### BIPB69 — Peros Banhos
**CRD page:** [Peros Banhos Coral Surveys (16–17 Apr 2015)](https://coralreefdiagnostics.com/peros-banhos-20150416-to-0417)

**16 Apr, 17 Apr and 2 May 2015 — three separate dives**
**GPS:** 5.29642°S, 71.76703°E
**Setting:** protected submerged lagoonal patch reef (inside the Peros Banhos lagoon); **site depth is blank in the published table** — no site mean is reported and none is invented here
**Coral cover:** 64.7% (4 depth strata, D2–D5, pooled from 22 individual transects; **algal cover 26.0%**) · **published image bleaching, `% bleaching (2nd training)`, reported per dive: 0% (16 Apr, 6 frames) · 12.05% (17 Apr, 69 frames) · 36.47% (2 May, 47 frames)**. The site-level `Mean(%bleaching)` cell for BIPB69 is **blank**; nothing is computed to fill it.
**Live hard coral cover across methods:** formal transects 64.7% · CoralNet points 36.7% · Coralscapes masks 37.3% (112 frames — see Flag 3, this pool necessarily spans April and May). These products have different sampling footprints and denominators and should not be averaged.
**Temperature:** 30.3 °C · **Salinity:** 34.4
**Photo review:** **all 6 of 6** frames of the 16 Apr dive; **11 of 69** frames of the 17 Apr dive (systematic, every 7th plus the final frame: P4170663, 708, 715, 727, 740, 747, 754, 762, 777, 784, 789); **11 of 47** frames of the 2 May dive (systematic, every 5th plus the final frame: P5020837, 882, 907, 919, 930, 935, 941, 947, 952, 957, 958). 28 of 122 stills in total; the 17 Apr folder also holds one unreviewed video clip (P4170761.MOV).
**Representative photos:** P4160782.jpg, P4170727.jpg, P4170777.jpg, P5020935.jpg, P5020957.jpg

BIPB69 is one of the highest-cover reefs in the archipelago: **64.7% living coral against 26.0% algae, ranking 5th of 109 Chagos sites for coral cover and 98th of 109 for algal cover** — that is, among the ten lowest algal covers recorded. The whole benthos closes cleanly: 64.73% coral plus 7.84% bare substrate, 1.46% non-coral invertebrates and 25.97% algae sums to exactly 100%. The algal budget is thin in every category — crustose coralline 9.40% (84th of 109), erect algae 7.48%, fleshy turf 7.24% (79th of 109), macroalgae 1.52%, with trace *Turbinaria*-type and cyanobacterial cover. This is a protected, submerged lagoonal patch reef where nearly two-thirds of the bottom is live coral and very little is either bare or weedy.

Twenty-five genera are non-zero — **53rd of 109 for genus richness, squarely mid-pack** — and the 25 values sum to 64.7333% of the benthos, an exact match to the published cover figure. All genus values below are percent of total benthos, not percent of coral. ***Acropora* alone holds 45.14% of the benthos — 70% of all the coral recorded at the site.** ***Echinopora* is a distant but genuine second at 8.88%** (14% of the coral), then *Favia* 3.63%, *Porites* 1.62%, *Heliopora* 1.00% (the HELI code — a blue coral, not the free-living mushroom *Heliofungia* the LOF codebook label suggests), *Platygyra* 0.61%, *Pocillopora* 0.55%, *Goniastrea* 0.53%, *Lobophyllia* 0.49%, *Millepora* 0.35%, *Fungia* 0.33%, *Mycedium* and *Seriatopora* 0.25% each, *Stylophora* 0.22%, *Symphyllia* 0.15%, *Favites* 0.13%, *Cyphastrea* 0.11%, and a tail below 0.09% — *Astreopora*, *Gardineroseris*, *Pachyseris*, *Montipora*, *Physogyra*, *Hydnophora*, *Montastrea* and *Ctenella* (CNEL — the Chagos-endemic monotypic genus, not *Cynarina*). Genus richness here is ordinary; the cover is not.

