Dark, Dark Oxygen
How a Nature paper commissioned by its own detractors and the credulous science reporting that followed sank the prospects of deep-sea mining.
Setting: The Abyssal Plain of the Clarion-Clipperton Zone
I have nothing against the bacteria living on the seafloor of the abyssal plain. All else being equal, I would love for these bacteria to live out their idyllic life, at a crushing five-thousand-meter depth under the sea. However, if you forced me to choose between, say, the Springboks of South Africa or the Wallabies of Australia, and the microbiome of the abyssal plain of the Pacific Ocean, I can safely say I’d side with kingdom Animalia over Archaea.
This was not the choice taken by the International Seabed Authority. Established by the UN’s 1982 Convention on the Law of the Sea, and operationalized a decade later, the ISA is an intergovernmental organization to which most countries (notably not the United States) are party. The ISA has jurisdiction over the development of deep-sea mining in international waters. For about a third of a century, the ISA has convened on an annual basis, mostly achieving mind-bogglingly incremental progressions in the language used in their reports, deliberations, and resolutions. The kind of achievements the ISA touts generally sound something like, “the delegates agreed that a mutual acknowledgement of the common heritage of the international seabed should lay the groundwork for the implementation of a global framework for the shared protection of this critical resource.”
Of course, the ISA was originally tasked to hash out an international agreement on how nations might share and mine resources on the seafloor. It has not done so. In thirty years, it has managed to permit a handful of ventures to conduct exploration of these resources, with basically no further progress towards commercial mining.
Protecting the deep seafloor, one of the last untouched reaches of our world, certainly seems admirable. However, most are of the opinion that to limit the environmental footprint of resource extraction, deep-sea mining is substantially preferable to the alternative: mining on land. The most widely proposed approach to seabed mining is surprisingly minimally disruptive. It involves robotically harvesting polymetallic “nodules” (basically little rocks) that are sitting on the seafloor. Robots would collect these nodules and bring them to the surface. There aren’t any proposed methodologies that would involve large-scale subsurface drilling, for example. Gathering little stones on the seafloor seems vastly preferable to large open pits on land that generate large volumes of toxic and non-toxic waste. And these polymetallic nodules would be great sources of cobalt, manganese, copper and other elements that are very useful to the buildout of renewable energy infrastructure: solar and wind infrastructure and especially batteries for energy storage.
Opposition to seabed mining focuses on the disruption to the environment. It may cause noise and light pollution, dust generation, and physical interactions between the machinery and organisms. But the standards to which seabed mining is held are outrageously high, a bar that most human practices in the oceans would fail to clear. We kill and eat trillions of fish per year (bottom trawling for fish is a similar process to the one proposed for seabed mining, but far more widespread, invasive, and damaging to ocean life). We dump plastics into the oceans with impunity. We drill for oil in the oceans, another process far more disruptive in virtually every way than harvesting polymetallic nodules from the abyssal plain!
The opposition often exhibits a total refusal to reckon with the opportunity cost of failing to expand deep-sea mining. In its place, there will be an expansion of conventional mining, as well as the less efficient and more environmentally disruptive harvesting of less mineral-dense resources in national waters, which are not subject to the UN’s Convention on the Law of the Sea. Both of these types of ventures will be far more damaging to ecosystems much richer in biomass than the inhospitable depths of the ocean.
Act One: The Canadian Miners
Canadian firm, The Metals Company, is one of several stakeholders that would really like to be able to harvest these nodules. Here’s how they describe the process on their website.
“The Metals Company plans to lift polymetallic nodules to the surface, take them to shore, and process them with near-zero solid waste, no tailings or deforestation, and with careful attention not to harm the integrity of the deep-ocean ecosystem.
Our production process begins with the collector, designed to collect polymetallic nodules from the abyssal seafloor. Over 90% of the entrained sediment is expected to be separated from the nodules inside the collector and discharged behind it, with most sediment settling back to the seafloor within a few hundred meters. From the collector, nodules travel up a riser system to our production vessel. Once aboard, nodules get dewatered and residual water, sediment and nodule fines will be returned below the photic zone to a depth scientifically chosen to have minimal impact on the collection area.”
