The Japan, Izu-Ogasawara and Ryukyu trenches (Fig. 1), all located within the Japanese exclusive economic zone (EEZ), were surveyed during the ‘Ring of Fire Expedition’ in August and September 2022 on DSSV Pressure Drop.

Japan Trench

Located parallel to the east coast of the Japanese mainland of Honshu, is the Japan Trench (Fig. 1A). It extends 611 km and covers an area of 37,854 km² beyond 6000 m depth. The Japan Trench is formed as the Pacific Plate subducts beneath the Okhotsk (minor) plate at a convergence rate of approximately 9 cm/yr (Nishikawa et al. 2023). The deepest point is 8001 m at 36.0872° N, 142.7272° E (this study). In the north, a subducting seamount partitions it from the Kuril-Kamchatka Trench and, in the south, the westernmost seamount of the Joban Seamount Chain separates it from the Izu-Ogasawara Trench (Yamazaki and Okamura 1989). The Japan Trench is seismically active and known for tsunamigenic megathrust earthquakes (Ueda et al. 2023) that result in large mass wasting events at depth, resulting in large volumes of re-suspended sediments (Itou et al. 2000, Oguri et al. 2013, Kioka et al. 2019).

Izu-Ogasawara Trench and Boso triple junction

The Izu-Ogasawara trench (or Izu-Bonin Trench; Plank et al. (2007)) connects to the Japan Trench at the intersection of the Joban Seamount Chain in the vicinity of the Boso triple junction (trench-trench-trench or TTT type; Ogawa and Yanagisawa (2011)) (Fig. 1C). The Izu-Ogasawara Trench is formed as the pacific Plate is subducted beneath the Philippine Plate at a convergence rate of approximately 3-6 cm/yr (Stern et al. 2003). The triple junction is where the Pacific, Eurasian and Philippine tectonic plates converge (Soh et al. 1988) (Fig. 1B). The trench extends between the Japan Trench and the Izu-Ogasawara Plateau for 1122 km and hosts 105,328 km² of seafloor > 6000 m. It has a maximum depth of 9775 m at 29.4262° N, 142.7075° E (this study). The northern part of the trench subduction occurs largely by aseismic slip, resulting in no great earthquakes (Fukao et al. 2021). In contrast, the southern sector of the trench is characterised as a double seismic zone (DSZ) (Nakata et al. 2019).

Ryukyu Trench

Parallel to the Japanese Nansei-Shotō Islands between Taiwan and Kyūshū is the Ryukyu Trench (or Nansei-Shotō Trench; Nishizawa et al. (2017)) (Fig. 1D). This trench is 1012 km long and covers a surface area of 39,073 km² > 6000 m. The deepest point is 7339 m at the south-western end of an elongated depression at 24.4752° N, 127.4088° E (this study). The trench is formed as the Philippine Plate is obliquely subducted beneath the continental crust of the Eurasian Plate at a rate of approximately 52 mm per year (Nishizawa et al. 2009). The trench exhibits very low-frequency earthquakes (VLFEs; Ando et al. (2012)) and is known to produce frequent slow slip events (SSEs) that can last for ~ 1 month (Heki and Kataoka 2008).

Survey areas were first mapped using a hull-mounted EM124 multibeam echosounder (MBES; Kongsberg, Norway; Bongiovanni et al. (2022)). In-situ observations were made using three identical free-fall landers (Skaff, Flere and Closp) equipped with high-definition video cameras (IP Multi SeaCam 3105; Deep Sea Power and Light, San Diego, CA) and conductivity, temperature and depth (CTD) probes (SBE 49 FastCAT, SeaBird Electronics, Bellevue, WA). Additionally, video transects were conducted using the full ocean depth submersible Limiting Factor, with footage recorded via two forward- and downward-facing cameras. Depth was recorded using two CTD probes. Cameras and CTD probes on the submersible were the same as on the landers. Full details of lander and submersible deployments are provided in Table 1 and Suppl. material 1, and information on landers and the submersible Limiting Factor can be found in Jamieson et al. (2019).

In the Japan Trench (Fig. 1A), four submersible dives were undertaken, one at its deepest point (8002 m, 36.087°N, 142.727°E) and three around the location of the 2011 Tōhoku earthquake epicentre. Of these, one was located on the trench axis (7529 m, 38.045°N, 143.999°E), one on the adjacent forearc (7324 m, 38.594°N, 144.026°E) and one on the adjacent downriding slope (7381 m, 38.571°N, 144.169°E). Thirty-two successful lander deployments were conducted in the Japan Trench, of which six were around its deepest point and the remaining 26 were around the location of the Tōhoku earthquake epicentre (Ueda et al. 2023) (Fig. 1A).

