For over a decade, marine geochemists have measured the respiration rates of seafloor sediments to estimate how much carbon is buried or released. These numbers feed into global carbon budgets, climate models, and our understanding of ocean productivity. But a new analysis of 20 published studies reveals that a single, seemingly minor procedural choice — how sediment cores are handled with respect to oxygen — systematically warped the results in 13 of them. The bias, which can shift rates by a factor of two to five, has likely propagated through the literature, affecting everything from regional carbon balances to global model calibrations.

The finding, published in Nature Geoscience, is a stark reminder that the devil in Earth science often hides in sample handling, not in the grand theory. And it raises uncomfortable questions about how funding structures and publication incentives have allowed such a basic methodological flaw to persist.

A Hidden Variable Skewed a Decade of Seafloor Carbon Data

The study, led by researcher Laura M. Treude at the Max Planck Institute for Marine Microbiology, compared respiration rates from sediment cores processed under different oxygen protocols. The team reanalyzed data from 20 peer-reviewed papers published between 2012 and 2023, covering sites from the abyssal plains of the Pacific to coastal margins of the Atlantic. In 13 of those studies, the cores had been exposed to ambient air during subsampling or initial handling — a step that introduces oxygen into an environment where it is normally absent below the top few millimeters.

When oxygen penetrates deeper, it can trigger a metabolic shift in the microbial community. Aerobic respiration kicks in, inflating the measured oxygen consumption rate relative to what would occur in situ. In some cases, the rates doubled; in others, they increased fivefold. The effect was strongest in sediments with high organic content, where microbes are primed to respond quickly to oxygen availability.

The finding does not mean the original studies are worthless, but it does mean their absolute respiration rates may be systematically high. For scientists trying to close the marine carbon budget — which currently has a mismatch between estimated inputs and outputs — this bias could explain part of the discrepancy. As Treude put it in a press briefing, “We’re not saying the papers are wrong; we’re saying the numbers need to be re-evaluated with this protocol effect in mind.”

The work echoes a broader reproducibility crisis in science, but with a twist: here the error is not in statistical analysis or p-hacking, but in a routine physical step that no one thought to standardize. As one commenter noted, it is a “quiet bias” that crept in because fieldwork constraints often push researchers to cut corners.

How a Simple Lab Step Introduced a Pervasive Bias

To understand the bias, one must picture the typical workflow for measuring sediment respiration. A core is retrieved from the seafloor — often by a multicorer that preserves the sediment-water interface — and then brought to the ship’s lab. There, subcores are taken, and the sediment is incubated to measure oxygen consumption over time. The critical moment is when the core is opened and subsampled. If this is done in air, oxygen diffuses into the sediment, especially if the core is not immediately capped or if the overlying water is drained.

Microbes in the sediment are adapted to low-oxygen or anoxic conditions. When suddenly exposed to oxygen, they can ramp up respiration within minutes. This “oxygen priming” effect has been known in soil science for decades, but its impact on marine sediments had not been systematically quantified until now. Treude’s team performed controlled experiments comparing cores handled under anoxic conditions (in a nitrogen-filled glove bag) with those exposed to air. The difference was dramatic: respiration rates in air-exposed cores were consistently higher, sometimes by 300%.

The 13 affected studies all used protocols that involved some air exposure — whether during slicing, sieving, or loading incubation chambers. The seven studies that avoided the bias either used strict anoxic handling or had very short exposure times (under 30 seconds) that minimized oxygen intrusion. The variation across labs was wide: some groups used glove bags, others worked on a bench top with a fan blowing, and a few even bubbled air through the overlying water during incubation, inadvertently increasing oxygen penetration.

The lack of a standard protocol is partly historical. Marine sediment respiration measurements evolved from limnology and soil science, each with its own conventions. International programs like the Joint Global Ocean Flux Study (JGOFS) provided guidelines, but they did not prescribe oxygen handling in detail. As a result, each lab developed its own “standard” method, and peer reviewers rarely questioned the step.

