Dark Oxygen Production (DOP) was a hypothesis published by Sweetman et al. (2024) to explain anomalous data from experiments in the Clarion-Clipperton Zone, Pacific ocean. These experiments recorded rising oxygen for the first few hours, when previous experiments of the same type had only ever recorded reducing oxygen due to consumption by benthic organisms. In order to explain this phenomenon, Sweetman et al. claimed that manganese nodules, which were present in many of the experiments, must be responsible for the oxygen production. It was thus given the name "dark oxygen production", borrowing the term "dark oxygen" that had previously been used for oxygen production by bacteria without sunlight. Sweetman et al. claimed that in this case the oxygen was being produced by electrolysis, using measured voltages from the surface of the nodules as evidence. The article, published in Nature Geoscience, was widely reported in mainstream media and social media, and contributed to the ongoing debate around deep sea mining. However, it has since emerged that there are many problems with the theory, and may even be an example of scientific misconduct since contrary results and pertinent information were omitted from the article. The omitted data showed that manganese nodules could not have been responsible for the phenomenon. A collaborative rebuttal published by leading scientists and industry experts instead claims that the rising oxygen levels are just experimental artefact, and not due to any natural phenomenon. Sweetman et al. have publicly stepped back from the two main conclusions in the article; that manganese nodules produce oxygen, and that they do this by electrolysis. Retraction of the paper has been called for by independent scientists.
On the 8th of April, 2026, Nature Geoscience posted an Editor's Note on the article, stating: "Readers are alerted that aspects of this article are subject to concerns that are being considered by the Editors. A further editorial response will follow the resolution of these issues."
The hypothesis
Benthic chamber landers are a form of benthic lander where a box, or chamber, is lowered to the seafloor and partially penetrates the sediment, creating a sealed volume of water which is then separate from the surrounding water column. In this chamber, various sensors and sampling devices are used, including oxygen sensors which measure the uptake of dissolved oxygen from the seawater by organisms within the chamber. However, Sweetman et al. published data from benthic chamber landers that appeared to show oxygen rising in many of their chambers during the first few hours of the experiment. Subsequently, further experiments were done by taking samples of the seafloor and incubating them in the lab, and these appeared to confirm rising oxygen in the presence of manganese nodules. This led to the hypothesis that manganese nodules were somehow causing or producing the rising oxygen. The mechanism was not explained until voltages were measured on the surface of the nodules, and a reported voltage of 0.95 V was used as evidence that the manganese nodules were splitting water via electrolysis, producing oxygen and, presumably, hydrogen (although no data on hydrogen was released). And so the theory was formed, that manganese nodules can split water by electrolysis and produce oxygen, and this was published in Nature Geoscience in 2024. The article made bold claims about how this could influence our understanding of the oxygenation of the Earth, the emergence of life and even life on other planets. It quickly attracted a lot of attention, and the article was widely reported, becoming one of Nature Geoscience's most read articles.
Rebuttals
The Dark Oxygen hypothesis quickly attracted doubt from industry leaders and independent scientists, stating numerous issues with the research. Some of the initial issues included that no energy source for electrolysis had been identified, and no other studies had observed the same phenomenon despite being conducted in similar places and in the presence of manganese nodules. Shortly, pre-print rebuttals emerged which brought solid claims against the findings, including that crucial data was omitted from the article in order to support the dark oxygen hypothesis. Direct rebuttals to the article are summarised here.
Downes et al. 2024 revealed that some of the lander experiments conducted in 2018 did not contain manganese nodules, and yet still recorded rising oxygen. This directly contradicts the hypothesis that manganese nodules were responsible for the rising oxygen. They also revealed a control experiment where rising oxygen was recorded in the absence of sediment and nodules, further contradicting the hypothesis. Nakamura 2024 noted that there is no "missing" oxygen source in the deep sea that would require such a process to exist, and there isn't elevated oxygen above manganese nodule fields. Tengberg et al. 2024 also highlight that earlier experiments that recorded increasing oxygen did so without nodules present. They further analyse the data from all the benthic chamber landers and find that only 2 out of 32 might have produced reliable data. They note that rising oxygen is a known experimental artefact, and an indication that the data from that chamber should be discarded. They explicitly call on Nature Geoscience to retract the article. Cuesta and Jaspars, 2025 go into great detail about the thermodynamics of the electrolysis hypothesis, concluding that it violates the 1st and 2nd laws of thermodynamics. Trellevik et al. 2024 note many flaws in the data that are apparent, including inconsistencies in the oxygen measurements and that the voltages recorded were not sufficient to split water.
