Heated and control plots in a long-term soil warming study at Harvard Forest, Petersham, Mass. Jerry Melillo of the Marine Biological Laboratory, Woods Hole, Mass., and colleagues began the study in 1991. (Audrey Barker-Plotkin)

One of the regular complaints of climate change doubters and skeptics is that scientific projections of a dire future are too heavily based on computer simulations, or models, which — they say — rest on a variety of questionable assumptions.

But a major climate change study published Thursday relied not on models but experimental data — a 26-year record of observations, no less — to reach a conclusion perhaps just as worrying. The research, tracking the emissions of carbon from artificially heated plots of a forest in Massachusetts, reinforces fears about the possibility of a climate change “feedback” involving the planet’s soils, one that could pile on top of and substantially worsen the ongoing warming trend triggered by the burning of fossil fuels.

“The study is one of the longest if not the longest climate change ecosystem experiment, beyond the one we are running in our own planet,” said Pep Canadell, an expert on the Earth’s carbon cycle at the Commonwealth Scientific and Industrial Research Organization in Australia. He was not part of the research.

Starting in 1991, a team of researchers have been studying the same 18 plots of forest soil in the Harvard Forest in Massachusetts. Six of the plots are entirely undisturbed, representing the natural state of the forest floor; six are artificially heated through underground cables to 5 degrees Celsius (9 degrees Fahrenheit) above the normal temperature; and six are “disturbed,” meaning that they contain heating cables, but the cables are not actually powered, so the temperature is not altered.

The researchers involved — affiliated with the Marine Biological Laboratory at Woods Hole, Mass., the University of New Hampshire, the University of Massachusetts at Amherst and New Hampshire-based Research Designs — have continually measured the difference in carbon dioxide emissions emanating from the different plots. The hypothesis is that warmer temperatures would lead microorganisms in the soil to become more active in breaking down plant matter and other materials. These microbes would then release more soil carbon into the atmosphere in the form of carbon dioxide gas, in a process known as respiration.

Now, in the latest update of the findings published in the journal Science, the team reports that 17 percent of the carbon has now been lost from the upper layer of soil of the heated plots over the period of the study. Extrapolating to the scale of the planet, the researchers suggest, that could be a big deal.

“This magnitude of loss could amount to hundreds of petagrams — billions of metric tons — of carbon fluxing from the world’s forest soils to the atmosphere, if those soils responded to warming like the Harvard forest soils have done over the experimental period,” said Jerry Melillo, a scientist at the Marine Biological Laboratory in Woods Hole who has overseen the research for nearly three decades. (A petagram is equivalent to a billion metric tons.)

The study suggests a possible release of about 190 petagrams over the course of the century from the top meter of the Earth’s soil, which it calculates is “equivalent to the past two decades of carbon emissions from fossil fuel burning.” And Melillo said that the possibility of this occurring isn’t adequately taken into account in future climate change projections.

“I think that everyone agrees that we don’t include the microbial feedback mechanisms very well in climate models,” Melillo said.

Below the canopy of a mid-latitude hardwood forest at Harvard Forest, Petersham, Mass. (Audrey Barker Plotkin)

That said, the Harvard Forest study contains some subtleties and surprises. Notably, the soil carbon did not stream out of the ground continuously over the 26-year period. Rather, the release occurred in four phases.

First there was a large burst of carbon that happened over the initial 10 years. Then, the carbon releases ceased for about the next seven years, as the heated plots ceased to show any difference from the non-heated ones. Then the emissions resumed — and then, most recently, stopped again.

The pattern appears to reflect the kinds of organisms in the soil changing so as to favor organisms that consume carbon in different ways. “One of the things that became very apparent is that we were changing in significant ways the structure of the microbial community,” Melillo said.

The research arrives a little under a year after another influential study compiled the results of 49 separate field studies on soil carbon emissions and concluded that this feedback is indeed real, based on the balance of the research. That study concluded that the feedback could lead to 50 petagrams of emissions just by 2050.

Still, there are several important cautions about the latest research, noted Canadell, who also heads up the Global Carbon Project, which annually releases figures on how much carbon fossil fuels and other parts of the Earth’s system have put in the atmosphere.

“By definition, manipulative experiments like this one, are systems treated with a shock perturbation overnight,” he said by email. But a sudden soil temperature increase of 5 degrees Celsius “is unrealistic to what is happening or will happen the real world,” Canadell continued, noting that global warming actually proceeds “at several decimal points of a degree warming per decade.”

Canadell also questioned the 190 petagram figure for possible global emissions by 2100, noting that the study “is in one type of ecosystem and in one single place in the world (a few hundreds of square meters), and the one thing we know is that different biomes and ecosystems are very different, and within, there are no two soils patches the same.”

“This negative assessment, however, doesn’t detract one bit from the extraordinary relevance of the data set and new understanding produced by Melillo’s experiment,” Canadell continued.

However the results ultimately get incorporated into our understanding of the planet, Melillo said, “I’m really quite thankful that there is a place in science for some of these long-term experiments at the systems level, where mechanisms unfold slowly over time.”

Read more at Energy & Environment:

Interior Department whistleblower resigns, calling Ryan Zinke’s leadership a failure

Scientists mapping Greenland have produced some surprising — and worrying — results

One of the most bizarre ideas about climate change just found more evidence in its favor

Rick Perry just proposed sweeping new steps to help struggling coal and nuclear plants

For more, you can sign up for our weekly newsletter here, and follow us on Twitter here.