To illustrate the diversity, consider three contrasting cases. One study from the Baltic Sea used a nitrogen-flushed glove bag for all subsampling and reported respiration rates roughly 0.2–0.4 mmol O₂ m⁻² d⁻¹. A second study from the same region, but using bench-top handling with a fan, reported rates of 0.7–1.0 mmol O₂ m⁻² d⁻¹ — nearly three times higher. A third study, from the oligotrophic South Pacific Gyre, employed a brief air exposure of about 45 seconds and obtained rates of 0.05–0.08 mmol O₂ m⁻² d⁻¹. When Treude’s team reanalyzed the Baltic Sea bench-top data with anoxic corrections, the adjusted rates fell to 0.3–0.5 mmol O₂ m⁻² d⁻¹, bringing them in line with the glove-bag study. The South Pacific data, however, showed minimal change, likely because the low organic content limited microbial response. These examples highlight that the bias is not uniform and depends on local sediment characteristics and handling details.

The Funding and Incentive Structure That Let It Slide

Why did this bias persist for over a decade? The answer, according to many scientists interviewed for this article, lies in the incentive structure of research funding and publishing. Grants are awarded for novel hypotheses and exciting results, not for method harmonization. Researchers are rewarded for collecting data from new sites or under extreme conditions — deep-sea vents, polar regions, oxygen minimum zones — rather than for replicating earlier work or testing methodological assumptions.

Peer review, the gatekeeper of scientific quality, has also failed to catch this issue. Reviewers typically focus on the scientific narrative and statistical analyses; sample handling details are often skimmed or assumed to be standard. A survey of 50 published papers on sediment respiration found that fewer than half described their oxygen handling protocol in enough detail to replicate it. Without explicit reporting, the bias remained invisible.

Infrastructure costs play a role too. Setting up anoxic handling facilities — glove bags, nitrogen lines, oxygen sensors — adds thousands of dollars to a shipboard lab budget. For a typical research cruise costing $30,000–$50,000 per day, any extra equipment or time is a hard sell. Principal investigators, already stretched, may choose to forgo the precautions, rationalizing that “everyone does it this way.”

Publication pressure compounds the problem. Early-career researchers, postdocs, and graduate students are under immense pressure to produce results quickly. If a method works well enough to get a paper accepted, there is little incentive to question it. As one postdoc, who asked not to be named, told me: “You don’t want to be the one who says, ‘Wait, maybe our data are wrong.’ That doesn’t get you a Nature paper.”

The result is a system that inadvertently rewards methodological sloppiness as long as it yields publishable numbers. Treude’s study is a corrective, but it also highlights how difficult it is to change entrenched practices without top-down pressure from funders or journals.

What the Correction Means for Global Carbon Budgets

If 13 of 20 studies overestimated respiration rates, what does that mean for the global carbon budget? The marine sediment respiration flux — the amount of organic carbon consumed by microbes on the seafloor — is a key term in the ocean carbon cycle. Current estimates place it at roughly 0.5–1.5 petagrams of carbon per year, but the uncertainty is large. A systematic overestimate of 20–50% would shift the budget significantly, potentially requiring revisions to how much carbon is buried versus recycled.

Ocean carbon models, which simulate the cycling of carbon from surface waters to the deep sea, use these respiration rates as input parameters. If the rates are biased high, models may predict too much carbon remineralization in the water column or too little burial in sediments. That, in turn, affects projections of how much anthropogenic CO2 the ocean can absorb in the coming centuries.

Three marine labs — at the University of Washington, the Alfred Wegener Institute, and the Japan Agency for Marine-Earth Science and Technology — have already begun reanalyzing their past data using corrected protocols. Preliminary results suggest that the bias is not uniform: it is larger in organic-rich coastal sediments and smaller in oligotrophic abyssal plains. A global reanalysis, led by Treude’s group, is underway and expected to be published in 2027.

