Part of the regional summaries: Indonesia - Halmahera and Indonesia - Maluku

Kusu Drift's camera flooded partway through the dive, leaving water intrusion visible in most later frames, but the reef itself runs from a sponge- and soft-coral-dominated deep wall into a dense mixed Acropora garden at the crest. One large meandroid colony shows a sharply bounded, roughly one-third bleached-but-living sector, and one branching colony has bleached upper branches. A blind Claude computer-vision point count puts coral cover around 26%.
Kamera di Kusu Drift kemasukan air di tengah penyelaman, sehingga rembesan air terlihat pada sebagian besar bingkai berikutnya, tetapi terumbunya sendiri membentang dari dinding dalam yang didominasi spons dan karang lunak menuju taman Acropora campuran yang rapat di tubirnya. Satu koloni meandroid besar memperlihatkan sektor yang berbatas tegas, kira-kira sepertiganya memutih namun masih hidup, dan satu koloni bercabang memiliki cabang atas yang memutih. Penghitungan titik buta visi komputer Claude memperkirakan tutupan karang sekitar 26%.
Terjemahan ringkasan ini oleh AI; versi bahasa Inggris yang berlaku.
These are unedited originals; edited versions will replace them.
This page shows 50 of the 147 photographs from this site, including every one the text cites.
31 of the 50 photographs on this page are shown beside the text that discusses them under "Site description" below.
The deep drift wall is sponge and soft coral rather than hard coral — maroon and ochre encrusting sponges, tan barrel sponges, jet-black bushy antipatharian (black coral) colonies, pale blue-gray and lavender Sinularia/Nephthea-type digitate soft corals, crimson Dendronephthya, and dense white and black feather stars, all under a continuous curtain of anthias and fusiliers (e.g., Image 4, Image 46).
The mid-slope terrace carries broad foliose and laminar plate fields (Montipora, Echinopora, Pachyseris-type), enormous single lobate Turbinaria/Montipora-type plates several meters across, tabulate Acropora laid flat on coralline pavement, and massive Porites heads riddled with crevices (e.g., Image 1, Image 3, Image 14, Image 18, Image 30, Image 31, Image 54).
The upper slope and reef crest is a dense mixed garden: mauve, pink, and lilac corymbose and staghorn Acropora, brown and tan Pocillopora, Stylophora, blade-like and digitate Millepora-type fire coral, orange, ochre, and rust-red arborescent soft corals, and green-brown Sinularia mats, running right up under the surface (e.g., Image 5, Image 6, Image 16, Image 21, Image 32, Image 47, Image 50).
One large meandroid colony consistent with Lobophyllia/Symphyllia carried a stark-white, sharply bounded bleached sector covering roughly a third of its surface. The white tissue is smooth, swollen, and un-fouled, the meandroid ridges intact, and it grades into normal dark brown through a narrow tan band — the signature of bleached-but-living tissue rather than recently killed skeleton, which would be dull, fouled, and showing exposed septa. The rest of the colony and every neighboring coral in the frame is normally pigmented. A second, arborescent colony has stark-white upper branches with intact tissue above its brown, white-tipped lower branches: partial bleaching (e.g., Image 27).
No crown-of-thorns, no Drupella aggregations, and no disease lesions were recorded in the 28 frames examined (a random subset of the 143 unedited frames).
A blue-gray moray photographed twice from its crevice among massive Porites; an anemone with resident anemonefish on the deep wall; Moorish idols, anthias, fusiliers, chromis, damselfish, and wrasse throughout (e.g., Image 2, Image 26, Image 52, Image 55).
The sharks in the dive log are not resolved in the sampled stills.
We found no other surveys at this site as of 2026. If you would like to share yours, please email us.
Kusu Drift was dived on 10 August 2026 during the Halmahera crossing aboard the liveaboard Ambai (Wallacea Dive Cruises), at 25–26°C.
Anderson's field note reads: "camera flooded at 35 m, lots of sharks", and the flood is written straight into the photographic record. From partway through the dive onward, irregular black lobes intrude from the edges of almost every picture — water inside the housing, not vignetting — and in the later frames the photographer's own fingertip appears at the lower left where the camera was being handled (e.g., Image 5, Image 7, Image 8, Image 10, Image 11, Image 13, Image 15, Image 19, Image 24, Image 25, Image 33).
The depth shown is Anderson's dive computer's logged maximum (24.4 m); the 35 m in the field note is quoted as written and is not used as the dive depth.
| Method | Hard coral cover (%) | Photos (n) ± SE |
|---|---|---|
| In-house DINOv3 | 27.6 | n = 143 · SE 0.9 |
| CoralSCOP unedited | 33.9 | n = 143 · SE 1.4 |
| Claude | 26.1 | n = 143 · SE 2.0 |
The two computer-vision methods agree within 6.3 pp (27.6–33.9%) and read 1.5–7.8 pp above Claude's blind point count (26.1%); the headline is Claude's blind point count because no transect or manual count exists here.
See a more detailed, stand-alone version of this table.
All images from Coral Reef Diagnostics. Data are from Coral Reef Diagnostics.
© Coral Reef Diagnostics.