Mycelium Networks — Nature's Underground Internet

What Is Mycelium?

Fungi occupy their own kingdom of life, distinct from plants, animals, and bacteria. The vegetative body of a fungus — the part doing the real work of nutrient acquisition — is the mycelium: a branching, anastomosing mass of filaments called hyphae. Individual hyphae are typically 1–10 micrometres in diameter (roughly the width of a bacterial cell), yet they can extend centimetres per day, collectively covering vast distances underground.

The mushrooms and brackets we see above ground are merely fruiting bodies — temporary reproductive structures. The permanent, metabolically active organism is the mycelial network hidden beneath the surface, decomposing organic matter, intercepting mineral ions, and exchanging resources with plant partners.

A critical structural feature is anastomosis: the fusion of adjacent hyphae to create closed loops in the network. Unlike a simple branching tree, a mycelial network with anastomosis can reroute flow around damaged sections, redistribute resources dynamically, and maintain connectivity even when parts of the network are disrupted. This topological property makes mycelium far more robust than any branching structure.

Fungi feed in three primary modes:

• Saprophytic: decomposing dead organic matter, cycling carbon and nutrients back into ecosystems
• Parasitic: extracting nutrients from a living host (often a plant or another fungus) at the host's expense
• Mycorrhizal: forming mutualistic associations with plant roots, providing minerals in exchange for photosynthetically fixed carbon

The sheer scale fungal networks can achieve is extraordinary. A single Armillaria ostoyae (honey fungus) colony in Malheur National Forest, Oregon, covers approximately 9.6 square kilometres and is estimated to be around 2,400 years old — making it one of the largest and oldest known organisms on Earth. The entire colony is genetically identical, connected by underground mycelial mats colonising tree roots across this enormous area.

The Wood Wide Web: Mycorrhizal Networks

The term Wood Wide Web — coined by science writer Merlin Sheldrake and popularised by Suzanne Simard's research — refers to the underground mycorrhizal networks that link the roots of multiple trees through shared fungal intermediaries. These networks form what ecologists call a common mycorrhizal network (CMN).

Two main types of mycorrhizal fungi structure these networks:

• Ectomycorrhizal fungi (EMF): sheaths around root tips without penetrating cells; common in temperate forests (oaks, pines, beeches)
• Arbuscular mycorrhizal fungi (AMF): penetrate root cells, forming arbuscles; the dominant type globally, found in most grasslands and tropical forests

The landmark scientific moment came in 1997, when Suzanne Simard and colleagues published results in Nature showing that Douglas fir (Pseudotsuga menziesii) and paper birch (Betula papyrifera) trees exchanged carbon bidirectionally through their common mycorrhizal network. Shaded birch seedlings received net carbon from the fir; shaded fir seedlings received it from the birch. The direction of flow depended on which tree had surplus photosynthate — suggesting the network was a resource-sharing system, not merely a passive conduit.

Network analysis of CMNs reveals a scale-free topology: most trees are connected to a few others, but a small number of highly connected hub trees — called mother trees by Simard — link vast numbers of neighbours. These hub trees anchor network connectivity: removing them disproportionately fragments the network, much like removing major internet exchange points collapses routing paths.

Resources transferred through CMNs include not only carbon (as sugars) but also nitrogen, phosphorus, water, and — as more recent research suggests — defence signalling compounds. When one tree is attacked by pathogens or herbivores, chemical warning signals may propagate through the network, priming neighbours to upregulate defences before attack arrives.

Nutrient Transport: Mass Flow and Diffusion

Moving nutrients across a network that may span kilometres requires more than passive diffusion. Mycelium employs two complementary mechanisms: cytoplasmic streaming (bulk flow) and Fickian diffusion for small molecules.

Cytoplasmic Streaming

Cytoplasmic streaming — also called mass flow — drives bulk cytoplasm through hyphal tubes using turgor pressure gradients. The cytoplasm in nutrient-rich zones is loaded with solutes (sugars, amino acids, ions), raising osmotic pressure and drawing water in, which drives cytoplasm outward through the network toward lower-pressure growth tips or nutrient sinks. Observed streaming speeds in fungal hyphae range from 5 to 10 µm/s, though oscillating bidirectional flows have been recorded.

The physics of flow through hyphal tubes follows the Hagen-Poiseuille equation, the same law governing laminar viscous flow through cylindrical pipes:

The critical insight is the r to the power of 4 dependence: doubling the radius of a hypha increases its flow capacity by a factor of 16. This is why fungi invest in wide "trunk" hyphae for long-distance transport and use fine exploratory hyphae only at the foraging frontier. The network architecture — thick central cords tapering to fine tips — directly reflects this hydraulic constraint.