In the Izu-Ogasawara Trench (Fig. 1B and C), two submersible dives were undertaken, one in the southern end at the deepest point (9775 m, 29.426°N, 142.708°E) and one in the north at the base of the Boso triple junction (9137 m, 34.288°N, 141.870°E). A total of 21 successful lander deployments were also conducted along the length of the Izu-Ogasawara Trench, of which seven were located in the vicinity of the Boso triple junction (34°N), three at 32°N and the remaining 11 at 29°N (Fig. 1B and C).

In the Ryukyu Trench (Fig. 1D), one submersible dive was undertaken to its deepest point (7322 m, 29.475°N, 127.409°E). Six lander deployments were conducted on the overriding plate, of which three were located around the deepest point of the trench and three on the landward slope (Fig. 1D).

Video footage recorded in-situ from the baited landers and submersible dives was continuously annotated using EventMeasure (SeaGIS Pty Ltd, version 6.42) to generate records of taxa present. Submersible dive footage was only annotated if the submersible was on-bottom. Lander video footage was only annotated when the lander was on the seafloor (bottom time) using MaxN (also known as N_max) which is a conservative metric of relative abundance and refers to the maximum number of individuals of each taxon observed at a single point in time (Priede et al. 1994).

All visible mobile fauna were identified to the lowest possible taxonomic rank using morphological characteristics. All identifications were derived from review of video footage from which still images were extracted, allowing us to distinguish morphotaxa using a combination of morphology, movement and behaviour. Where species-level identifications were not possible, taxa were placed in putative ‘morphotaxa’. The level of taxonomic precision achieved was indicated using the open taxonomic nomenclature signs recommended for image-based identifications by Horton et al. (2021). A suffix was added to specify the taxonomic rank (e.g., “fam.”, “gen.”, “sp.”). The designation “indet” was used when a single observed form could not be reliably identified beyond a given taxonomic level because diagnostic characteristics were unresolvable from the imagery. The designation “spp.” was used when multiple visual forms were observed within a given taxonomic rank, but lacked sufficient detail to be consistently separated across all annotated imagery, and therefore grouped (e.g. “Mysidae spp.”). Representative reference images were extracted for each morphotaxon; however, images of very poor quality were excluded from the ID guide and marked as ‘ns’ (not shown).

Submersible and lander datasets were combined to provide an overview of morphotaxon occurrences across trenches (Table 2, Suppl. material 2). Depth ranges for each observed morphotaxon were determined, based on the minimum and maximum depths of observations.

Of the 18 morphotaxa identified in the Annelida phylum, 15 showed morphological features that allowed identification beyond phylum level; however, three morphotaxa could not be assigned confidently beyond Annelida indet. (Fig. 2E-G). Of these, Annelida cls. indet. 1 and Annelida cls. indet. 2 (Fig. 2E and 2F) appeared morphologically similar however, their swimming styles differed. Annelida indet. 1 exhibited torsional rotations of the body axis while indet. 2 moved with lateral flexion from side-to-side. The other Annelida were represented by four identifiable orders (Echiuroidea; Fig. 2A-D, Phyllodocida; Fig. 2I-L, Sabellida; Fig. 2M, N and Terebellida; Fig. 2H-P). Six morphotaxa were identified in the Echiuroidea order (with an additional Echiuroidea spp. comprising small indeterminable morphospecies), four in the Phyllodocida order (subfamilies Macellicephalinae and Lepidonotopodini) and two in the Terebellida (families Acrocirridae and Flabelligeridae) order, the others being two morphotaxa in the Sabellidae order. None of the annelid morphotaxa were determined to the species level and, whilst observations of annelids were abundant, their small size complicated confident assignment of morphotaxa based on morphology. For this reason, several annelid morphotaxa identifications relied on movement and behaviour. Apart from two, all annelids observed were present in the Japan Trench, while seven were observed in the Ryukyu Trench and six in the Izu-Ogasawara Trench. All were observed at hadal depths. All were benthic taxa, except for polynoids in the subfamily Macellicephalinae, which included Flabelligeridae (possibly including the genus Flota; Fig. 2H) and three unidentified morphotaxa of annelids that were observed swimming off the bottom (Fig. 2E-G). Of the three echiuroids, Echiuroidea fam. indet. 1 (Fig. 2A) and Echiuroidea fam. indet. 3 (Fig. 2C) were frequently observed appearing from burrows; however, the size of their burrow and proboscis shape were consistently different from one another. Echiuroidea fam. indet. 2 (Fig. 2B) resembled Urechidae, but family-level identification could not be confirmed with certainty. All Annelida were mobile, except for the two Sabellidae morphotaxa which were observed in a ridged or semi-ridged tube protruding vertically from the seafloor. Their slow mobility or sessile nature meant that the majority were observed by submersible rather than from the lander.

