The Hadal Zone: A Ceiling Predicted, And Then Tested

Pressure rises by about one atmosphere for every ten metres, so full ocean depth is roughly 1,100 atmospheres, 110 megapascals. Temperature is 1 to 4 degrees Celsius throughout, except near hydrothermal vents. The trenches are largely cut off from ocean circulation and hold their own water masses. Food arrives as marine snow drifting down from the surface, plus occasional large carcasses, plus, in some trenches, seismically triggered turbidity currents that sweep sediment and organic matter down into the axis.

Deep-sea animals carry trimethylamine N-oxide, TMAO, which stabilises proteins against being pulled apart by pressure. It rises with depth. Paul Yancey, Mackenzie Gerringer, Jeffrey Drazen, Ashley Rowden and colleagues published the arithmetic in PNAS in 2014. Verbatim from the abstract: "The common osmolyte trimethylamine N-oxide (TMAO) stabilizes proteins against pressure and increases with depth, going from 40 to 261 mmol/kg in teleost fishes from 0 to 4,850 m." They caught five hadal snailfish at 7,000 metres in the Kermadec Trench and measured them: "We found their muscles to have a TMAO content of 386 ± 18 mmol/kg and osmolality of 991 ± 22 mOsmol/kg. These data fit previous extrapolations and, combined with new osmolalities from bathyal and abyssal fishes, predict isosmotic state at 8,200 m."

Our research file says: no fish have been found below about 8,400 metres, a physiological limit hypothesised to relate to TMAO. Yancey's paper keeps those apart and so should we. 8,400 METRES IS WHERE THE OBSERVED ABSENCE BEGINS, the top of the deepest quarter of the ocean where nobody has ever seen a fish. 8,200 METRES IS THE PREDICTED ISOSMOTIC POINT, calculated from measured chemistry. One is a fact about the record of sightings. The other is a calculation about what a body can do. 8,200 and 8,400 are not the same claim.

Our file's summary says the snailfish was "identified at 8,178 m in the Izu-Ogasawara Trench, the deepest fish ever recorded, 2023". Its own section 1.2 says the team FILMED one at 8,336 metres in 2022 and COLLECTED a specimen at 8,178 metres. Those are two different events, a sighting and a capture, and the summary keeps the shallower number and calls it the record. The 8,336 figure is the one printed in the 2023 paper's title. The 8,178 collection depth is carried here as our file's own and was not independently verified for this page.

Hadal communities run on scavenging. Amphipods are the signature animal: Hirondellea gigas and relatives appear at every trench ever surveyed, thousands of them per bait fall, carrying pressure-adapted enzymes and the same TMAO that limits the fish. Xenophyophores, single-celled foraminiferans up to 20 centimetres across, carpet the sediment in some trenches, with species endemic to individual ones. And the microbes are busier at the very bottom than on the plain above: Ronnie Glud and colleagues measured oxygen consumption in Challenger Deep sediment in 2013 and found activity twice that of the adjacent abyssal plain at 6,000 metres, most likely because slope failures funnel organic matter into the axis.

Trenches are separated by thousands of kilometres of abyssal plain at 4,000 to 6,000 metres, which to a hadal-adapted animal is a shallow-water barrier. Jamieson proposed in 2015 that each major trench works as a hadal island with its own amphipod assemblage, and Kermadec and Tonga, adjacent and linked by the Louisville Seamount Chain, do carry distinct species. But Hirondellea gigas turns up in several Pacific trenches. Our file is careful here and it is right to be: molecular work is early, sampling is brutally difficult, and whether trenches are evolutionary dead ends or cradles of speciation is genuinely open.

There are no undiscovered monsters in the trenches, and our file's argument for that is the good one rather than the dismissive one. Organic carbon reaching the trench floor is typically under 1 to 5 grams per square metre per year. That does not feed a large predator. Every hadal animal anyone has found is small, amphipods 2 to 10 centimetres and snailfish 15 to 30, which is what a severely energy-limited place produces. New species turn up regularly and they are invariably small invertebrates, microbes or fish.

Jamieson and colleagues found PCBs, banned since 1979, and PBDEs in amphipod tissue from the Mariana and Kermadec trenches at up to 905 nanograms per gram of lipid, exceeding levels in surface crustaceans from industrialised coastal waters. Peng and colleagues found microplastic particles in Mariana Trench sediment at 10,890 metres in 2018. The routes in are sinking contaminated marine snow, lateral drift from continental margins and bioaccumulation up the food web. What it does to these animals is unknown, and our file says so plainly: they have no evolutionary exposure to synthetic chemicals at all.