But the correction is not straightforward. Simply subtracting a percentage from previous measurements is not valid, because the magnitude of the bias depends on local sediment properties, oxygen exposure time, and microbial community composition. Each study may need to be re-evaluated individually, a painstaking process that requires access to original core samples — many of which have been discarded.

As a concrete example, consider the dataset from the California Margin. A 2019 study reported respiration rates of 1.2–1.8 mmol O₂ m⁻² d⁻¹ using bench-top handling. Treude’s team reanalyzed archived cores from the same site using anoxic protocols and obtained rates of 0.6–0.9 mmol O₂ m⁻² d⁻¹ — a 40–50% reduction. In contrast, a 2021 study from the abyssal North Pacific, which used a glove bag, reported 0.08–0.12 mmol O₂ m⁻² d⁻¹, and reanalysis confirmed those values within 10%. These site-specific corrections will be essential for updating regional carbon budgets, such as those for the California Current System, where the bias may have led to overestimates of benthic carbon cycling.

The uncertainty range for benthic respiration may widen before it narrows. As Treude noted, “We now know that our past numbers are less certain than we thought. That is uncomfortable, but it is also progress.”

Lessons for Fieldwork and Data Reusability

The sediment respiration case is not an isolated incident. Similar methodological blind spots have emerged in other fields. A study of antibiotic susceptibility testing found that using a single unindexed vial warped 14 of 20 curves (related article). In neuroscience, an uncalibrated optogenetic laser swapped 12 of 20 place cell maps (related article). And in exoplanet science, a single filter choice skewed 14 atmosphere spectra (related article). These examples share a common theme: a small, unexamined procedural detail can propagate through the literature, undermining the reliability of downstream analyses.

What can be done? One proposal gaining traction is the creation of open-source protocol registries, where researchers deposit detailed step-by-step methods before data collection begins. Such registries would allow reviewers and readers to check for potential biases. A pilot registry for marine sediment respiration is being developed by the Scientific Committee on Oceanic Research (SCOR) and could be operational by 2028.

Low-cost modifications can also prevent oxygen artifacts. Simple changes — such as using a nitrogen blanket during subsampling, reducing exposure time, or capping cores immediately — can make a big difference. Treude’s team has published a set of “best practice” guidelines that cost less than $500 to implement on a typical research vessel.

Long-term observation networks, such as the Ocean Observatories Initiative and the European Multidisciplinary Seafloor and water column Observatory, are beginning to audit their methods. They now require that all sample handling steps be documented in a machine-readable format, so that future researchers can assess the potential for bias. This is a step toward data reusability — a key goal of the FAIR data principles.

Funding agencies, including the U.S. National Science Foundation and the European Research Council, are starting to require method checklists in grant proposals. The idea is to force researchers to think about potential biases before they go to sea. As one program officer put it, “We don’t want to fund data that will be questioned a decade later.”

But there are limits. Standardization can stifle innovation, and some researchers argue that rigid protocols are not suitable for the diverse conditions found in the ocean. A method that works in the cold, organic-rich sediments of the Baltic Sea may not be appropriate for the hot, carbonate sands of the Bahamas. The challenge is to develop flexible guidelines that allow for site-specific adaptations while maintaining core principles.

For instance, a one-size-fits-all rule like “always use a glove bag” may be unnecessary in low-organic abyssal sediments where oxygen exposure has minimal effect. A more nuanced approach would require researchers to document oxygen exposure time and sediment organic content, and to include a control experiment to assess the magnitude of the bias at their specific site. This would allow the scientific community to apply corrections retroactively, rather than discarding valuable data.

In the end, the sediment respiration story is a cautionary tale about the hidden costs of scientific convenience. It shows that even the most mundane decisions — how you open a core, how long you leave it exposed — can ripple through the literature, shaping our understanding of the planet. The fix is not glamorous, but it is essential: slow down, document, and question every step. As Treude said, “We need to make our methods as robust as our theories.”