Historical records show two unknown species of polynoid, Macellicephala “sp. A” and “sp. B” from 7420-7450 m on the axis and 6700-7400 m on the adjacent slope of the Japan Trench (Horikoshi et al. 1990). The genus Flota has previously been reported at a maximum depth of 6444 m in the Japan Trench (Lindsay 2005).

The Arthropoda were represented by two classes (Malacostraca and Copepoda), five orders, ten families (plus two unknown) forming 13 morphotaxa. Within the Malacostraca, five orders were observed: Amphipoda, Decapoda, Isopoda, Mysida and Tanaidacea (Fig. 3, Table 2).

In the Amphipoda order, multiple species were present within the Alicelloidea and Lysianassoidea superfamilies. Due the typically small body size of abyssal and hadal amphipods and their propensity to aggregate at baited lander in high numbers complicating identifications as well as counts, these amphipods were collectively listed here as Amphipoda spp. Known species within these trenches and these depths are Hirondellea gigas (Birstein and Vinogradov, 1955), Paralicella microps (Birstein and Vinogradov, 1958), Eurythenes gryllus (Lichtenstein in Mandt, 1822) and Bathycallisoma schellenbergi (Birstein and Vinogradov, 1958), although there are no records from the Ryukyu Trench (Jamieson and Weston 2023). Due to their abnormally large body size, the ‘supergiant amphipod’ Alicella gigantea Chevreux 1899 (Fig. 3A), the ‘giant amphipod’ Eurythenes sp. indet. (Fig. 3B) and the predatory amphipod Princaxelia jamiesoni Lörz 2010 (Fig. 3C) were identified from both landers and submersible footage. A. gigantea was observed in all three trenches for the first time (total depth range = 6529-8022 m), around their known depth range (6253-8903 m; Jamieson and Weston (2023)). A. gigantea was most conspicuous in baited lander footage, whereby they approach the bait by swimming close to the bottom and ultimately consuming the bait. Eurythenes sp. indet. was absent in the Izu-Ogasawara Trench between 6941 m and 7310 m, close to the depth Belyaev (1989) reported E. gryllus, which is likely sensu lato (Jamieson and Weston 2023). Their behaviour was akin to that of A. gigantea. Princaxelia jamiesoni was absent from Ryukyu Trench which is in accordance with collections and observation of Lörz (2010) and Jamieson et al. (2012b), respectively. They were observed in baited camera footage preying upon smaller scavenging amphipods (as per Jamieson et al. (2012b); however, in submersible footage, they were often seen stationary on rocks or hard substrate.

The Decapoda were dominated by two species; Cerataspis monstrosus Gray 1828 (Aristeidae) between 4534 and 6692 m and Benthesicymus crenatus Spence Bate, 1881 (Benthesicymidae) between 4534 m and 7571 m (Fig. 3F and G). The former is a scavenging species, while the latter preys upon scavenging amphipods and both are most readily recorded using baited cameras. These types of feeding behaviour and depth ranges are typical for these two species. Although they were only known from the Japan Trench (Swan et al. 2021), they were observed in all three in this study. Heterogenys sp. indet. 1 of the Acanthephyridae family was also observed between 4534 m and 6692 m in all trenches (Fig. 3E). A fourth morphotaxon was observed at 5941 m in the Ryukyu Trench, but was too small for species-level identification.

One family-level grouping (Munnopsidae spp.) and one known species of isopod (Rectisura herculea, Birstein 1957) were observed (Fig. 3H and I). Munnopsidae are a cryptic family and difficult to delineate to lower taxonomic levels from imagery, but were recorded between 4534 m and 9734 m across all trenches. Rectisura herculea (Birstein, 1957) were present in the Japan and Izu-Ogasawara trenches between 5525m and 8291 m. These epibenthic species have been recorded in these trenches at similar depths before (seeHorikoshi et al. 1990; note, the genus at the time was Storynthgura; and Jamieson et al. 2012a). Rectisura herculea was quite common and conspicuous in the Japan and Izu-Ogasawara trenches; therefore, its absence from the Ryukyu Trench is likely real.

Valviferan isopods (Antarcturidae gen. indet.) were seen between 7165 m and 7345 m in the Japan Trench (Fig. 3J). They were seen perched along the edges of rocks protruding from the seafloor with their pereiopods elevated in the current.

Mysids (Mysidae spp.) were observed at all depths across all trenches (Fig. 3K). They typically swim close to, but above the seafloor, often in swarms of scavenging amphipods around bait or resting on the seafloor in submersible transects. Usually, they are either white or red, but their small size made it difficult to determine if body colour indicates two different morphotaxa.

