Global Map Reveals Vast Scale of Underground Fungal Networks and Their Carbon Role

A study published in Science has produced the first global estimate of the physical extent of arbuscular mycorrhizal (AM) fungal networks in the top 15 centimeters of soil. Led by the Society for the Protection of Underground Networks (SPUN) and drawing on more than 16,000 soil cores worldwide, the research combined field data with machine-learning models trained on ecosystems ranging from deserts to tundra. The models were calibrated using robotic imaging of over 300,000 living AM fungal hyphae grown in laboratory conditions at the AMOLF Biophysics Institute in Amsterdam.

The resulting estimate suggests roughly 110 quadrillion kilometers of hyphae — thread-like structures each about a tenth to a twentieth the width of a human hair — exist in the upper soil layer globally. These networks carry an estimated 300 megatons of carbon and move approximately 4 billion tons of carbon dioxide equivalent into soils each year, an amount equal to roughly 11% of annual human-related CO2 emissions. AM fungi form symbiotic partnerships with about 70% of plant species, extending root reach by up to 100 times and trading soil nutrients such as phosphorus and nitrogen for plant-produced carbon.

The mapping reveals that wild grasslands hold an estimated 40% of the planet's AM fungal biomass, with particularly dense networks predicted in South Sudan's flooded grasslands, Florida's Everglades, and the Tibetan Plateau. However, grasslands are among Earth's least protected ecosystem types and are being converted to farmland roughly four times faster than forests are being cleared. Croplands showed roughly half the fungal network density of wild ecosystems, a 47% reduction where native grasslands had been converted to agriculture.

Researchers suspect that tilling physically disrupts hyphal networks, while synthetic fertilizer may reduce plants' reliance on fungal partnerships for nutrient uptake. Lead author Justin Stewart suggested that Indigenous farming methods, no-till practices, and organic growing may preserve more underground infrastructure, though the study cautions that definitively linking specific farming practices to fungal health requires further research.

The conservation gap emerges as a central finding. Earlier SPUN research published in Nature showed that fewer than 10% of AM fungal biodiversity hotspots fall within existing protected areas. The new mapping demonstrates that this protection gap extends to the sheer mass and density of fungal networks themselves, not just species richness. Protected area boundaries drawn to safeguard visible wildlife and forest cover largely miss the underground systems those ecosystems depend on.

Significant uncertainty remains in the methodology. Regions such as the Sahara Desert and Greenland lack sufficient soil sampling data for confident estimates, leaving blank spots in the global model. The 110-quadrillion-kilometer figure is a machine-learning extrapolation from thousands of sampled points, not a direct measurement, and carries the same caveats as any large-scale ecological model built substantially on interpolation.

The study does not by itself slow grassland conversion or change agricultural policies that drive reduced fungal density on farmland. However, it provides a detailed baseline against which future soil disturbance, land conversion, or ecosystem loss can be measured, turning a previously invisible planetary system into something scientists and policymakers can track over time.