**The depth profile is the cleanest shallow-to-deep genus handover in the Chagos dataset, and it is the real finding at this site.** D2 carries 74.5% coral on 23.4% algae, and ***Acropora* alone is 71.25% of the benthos — 96% of every coral point on that transect**; nine other genera are present and none exceeds 0.875%. D3 is the site's highest-cover stratum at **78.1% coral on only 13.5% algae**, still overwhelmingly *Acropora* at 67.30% (86% of its coral), but with the widest genus list of the profile — 17 non-zero genera, including *Favia* 2.70%, *Echinopora* 1.30%, *Lobophyllia* 1.30%, *Pocillopora* 1.20% and *Platygyra* 1.00%. D4 breaks the pattern: cover falls to 61.3% and algae rises to 29.0%, *Acropora* drops to 32.00% — **just 52% of the coral present** — and *Favia* surges to 11.33% with *Echinopora* 6.33% and *Goniastrea* 2.00% behind it. D5 completes the handover: cover falls to 45.0% on the profile's highest algal cover (38.0%), and ***Echinopora* takes over outright at 27.00% — 60% of all coral on that transect — while *Acropora* collapses to 10.00%**, with *Heliopora* and *Porites* at 4.00% each and nothing else recorded at all. Bare substrate rises monotonically down the profile — 0.4%, 4.3%, 9.7%, 17.0% — and crustose coralline climbs from 8.9% at D2 to 18.0% at D5, so the deep stratum is a lower-cover, coralline-pavement reef with a different dominant genus. **The site-level figures of 64.7% cover and 45.14% *Acropora* describe no single transect.** They are the unweighted mean of the four depth strata — (74.5 + 78.1 + 61.3 + 45.0) / 4 = 64.73, and (71.25 + 67.30 + 32.00 + 10.00) / 4 = 45.14, both exact — which means a near-monospecific tabular *Acropora* stand and an *Echinopora*-dominated deep zone are averaged together at equal weight despite resting on very unequal survey effort (see Flag 1).

**The 16 April 2015 dive (6 frames, 16:10–16:16)** is a five-minute burst, the shortest photographic record at the site, and it documents a mixed, structurally complex zone rather than the tabular monoculture. P4160782 and P4160790 show plate and table colonies interleaved with dense bright-green fine-branched erect algae and maroon-red turf — a direct visual counterpart to the site's 7.48% erect and 7.24% turf categories — plus orange encrusting sponge and pale, lettuce-form foliose plates consistent with *Echinopora lamellosa*-type growth. P4160791 is a wide view of overlapping plates, corymbose colonies and grey-lavender foliose whorls; P4160792 a single large olive-green corymbose table filling the frame. P4160784 is a close-up of a discoid, uniformly lavender-grey massive colony with fine even corallites and dark maroon branching structures on its surface (consistent with a boring or encrusting associate rather than tissue loss), not confidently assignable to a genus. Several plate and table surfaces in P4160782, P4160790 and P4160791 read as pale to white, which sits awkwardly against the published 0% figure this dive carries under both training runs (Flag 5).

**The 17 April 2015 dive (69 frames, 08:05–09:06) is the tabular *Acropora* reef the transects describe.** P4170727 is the defining frame: tiered brown and olive tables tiling the reef to the limit of visibility, with a further tier stepping away up-slope — the visual match to a stratum carrying 96% of its coral in one genus. P4170708 and P4170762 repeat that structure in cleaner water, plate margins stacked several deep over hard pavement with massive lobate heads beneath the canopy. P4170777 pairs a large whorled, spiral-tiered table against an extensive arborescent staghorn thicket behind it, separated by a sand channel; P4170740 and P4170754 add a dense staghorn stand under a tiered table field, with massive domes (consistent with *Porites*-type growth, not confidently assignable) and chromis, damselfish and a butterflyfish working the structure. P4170747 is the useful counter-frame — whorled, tiered grey-brown foliose plates occupy the centre, the growth form expected of the 8.88% *Echinopora* component. **P4170784 is anomalous**: an entire reef surface of small, knobby, nodular pale-grey colonies with dark pits extending to the limit of visibility, growth form unassignable and pallor of unresolved cause (Flag 6). P4170715 and P4170708 each carry a stark-white tabular colony directly beside fully pigmented yellow-green and olive tables of the same growth form — colony-scale, within-frame contrast that rules out a lighting artefact and is consistent with the 12.05% this dive carries under the 2nd training run. P4170789 shows three divers over the plate field above a large massive dome, with survey gear on the bottom at frame left.

