### BISP90 — Speaker's Bank
**CRD page:** [Speaker's Bank Coral Survey Data](https://coralreefdiagnostics.com/speakers-bank)

**24 Apr 2015**
**GPS:** 4.95045°S, 72.41215°E
**Setting:** protected (leeward) submerged bank reef, inside the bank, 24.9 m
**Coral cover:** 39.8% (2 depth strata over 16 formal transects, D3–D4; **algal cover 43.4%**) · **published image bleaching (`Mean(%bleaching)` column):** 2.1%
**Live hard coral cover across methods:** formal transects 39.8% · CoralNet points 23.3% · Coralscapes masks 22.3% (8 frames). These products have different sampling footprints and denominators and should not be averaged.
**Temperature:** 29.6 °C · **Salinity:** 34.4
**Photo review:** all 8 of 8 frames examined (the entire retained set for this site)
**Representative photos:** P4240908.jpg, P4240917.jpg, P4240924.jpg, P4240930.jpg, P4240943.jpg

BISP90 sits on a submerged bank reef at Speaker's Bank, the northernmost of the Chagos survey areas, at 24.9 m — the deepest of the three lagoon-and-bank sites in this group. It is a mid-table site on both headline numbers: **39.8% coral against 43.4% algae, ranking 53rd of 109 Chagos sites for coral cover and 40th of 109 for algal cover.** The algal budget is unusually evenly split rather than dominated by any one category: macroalgae (PMTS) 12.6%, fleshy turf (PTTS) 11.4% (43rd of 109), crustose coralline algae (PCCATS) 10.3% (75th of 109), turf-on-sand (PTSTS) 6.7% and encrusting algae (PETS) 2.3%. Bare substrate is 6.8% and non-coral invertebrates 10.0%. Nothing here is extreme; the interest at this site is in how the assemblage reorganises with depth.

The coral is spread across **27 nonzero genera (40th of 109 Chagos sites for genus count)**, and the 27 genus percentages sum to 39.795% of the benthos — an exact match to the published cover figure. All values below are percent of total benthos, not percent of coral. ***Acropora* leads at 11.64% of the benthos, 29% of all coral present**, followed by *Porites* 5.77% (14%), *Montipora* 4.61% (12%), *Pocillopora* 3.02% (8%) and *Goniastrea* 2.26% (6%) — together 69% of the coral at the site. *Stylophora* (1.96%), *Echinopora* (1.83%), *Lobophyllia* (1.65%), *Favia* (1.51%) and *Favites* (1.17%) follow. The tail includes **the Chagos-endemic brain coral *Ctenella chagius* at 0.27% of the benthos**, alongside *Cantharellus*, *Platygyra*, *Pavona*, *Leptastrea*, *Hydnophora*, *Seriatopora*, *Cyphastrea*, *Astreopora*, *Gardineroseris*, *Montastrea*, *Fungia*, *Alveopora*, *Symphyllia*, *Psammocora*, *Millepora* and *Galaxea*.

**The two depth strata document a clean handover from an *Acropora*-dominated upper bank to a *Montipora*/*Goniastrea*/*Echinopora* assemblage at depth, and the site-level genus figures describe neither stratum.** D3 carries **47.1% coral against 39.4% algae** across 26 nonzero genera, and it is emphatically *Acropora* ground: ***Acropora* alone is 18.14% of the benthos there — 38% of all the coral on that stratum** — with *Porites* 7.00%, *Pocillopora* 4.57%, *Montipora* 3.68%, *Lobophyllia* 3.31% and *Stylophora* 3.05%. D4 falls to **32.5% coral against 47.3% algae** across 20 genera, and the ordering inverts: ***Montipora* takes the lead at 5.53%, *Acropora* collapses to 5.13%** (from 38% of the coral to 16%), and two genera that are effectively absent shallow become leading contributors — ***Goniastrea* rises from 0.46% to 4.07% and *Echinopora* from 0.06% to 3.60%**, roughly nine- and sixty-fold increases. *Cantharellus* makes the same jump (0.14% to 1.20%). The published site figures of 2.26% *Goniastrea* and 1.83% *Echinopora* therefore describe no depth at this reef: they are the average of near-absence at D3 and prominence at D4. Bare substrate and non-coral invertebrates both roughly double with depth (4.4% to 9.2%, and 9.1% to 11.0%), so the deeper stratum is more open ground as well as a different coral assemblage. **The D4 stratum rests on 3 transects against 13 at D3, but the site figure weights the two strata equally**; transect-weighting the same data gives 44.38% instead of the published 39.79%, a 4.6-percentage-point swing.

**All eight retained frames were examined, and they document a low- to moderate-relief bank-top mosaic in exceptionally clear water rather than a high-relief slope.** The dominant visual is pink-to-mauve coralline-encrusted pavement carrying a densely packed but low-profile mosaic of small corymbose and table *Acropora*-type colonies, compact branching colonies, brown foliose plates, small massive and meandroid heads, black sponges or hydroids, and abundant small sessile invertebrates. **P4240908 and P4240909 show the most densely occupied ground**, a near-continuous carpet of small colonies of many growth forms with almost no open substrate, including a conspicuous yellow-green table colony in P4240908. **P4240930 is the highest-cover frame in the set** and is close to a monospecific stand: cream-to-tan corymbose branching colonies with white growing tips fill the frame edge to edge, with mauve-pink compact branching colonies and coralline pavement in the gaps — a direct photographic counterpart to the 18.14% *Acropora* figure at D3. **P4240916, P4240917, P4240924 and P4240931** show the same architecture at lower packing density, with larger elements appearing: brown foliose and scroll-shaped plate colonies, large meandroid brain colonies (one at the left edge of P4240917 with the fine, tightly-packed meanders characteristic of *Ctenella chagius*, the endemic recorded on the transects, though a single frame cannot confirm it), tan lobed and encrusting colonies in the foreground of P4240924, large tan massive mounds and a dark foliose plate in P4240931, and a diver silhouetted in the upper right of P4240931. **P4240943 is the most open frame**: extensive pink-mauve coralline pavement with only sparse small colonies and one large mustard-yellow foliose plate colony right of centre — a plausible visual analogue of the lower-cover, more open D4 stratum. The eight-frame sequence therefore shows genuine small-scale variation rather than uniformly continuous coral.