Fickian Diffusion

For small signalling molecules and ions, passive diffusion operates alongside bulk flow. Fick's first law describes the diffusive flux:

In practice, the two mechanisms act together: bulk flow delivers concentrated nutrient packages rapidly over long distances, while diffusion handles short-range redistribution within and between cells. The network continuously balances these flows, diverting resources from nutrient-rich source zones to actively growing sink zones — a dynamic that mirrors source-sink economics in plant physiology.

Physarum polycephalum: The Optimal Network Builder

Physarum polycephalum is a slime mould — technically not a fungus but a member of Mycetozoa — yet it offers the clearest window into how biological networks solve optimisation problems. When foraging, Physarum forms a network of cytoplasm-filled veins connecting food sources. The network is not random: it converges on solutions that balance path efficiency, fault tolerance, and material cost.

The famous Tokyo rail network experiment (Toshiyuki Nakagaki, 2000, published in Nature) placed oat flakes on a map in the geographic positions of Tokyo's main cities and towns, then inoculated the map with Physarum at the Tokyo position. Over 26 hours, the slime mould extended, explored, and then retracted, leaving a network of reinforced veins connecting the food sources. The final network closely resembled the actual Tokyo metropolitan rail network in both topology and efficiency — a structure designed by engineers over decades, reproduced by a brainless organism overnight.

The Reinforcement Rule

The mechanism behind this optimisation is a simple local feedback rule. Each vein carrying flux Q tends to widen; veins carrying little flux shrink:

Combined with Kirchhoff's current law at each node (flow in = flow out) and Hagen-Poiseuille flow through each vein, this produces a linear system of equations for pressure at each node. Solving iteratively for pressures, updating flows, then updating conductances — repeated hundreds of times — the network self-organises toward a near-minimum spanning tree structure with added redundancy for fault tolerance.

The resulting networks exhibit three desirable properties simultaneously:

• Low cost: total tube material is minimised
• Short paths: travel time between any two nodes is close to minimum
• Fault tolerance: the network remains connected even if random links are removed

Electrical Signalling in Fungi

The idea that fungi might communicate electrically seemed fringe science until a series of experiments by Andrew Adamatzky and colleagues at the University of the West of England, published in 2022, recorded clear electrical potential oscillations in live mycelium of multiple species including Ghost fungi and Caterpillar fungi.

The measured signals share characteristics with neuronal action potentials:

• Spike amplitude of approximately 50 mV
• Spike frequency up to approximately 1 Hz
• Propagation along individual hyphae
• Stimulus-response coupling: adding nutrients or damage near one end of the mycelium triggered spike propagation toward the stimulus

The mathematical description of signal propagation along a hypha borrows from neuroscience. A simplified cable equation describes how membrane potential V decays and spreads along a cylindrical conductor:

Unlike neurons, hyphae have no axon, no myelin sheath, and no synaptic cleft. Electrical coupling occurs through the shared cytoplasm and possibly through ion channels in the hyphal membrane. The signals more closely resemble plant action potentials (as seen in Mimosa pudica or the Venus flytrap) than animal neural signals.

The scientific debate centres on interpretation: are these spikes a genuine information-carrying communication system, coordinating nutrient allocation or defence responses across the network? Or are they a metabolic by-product — electrochemical noise from ion transport processes with no signalling role? The stimulus-response experiments suggest intentionality, but distinguishing signal from correlated noise in a living system remains methodologically challenging.

Analogies with plant signalling suggest that even if the mechanism is not "intelligence" in any cognitive sense, it may constitute a form of distributed sensing and response that allows a mycelial network to integrate information about its environment across spatial scales that pure diffusion or bulk flow could not achieve on biologically useful timescales.

Mycelium as a Computing Substrate

Adamatzky's electrical signalling findings opened a new research direction: could mycelium serve as a literal computing substrate — a biological hardware on which logical operations are performed? Early results are promising, if preliminary.

By placing electrodes at multiple points in a growing mycelial mat and applying stimuli (electrical, chemical, light), researchers have recorded outputs consistent with AND gates (output spike only when both inputs are active) and OR gates (output when either input is active). The logic is implemented by the network's own dynamics rather than by designed circuitry — an emergent computation.

Reaction-Diffusion Computing

A parallel thread of research treats mycelium as a reaction-diffusion medium. Chemical waves propagating through the network can encode information in their timing, amplitude, and collision patterns. This is conceptually analogous to Belousov-Zhabotinsky reactions in chemistry, which have been demonstrated to solve maze problems.