In the Japan and Ryukyu trenches, the giant tanaid Gigantapseudes sp. indet. was seen between 6937 m and 7345 m in submersible videos (Fig. 3L). They were visible on the seafloor, often nearby a burrow as Kakui and Fujiwara (2020) reported from the Ryukyu Trench.

Copepods of unknown order(s) were seen between 4913 m and 7621 m in the Japan and Izu-Ogasawara trenches (Fig. 3M) which is a depth range similar to that of other copepods found in neighbouring trenches (Markhaseva and Renz 2025).

The Chordata were represented by three classes: Appendicularia, Ascidiacea and Teleostei (Figs 4, 5, Table 2). The Appendicularia were represented by three morphotaxa in the Copelata order: Fritillariidae spp. (5042-6579 m Japan and Izu-Ogasawara trenches; Fig. 4C), Oikopleuridae spp. (4534-5941 m, Izu-Ogasawara and Ryukyu trenches; Fig. 4B) and Copelata spp. (4534-8336 m, all trenches, Fig. 4A). All were observed from baited landers, drifting past the camera in the near-bottom currents. If the housing was ovoid, showed two distinct chambers and was centrally positioned, an individual was classified as Oikopleuridae; if it was elongate and bilaterally symmetrical, it was identified as Frittilariidae (Alldredge 1976, Bone 1998); otherwise, they were left at the order level. Five Ascidiacea morphotaxa were observed. One was too cryptic to determine individual morphotaxa (Ascidiacea spp.; 7096-7345 m; Japan Trench; Fig. 4D) and one was of an unknown species (Octacnemus sp. indet.; 6606-7344 m; Izu-Ogasawara and Japan trenches; Fig. 4E) (Mandre and Rouse 2025). The other three were of the Octacnemidae (Phlebobranchia) family (Mandre and Rouse 2025); Octacnemidae gen. indet. 1 (7421-7290 m; Japan Trench; Fig. 4F), Octacnemidae gen. indet. 2 (7170-8077 m; Fig. 4G) and Octacnemidae gen. indet. 3 (7345-8077 m; Izu-Ogasawara and Japan trenches; Fig. 4H). No Ascidians were observed in the Ryukyu Trench.

The bony fish (Teleostei) were represented by five orders and five families. Twelve morphospecies or species were identified following Jamieson et al. (2021). The synaphobranchid eel Ilyophis sp. indet. (Synaphobranchidae) was seen at 4534 m in the Izu-Ogasawara Trench (Fig. 5A). The pricklefish Abyssoberyx sp. indet. (Stephanoberycidae) was seen at 5459 m in the Izu-Ogasawara Trench (Fig. 5B). The grenadier Coryphaenoides armatus (Hector, 1875) (Macrouridae) was seen in the Izu-Ogasawara Trench between 5042 m and 5459 m (Fig. 5C). Another macrourid, C. yaquinae (Iwamoto & Stein, 1974), was seen in the Japan and Izu-Ogasawara trenches between 4534 m and 7259 m (Fig. 5D). The deepest observation of C. yaquinae represents the deepest known use of a gas-filled swimbladder in fishes (Priede et al. 2024). Five morphospecies of cusk eel (Ophidiidae) were seen, with Barathrites iris Zugmayer, 1911 observed at 5110-5941 m in the Ryukyu Trench (Fig. 5E). The other four were morphospecies of the genus Bassozetus, labelled here as Bassozetus sp. indet. 1, 2, 3 and 6 (Fig. 5F-I) in keeping with the categorisation of Jamieson et al. (2021). Three morphospecies of snailfish (Liparidae) were seen: Pseudoliparis sp. indet. 1, 2 and 3 (Fig. 5J-L; corresponding to Liparid sp. indet. 1, 3 and 2 in Jamieson et al. (2021)). They were found in all trenches, except Pseudoliparis sp. indet. 2, which was absent from the Ryukyu Trench. Their depth ranges were 6023-8336 m, 6579-7621 m and 6860-8022 m, respectively. Pseudoliparis sp. indet. 1 at 8336 m in the Izu-Ogasawara Trench is the deepest recorded fish (Jamieson et al. 2023b). Another putative morphospecies of liparid is discussed further in the Discussion under the section ‘Rarities and Peculiarities’, but is not included in the species list.

All fish were observed by baited camera; however, the three Pseudoliparis and four Bassozetus morphospecies attended the bait to prey upon scavenging amphipods. Barathrites iris, Ilyophis sp. indet. and the two macrourids all scavenged upon the bait. Abyssoberyx sp. indet. was observed incidentally, irrespective of the bait.

The Cnidaria were represented by three classes: Hexacorallia, Hydrozoa and Scyphozoa (Figs 6, 7, Table 2).

