.02% of mask area. They answer different questions and are not interchangeable. **A single bleached point underpins the human audit at this site.** (7) **The one condition feature visible in the imagery is an old lesion, not fresh bleaching**: the pale bare crescent on the large tabular *Acropora* in P3240478 is already colonized by crustose coralline algae, which is inconsistent with recent tissue loss. Several pale-to-white branch tips in the same frame's foreground are unresolved between paling and recent breakage. (8) **No COTS, disease or predation scarring is visible in either frame** — recorded as an explicit absence, but from a two-frame sample it is weak evidence. (9) **There are no D1 transects at this site**; the shallowest stratum sampled is D2, so the shallow reef is unsampled. (10) **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*) — here, of just two images. 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`). The site table's `Mean(%bleaching)` column is not a mean — it equals the 1st CoralNet training run — so any rank taken off it is approximate; it also carries at least one known bad value elsewhere in the archipelago (BITB58 listed at 22.22% against a corrected 0.77%). Never read this number as a diver's field estimate.

**Benthic data:** [Download BITB47 benthic survey data (CSV)](../benthic_data/BITB47_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).