The distinction matters because soils hold more carbon than the atmosphere and vegetation combined, much of it in mineral-associated organic matter, or MAOM. This carbon can remain protected for centuries to millennia when organic molecules become attached to clay minerals, iron compounds and other reactive surfaces. Yet soil is not chemically static. Rainfall, flooding, drainage, root activity and microbial respiration can repeatedly switch microscopic environments between oxygen-rich and oxygen-poor states. Those changes alter the oxidation state of iron, the solubility of minerals and the ability of microorganisms to break down organic compounds. When iron-bearing minerals are reduced, they may dissolve and release previously protected carbon. When oxygen returns, minerals can re-form and potentially capture carbon again. The new experiment tested whether these redox oscillations—the chemical equivalent of repeatedly flipping an oxygen switch—interact differently with carbon molecules that have contrasting properties.

The researchers chose glucose and oxalate because they represent two common forms of low-molecular-weight carbon released by plants and roots. Glucose is relatively chemically reduced and yields substantial energy when microorganisms oxidize it to carbon dioxide. Oxalate, an organic acid with two carboxyl groups, is more oxidized and provides less energy during further oxidation, but it has a strong affinity for binding to soil minerals. In simplified terms, glucose is attractive food for microbes but a relatively weak mineral glue, whereas oxalate is less rewarding food but a powerful competitor for mineral binding sites. The study therefore set up a direct test of two possible routes to long-term soil-carbon storage: microbial processing followed by incorporation into organic matter, or direct sorption of an intact molecule onto mineral surfaces.

For the experiment, the scientists collected silty clay loam from the A and upper B horizons of a mixed deciduous forest soil at Margery Gallogly Nature Preserve on Grand Island, New York. The soil contained 4.8 percent organic matter and substantial free iron, providing the clay and iron-rich surfaces needed for mineral-organic interactions. They added either glucose or oxalate labeled with carbon-13, a non-radioactive isotope that allowed the researchers to distinguish newly added carbon from carbon already present in the soil. Each gram of dry-equivalent soil received 2 milligrams of labeled carbon—about 4 percent of the soil’s existing carbon content. The amended material was sealed in airtight glass jars and incubated under four regimes: continuously aerobic, oxygenated and oxygen-free on a weekly cycle, oxygenated and oxygen-free on a two-week cycle, or continuously anaerobic. Ten replicate jars were prepared for every substrate and oxygen combination, alongside unamended controls.

Over 20 weeks, the team repeatedly sampled gases in the jar headspaces, measuring carbon dioxide, methane and carbon dioxide containing the carbon-13 label. At the end of five months, the researchers separated the remaining soil carbon into chemically and physically distinct pools. A density separation divided relatively light particulate material from the heavier MAOM fraction. The heavy fraction was then subjected to sequential chemical extractions designed to identify carbon associated with polyvalent cation bridges, reducible minerals, non-reducible short-range-order minerals and crystalline mineral phases. Water extractions measured carbon that remained dissolved or readily mobilized. This approach did not simply ask how much carbon remained in the soil; it tracked where the added carbon went, whether it was respired as gas, dissolved in water, loosely retained in particulate matter or attached to mineral surfaces. Statistical comparisons used two-way analyses of variance, testing the independent and interactive effects of carbon substrate and redox regime.

The central result was stark. Under continuously aerobic conditions and under both intermittent oxygen treatments, about seven times more added glucose carbon was retained as soil organic carbon than added oxalate carbon. The difference appeared in both the light fraction and MAOM, with glucose retention averaging 7.4 times higher in the light fraction and 6.4 times higher across MAOM fractions. Oxygen did not need to be continuously present for this pattern to emerge. Weekly or biweekly anaerobic intervals did not significantly change the total amount of glucose or oxalate carbon stabilized compared with persistent aerobic incubation. The researchers interpret this as evidence that anaerobic periods temporarily pause microbial transformation and mineral-associated carbon formation, rather than creating a lasting storage advantage. Once oxygen returns, aerobic metabolism appears to dominate the overall carbon trajectory.

The explanation lies partly in microbial carbon-use efficiency, a measure of how much consumed carbon microorganisms convert into biomass rather than releasing as carbon dioxide. Oxygen is an energetically favorable terminal electron acceptor, enabling microbes to extract more energy from many substrates. Glucose, with its relatively high bioenergetic yield, can therefore support greater microbial growth under oxygenated conditions. Microbial cells and their residues are increasingly recognized as important sources of MAOM: organisms consume plant-derived compounds, build biomass and eventually leave behind chemically altered remains that bind to minerals. Oxalate follows a different path. Although it can bind strongly to minerals, its lower energy yield limits the microbial biomass produced per unit of carbon under aerobic conditions. In the experiment, much more oxalate remained dissolved in water—about five times more than glucose—and more was respired under continuously aerobic conditions. Its mineral-binding ability did not compensate for its weaker capacity to fuel microbial processing.

