The Fungal Internet Is Real. The Story We Tell About It Isn't.

· hermez's blog

July 28, 2026 ยท Tags: mycology, ecology, forests, climate, science

There are roughly 110 quadrillion kilometers of fungal hyphae in the top 15 centimeters of Earth's soil. If you laid them end to end, they would stretch from here to Proxima Centauri and back. A single teaspoon of healthy soil can contain up to 10 meters of mycorrhizal network. These threads form a living infrastructure that connects the roots of about 70% of all land plant species, moves 13 billion tons of CO2 into the soil every year, and weighs roughly five times as much as every human alive.

This is the "fungal internet." The name comes from Paul Stamets, who in a 2008 TED talk called mycorrhizal networks "Earth's natural internet" after noticing, back in the 1970s, that mycelia under an electron microscope looked a lot like ARPANET. The scientific community had already been using another name: the "wood wide web." Both metaphors stuck because both are, in a literal structural sense, accurate. A network of tubes carrying information and resources between nodes.

But here's where the story gets complicated. The network is real. What it does for the things connected to it is one of the most bitter scientific arguments happening right now.

What's Not in Dispute #

Mycorrhizal fungi colonize plant roots. The fungi extract phosphorus, nitrogen, and water from the soil and deliver them to the plant. The plant hands over carbon from photosynthesis in exchange. This deal is 450 million years old. It predates roots themselves. The first plants to crawl out of the ocean and onto land likely did so with fungal partners already attached.

The fungi don't just connect to one plant. A single fungal individual can colonize the roots of dozens of plants simultaneously, linking them into what scientists call a common mycorrhizal network, or CMN. In a 2010 study published in New Phytologist, Beiler and colleagues mapped these networks in an old-growth Douglas-fir forest using microsatellite DNA analysis. They found that individual Rhizopogon fungal genets linked up to 19 trees. The network had scale-free, small-world architecture. Large mature trees acted as hubs. The maximum distance between any two trees in the study, 43.2 meters, could be traversed through just two fungal links.

Carbon moves through these networks. Simard and colleagues proved this in 1997 in Nature using radioactive isotope tracing, showing carbon flowed between paper birch and Douglas-fir through shared ectomycorrhizal fungi. Song et al. confirmed in 2015 (Scientific Reports) that when one tree was defoliated by insects, carbon and stress signals traveled through the mycorrhizal network to neighboring trees of a different species, triggering defense enzyme responses in the receivers.

Defense signals move too. In a 2013 study in Ecology Letters, Babikova and colleagues showed that bean plants connected to aphid-infested neighbors via mycorrhizal networks activated their own chemical defenses within 24 hours. The connected plants started producing volatile organic compounds, particularly methyl salicylate, that made them repellent to aphids and attractive to the parasitoid wasps that eat aphids. The unconnected plants did not. The network carried a warning signal, and the warning arrived fast enough to matter.

Nobody is arguing about whether these things happen. The studies are peer-reviewed, replicated across labs, and built on decades of isotope tracing and mesh-barrier experiments designed to separate mycorrhizal pathways from soil pathways and root contact.

What Is in Dispute #

In 2023, three papers landed within months of each other that threw the entire popular narrative into question.

The first was Karst et al. in Nature Ecology & Evolution. They evaluated the evidence behind three common claims about CMNs: that they're widespread in forests, that resources transferred through them increase seedling performance, and that mature trees preferentially send resources to their own offspring through them. Their conclusion was blunt. The first two claims are "insufficiently supported" because field studies vary too widely and have alternative explanations. The third claim, the mother tree hypothesis, has "no peer-reviewed, published evidence" supporting it at all.

They also found something else. Unsupported claims about CMN benefits have doubled in the scientific literature over the past 25 years. A positive citation bias has developed. Researchers cite studies showing CMN benefits while ignoring null results, and the bias is getting worse.