Sounds reasonable? From where I stand there are primarily two concerns with this approach, with regard to environmental harm. Firstly, there’s the sediment generation: the sediment could disrupt, settle on top of, get consumed by, or interfere with the sensing of organisms in the ocean. However, I think the sediment generation from this deep-sea mining approach is very likely to be minimal compared to natural sources of sediment generation. A single seismic event on the seafloor would generate, realistically, several orders of magnitude more sediment and dust than these mining ventures could hope to produce.
Secondly, you have the concern that these polymetallic nodules are relevant for the ecosystems on the seafloor. Now, realistically, only a very small fraction of these nodules will end up being mined—the oceans are massive. And it’s unlikely that there are, say, organisms living directly on these nodules, or feeding off them or something. The surface of mixed metal oxide ore is not a particularly hospitable environment. The Metals Company surely thought that a rigorous scientific analysis of the role of these nodules on the seafloor would only benefit their case. So, they funded research to look into this. Oops!
Act Two: The Research Collaboration
Researchers from The Scottish Association for Marine Science, Boston University, Northwestern University (I actually worked down the hall from one of the researchers on this collaboration), and several other UK and German institutions collaborated for several years to investigate the potential production of oxygen from these polymetallic nodules. They put out a paper in Nature Geoscience last July.
Here’s my TL;DR of that paper:
They wanted to investigate whether the polymetallic nodules that mining companies want to harvest are implicated in the production of something called “dark oxygen”, potentially abiotically-produced oxygen that exists at the seafloor (really within the sediment of the seafloor) at (maybe) higher-than-expected concentrations, and thus supports the limited ecosystems down there (microbes). To do this, they lowered sophisticated apparatuses, or “benthic chambers”, to the bottom of the ocean, sealed them over nodules sitting on the seafloor, and used optical sensors to detect the change in oxygen concentration. They claim to have detected increasing oxygen levels which would indicate that these nodules are indeed producing oxygen.
After I read over the paper last summer, I had some serious concerns. I don’t want to be overly critical or harsh to the researchers. They likely have far more domain expertise than I do in the field of benthic measurements. This type of experiment is also unbelievably challenging to do and requires a ton of coordination. There’s also, as always, a distinct possibility that I’m missing something important, and if anyone reading this knows what that is, please point it out to me! If someone went through one of the papers I’ve written, piece by piece, and pointed out every possible flaw, I’m sure they’d have a lot to find. But my papers also haven’t been directly used by a United Nations body to make a decision that affects global supply chains, and if they were, they should face similar levels of scrutiny.
Disclaimers aside, this paper should never have been published in the form that it’s in, let alone in a journal as prestigious and known for scientific rigor as Nature Geoscience. Let’s dig in.
Say you wanted to know whether a certain kind of rock at the bottom of the ocean leads to higher oxygen levels that can support an ecosystem down on the abyssal plain. Questions you might have are:
Do areas of the seafloor with polymetallic nodules have higher oxygen concentrations than areas without polymetallic nodules? And if so, is this because high oxygen concentrations cause the formation of the nodules, or because the nodules cause the generation of oxygen?
How much oxygen are polymetallic nodules responsible for, vs other sources of oxygen in these ecosystems? Are they contributing 0.01%, 1%, or 100% of the (excess) oxygen levels observed at the seafloor, or in sediments at varying depths below the seafloor?
Do polymetallic nodules produce more oxygen than other rocks/sand/sediment at the seafloor?
This paper answers none of these questions. It doesn’t actually report any ambient oxygen levels at the seafloor, in fact. It reports oxygen levels within the artificial environments of the sealed benthic chambers.
It doesn’t estimate how much oxygen is being created, or what fraction of the known oxygen in these environments are the polymetallic nodules responsible for.