The picture changed completely in the permanently anaerobic jars. Respiration was strongly suppressed for both substrates, and much larger amounts remained in the aqueous phase: roughly 108 times more glucose carbon and 63 times more oxalate carbon remained dissolved than under the other redox regimes. Under these oxygen-free conditions, the difference in total solid-phase carbon between glucose and oxalate vanished. Oxalate-derived carbon entering MAOM increased compared with aerobic treatments, while glucose-derived carbon entering MAOM decreased. The researchers propose that suppressed microbial uptake under sustained anaerobiosis gave oxalate more time to associate directly with mineral surfaces. Because oxalate is a strong sorber, it may bind to clay or iron-containing minerals even when microbial transformation is energetically constrained. Glucose, by contrast, lost the microbial advantage that helped stabilize it under oxygenated conditions. The authors caution that isotope measurements cannot prove that labeled carbon recovered from MAOM remained chemically intact as oxalate or was converted into microbial residues, but direct sorption is a plausible explanation.

The experiment also exposed a potentially troubling side effect of oxalate. Under aerobic and fluctuating conditions, adding oxalate caused a strong priming effect: it stimulated the breakdown of carbon that had already been present in the soil. Compared with unamended controls, oxalate increased respiration of existing soil carbon by 1.8 times under persistent aerobic conditions and by about 1.9 times under the intermittent regimes. It also released far more pre-existing carbon into soil solution—9.1 times more under continuous oxygen, 24.7 times more under weekly cycling, 17.3 times more under biweekly cycling and 4.3 times more under persistent anaerobiosis. One likely mechanism is that oxalate’s small size and carboxyl groups allow it to displace older carbon from mineral surfaces, a process sometimes compared with molecular “unbuttoning” of mineral-organic associations. Once released into solution, that carbon becomes accessible to microbes. Fluctuating oxygen may intensify the process by dissolving iron-associated carbon during anaerobic intervals and exposing it to oxidation and microbial consumption when oxygen returns.

Glucose did not produce the expected positive priming effect. It neither increased respiration nor substantially increased solubilization of existing soil carbon under aerobic or fluctuating conditions, and it actually suppressed respiration of existing carbon under persistent anaerobiosis. The researchers suggest that microbes may have preferentially consumed the added glucose, reducing their need to attack older carbon. This behavior could be especially pronounced when oxygen and other electron acceptors are scarce. The results therefore complicate broad claims that adding labile carbon will universally accelerate the loss of stored soil carbon. The outcome depends on the molecular identity of the input, the minerals present, the availability of oxygen and the microbial community’s energetic constraints. A root exudate rich in organic acids may destabilize existing mineral-associated carbon even as it adds new carbon, while a sugar may be routed more efficiently through microbial biomass.

The study’s controlled jars cannot reproduce the full complexity of a living forest soil, where roots, fungi, fauna, water movement, temperature and nutrient availability interact over seasons and decades. The carbon addition was also deliberately high to ensure that the carbon-13 label could be recovered after five months. Even so, the findings offer a mechanistic warning for soil-carbon models and climate strategies: oxygen availability alone is not enough to predict whether new carbon will persist. Models must also account for the chemical identity of plant inputs and for the contrasting behavior of carbon compounds during microbial uptake, mineral sorption and priming. In particular, wetland soils, flooded agricultural fields, compacted soils and periodically saturated landscapes may respond differently depending on whether anaerobiosis is brief or sustained. The authors’ broader message is that soil carbon is not a single pool with a single fate. It is a moving network of molecules, microbes, minerals and redox reactions—and a small change in molecular structure can determine whether carbon is stored, dissolved, respired or used to unlock older reserves.

A groundbreaking study, recently published by Seabloom, Hobbie, MacDougall, and colleagues in Nature Geoscience (2025), now sheds light on an enduring mystery at the intersection of nutrient pollution and carbon sequestration. This seminal work probes into the longevity of soil carbon gains induced by nutrient enrichment, particularly after cessation of fertilizer application and tillage on marginal farmland retired from active cropping. Their findings unveil the surprising persistence of enhanced soil carbon stocks for multiple decades — an insight that holds profound implications for land management strategies aimed at mitigating climate change.

For decades, the role of nutrient enrichment in promoting soil carbon accumulation has been recognized but poorly understood in terms of its temporal sustainability. Nutrients such as nitrogen can stimulate plant growth and increase biomass production, which underlies carbon inputs into soils through root systems and litter deposition. However, concerns loomed that halting nutrient inputs might trigger a rapid loss of these carbon stores, potentially negating any benefits accrued during fertilization. Seabloom et al.’s long-term experimental landscape, nested in retired cropland previously subjected to intensive tillage and fertilization, provided an unparalleled setting to address this critical knowledge gap.

The experiment spanned an impressive four decades, beginning with the cessation of both fertilizer application and tilling on plots once used for conventional agriculture. By meticulously monitoring soil carbon content, plant biomass, and community composition over this time frame, the researchers could disentangle the lasting impact of prior nutrient enrichment from immediate ecosystem responses. Their results were unequivocal: the soil carbon accumulated during fertilization did not simply vanish with the stopping of nutrient inputs; rather, these gains remained stable for over thirty years.