Actiniarians (Actinaria spp.) were observed in all three trenches between 6485 m and 9617 m. They were often too small and cryptic to assign a specific morphotaxon. Nonetheless, seven actiniarian morphotaxa were identified, mainly in the Izu-Ogasawara and Japan trenches. Morphotaxa were distinguished through consistent differences in column length and width, column diameter to oral disc ratio, whorl count and tentacle number and morphology.

Actiniaria fam. indet. 1, 2, 3, 4 and 7 were all observed at similar depth ranges in the Izu-Ogasawara and Japan trenches (7528-9775 m, 7526-9137 m, 7095-9745 m, 7082-9734 m and 7168-9136 m, respectively; Fig. 6A-D, G). Actiniaria fam. indet. 1 (Fig. 6A) was distinguished by the presence of a very extended column, whereas a column was practically absent in Actiniaria fam. indet. 2 (Fig. 6B). Actiniaria fam. indet. 3 and 4 (Fig. 6CD) appeared to have similar basal disc morphology, but differed in an increased column length to oral disc ratio and tentacle count in Actiniaria fam. indet. 4. Actiniaria fam. indet. 7 (Fig. 6G) had the greatest column diameter to oral disc ratio of morphotypes observed. Actiniaria fam. indet. 5 and 6 were only seen in the Japan Trench at 7180-7363 m and 7527-7528 m, respectively (Fig. 6E and F). Compared to other morphotypes, Actiniaria fam. indet. 5 (Fig. 6E) has a high density of closely-spaced tentacles exceeding the length of the column. Actiniaria fam. indet. 6 (Fig. 6F) featured a tentacle arrangement in multiple whorls, extending into the oral disc.

Other morphotaxa observed in the Actiniaria included Galatheanthemum sp. indet. 1 of the Galatheanthemidae family, seen between 7300 m and 7500 m in the Japan Trench (Fig. 6H). Anthozoa spp. were observed on soft sediments in all trenches (7309-9745 m; Fig. 6I).

The Hydrozoa were represented by three known orders (Anthoathecata, Narcomedusae and Trachymedusae) and one indeterminable order. Ten hydrozoan morphotaxa were identified.

Within the Anthoathecata, there were two hydrozoan morphotaxa in the Corymorphidae family: the giant hydroid Branchiocerianthus sp. indet at 8022 m and Corymorphidae gen. indet. at 4913 m, both in the Japan Trench (Fig. 7B and C, respectively). Two morphotaxa of were observed in the Narcomedusae order, the first being small individuals grouped under Aeginidae spp. between 6825 m and 9773 m in the Izu-Ogasawara and Japan trenches and Narcomedusae spp. between 5110 m and 8000 m in the Japan and Ryukyu trenches (Fig. 7D and E).

The Trachymedusae Benthocodon spp. (Rhopalonematidae) were found at depths ranging from 7183 m to 9773 m in the Izu-Ogasawara and Japan trenches (Fig. 7F and G). Other Rhopalonematidae were observed between 5459 m and 9773 m in the Izu-Ogasawara and Japan trenches (Fig. 7H–M). However, due to poor image quality, it was not possible to clearly identify morphological features, such as centripetal canals stemming from the ring canal and gonad positions, which are key for distinguishing genera. The Rhopalonematidae spp. group includes species with a peduncle, such as the Benthocodon and Pectis genera (e.g. Fig. 7H, J and L), as well as other individuals where even the presence of a peduncle could not be ascertained. Between 5110 m and 6023 m in the Ryukyu and Izu-Ogasawara trenches, the Trachymedusa Crossota spp. were observed (Fig. 7N and O). In some cases, they resembled C. millsae Thuesen 2003, but exact species identities could not be confirmed in all instances.

A single scyphozoan of the Coronatae order (Periphyllidae or Paraphyllinidae) was seen at 5993 m in the Japan Trench (Fig. 7P).

Ctenophores were observed in the Tentaculata class, belonging to the orders Platyctenida, Cydippida and Lobata (Fig. 8, Table 2). Lobata fam. indet. were seen drifting past baited cameras close to the seafloor between 5042 m and 6177 m in all trenches (Fig. 8A). In the Cydippida order, video data supported Cydippida fam. indet. 1 (Fig. 8B) as the same morphotype as reported in the Ryukyu Trench by Lindsay and Miyake (2007). This study represents its first records in the Japan and Izu-Ogasawara trenches, thereby also extending its known depth range to 9136 m. Other Cydippids (Cydippida fam. indet., Fig. 8EF) were observed in all trenches at depths between 4534 m and 7519 m. The benthic ctenophore Lyroctenidae gen. indet. (Platyctenida, 7091-8001 m) was frequently seen attached to hard substrate in the Japan Trench (Fig. 8C and D). Additionally, smaller individuals were observed in the Japan Trench at 8000 m, but, as these could not be identified to lower taxonomic level, these were collectively classified as Ctenophora spp.