But worst of all, it doesn’t present any real controls! The paper executes a sort of sleight of hand, where they carry out an experiment where they introduce small amounts of dried algae, saline solution, and cold seawater from the surface through injections into the benthic chambers. And then they also have “controls” where they don’t do any injections. But this isn’t the control that anyone would be interested in, and it isn’t relevant for supporting their claim that these nodules are producing oxygen. And they still find that there’s no significant difference between any of these conditions.
They also, mysteriously, present a “control rock” which is made of metamorphosed carbonate, solely for their measurements of the voltage potentials at the surfaces of the nodules. So we know they did test controls, but they just didn’t… report the change in oxygen levels from those measurements in their paper?
The most obvious experiment one might contrive is something like this:
Place benthic chambers over a dozen nodules. Place identical benthic chambers over a dozen locations without nodules. Plot the average change in oxygen concentration for the nodules against the controls. See if there is a difference. If you’re being even more rigorous, you’d probably want to do this with several types of controls: bare sediment, other non-nodule rocks, the exact location where the nodule was after you’ve removed the nodule, the seawater ten meters above the seafloor, etc.
Instead, the researchers reported none of those controls. They also presented the change in oxygen concentrations that they claim the nodules are responsible for in a really bizarre and confusing way. I hate to be nitpicky, but the graphs are poorly formatted, the samples are labeled things like “AKS286-Ch.2”, the color scheme looks like it was selected deliberately to piss off anyone trying to distinguish between the samples, and there are multiple compounding data artifacts clearly visible, which they don’t fully explain. Most critically, they provide no explanation for the incomprehensible labeling scheme for their samples, and you have to dig into their source data to even understand what each sample label is referring to.
Aside: For a 16-author collaboration that required multiple oceanic cruises to gather the data, you’d think the least they could do would be to spend more than 15 minutes plotting the data before submitting it to Nature. It’s actually bizarre to me that Nature Geoscience’s peer reviewers didn’t force them to re-label their data before accepting the paper. And I’m not exactly someone with a high regard for the peer review process.
None of these issues stopped this research from going viral.
Exeunt Scientists, Enter Journalists
After the paper came out, it got a lot of attention:
As is typical with coverage of new, high profile science, the actual findings of the paper are exaggerated by a substantial factor, there’s little nuance in dealing with the uncertainty of the results, there’s no effort to place these results into the recent history of the field, and the journalists essentially accept everything the authors say in their interviews without pushback or consulting others with different viewpoints.
Even from some of the better, and slightly more skeptical press coverage, the quotes from the authors seem highly ideological about mining in ways that their experiments are not at all empowered to support. From Scientific American…
These findings are “another thing that we now need to take into account when it comes to deciding, ‘Do we go and mine the deep ocean, or don’t we?’” Sweetman says. “To me, that decision needs to be based on sound scientific advice and input.”
…and from the Washington Post:
But if that finding is borne out, “we need to rethink how to mine” materials like cobalt, nickel, copper, lithium and manganese underwater, “so that we do not deplete the oxygen source for deep-sea life,” said Franz Geiger, a professor of chemistry at Northwestern University and one of the co-authors of the study, in the release.
Further on in the Washington Post piece, Geiger claims that the benthic sea floor contains more faunal biodiversity than tropical rainforests. I looked into this claim and found no unambiguously supporting evidence for this. There was a study from the 1960s and another from the 1980’s that make slightly weaker claims, comparing the deep-sea floor to a rainforest. However, both primarily describe sampling methods that involve trawling with a sled across hundreds of meters of seafloor, where they find hundreds of species worth of microbial organisms, and extrapolate some biodiversity metric from this. I don’t think this sampling method is analogous to something that could be done on land, but I’d be willing to bet a lot that you could find far more than hundreds of species if you were able to trawl across hundreds of meters of soil, dirt, shrub, and tree in a tropical rainforest. These species might also be more interesting and worthy of preservation than nematodes, but perhaps my bias for organisms visible to the naked eye is starting to creep in here. Regardless, “biodiversity” as a single figure of merit that an ecosystem can have “more” or “less” of is not a rigorously defined property.
Now, none of this is the authors’ fault. If I were interviewed about my research by the Washington Post, I’d make similar types of statements. It’s the job of the science journalists writing the piece and their editors to recognize what is real and what is embellishment.
