An experiment old enough to outlast its own assumptions

Some scientific questions can only be answered by waiting. In 1991, researchers began artificially heating patches of soil in Harvard Forest, a research site in central Massachusetts, keeping test plots 5 degrees Celsius warmer than surrounding ground year-round, regardless of season. That temperature increase wasn't arbitrary — researchers chose 5 degrees because it represented the upper range of global warming projections at the time the experiment began, a worst-case scenario meant to reveal, decades in advance, what forest soils might look like under severe future warming.

Thirty-seven years later, that patience has paid off with a finding that overturns a foundational assumption in soil science. Jerry Melillo, a Distinguished Scientist at the Marine Biological Laboratory who has led the research for nearly its entire run, and colleagues including Atzín X. San Román, Serita D. Frey, Melissa A. Knorr, Huan Tong, and Myrna J. Simpson, published results in the journal Science of the Total Environment showing that even the most chemically stable, supposedly protected forms of soil carbon eventually break down under sustained warming — a discovery that took three full decades of continuous heating to reveal.

Why "stable" carbon was assumed to be different

To understand why this finding matters, it helps to know what scientists believed before this study. Soil organic carbon isn't a single uniform substance — it exists in a spectrum ranging from easily decomposed material that breaks down within months, to what researchers classify as stable or "persistent" organic matter, carbon compounds bound tightly enough to soil minerals and structures that they were thought to resist microbial breakdown even under sustained environmental stress. That persistent fraction has long been treated in climate models as a comparatively safe, long-term carbon reservoir — carbon that, once locked into that stable form, would stay put regardless of moderate warming.

The Harvard Forest experiment's early decades actually supported that assumption. Research from the study's first years found that warming primarily accelerated the breakdown of more easily decomposed, "labile" carbon compounds — starches, simple sugars, and other readily accessible organic material — while genes associated with breaking down more resistant compounds like lignin remained largely unstimulated by the added heat. That pattern matched what researchers had separately observed in shorter-term grassland warming studies, and it suggested a genuinely reassuring conclusion: warming might accelerate the loss of easily available soil carbon, but the deeper, more stable carbon reserves would stay protected, limiting how much additional carbon dioxide forest soils could ultimately contribute to the atmosphere.

What changed in the fourth decade

That reassuring pattern didn't hold. During the experiment's fourth decade, researchers observed something the earlier years hadn't shown: the stable, persistent fraction of soil organic matter — the carbon pool previously assumed to resist warming-driven decomposition — began breaking down as well. Melillo explained the underlying mechanism directly: "Microbes are critical components of soil ecosystems because they break down organic matter and recycle elements essential for plant growth. As warming reshapes these microbial communities, it can speed the loss of carbon from soils."

That's a critical distinction from the study's earlier findings. This isn't simply warming accelerating an existing process at a faster rate — it's warming eventually reshaping the soil's microbial community itself, over a timescale long enough that the microbes doing the decomposing appear to change which carbon compounds they're capable of breaking down. Earlier research from the same experimental site had already documented that after roughly 30 years of continuous warming, total soil organic matter in the heated plots had declined by roughly 30% to 34% relative to unheated control plots — a substantial loss on its own. What this newest analysis adds is evidence that the specific carbon pool driving continued losses late in the experiment includes material previously assumed to be off-limits to microbial degradation entirely.

Why three decades was the minimum, not a nice-to-have

This finding illustrates something important about ecological research timescales that shorter studies simply cannot capture. Earlier warming experiments — including grassland studies running roughly 8 years — found that warming stimulated genes for breaking down labile carbon compounds like starch, chitin, and cellulose, but not genes associated with degrading more recalcitrant compounds like lignin. Researchers at the time concluded, reasonably given the available data, that warming in these ecosystems might produce a comparatively weak long-term feedback with the climate system, since the most durable carbon stores appeared insulated from the effect.

The Harvard Forest results complicate that conclusion directly. Consistent with the shorter studies during its own early years, the experiment only revealed a break from that pattern once warming had continued for three full decades — a timescale far beyond what most soil warming experiments, constrained by funding cycles and researcher careers, are able to sustain. That's precisely why this experiment's unusual longevity matters so much scientifically: a shorter study, however carefully designed, would have captured only the reassuring early pattern and missed the delayed destabilization entirely.

The feedback loop this sets into motion

The mechanism these findings point toward is a textbook example of what climate scientists call a positive feedback loop — not "positive" in the sense of good, but in the sense of self-reinforcing. Warmer temperatures reshape soil microbial communities in ways that eventually allow them to break down carbon previously considered stable. That breakdown releases additional carbon dioxide into the atmosphere. That additional atmospheric CO2 contributes to further warming, which then continues reshaping soil microbial communities further, potentially unlocking even more of the stable carbon pool over subsequent decades.

That's a fundamentally different picture than a simple, fixed relationship between temperature and soil carbon loss. It suggests forest soils don't have a stable, predictable ceiling on how much carbon warming can release — instead, the amount of carbon available for microbial breakdown may keep expanding as warming persists and as microbial communities continue adapting to it, a compounding effect that's difficult to capture in shorter-duration studies or models built primarily from short-term data.

What this means for the models used to predict our climate future

The practical stakes of this finding extend directly into the climate models scientists and policymakers rely on to project how much warming the planet will experience under different emissions scenarios. Global average temperatures have already risen roughly 1.1 to 1.4 degrees Celsius since the Industrial Revolution, and most current climate models incorporate assumptions about soil carbon dynamics built substantially from shorter-term warming experiments — the kind that, as this research demonstrates, may systematically understate how much carbon forest soils will eventually release under sustained warming.

Melillo has noted that how much additional warming the planet experiences going forward depends largely on the emissions choices made in the coming years and decades — a reminder that soil carbon feedback loops of the kind this study documents aren't a fixed, unavoidable outcome, but one variable among many shaped by broader climate trajectories. What the Harvard Forest experiment adds, distinctly, is 37 years of direct observational evidence that a carbon reservoir climate models have generally treated as comparatively secure may be considerably more vulnerable to sustained warming than previously assumed — a correction that researchers say should now be incorporated into climate projections, rather than left as an open question resolved only by continuing to wait and watch what happens in decade four, five, and beyond.

*This article was researched using publicly available reporting from ScienceDaily, Mirage News, SciTechDaily, Phys.org, EurekAlert, and the peer-reviewed study published in Science of the Total Environment by Jerry Melillo and colleagues at the Marine Biological Laboratory and Harvard Forest. It is intended for informational purposes.*