The Echinodermata comprised five classes: Asteroidea, Crinoidea, Echinoidea, Holothuroidea and Ophiuroidea (Fig. 9, Table 2).

The class Asteroidea was represented by Porcellanasteridae gen. indet. (Fig. 9A, 7309-7310 m, Ryukyu Trench), Hymenaster sp. indet. (Fig. 9B, 6938-7369 m, Japan Trench) and a third indeterminable morphotaxon (7453 m, Japan Trench).

Two morphotaxa of crinoids (Crinoidea) were observed, present in the Japan Trench between 7121 m and 7334 m (Bathycrinus sp. indet. 1, Fig. 9C) and in the Izu-Ogasawara Trench between 8975 m and 9736 m (Bathycrinus sp. indet. 2., Fig. 9D). At the deepest base of the Boso triple junction, Bathycrinus sp. indet. 2 were observed in large numbers, forming dense aggregations along terraces and on rocks, as reported by Oji et al. (2009). Most individuals spread their arms in conical fans with the oral surface facing down-current, which is a typical feeding posture in near-bottom currents (Macurda and Meyer 1974).

The most diverse Echinodermata class were the Holothuroidea. Twenty morphotaxa spanning at least three orders and four families were observed (Note, there is currently no assigned family for the genus Benthothuria, but it is the only genus from the Persiculida order in this study and, therefore, a fourth family is assumed). The most diverse (ten morphotaxa) was the Elpidiidae family. These included four morphotaxa in the genus Elpidia: Elpidia. sp. indet. 1 (cf. birsteini) (Fig. 9E, 7001-7311 m, Japan Trench), Elpidia sp. indet. 2 (Fig. 9F, 7286-9037 m, Izu-Ogasawara and Japan trenches), Elpidia sp. indet. 3 (cf. kurilensis) (Fig. 9G, 8974-9047 m, Izu-Ogasawara Trench) and Elpidia sp. indet. 4 (cf. longicirrata) (Fig. 9H, 9025-9623 m, Izu-Ogasawara Trench). Three morphotaxa in the genus Peniagone were observed: Peniagone sp. indet. 1 (Fig. 9J), Peniagone sp. indet. 2 (Fig. 9K), Peniagone sp. indet. 3 (Fig. 9L). Three further morphotaxa were defined in the Elpidiidae family, but could not be confidently identified to lower taxonomic levels: Elpidiidae spp. (4913-8336 m, all trenches), Elpidia spp. (6819–9745 m, Izu-Ogasawara and Japan trenches) and Peniagone spp. (4913 -9775 m, all trenches, example shown in Fig. 9I). A further species of Elasipodida, Enypniastes eximia Théel 1882, was observed in all trenches between 5110 m and 6883 m, sometimes on the seafloor and, in other instances, swimming in the water overlying the seafloor (Fig. 9M). Mesothuria spp. (Mesothuridae) were observed in all trenches between 7109 m and 9137 m and, although morphological variation was observed (Fig. 9N-P), different morphotaxa could not be consistently identified from video data. Finally, Benthothuria sp. indet. (Fig. 9Q) was seen in video data from all trenches between 5932 m and 6824 m.

The only representative of the class Ophiuroidea was an indeterminable morphotype with a depth range of 4913-7339 m seen in all three trenches (Fig. 9R).

Lastly, one individual of the class Echinoidea (Echinothurioida fam. indet.) was observed and only found at 5110 m in the Ryukyu Trench (Fig. 9S). This individual occurred in the far field of view of the camera and identification is therefore limited, but video imagery was used to confirm the presence of elongate spines which it used for locomotion, characteristic of an echinoid.

Tentatively identified foraminifera (Monothalamea) were observed in irregular patches on the seafloor, with many clusters forming around obstacles such as rocks, metallic anthropogenic debris or wood debris (Fig. 10AB, all trenches between 6938 m and 9762 m). These patches often appeared as disorganised, irregularly reticulated masses lacking any consistent pattern, possibly having disintegrated over time. The disorganised and fragmented nature of these structures makes definitive identification challenging without physical samples.