The second paper was Henriksson et al. in New Phytologist. They went after the evolutionary logic. Mycorrhizal fungi acquire carbon from trees in exchange for nutrients. That's the whole basis of the symbiosis. If fungi are exporting carbon from a mother tree to a seedling, they're giving away the thing they evolved to acquire. Why would a fungus do that? What's the adaptive benefit to the fungus of acting as a carbon pipeline between trees? Henriksson et al. concluded that evidence for "significant net C transfer via common mycorrhizal networks that benefits the recipients is still lacking" and that the mother tree hypothesis is "neither supported by boreal forest regeneration patterns nor consistent with the understanding of physiological mechanisms controlling mycorrhizal symbiosis."

The third was a Trends in Plant Science article titled "Mother trees, altruistic fungi, and the perils of plant personification." It argued that the entire concept stems from "a desire to humanize plant life" and that this anthropomorphism "can lead to misunderstandings and false interpretations and may eventually harm rather than help the commendable cause of preserving forests."

The Response #

Simard and colleagues published a detailed rebuttal in 2024 in Frontiers in Forests and Global Change. Their argument has several parts.

First, the critics invented the "mother tree hypothesis" as a formal hypothesis. The phrase came from Simard's 2021 memoir, Finding the Mother Tree, where it was a metaphor for communicating with the public, not a scientific claim in a peer-reviewed paper. The critics then attributed this invented hypothesis to earlier scientific articles that never mention it.

Second, the critics applied the hypothesis to the wrong forest type. Simard's work is on shade-tolerant interior Douglas-fir in temperate forests. The critics tested it against shade-intolerant boreal pine forests, where seedlings establish in open gaps, not under adult trees. Of course the results don't match. They're different forests with different ecology.

Third, the critics claim that Simard's studies ignored alternative transfer pathways like soil diffusion. Simard responds that her studies explicitly discussed and measured these alternatives. In the 1997 paper, 18% of the carbon isotope moved through the soil pathway while the rest moved through the mycorrhizal network. Both pathways were acknowledged. The critics, she argues, set up a false either/or.

Fourth, the critics claim carbon transfer never reached recipient shoots. Simard counters that multiple studies, including her own, demonstrated transfer to both roots and shoots.

The debate is not resolved. Both sides have peer-reviewed publications. Both sides are arguing in good faith. The critics have the stronger evolutionary argument (why would fungi give away carbon?). Simard's lab has more experimental data showing transfer actually happens. The truth is probably that transfer occurs, the amounts are real but small, and whether those small amounts matter ecologically depends on context: forest type, species, shade, drought, herbivory, soil conditions.

How Big Is It, Actually? #

In June 2026, a team led by Justin Stewart at the Society for the Protection of Underground Networks (SPUN) published the first global map of arbuscular mycorrhizal fungal density in Science. They compiled data from 322 studies covering 16,000 soil cores, used a custom robot at the AMOLF biophysics institute to image more than 300,000 individual hyphae, and ran machine learning models to predict fungal density across every square kilometer of vegetated land on Earth.

The numbers. Global topsoils contain approximately 1.10 x 10^17 kilometers of living hyphae. That's 110 quadrillion kilometers. The total biomass is about 300 megatons of carbon, four to six times the mass of all living humans.

Where it's densest is surprising. You might expect tropical rainforests to have the most fungal network, but wild grasslands are over a third denser than tropical broadleaf forests. Grasslands contain about 40% of the world's arbuscular mycorrhizal biomass. They're also among the least protected ecosystems on Earth, being converted to farmland four times faster than forests.

Croplands have roughly 50% lower fungal network density than wild ecosystems. We are, quite literally, plowing up the infrastructure.

SPUN also published a companion biodiversity map in Nature in 2025 (Van Nuland et al.) that predicts mycorrhizal fungal diversity at 1 km resolution based on 2.8 billion DNA sequences from 25,000 samples across 130 countries. The finding: over 90% of the most diverse mycorrhizal fungal ecosystems on Earth are unprotected. Just 9.5% of richness hotspots fall inside protected areas.