Subsequently, many of the authors involved in the rebuttals collaborated to produce a comprehensive and peer reviewed critique of the original article, published in Frontiers in Marine Science in 2025.
In August 2026, a research article was published that re-analysed some data from Sweetman et al. (2024). It noted that there is one key piece of evidence that linked polymetallic nodules to oxygen production, which was a correlation between nodule surface area (as indicated from photographs of the benthic chambers) and total oxygen production. However, the authors noted that this correlation is incompatible with the observation that DOP was seen in chambers that didn't include any nodules at all. They also noted that three chambers appeared to have been excluded from the surface area analysis of Sweetman et al. (2024) even though the oxygen production rates from those chambers were published and the chamber photographs were available. When including these three additional chambers in the analysis, the correlation was found to disappear, and the authors question why these chambers were excluded from the analysis of Sweetman et al. (2024). They also include an analysis of the weight of nodules collected from the benthic chambers after recovery. They note that there is no correlation between nodule weight and oxygen production. The authors conclude the oxygen concentration increases are not related to polymetallic nodules and they question the experimental timeline, where chambers without nodules were seen to experience oxygen concentration increases as early as 2018.
Grounds for Retraction
Retraction of the paper has been called for. Nature Geoscience, who published the article, are a member of the Committee on Publication Ethics (COPE). They say that retraction of a paper may be warranted if "the work contains such seriously flawed or erroneous content or data that its findings and conclusions cannot be relied upon". The rebuttals indicate that several severe methodological flaws are present in paper, including that oxygen rose in chambers without nodules, and that oxygen concentration measurements were unreliable and indicate contamination. These lead to the conclusion that the findings of the paper cannot be relied upon, since they show that manganese nodules cannot be responsible for the rising oxygen, and that the rising oxygen is just experimental artefact.
Another reason given by COPE for retraction is for any form of misrepresentation, including data falsification, such as "omitting data or results so the research is not accurately represented in the record". The rebuttals indicate that at least two important pieces of information were omitted - that oxygen rose in experiments without manganese nodules present, and that oxygen rose in a control experiment conducted without nodules. These two pieces of omitted information invalidate the main claim of the article, that manganese nodules are responsible for the oxygen concentration increases.
Problems
This is a list of some of many issues raised with the research.
Some landers did not contain manganese nodules. In probably the leading problem for the hypothesis, which relies on manganese nodules being present, some of the lander experiments conducted in 2018 either didn't contain any nodules or only contained small manganese oxide particles: "No nodules were collected as they were either not present or were too small to be relevant". Manganese oxide particles are extremely common in deep sea oxic sediments. The fact that some of the landers did not contain nodules was not discussed or disclosed in the article, and the research that states this fact was not cited. Another paper describes the lack of nodules at one of the sites: "no nodules were observed in APEI-7" (referring to one of the areas where DOP was measured). Since some of the landers that recorded increasing oxygen didn't contain manganese nodules, then manganese nodules cannot be responsible for the rising oxygen. The timeline of events suggests a deliberate decision to attribute dark oxygen production to manganese nodules. The landers that recorded rising oxygen without manganese nodules were conducted in 2018, and yet further work was conducted on at least four separate expeditions in 2021, 2022 and 2023, including shipboard incubation of manganese nodules. The fact that rising oxygen was observed without nodules in 2018 was seemingly disregarded during this further work, as other scenarios such as sediment-only incubations were not conducted, or if they were then they were not presented. This demonstrates a bias towards finding manganese nodules responsible for rising oxygen, when evidence pointed to the contrary. Furthermore, the lack of citation for the article that stated there were no nodules present, and the lack of discussion of this point, demonstrates a concerted effort to hide the fact that oxygen rose without nodules present. This has been suggested to be an example of data falsification, where data is manipulated, cherry picked or omitted to support a certain conclusion - a form of scientific misconduct. Control experiments were omitted. On at least one occasion, the benthic chamber lander was lowered to just above the seafloor, the chamber doors were closed, and the incubation proceeded with just seawater present. Two of the three chambers recorded rising oxygen levels in this scenario. This control experiment ruled out sediment and manganese nodules causing the rising oxygen. However, this control experiment was not included in the article by Sweetman et al., and thus important information that rejects the primary hypothesis was omitted. The findings are anomalous in comparison to previous work. The paper states "Our findings contrast with all published deep-sea benthic O2 flux studies" and yet this inconsistency is not expanded or commented upon. No attempt is made to explain why rising oxygen was seen in this case but not in others, despite the fact that anomalous results are an indication of experimental artefact. See previous work for more details.