The organisms from the Izu-Ogasawara Trench, particularly near the Boso triple junction (Fig. 10C), can tentatively be classified as xenophyophores, exhibiting morphological features similar to known species such as Shinkaiya lindsayi (described from Japan Trench specimens at approximately 5435 m depth, Lecroq et al. (2009)) or Syringammina spp. (pers. comm. Andrew Gooday, NOCS, UK). These organisms also resemble presumed xenophyophores documented in seafloor imagery from the NE Atlantic at around 3000 m depth, as discussed by Tendal and Gooday (1981). The branched forms, especially those depicted in Fig. 10B, display complex, irregular reticulated structures akin to those observed in the Japan Trench, but are not always clearly associated with physical structures (Fig. 10D).

In the Ryukyu Trench, patches resembling xenophyophores were observed throughout the dive, notably in areas close to and down-current from discarded metal canisters (Fig. 10E). These patches were generally more compact and dispersed compared to those near the Boso triple junction. While some displayed irregularly reticulated patterns, many appeared fragmented or disconnected, suggesting possible disintegration or varying growth stages.

Additionally, in the Ryukyu Trench, an area roughly 2 m in diameter was noted where multiple evenly spaced, spherical “xenophyophores” were observed on the seafloor (Fig. 10F). These distinctive organisms had smooth surfaces and were surrounded by a distinct moat (see Fig. 10F inset) and may feature faint radial markings or grooves within and on parts of the margins of the moat.

Several phyla were represented by rare observations. These included the Hemichordata, with two enteropneust morphotaxa in the Torquatoridae family observed in the Japan Trench at 7526–7528 m (Torquatoridae gen. indet. 1, Fig. 11A) and 6819 m (Torquatoridae gen. indet. 2, Fig. 11B). Mollusca were represented by the Cephalopoda (Octopoda) and Gastropoda (Neogastropoda). A solitary individual octopod of the Cirrata suborder was seen travelling high over the bait in a lander deployment at 5941 m in the Ryukyu Trench (Fig. 11C). Deep cirrated octopods are known from similar depths, but the image quality was not sufficient to determine whether this individual was Grimpoteuthidae or Cirroteuthidae, as discussed by Jamieson and Vecchione (2022). Gastropods (Buccinidae) were seen between 5525 m and 7529 m in the Japan and Ryukyu trenches (Fig. 11D). Nemertean worms were observed crawling on the sediment surface or swimming close to the sediment surface. Two morphotaxa were observed: Nemertea cls. indet. 1 (all trenches, 4913–8291 m, Fig. 11E) and Nemertea cls. indet. 2 (Japan and Izu-Ogasawara trenches, 5445-8000 m, Fig. 11F). Nemertea cls. indet. 1 was dorsoventrally flattened and exclusively observed crawling on the sediment surface, whereas Nemertea cls. indet. 2 had an elongated body and moved using high amplitude undulations close to the sediment surface.

The only morphotaxon identified within the Porifera phylum was a demosponge belonging to the Cladorhizidae family (Fig. 11G). It was observed in the Izu-Ogasawara Trench at depths between 9568 m and 9744 m, marking the deepest in-situ observations of carnivorous sponges to date. Additionally, Chaetognatha of indeterminate class(es) (Fig. 11H) were seen in all trenches at depths ranging from 4913 m to 6870 m, swimming above the seafloor. Intriguingly, an organism of unknown phylum was recorded by a lander at 8022 m in the Japan Trench and by a submersible at 9,137 m in the Izu-Ogasawara Trench (Fig. 11I). In both instances, the organism slowly glided to the seafloor. However, due to considerations discussed below, it has not been possible to assign a definitive taxonomic classification, based on expert consensus.

In addition to the established approach of baited cameras for documenting mobile fauna and exploratory vehicles for documenting sessile fauna, this study further highlights how different survey methodologies, in this case, landers and submersibles, influence both observed biodiversity and behavioural patterns. Perhaps the starkest difference are the large aggregations of scavenging amphipods that rapidly consume bait from a baited system, yet are seldom observed during a submersible dive. While baited cameras provide critical insights into mobile scavenger assemblages, they target mobile species attracted to bait. On the other hand, non-baited video transects can capture sessile and other non-bait-attending organisms across multiple habitats. In addition, larger more easily identifiable species like Princaxelia jamiesoni and mysids behave quite differently, depending on the observation method. Princaxelids are seen crawling and swimming in pursuit of smaller Lysianassoid amphipods when attending baited systems, but in non-baited video transects are easily identified sitting stationary on rocks (as on the left side of the rock in Fig. 12B). Likewise, mysids are often seen swimming within these aggregations of scavenging amphipods, yet from submersible observations are often seen sitting idly on the seafloor facing into the current.