A particularly striking aspect of this study lies in the observed decoupling between plant community dynamics and soil carbon persistence. While plant biomass quickly reverted back to pre-fertilization levels and the species composition of vegetation shifted in response to nutrient cessation, the soil carbon stores retained their heightened state. This finding overturns the simplistic assumption that aboveground vegetation immediately drives belowground carbon processes, highlighting a temporal lag and complex interactions within soil ecosystems that buffer against rapid carbon loss.

The underlying mechanisms for this multidecadal carbon persistence appear linked to soil structure and microbial processes altered during the fertilization period. Intensive nutrient input and tillage had modified the soil matrix and microbial communities in ways that favored stable carbon accrual, possibly through enhanced aggregation and chemical stabilization of organic matter. Post-fertilization, the cessation of tillage likely played a critical role in maintaining these stable carbon pools by minimizing physical disturbance and erosion.

Importantly, this research highlights the pivotal importance of land-use history and soil management practices in governing ecosystem carbon dynamics. The legacy of intensive agriculture and subsequent rehabilitation on retired croplands is not erased quickly with changes in nutrient policy or cessation of fertilization. Instead, soil carbon stocks can retain the imprint of past management decisions for decades, offering a window for climate mitigation strategies that integrate nutrient management with soil conservation measures.

These findings have direct policy relevance amid global efforts to curb nutrient pollution, which often promulgate drastic reductions in fertilizer use to mitigate eutrophication and biodiversity loss in aquatic and terrestrial systems. A key concern among stakeholders has been whether such measures might inadvertently reverse beneficial carbon sequestration on formerly fertilized lands. The demonstration that soil carbon gains can endure for decades post-fertilizer cessation provides a hopeful counter-narrative, underscoring the resilience of soil carbon ecosystems when left undisturbed by tillage.

Moreover, this study advances our understanding of the timescales over which nutrient and carbon cycles interact within terrestrial ecosystems. The multidecadal persistence of soil carbon beyond the lifespan of elevated nutrient inputs challenges current carbon turnover models that often emphasize shorter-term processes. It prompts a reevaluation of how soil carbon feedbacks are incorporated into earth system models predicting future climate trajectories under varying land-use and nutrient scenarios.

The experimental design itself, characterized by long-term, field-scale manipulation, sets a new benchmark for ecological research. Most prior studies investigating nutrient effects on carbon stocks have been constrained by short durations and confounding variables. The robust dataset compiled here fills a critical void, providing empirical evidence for veritable longevity of carbon storage behavior in retired croplands, a widespread yet underappreciated landscape category globally.

Intriguingly, the persistence of soil carbon storage despite the reduction of nutrient inputs also hints at the potential of natural ecosystem recovery following anthropogenic perturbations. Even in landscapes heavily modified by agriculture, soil processes can slowly rebound to create conditions favorable for carbon retention. This insight strengthens calls for restoration-focused agricultural policies, advocating for longer fallow periods, cessation of unnecessary fertilizer inputs, and strategies aimed at reducing tillage intensity.

Nevertheless, the researchers caution against overgeneralization of their findings to all soil types and biomes. The specific soil texture, climate, prior management, and crop types likely modulate the degree and duration of carbon persistence across different systems. Further research extending these long-term analyses to diverse landscapes will be crucial to tailor regionally appropriate management recommendations.

In conclusion, the work of Seabloom and colleagues provides a compelling narrative that soil carbon gains induced by nutrient enrichment on retired cropland are not transient phenomena destined to dissipate rapidly once fertilization halts. Instead, these carbon stocks exhibit multidecadal resilience, anchored by soil biological and physical processes that stabilize organic matter. Such knowledge elevates soil carbon as a robust and meaningful carbon sink in the broader climate mitigation portfolio, contingent upon thoughtful land management that acknowledges both legacy effects and future ecosystem trajectories.

As global society confronts the twin challenges of reducing greenhouse gas emissions and mitigating nutrient pollution, these findings present a nuanced pathway forward. By integrating nutrient management with land retirement and reduced disturbance strategies, it may be possible to sustain and amplify soil carbon storage—turning a historical environmental dilemma into an unexpected climate asset. Continued vigilance and innovative stewardship of soil resources will be indispensable as we seek a harmonious balance between food production, ecosystem conservation, and climate stability.

Subject of Research: Soil carbon persistence and nutrient effects on retired cropland ecosystems

Article Title: Multidecadal persistence of soil carbon gains on retired cropland following fertilizer cessation

Article References:
Seabloom, E.W., Hobbie, S.E., MacDougall, A.S. et al. Multidecadal persistence of soil carbon gains on retired cropland following fertilizer cessation. Nat. Geosci. (2025). https://doi.org/10.1038/s41561-025-01801-5

Image Credits: AI Generated