PETERSHAM, MASSACHUSETTS
Buried in the quiet woods of central Massachusetts sits one of the longest-running science experiments on Earth, and it just delivered a verdict that should worry anyone who has ever trusted a climate model. For 37 years, researchers at Harvard Forest have been artificially heating patches of forest soil, year-round, by five degrees Celsius — a real-world simulation of the upper range of global warming projections scientists feared back when Ronald Reagan was still in the White House. Now, after nearly four decades of continuous heating, the experiment has produced a result that upends a core assumption baked into virtually every climate model used today: carbon that was supposed to stay locked in the ground is starting to break free.
The findings, published this year in the journal Science of the Total Environment, come from a team led by Jerry Melillo, a distinguished scientist at the Marine Biological Laboratory in Woods Hole, who has personally overseen the heated plots for most of his career. What he and his colleagues discovered is that the most stubborn, supposedly "stable" fraction of organic carbon in forest soil — the deep reserve that climate scientists have long treated as a safe, essentially permanent carbon vault — has begun decomposing under sustained warming, releasing extra carbon dioxide into the atmosphere that models never accounted for.
Four Decades Of Cooking The Ground On Purpose
The experiment itself sounds almost absurdly simple: bury heating cables in the soil, keep a patch of forest floor running 5 degrees Celsius warmer than the surrounding, untouched ground, year-round, and just... wait. And wait. And keep waiting, through multiple presidential administrations, through the entire rise of the modern internet, through decades in which "climate change" went from a fringe scientific worry to a mainstream, unavoidable reality.
Melillo's team chose that 5-degree figure deliberately. When the experiment began, it represented the upper edge of what climate scientists projected the planet could plausibly warm by under high-emissions scenarios. The idea was to essentially fast-forward a patch of New England forest floor into a worst-case climate future and watch, in real time, what happened to the soil underneath it — rather than trying to model or guess.
For years, the early results seemed almost reassuring. Soil microbes initially responded to the extra heat exactly as expected: they woke up, sped up their metabolism, and began releasing more carbon dioxide as they broke down surface organic matter faster. But that was the "easy" carbon — the stuff scientists always expected to be vulnerable to warming. The deeper, older, chemically stubborn carbon compounds were thought to be different: protected by their molecular structure, bound up in soil minerals, or simply too chemically inert for microbes to bother with, even in a warmer world. That assumption is what just broke.
When The "Safe" Carbon Stopped Being Safe
According to the new findings, after decades of sustained warming, the microbial communities living in the heated soil plots didn't just speed up — they fundamentally changed. The experiment documented a 45% average annual increase in net nitrogen mineralization, the process by which microbes break down organic material and release nitrogen in forms plants can use. Nitrification — a related microbial process — nearly tripled, with roughly a quarter of all mineralized nitrogen in years five through seven of one measurement period getting converted through that pathway.
Those shifts in nitrogen cycling turned out to be the canary in the coal mine for something bigger. As the soil's microbial ecosystem reorganized itself around the new, warmer normal, it eventually began attacking carbon reserves that had previously been left alone — the "recalcitrant," long-buried organic matter that had survived essentially undisturbed for centuries or longer. Once microbes adapted to exploit it, that older carbon began breaking down and releasing as carbon dioxide, just like the more easily accessible surface carbon had decades earlier.
"Microbes are critical components of soil ecosystems because they break down organic matter and recycle elements essential for plant growth," Melillo explained, describing how warming reshapes those communities over time and, in turn, accelerates the loss of carbon that had been sitting undisturbed in the ground. The unsettling part, in his telling, is how this behaves once it starts. "Once this self-reinforcing feedback begins, there is no easy way to turn it off," Melillo said. "There is no switch to flip."
Why A Feedback Loop Is The Scariest Kind Of Climate Problem
Climate scientists have spent decades worrying about a specific category of danger known as a feedback loop — a process where warming itself triggers changes that cause even more warming, independent of anything humans do afterward. Melting Arctic permafrost releasing trapped methane is the feedback loop most people have heard of. The Harvard Forest findings suggest forest soil carbon may represent another one, hiding in plain sight in ordinary temperate forests, not just the dramatic permafrost of the Arctic.