These observations suggest that integrating multiple survey methods supports a holistic assessment of hadal biodiversity, as different methodologies capture different aspects of community composition and species behaviour. Comparisons with trawl sampling at these depths further underscore this. For example, Horikoshi et al. (1990) collected 6193 specimens of eight species during three trawls (two between 7420 m and 7450 m on the trench axis and one between 6700 m and 7400 m on the adjacent slope). A total of 81% of the samples collected were the bivalve Kelliella sp. (Venerida), although none were collected in the trawl from the slope. This suggests a significant presence of a bivalve community that is not detectable by visual assessments alone. Conversely, these trawls did not capture many of the > 100 morphotaxa from the Japan Trench in this study, except for snailfish, polynoid annelids, isopods and elpidid holothuroids, suggesting that there are easily observable sampling biases when comparing both methods.

Other interesting observations relate to rarity and unusually high abundance of certain taxa. Two particularly striking examples of high abundance were recorded in the Izu-Ogasawara Trench. At the northern end, during a submersible dive to the base of the Boso triple junction at 9137 m in the northern end of the Izu-Ogasawara Trench, a dense aggregation of crinoids was observed – referred to as a "crinoid meadow" by Oji et al. (2009) - comprising 1524 individual crinoids attached to terraces and rocks over a 4 h transect (Fig. 12A and B). Such aggregation patterns are often attributed to localised habitat conditions, such as the presence of highly topographically complex features, including rocks and fault scarps, which provide enhanced substrate stability and facilitate attachment of sessile organisms (Oji et al. 2009). At the southern end of the Izu-Ogasawara Trench, near its deepest point, dense aggregations of cladorhizid sponges (Fig. 11G) were observed. These sponges were conspicuously absent from other surveyed locations. Their presence may reflect favourable conditions, such as elevated POC flux or fine-scale current dynamics that facilitate passive prey capture, a key factor in cladorhizid feeding ecology (Vacelet and Boury-Esnault 1995).

In contrast, substrate instability following disturbance events can also shape species aggregation patterns. An example is the dense aggregations of echiuroids on the floor of various fissures on the fault scarp generated by the 2011 Tohoku-Oki earthquake (Ueda et al. 2023; Fig. 10C). It may be that the echiuroids would have been distributed more evenly across the seafloor had they not fallen into these fissures, resulting in a usually patchy distribution, specific to this area.

Valviferan isopods (Antarcturidae gen. indet.) were only observed lined up along the acute edges of rocky outcrops (Fig. 12DE). Many valviferan isopods have a somewhat elevated anterior body that can flex to elevate the filtering setae on pereiopods 2–4 away from the substrate, a posture that is particularly pronounced in Antarcturus sp. indet. during feeding (Poore 2001). Based on this morphology and behaviour, we infer that the isopods may use the outcrop edges as vantage points for filter feeding, by allowing greater exposure of setae to currents or particle flow.

In the case of rarities, across the entire dataset of ~ 460 h of footage, some whole phyla or classes were represented by just one individual, for example Enteropneusta, Octopoda and Echinoidea. While acknowledging that highly mobile Octopoda may be elusive due to vehicle avoidance, as well as potentially low in abundance, the other phyla are slow-moving epibenthic groups that should exhibit no known avoidance behaviour to sampling gear and, therefore, are likely to be extremely low in abundance. Their presence is only known from collecting a large amount of footage.

In the case of both rare and peculiar, there were at least two instances of snailfish in the Japan Trench that appeared to lack eyes (Fig. 13AB). The typical black eyes of hadal snailfish are usually very prominent (see Fig. 5J-L; see Jamieson et al. (2023b) for comparison). In this instance, the individuals have body morphologies indistinguishable from Pseudoliparis sp. indet. 2, except for the distinct lack of eyes, nor does it appear to have visible eye sockets. Whether this is an unfortunate imaging obscurity or the result of mutation or disease, we highlight it here in the hope it can be addressed in future studies.

Another example highlighting the importance of physical sampling for definitive taxonomic classification is the identification challenge posed by an unknown metazoan of uncertain phylum, designated here as Animalia incerta sedis (Fig. 11I, Fig. 13C-E). Initially, the authors speculated that this organism might be a nudibranch. This was based on morphological features, such as bilateral symmetry, appendages akin to rhinophores, ‘leaf-like’ cerata that were longer at the anterior, similar to those of Dirona albolineata MacFarland, 1905 (Dironidae). In addition to the authors, several taxonomists (listed in the acknowledgements) were consulted. Some noted that the appendages appeared too rigid to belong to a nudibranch, while others speculated that they appeared to be of “molluscan morphology”, but could not speculate beyond that. Another issue regarding it being a nudibranch was that the depth of the deepest individual (9131 m) in this study was 4722 m deeper (more than double) than the next deepest known nudibranch (4435 m; Valdés (2002)). Other experts speculated that it could be a holothuroid, although they did not resemble any other deep-sea holothuroid, nor did it undulate in the water column in the way benthopelagic holothuroids do.