The branching patterns of mycelium itself are governed by activator-inhibitor dynamics — the same class of equations Alan Turing proposed in 1952 to explain biological pattern formation. An activator chemical promotes branching at a tip; an inhibitor diffuses ahead and suppresses branching in nearby zones, creating the regular spacing of hyphal tips:

These equations are mathematically identical to those used in some neuromorphic computing architectures and spiking neural network models. The parallel is more than superficial: both mycelium and neuromorphic chips exploit local, parallel, low-energy computation rather than centralised sequential processing.

Practical constraints on mycelium computing are substantial. Biological processes operate on timescales of seconds to hours — millions of times slower than silicon. The substrate is non-reprogrammable in any conventional sense, sensitive to temperature and humidity, and difficult to interface reliably with electronic systems. Nevertheless, mycelium computing research contributes to a broader understanding of what computation means beyond silicon, and may eventually yield hybrid bio-electronic devices for environmental sensing applications.

Sustainable Materials and Biotechnology

Beyond information and ecology, mycelium has become a material science platform. The same properties that make fungal networks resilient — structural strength, chemical versatility, ability to colonise and bind substrates — translate directly into industrial applications.

Mycelium Composites

Mycelium composites are manufactured by inoculating agricultural waste substrates (hemp hurds, corn stalks, sawdust, straw) with selected fungal strains, then allowing the mycelium to colonise and bind the particles over several days. The resulting block is dried (killing the organism) and can be moulded into almost any shape during the growth phase.

Material properties compare favourably with conventional foams:

• Thermal insulation: R-value approximately 3.5 per inch, comparable to expanded polystyrene
• Compressive strength: up to 200 kPa for dense formulations
• Fire resistance: naturally self-extinguishing due to chitin content
• End of life: fully home-compostable within weeks

Commercial pioneers include Ecovative Design (packaging inserts, replacing polystyrene; leather-like sheets under the brand "Forager") and MOGU (acoustic and thermal building insulation panels from agricultural waste).

Mycoremediation

Mycoremediation exploits the enzymatic machinery fungi evolved to decompose lignin — one of the toughest polymers in biology. The same extracellular enzymes (laccases, manganese peroxidases, lignin peroxidases) that white rot fungi like Pleurotus ostreatus (oyster mushroom) use to break down wood also degrade structurally similar environmental pollutants:

• Petroleum hydrocarbons (diesel, PAHs, benzene derivatives)
• Synthetic dyes from textile effluent
• Some chlorinated pesticides and herbicides

Heavy metals cannot be enzymatically broken down, but fungal cell walls (rich in chitin and glucans) can biosorb metal ions — sequestering lead, cadmium, arsenic, and mercury from contaminated water and soil, concentrating them in fungal biomass that can then be harvested and safely disposed of.

Mycelium Leather

Bolt Threads produces "Mylo" — a leather substitute grown from Ganoderma mycelium on corn waste. The material mimics the collagen fibre structure of animal leather at the microscopic level, producing a supple, durable sheet without slaughtering animals or using chromium-based tanning chemicals. Several fashion brands have produced limited runs of bags and shoes from Mylo.

Network Science Lessons from Mycelium

Mycelium is not merely a biological curiosity — it is a working proof-of-concept for a set of network design principles that engineers and computer scientists increasingly want to replicate.

Fault Tolerance Through Redundancy

Anastomosis creates redundant paths between any two points in the network. In graph theory terms, the network has high edge connectivity: many edges must be removed before the network disconnects. This is the same design principle behind mesh networking in telecommunications and the original ARPANET design philosophy.

Adaptive Resource Allocation

The positive feedback loop — more resources flowing through a tube causes the tube to widen, which increases flow further — is a biological implementation of reinforcement learning. The network continuously adapts its topology to match current demand without any global planner. Tubes that are not used shrink (a form of pruning that reduces maintenance cost), while high-demand routes grow. The system finds efficient configurations automatically.

Decentralised Control

There is no central node in a mycelial network — no "brain" fungal cell that coordinates the others. Network behaviour emerges from local rules applied simultaneously at every node and edge. This makes the system intrinsically scalable (adding more network does not bottleneck any central processor) and robust to targeted attack on any single node (since no node is uniquely essential).

Quantifying Network Vulnerability

Network scientists measure node importance using betweenness centrality — the fraction of all shortest paths in the network that pass through a given node:

In forest CMNs, hub trees (mother trees) have disproportionately high betweenness centrality. Removing them does not just disconnect their direct neighbours — it fragments distant parts of the network that relied on those nodes as relay points. This has conservation implications: protecting hub trees may be more important for network resilience than protecting an equal number of randomly chosen trees.