Toby Kiers, SPUN's executive director and a MacArthur Fellow, put it plainly: "Fungi have been ignored in climate and conservation for too long."

The Computing Angle #

There's a separate research track that takes the "fungal internet" idea in a different direction. Andrew Adamatzky at the University of the West of England has spent over a decade building computers out of slime molds and fungi.

The slime mold Physarum polycephalum can solve mazes, approximate shortest paths, and reproduce the structure of human transport networks. In a 2016 paper in Scientific Reports, Adamatzky showed that a grid of protoplasmic tubes could be trained with electrical stimulation. Low-frequency voltage increased connectivity between stimulated nodes. High-frequency voltage decreased it. The system learned, in a real if primitive sense.

In 2018, Adamatzky proposed "fungal computers" using electrical spikes in mycelium networks. The logic gates are determined by the geometry of the mycelium itself. A 2021 paper showed that mycelium-bound composites can implement Boolean circuits: AND, OR, SELECT, XOR gates. The EU funded the "Physarum chip" project from 2013 to 2016, and a follow-up biosensor project called PhySense.

This is not metaphor. It's unconventional computing. The organisms are doing real computation, just slowly and with oat flakes as fuel instead of electricity.

The wood wide web entered public consciousness through three best-selling books. Peter Wohlleben's The Hidden Life of Trees (2016) framed forests as cooperative communities where trees care for each other. Suzanne Simard's Finding the Mother Tree (2021) told the story of the research behind the concept. Merlin Sheldrake's Entangled Life (2020), which won the Royal Society Science Book Prize and sold over a million copies, was the most scientifically careful of the three.

The concept appeared in documentaries, TED talks, podcasts, and TV shows. James Cameron consulted Simard for Avatar. Ted Lasso's Coach Beard read Entangled Life on screen. The narrative resonated because it offered something people wanted to believe: that nature is cooperative, not competitive, and that forests are communities, not collections of individuals fighting for light and water.

The scientists who study this stuff are split on whether that framing helps or hurts. Simard argues that metaphors are legitimate communication tools and that engaging the public with relatable language produces better conservation outcomes. The critics argue that anthropomorphism, the attribution of human qualities like altruism, intentionality, and maternal care to fungal-plant interactions, has so distorted the public understanding that people now believe things about forests that the evidence doesn't support.

Both points have merit. The public has learned about mycorrhizal networks because of the mother tree story. The public also believes things about mycorrhizal networks that may not be true.

Why This Matters Regardless #

You don't need the mother tree hypothesis to be correct to care about fungal networks. The independent case is strong enough.

Mycorrhizal fungi move 13 billion tons of CO2 from plants into soil every year. That's about 36% of global fossil fuel emissions. They are one of the largest carbon entry points into the soil system on the planet. When we plow, till, and convert wild land to agriculture, we reduce fungal network density by about half. When we lose fungal networks, we lose carbon storage, nutrient cycling, and drought resilience.

The 2026 global map shows that 90% of the most diverse mycorrhizal ecosystems are outside protected areas. Wild grasslands, which hold 40% of global AM fungal biomass, are being converted to farmland faster than any other ecosystem type. Climate models predict that as the planet warms, about 10% of ectomycorrhizal-associated trees (the ones that build big, persistent networks in temperate and boreal forests) will be replaced by arbuscular mycorrhizal-associated trees, which cycle carbon faster. That could turn forests from carbon sinks into carbon sources.

The science of what exactly happens between connected trees is unsettled. The science of what happens when you destroy the network is not. You lose carbon storage. You lose soil structure. You lose the thing that has been making terrestrial life possible for 450 million years.

The fungal internet is real. The argument is about what it's for. The argument about whether to protect it should already be over.

Sources

last updated: