She Agreed to This Interview Because She Doesn't Exist

Margit Holmgren does not exist, which is the only reason she agreed to this interview.

Everything around her does. The Marcell Experimental Forest, 40 kilometers north of Grand Rapids, Minnesota, has collected continuous climate and hydrologic data since 1960.¹ The SPRUCE experiment, ten chambers each 40 feet in diameter testing what warming does to peatland carbon, concluded its planned run at the end of 2025.² The Grand Rapids lab connected to Marcell is on the closure list; the fate of the experimental forest itself remains officially undecided.³

Of the 633 STEM PhDs employed by the Forest Service in December 2024, 137 had left by the end of 2025.⁴ Margit, 52, soil scientist, 22 years at Marcell, Finnish-American, Iron Range native, took the buyout in January 2026. She is currently adjunct faculty at the University of Minnesota Duluth, which she describes as "a real job that pays like a hobby."

We spoke by phone on a Tuesday in May. She was eating a sandwich.

You spent twenty-two years measuring a bog. How do you explain that at parties?

Margit: You don't. You say "soil science" and people nod and change the subject. Which is fine. If I wanted to be interesting at parties I would have studied volcanoes.

The thing about peatlands, though. I know how this sounds. The scientist explaining why her obscure subject actually matters. But peatlands are three percent of the Earth's landmass holding a third of global soil carbon.⁵ More carbon than all the world's forests combined. The bogs formed over thousands of years because the cold, wet, acidic conditions slow decomposition to almost nothing. Dead plants just accumulate. Layer after layer after layer. You're standing on ten thousand years of stored carbon and it feels like regular ground.

What we were asking with SPRUCE was: what happens when you warm that up? When the conditions that kept all that carbon locked away start to shift? So we built ten enormous chambers in the bog, forty feet across, thirty feet tall, and we heated the peat and the air inside them to different temperatures. And we watched.⁶

What did you find?

Margit: That the carbon comes out. Faster than the models predicted. The warmed plots started releasing more CO₂ and methane almost immediately. Peat that had been a carbon sink for millennia was becoming a source.

And here's the part that keeps me up. The temperature increase we were simulating in the warmest chambers? The ambient temperature at Marcell has risen about 0.72 degrees Fahrenheit per decade since the 1960s.⁷ The real world is catching up to the experiment. We built a preview machine for the future and then we took the machine apart right as the preview started playing outside the windows.

The experiment ended its planned run. That was always the timeline?

Margit: The chambers, yes. But Isabella Garrioch, who did her undergraduate research at Marcell, put it well. She wrote that the work was never just the experiment. It was everything that comes after: the publications, the synthesis, the models, the students who carry the questions forward.³ Decommissioning the chambers is one thing. Closing the Grand Rapids lab is something else entirely. That's the loss of the people who know what the data means.

What does that look like, concretely?

Margit: We have hydrologic monitoring data going back to 1960. Some of it's digital. A lot of it is in field notebooks. Handwritten. The Forest Service Research Data Archive has been digitizing these records, but as of 2020 they'd transcribed 2,943 pages from eight research studies.⁸ Eight studies. We have hundreds.

And the pages aren't self-explanatory. You need to know that in 1978 a beaver dammed the outflow from S2 and the water table readings are wrong for three months. That Dr. Verry switched notation systems in 1985. That the sensor in bog plot 4 drifted for a full growing season in 2003 before anyone caught it.

That context lives in people. When the people leave, the numbers are still there but they stop meaning what they meant.

Has anyone told you what happens to the long-term monitoring stations?

Margit: No.

You took the buyout in January. What's it like watching fire season approach from the outside?

Margit: [long pause]

Marcell isn't a fire site, exactly. But the whole system is connected. The Forest Service is closing 57 of 77 research facilities.⁴ Alaska will have no research facility left. Nine million hectares of national forest, zero federal research capacity.⁹ Wildfire researchers at the Pacific Northwest station, the Rocky Mountain station, those people are gone or going. Fire season is coming. The snowpack is low. And the people who would normally be running the models and updating the risk assessments are packing boxes or already packed.

I keep checking the weather in Grand Rapids. Force of habit. Twenty-two years of checking the weather in Grand Rapids.