**Condition reads as good, with a localised bleaching signal that the imagery documents more visibly than the published site figure implies.** The published image bleaching value is **2.1% (`Mean(%bleaching)` column; 88th of 98 Chagos sites with a value — well into the lowest quartile of bleaching severity)**, and the two CoralNet runs agree to four decimal places, so this is not one of the divergent bleaching-onset dives. Against that, the earlier photo reviewer recorded high-confidence bleaching in two frames: **P4240908**, where several compact branching colonies are nearly white beside brown, yellow and lavender neighbours, and **P4240917**, where pale-white branches occur in the foreground and on the right while adjacent colonies retain pigment. Both observations are preserved here. This re-review of the same frames is more equivocal about them: the pale colonies are real, but at this site a large share of the coral is cream-to-white-tipped corymbose *Acropora* whose growing margins are naturally white (unmistakable across P4240930), so pale branch tips and cream colony colour are not by themselves diagnostic. P4240943 also carries scattered pale colonies, but distance and water colour make them less diagnostic still. The honest reading is that BISP90 carries a small number of genuinely pale colonies against a background of naturally pale-cream branching coral, consistent in magnitude with the low published figure. Pale substrate was not treated as bleached coral; texture and colony continuity were considered frame by frame. Fish occur sparsely in the wide views — a blue-and-yellow surgeonfish in P4240917, another in P4240924, and small dark fishes among the coral — but the set is not standardised for fish abundance and no count is inferred. **No crown-of-thorns starfish, diagnostic disease band, tissue-sloughing front or discrete feeding scar with a visible predator was identified in any of the eight frames.** Old partial mortality, broken branches and encrusted framework occur but cannot be assigned a cause from photographs.

<b>Assemblage composition.</b> Formal transects give 39.8% coral against 43.4% algae — an evenly divided algal budget of macroalgae, turf and coralline pavement in roughly equal thirds — across 27 nonzero genera led by *Acropora* (11.6% of benthos), *Porites* (5.8%), *Montipora* (4.6%), *Pocillopora* (3.0%) and *Goniastrea* (2.3%), and including the Chagos-endemic *Ctenella chagius* at 0.27%. The two-stratum depth profile shows *Acropora* dominance (18.1% of benthos, 38% of coral) at D3 giving way at D4 to a *Montipora*/*Goniastrea*/*Echinopora* assemblage on more open ground, with total cover falling from 47.1% to 32.5%. The eight photographs corroborate that picture directly: a low-relief, coralline-encrusted bank top carrying a dense small-colony mosaic, one near-monospecific corymbose *Acropora* stand, foliose plates, meandroid brain colonies and open pavement, with a small number of pale colonies against a background of naturally cream-tipped branching coral and no sign of COTS, disease or predation.

**Flags:** (1) **The two strata are weighted equally despite 13 transects at D3 and 3 at D4.** Transect-weighting the identical data gives 44.38% against the published 39.79% — a 4.6 pp swing. The published figure stands under the standing protocol; the caveat should travel with it. (2) **Two site-level genus figures describe neither stratum**: *Goniastrea* (2.26% site) is 0.46% at D3 and 4.07% at D4; *Echinopora* (1.83% site) is 0.06% at D3 and 3.60% at D4. Any downstream use of BISP90 genus proportions should use the depth rows, not the site row. (3) **Localised bleaching is recorded photographically at P4240908 and P4240917** (earlier reviewer, high confidence) against a published site value of only 2.1%. This re-review preserves those observations but rates them lower-confidence, because much of the coral at this site is naturally cream-to-white-tipped corymbose *Acropora* (see P4240930). Someone with in-situ knowledge of the dive should adjudicate. (4) **Multi-method deviation is large**: CoralNet (23.3%) and Coralscapes (22.3%) sit 16.5–17.5 pp below the transect figure of 39.8%. Do not average them; the transect value is ground truth for cover. (5) **The CoralNet points figure is a CV product, not verified ground truth** — point-annotation output corpus-wide is roughly 76% unconfirmed machine output, and the CV figures here rest on the 8-frame scored subset, which happens to equal the full folder at this site. (6) **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`). Named column: `Mean(%bleaching)`, which equals the 1st CoralNet training run; the two runs agree here (2.0833 vs 2.0833). Any rank off that column is approximate — it carries at least one known bad value (BITB58 listed at 22.22% against a corrected 0.77%). (7) **A possible *Ctenella chagius* colony appears at the left edge of P4240917.** The transects record the endemic at 0.27% of benthos (0.54% at D3), so the identification is plausible, but it is a growth-form call from one frame and is not confirmed. (8) **No COTS, disease or predation scarring was observed** across the full 8-frame review — an explicit absence-of-evidence finding. (9) **A diver is visible in P4240931** — standard field documentation, not a data concern. (10) **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 **2.08%, rank 96 of 110** (1st training: 2.08%, rank 97 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 BISP90 benthic survey data (CSV)](../benthic_data/BISP90_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).