Because if they don’t… John Oliver releases a 23 minute long diatribe against deep-sea mining, filled with far too many misleading claims to debunk in a single substack post. Most of these claims are in the form of talking points that have been passed through some crude game of telephone from the original research to some C-tier online publication with a name like “Global Science Updates Monitor” or “Climate Express News”.
Act Three: Defeat at the ISA
International delegates, buoyed by activist organizations touting the study as critical evidence for the harms of deep-sea mining, raised a storm during negotiations in 2024. Just a couple months prior, there was talk of momentum finally leading to a framework by which countries could begin transitioning from exploration of resources to commercial mining. Negotiations for this had been many years in the making, and most thought it was only a matter of time before the ISA got down to business.
Instead, immersed in the firestorm of publicity around the Dark Oxygen study, delegates elected the most staunchly anti-mining Secretary-General in their history, ground negotiations to a halt, and ended 2024 with the position of deep-sea mining looking more tenuous than before.
Look, there’s no such thing as free lunch. Rather than mining deep-sea deposits, mining firms will simply expand mining operations on land in countries like the DRC, Brazil, China, South Africa, and Australia. It’s hard to overstate this, but the environmental and human rights harms from cobalt mining in the DRC, for example, far outweigh the potential damages to bacterial life at the seafloor. The ISA could choose to, for example, preserve the biodiversity of these deep-sea ecosystems by preserving 80% while opening the other 20% to mining ventures. But they have not done so.
Aside: The United States is not a party to the UN Convention on the Law of the Sea. I think they should seriously evaluate the possibility of unilaterally mining cobalt-rich nodules from the deep-sea floor of the Clarion-Clipperton Zone, to reduce their reliance on conflict minerals.
The Metals Company was not happy with any of these proceedings. They released a formal rebuttal of the Nature Geoscience paper that they had funded (with an attached preprint).
They hit some of the same points I mentioned earlier, as well as a few more that are quite concerning. In their own words:
“To summarise, Sweetman et al. rely on two main lines of evidence to demonstrate the production of dark oxygen: 1) oxygen production from nodules present in the benthic chamber landers and 2) oxygen production during ex situ core incubations. However, the data that they omit from their study invalidate both lines of evidence. They omit and selectively report pertinent metadata and data from a previously published study and additional experiments conducted as part of this study that show oxygen concentrations rose in chambers that contained no nodules. Furthermore, they omit contemporaneous knowledge of bottom water oxygen concentrations in NORI-D and that T0 oxygen concentrations in ex situ core incubations did not represent this NORI-D bottom water. With these lines of evidence invalidated, the hypothesis that nodules can produce oxygen on the abyssal seafloor is completely unsupported.”
Notably, TMC seems to have found that the researchers selectively reported data and may have used duplicated data from samples without nodules present. If this is true, this would likely constitute serious scientific fraud.
They also bring up that the voltage potentials measured in the study are insufficient to allow for the hydrolysis of seawater into hydrogen and oxygen, a fact that was somewhat occluded by the confusing way that the researchers presented the data in the paper.
Other researchers have flagged similar concerns. In particular, this rebuttal preprint from researchers from the University of Gothenburg is pretty definitive, and if you’re not already persuaded, I’d consider reading it. Unfortunately, it’s far too late. The research achieved its goal: putting a damper on deep-sea mining for the foreseeable future. The new Secretary-General of the ISA will serve out a four year term, during which it seems quite unlikely that the ISA will make considerable progress towards commercial mining licenses. During those four years, the high costs and challenging supply chains for elements like cobalt and manganese will limit the build-out of clean energy infrastructure, setting back the fight against rising CO2 emissions and continuing to provide incentives for the unsavory status quo in cobalt mining practices. Hopefully, expanding capabilities for the recycling of elements like cobalt and nickel at high efficiencies from used batteries will help pick up the slack and set us on a path towards a more circular economy for minerals, but as I see it, this whole affair was a massive own-goal for the climate movement and efforts to make global supply chains more sustainable.