Our file's counter-arguments section does not hedge. It says hadal science is in a stamp-collecting phase, heavy on descriptive taxonomy and light on population dynamics, trophic structure, reproduction and larval dispersal. It says sampling bias is severe, with most data from about six Pacific trenches while the Atlantic and Indian Ocean trenches are barely studied. It says baited-camera data is biased toward scavengers and misses anything immobile or uninterested in bait. It notes that the pristine framing was already destroyed by the contamination work, which raises the question of whether deep-sea mining should ever be extended to trenches. And it points out that the hadal zone falls outside virtually every marine protected area and is covered by no specific international treaty.

Correct: the ISBN for Jamieson's 2015 book, and the identifiers for Jamieson on pollutants, Glud on microbial turnover, Linley on hadal fishes, Yancey on the TMAO limit, Peng on microplastics, Fujii on the liparid aggregation, Blankenship-Williams and Levin on trench ecology, Stewart and Jamieson on habitat heterogeneity, and Taira on currents in the Challenger Deep. Every one of those resolves to exactly the work it is attached to, with matching authors, journal, volume and pages. Three are wrong. AND ALL THREE ARE WRONG IN THE SAME WAY.

Entry 6 gives the Nunoura hadal biosphere paper as PNAS volume 115, 2018, DOI ending 1421816115. That returns nothing at all. The real paper is DOI ending 1421816112, PNAS VOLUME 112, ISSUE 11, 2015. A PNAS DOI suffix ends in the volume number, so THE SAME WRONG DIGITS APPEAR TWICE: 112 became 115 in the identifier and 112 became 115 in the volume. That is one slip surfacing in two fields, not two separate mistakes.

Entry 8 gives Don Walsh's memoir piece "In the Beginning..." as Marine Technology Society Journal 46(5):4-6, 2012. The identifier resolves to Robert Weisberg and colleagues, "A Critique of Alternative Power Generation for Florida by Mechanical and Solar Means", in the same journal, the same volume and the same issue, pages 12 to 23. The real Walsh piece is "In the Beginning... A Personal View", Marine Technology Society Journal 43(5):9-14, 2009. AND OUR FILE ALREADY POINTS CORRECTLY INTO THAT ISSUE: entry 11 is in the same MTSJ 43(5) and resolves exactly. One entry finds the right issue and the one next to it does not.

Entry 14 gives Jamieson and Stewart, "Hadal ecology and the identification of new deep-sea environments", Trends in Ecology and Evolution 36(12):1089-1098, 2021. The identifier resolves to a paper on how fruit chemistry shapes seed dispersal, same journal, same volume, same issue, pages 1113 to 1123. The issue is indexed complete, thirteen items running continuously from 1061 to 1163, and THERE IS NO SLOT AT 1089-1098: page 1089 sits inside an article running 1083 to 1092, page 1098 inside one running 1093 to 1101. The range straddles two real articles and matches neither, and there is no hadal paper in the issue. A title search finds nothing matching. Jamieson and Stewart are real and publish together often; no paper of this title was found.

Four failures across two research files audited on the same day, and not one of them is wrong at random. In every case the journal is right. In three of the four the volume is right too. In two of them the issue is right as well, and the identifier points at an article a few pages away. THESE ARE NOT RANDOM WRONG NUMBERS. They are what you get from a process that knows where a paper was published and does not know which paper it is. That matters for repair, because it means the venue is usually salvageable even when the pointer is not.

Entry 11 is a correct citation, but our file gives its first author as Lisa Blankenship-Williams. The record gives Lesley E. Blankenship-Williams. The second author is Lisa A. Levin. THE GIVEN NAME OF THE SECOND AUTHOR APPEARS TO HAVE MOVED ONTO THE FIRST. That is the third wrong author name found in this series, and it is the first one where the mechanism is visible in the citation itself.

Entry 9 is Lacey and colleagues, "The Mariana Trench", Oceanography 33(2):56-57, 2020, with no identifier at all. A title search returns nothing matching. THE PAGE-RANGE TEST DOES NOT WORK HERE, because the registry's records for that issue mostly carry no page numbers, so the volume cannot be checked for a gap. UNVERIFIED IS NOT DISPROVEN. This entry is recorded as unchecked, and it would be dishonest to file it with the three above.