**The 2 May 2015 revisit — the same reef fifteen days later, at bleaching onset.** Bleaching rose from **12.05% on 17 April to 36.47% on 2 May** under `% bleaching (2nd training)`; the earlier training run scored the same two dives at 3.17% and 13.65%, so both runs record the rise but disagree sharply on its size (Flag 6). Reef type, depth zonation and generic composition are described above from the April transects and are not restated; what changed in fifteen days was condition, and the photographs show it as a **mosaic sorted by growth form rather than a uniform whitening**. Tabular and plating *Acropora* is worst affected: **P5020935 shows a near-continuous field of cream and stark-white tables running to the limit of visibility**, and P5020947 sets white, cream and lilac tables against dark maroon plate surfaces in a single frame. P5020907 settles the question of artefact — a stark-white plating stand on the right and a large white table lower-left sit directly beside fully pigmented brown and olive tables mid-frame, a diver above. P5020952 carries the same contrast at close range: white and lavender-lilac tables and a pale massive brain-form colony in the foreground, an entirely pigmented dark-brown staghorn thicket immediately to their left. **The branching and corymbose stands are conspicuously the least affected.** P5020941 is a large, wholly pigmented brown staghorn thicket with a snapper over it and only pale tables behind; **P5020957 is the strongest single statement of the mosaic — a vast, dark red-brown branching *Acropora* stand filling two-thirds of the frame in full pigment, with bleached white plates immediately to its right.** P5020837 and P5020919 show lilac and lavender patches on otherwise pigmented plates, the chromatic pigmentation response typical of early thermal stress. P5020882 shows a large whorled foliose colony (*Echinopora lamellosa*-type) uniformly pale grey-brown beside white plating colonies, and P5020958 is the most varied frame — a stark-white table centre-frame beside a pale-green one, over dark maroon encrusting substrate carrying dozens of small corymbose and plate colonies in yellow, cream, lavender and olive. That dark red-maroon encrusting cover on plate-shaped surfaces recurs in P5020919, P5020947 and P5020958; it is consistent with either recently killed skeleton colonised by algae or pre-existing dead framework, and a still cannot separate the two, so **no mortality is claimed here**.

**The taxonomic exposure matters here.** The colonies photographed bleached on 2 May are overwhelmingly tabular and plating *Acropora*, and BIPB69's April transects record ***Acropora* at 45.14% of the benthos and 70% of all coral**, rising to 96% of the coral at D2 — a reef built this heavily from one thermally sensitive genus and growth form is maximally exposed to exactly this event. That is an observed association between the April assemblage and the May condition, not a tested causal claim. Note too that image-derived coral cover **rose** between the dives, 52.50% (17 Apr) to 57.29% (2 May) under the 2nd training run — a further reason not to read any image product here as a mortality signal (Flag 9).

<b>Assemblage composition.</b> Formal transects give 64.7% coral (5th of 109 Chagos sites) against 26.0% algae (98th of 109) across 25 genera, with *Acropora* at 70% of all coral and *Echinopora* a distant second. That average conceals the cleanest depth handover in the dataset: a near-monospecific tabular *Acropora* stand at D2–D3 (74–78% cover, 86–96% of coral in one genus) giving way through a *Favia*-inflected D4 to an *Echinopora*-dominated D5 at 45% cover where *Acropora* holds only 22% of the coral. Three dives over sixteen days document the reef's condition changing under it — a mixed algal-and-plate zone on 16 April, the tabular *Acropora* canopy in full on 17 April at 12.05% bleaching, and on 2 May a growth-form-sorted bleaching mosaic at 36.47% in which tables and plates whiten while the branching stands stay pigmented.