[sound of sandwich wrapper]

Sorry. I'm fine. I'm eating lunch.

What would you want someone who's never heard of Marcell to understand?

Margit: That the most important findings in long-term ecology come from contradiction. You set up a study expecting one thing and fifteen years later the data says something completely different. But you only see it if you were watching the whole time.¹⁰

I had a colleague, real person, I won't name them, who spent thirty years on a watershed study. In year twenty-six the trend reversed. Everything they thought about how that watershed responded to disturbance turned out to be true only on a twenty-five-year timescale. If they'd stopped at year twenty, they'd have published the wrong conclusion and everyone would have believed it because it looked complete.

We're losing the patience. The institutional willingness to watch something for thirty years before you understand it. That capacity took decades to build and it can be disassembled in a budget cycle.

Are you angry?

Margit: I'm Finnish. We don't do angry. We do quiet disappointment sustained over decades. It's practically a national art form.

[long pause]

Yes. I'm angry. I'm angry that nobody inventoried the scientific research needs of the National Forest System before dismantling the research capacity.¹¹ I'm angry that I spent twenty-two years building something that takes twenty-two years to build and someone who has never stood in a bog decided it wasn't worth keeping. I'm angry that I'm eating a sandwich in Duluth instead of reading sensors in the field.

But mostly I'm. The word isn't angry. It's watching a record you spent your life maintaining go illegible. Everyone's still alive. The numbers are still on a server somewhere. And none of it means what it meant.

What happens to the bog?

Margit: The bog doesn't care. It's been there for ten thousand years. It'll keep accumulating carbon or it'll start releasing it and either way it won't notice whether anyone's measuring.

We just won't know which.

The Marcell Experimental Forest's phased decommissioning began in 2026.² The Grand Rapids research lab is on the closure list. The fate of the experimental forest's long-term monitoring program remains undecided.³ The average temperature at the site continues to rise at approximately 0.72°F per decade.⁷

Margit Holmgren is not real. The 137 STEM PhDs who left the Forest Service in 2025 are.⁴

Footnotes

1. USFS Northern Research Station, "Marcell Experimental Forest." https://research.fs.usda.gov/nrs/forestsandranges/locations/marcell ↩

2. SPRUCE Experiment, Oak Ridge National Laboratory. https://mnspruce.ornl.gov/ ↩ ↩²

3. Garrioch, Isabella. "Trump's Forest Service cuts jeopardize a legacy of environmental research." Minnesota Star Tribune, April 15, 2026. https://www.startribune.com/trump-admin-forest-labs-ecology-conservation/601681770 ↩ ↩² ↩³

4. Stokstad, Erik. "U.S. Forest Service unveils extensive closures of research facilities." Science, April 3, 2026. https://www.science.org/content/article/u-s-forest-service-unveils-extensive-closures-research-facilities ↩ ↩² ↩³

5. BioScience, "Long-Term Research at the USDA Forest Service's Experimental Forests and Ranges." Oxford Academic, 2006. https://academic.oup.com/bioscience/article/56/1/39/224819; SPRUCE project background materials, ORNL. ↩

6. USFS Research and Development, SPRUCE Project. https://research.fs.usda.gov/nrs/projects/spruce ↩

7. Marcell Experimental Forest climate records, USFS Northern Research Station. https://research.fs.usda.gov/nrs/forestsandranges/locations/marcell ↩ ↩²

8. USFS Research Data Archive and EFR data digitization records. https://www.fs.usda.gov/rds/archive/ ↩

9. Alaska Beacon, "Alaska Forest Service facility slated for closure amid federal restructuring," April 15, 2026. https://alaskabeacon.com/2026/04/15/alaska-forest-service-facility-slated-for-closure-amid-federal-restructuring/ ↩

10. Springer Nature, USDA Forest Service Experimental Forests and Ranges: Research for the Long Term, 2013. https://link.springer.com/book/10.1007/978-1-4614-1818-4 ↩

11. Public Employees for Environmental Responsibility (PEER), "Forest Service Sheds Research Capacity in Move to Utah," April 2, 2026. https://peer.org/forest-service-sheds-research-capacity-in-move-to-utah/ ↩