The scale of what's potentially at stake is difficult to overstate. Scientists estimate that soils worldwide hold somewhere between 1,500 and 2,400 gigatons of carbon in just their top two meters — roughly two to three times as much carbon as currently exists in the entire atmosphere. Even a modest, single-digit percentage loss of that reservoir, compounded over decades, could rival the cumulative fossil fuel emissions of an entire major industrialized economy. Global climate models have generally assumed that a significant share of that soil carbon is essentially locked away, insulated from the warming happening above it. The Harvard Forest data suggests that assumption may simply be wrong, at least under sustained, realistic warming scenarios — and this experiment is the closest thing science has to a four-decade real-world test of that assumption, rather than a projection built on shorter, more speculative studies.
Not The First Warning, But The Longest-Running One
This is not entirely a bolt from the blue. Melillo and Harvard Forest researchers have published interim findings from this same experiment periodically for years, including earlier work showing that the site's forest soil carbon feedback responded to warming in complex, sometimes counterintuitive ways depending on the stage of the experiment. A related 2017 study using similar long-term soil warming data warned that carbon feedback from forest soils could measurably accelerate global warming beyond what models predicted. What makes the newest results notable isn't that the danger is brand new — it's that it took nearly four decades of patient, unglamorous, continuously funded fieldwork to finally observe the "stable" carbon reservoir itself start to give way, confirming a worst-case possibility that shorter experiments simply couldn't detect.
That's also what makes this kind of research so vulnerable in the real world: experiments like this require sustained funding and institutional patience across nearly forty years and multiple generations of graduate students, a luxury increasingly rare in a research funding landscape that prizes short-term, high-turnover grant cycles. Had Harvard Forest's warming plots been defunded or abandoned at year fifteen, or even year twenty-five, this particular discovery — the slow-motion failure of "stable" soil carbon — would likely still be invisible to climate science today.
What This Means For You
This isn't an abstract concern confined to a patch of forest in Massachusetts. Soil carbon dynamics like these play out in temperate forests across North America, Europe, and Asia — the same kinds of ecosystems that blanket much of the populated Northern Hemisphere. If similar carbon release accelerates broadly as global temperatures climb, it means the planet could warm measurably faster than current climate models predict, independent of any additional fossil fuel emissions humans produce. That has direct implications for how aggressively emissions need to be cut to hit any given global temperature target, because it suggests the "carbon budget" — the amount of CO2 humanity can still emit before crossing dangerous warming thresholds — may effectively be smaller than currently modeled, since nature itself could be adding a previously unaccounted-for source of atmospheric carbon.
It also reframes something people tend to take for granted: the ground beneath forests, farms, and backyards as an inert, stable feature of the landscape. This research suggests soil is a dynamic, living system that can flip from carbon sink to carbon source as temperatures rise, in ways that unfold slowly enough to be nearly invisible on human timescales — until, after decades, they suddenly aren't.
The Bigger Picture
The Harvard Forest results land at a moment when 2026 has already delivered no shortage of climate stress signals — from a Category 5 hurricane forming in the East Pacific for the first time since 2024, to record heat and flash flooding across parts of the United States, to accelerating deoxygenation documented in oceans and inland waters worldwide. Against that backdrop, a slow-building, self-reinforcing carbon release from ordinary forest soil is a quieter kind of alarm — but arguably a more structurally significant one, because it suggests some of the planet's warming math may need to be redone.
Scientists are now left grappling with how to fold a mechanism like this into the next generation of global climate models, and how urgently. Given that global soils store two to three times as much carbon as the atmosphere itself, even incremental miscalculations about how much of that reservoir stays locked away as the world warms could meaningfully shift projections for how hot, and how fast, the century ahead actually gets.
Conclusion
It took 37 years of patiently heating a patch of New England forest floor to find something that shorter experiments were never going to catch: the deep, supposedly untouchable carbon locked in soil is not as safe as climate science assumed. Jerry Melillo's decades of unglamorous fieldwork have handed the world an uncomfortable data point right as climate stresses are visibly intensifying elsewhere. As Melillo himself put it, once that feedback loop starts, there's no switch to flip to turn it back off — which means the real question left hanging over this research isn't whether the ground beneath us is starting to give back the carbon it has held for centuries, but how quickly the rest of the world's forests might follow Harvard Forest's lead.