**Flags:** (1) **The site figure equally weights four strata built on 8, 10, 3 and 1 transects.** `Chagos benthic data-all 1556 transects.csv` carries 22 individual transects for BIPB69 — D2 n=8, D3 n=10, D4 n=3, **D5 n=1** — but the site value is the unweighted mean of the four depth means (verified: 64.7333 and *Acropora* 45.1375 both reproduce exactly). **The *Echinopora*-dominated deep zone therefore rests on a single transect yet supplies a quarter of the site-level number.** Treat D5 as a real but singly-replicated observation; do not report the handover as if the strata were equally surveyed. (2) **`sitefacts.py` reports "transect rows: 0" for this site, which is wrong** — the 1556-transect file keys its site column `side code`, not `site code`, so the lookup silently returns nothing while all 22 rows are present. This affects every Chagos site and is worth fixing centrally. (3) **The multi-method cover figures pool the April and May dives.** `chagos_multimethod_cover_by_site.csv` records `n_photos = 112` against 6 + 69 + 47 = 122 stills across the three folders. 112 exceeds both the largest single dive (69) and April alone (75), so the annotated pool **necessarily contains 2 May frames alongside April frames**; the CoralNet (36.7%) and Coralscapes (37.3%) figures blend a pre-bleaching and a bleaching-onset reef and are not interpretable as either. 112 also matches no clean pair of folders (b+c = 116), leaving 10 frames unaccounted for. The 64.7% transect figure is April-only and unaffected. (4) **The published sheet's date column has the first two dives' dates swapped relative to EXIF.** Sheet row `BIPB69` is dated 17 Apr 2015 but carries `N Rows` 6; row `BIPB69B` is dated 16 Apr 2015 but carries `N Rows` 69. EXIF is unambiguous: `BiPb69a-finished` = **16 Apr**, 16:10:51–16:15:55, 6 frames; `BiPb69b-great dive-finished` = **17 Apr**, 08:05:21–09:05:33, 69 frames; `BiPb69c-awesome-finished` = **2 May**, 08:38:13–09:36:19, 47 frames. **The frame counts are the reliable link between sheet row and dive, not the sheet's dates**, and every figure in this shard is assigned by frame count. (5) **Two unresolved pallor observations in the April frames.** The 16 April dive is published at 0% bleaching under both training runs (one of only six zero rows in the archipelago, ranked joint-last, 110 of 110), yet several of its 6 frames show pale to white plate and table surfaces (P4160782, P4160790, P4160791) — at n = 6 frames this is one of the smallest annotated sets in the dataset and the 0% is a weak estimate. Separately, P4170784 (17 Apr) shows an extensive stand of small nodular pale-grey colonies of unassignable genus whose uniform pallor could be natural pigmentation or an early stress response; it should not be captioned either way, and it is a plausible driver of the run-to-run disagreement at that dive. (6) **A third, much lower bleaching figure exists for this site and it is a POOLED, undated one.** `chagos_cots_bleaching_log.csv` carries `BIPB69, (site aggregate), bleaching, 2.89, human-annotated` — "2.89% of 30-pt human annotations bleached; *Acropora* 2.11%, other hard coral 0.65%, *Pocillopora* 0.12%". That aggregate carries no date and cannot be reconciled with per-dive CoralNet values of 0% / 12.05% / 36.47%; if it pools all three dives it is dominated by the two April dives and is not a 2 May figure. **Do not quote 2.89% as this site's bleaching without saying which dives it covers.** The same file also logs a Coralscapes COTS detection on P5020948.jpg at 0.00% of frame, self-labelled "likely false positive"; no COTS was seen in the reviewed frames. (7) **The two CoralNet training runs diverge sharply here, in the bleaching-onset direction.** 17 Apr: 3.17% (1st) vs 12.05% (2nd); 2 May: 13.65% (1st) vs 36.47% (2nd). Corpus-wide the runs agree to <0.05 pp on 100 of 110 rows and diverge only on the bleaching-onset dives, always with the 2nd higher (`CHAGOS_CENTRAL_RANKS.json` `_provenance`). **Always name the column.** The dive-history map publishes the 2nd-training value; the site table's `Mean(%bleaching)` column, where populated, equals the 1st-training value and is not a mean. (8) **The site-level `Mean(%bleaching)` cell for BIPB69 is blank** in `Chagos benthic data by site (n=109).csv`; it is reported blank and nothing is computed to fill it. `CHAGOS_CENTRAL_RANKS.json` accordingly carries no site bleaching rank beyond the 6-frame row, so **no rank is quoted for the 12.05% or 36.47% values** — note only that the archipelago maximum under the same column is 38.22% (BIPB112), placing the 2 May figure within two percentage points of the highest value in the dataset. (9) **Do not read any image-derived cover change here as mortality.** Bleached coral is still live coral, and at this site image-derived cover *rose* from 52.50% (17 Apr) to 57.29% (2 May) under the 2nd training run — there is no image-based cover-loss signal in either direction. (10) **Dark red-maroon encrusting cover on plate-shaped surfaces** (P5020919, P5020947, P5020958) is consistent with either recently killed tabular skeleton colonised by algae or long-dead framework; fifteen days is fast for that transition and the stills cannot separate the two. No mortality is claimed. (11) **Site depth is blank** in the published table, and per-stratum meter depths for D2–D5 are absent from the source data. Do not invent them. (12) **The published bleaching figures are COMPUTER-VISION products, not in-situ observations** — 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`). Treat them as the weaker of the two published numbers, never as a diver's field estimate. (13) **The CRD page mapped to this site covers 16–17 Apr 2015 only**; the 2 May dive falls outside its date range and has no page of its own in `chagos_site_to_page.json`. (14) **Sampling and absences:** systematic sample for two of three dives (11 of 69, 11 of 47, against 6 of 6), frame lists in the header; one unreviewed video clip (P4170761.MOV) in the 17 Apr folder; divers and gear in P4170789, P5020907 and P5020930 but no transect tape in the reviewed May frames and no May transect data in the published tables; **no disease lesions, *Drupella* aggregations or predation scarring** seen in any of the 28 frames.

**Benthic data:** [Download BIPB69 benthic survey data (CSV)](../benthic_data/BIPB69_benthic.csv) — scleractinian cover by genus with full names, plus substrate and algal categories, from the LOF BIOT-1 survey (Mayfield & Dempsey 2026, *Oceans*, supplemental file S1).