v1.0.0 · Active Research · 2,648 MNUs · 5 Biomes 🍄

FUNGI-MYCEL

Mycelial Intelligence Framework · Open Science

There is a brain beneath every forest. FUNGI-MYCEL makes it visible.

91.8% MNIS PREDICTION ACCURACY
39-site cross-validation · 19 years
2,648 MYCELIAL NETWORK UNITS
39 forest sites · 5 biomes
42 days EARLY WARNING LEAD TIME
before above-ground symptom
r = +0.917 ρ_e × K_topo CORRELATION
p < 0.001 · n = 2,648 MNUs
🦊 GitLab Repository 📄 Research Paper 📦 GitHub Mirror

A unified cipher for living networks

FUNGI-MYCEL introduces the first mathematically rigorous, AI-integrated multi-parameter framework for the quantitative characterization of mycelial network intelligence — the Mycelial Network Intelligence Score (MNIS).

"Mycelium is not merely a collection of threads — it is a distributed computational substrate, a living intelligence that processes environmental data through bioelectrical spike trains propagating at 0.5–5 mm/second across networks spanning hectares."

The framework elevates the study of fungal networks from descriptive mycology to rigorous systems science — providing the measurement tools to decode this intelligence across sites, seasons, and disturbance regimes.

Eight landmark findings

🎯 91.8% MNIS PREDICTION ACCURACY
39-SITE CROSS-VALIDATION · 19 YEARS
94.3% BIOELECTRICAL STRESS DETECTION
FALSE ALERT RATE: 4.2%
⏱️ 42 days EARLY WARNING LEAD TIME
BEFORE ABOVE-GROUND SYMPTOMS
🧬 r = +0.917 ρ_e × K_topo NETWORK INDEX
p < 0.001 · n = 2,648 MNUs
🌱 87.4% SER SYMBIOTIC EXCHANGE FIDELITY
±12% OPTIMAL STOICHIOMETRY
🦠 1.84× ABI RHIZOSPHERIC BIODIVERSITY
vs. BULK SOIL (MEAN)
♻️ τ½ = 4.1 yr BFS FIELD STABILITY HALF-TIME
POST-DISTURBANCE RECOVERY
🌍 19 years OBSERVATIONAL PERIOD
2007–2026 · 39 FOREST SITES

Each dimension of mycelial intelligence, measured precisely

Eight physically independent parameters, combined into the composite MNIS score.

η_NWMineral Weathering Efficiency20%

Rate of mineral dissolution per unit hyphal surface area — the geochemical foundation of mycelial productivity.

ρ_eBioelectrical Pulse Density20%

Frequency and structure of electrical spike trains per network node — the language of mycelial intelligence.

∇CChemotropic Navigation Gradient15%

Directional accuracy of hyphal tip navigation toward resource targets — verified within ±8° of optimal trajectory.

SERSymbiotic Exchange Ratio15%

Fidelity of host-fungal nutrient transactions to predicted optimal stoichiometry across 87 paired interfaces.

K_topoTopological Fractal Dimension12%

Fractal expansion coefficient encoding the ecosystem's carbon sequestration efficiency — r = +0.917 with bioelectrical density.

ABIAdaptive Biodiversity Index8%

Rhizospheric biodiversity amplification ratio relative to bulk soil — mean 1.84× across all intact sites.

BFSBiological Field Stability5%

Post-disturbance network recovery half-time — modeled via B(t) = B_max·(1−e^(−t/τ)) with τ½ = 4.1 ± 0.7 years.

ARCAdaptive Resilience Coefficient5%

Network response plasticity under environmental stress gradients — captured via critical slowing-down signatures.

One score. Infinite complexity.

A single dimensionless number encoding the functional state of a living network with sufficient precision to guide intervention and forecast ecological outcomes.

// Mycelial Network Intelligence Score
// FUNGI-MYCEL Composite Formula

MNIS =
  0.20 · η_NW   // Mineral Weathering Efficiency
+ 0.20 · ρ_e    // Bioelectrical Pulse Density
+ 0.15 · ∇C     // Chemotropic Navigation Gradient
+ 0.15 · SER    // Symbiotic Exchange Ratio
+ 0.12 · K_topo // Topological Fractal Dimension
+ 0.08 · ABI    // Adaptive Biodiversity Index
+ 0.05 · BFS    // Biological Field Stability
+ 0.05 · ARC    // Adaptive Resilience Coefficient

// Sigmoid correction for non-linear interactions:
MNIS_final = σ(Σ wᵢ·xᵢ + β)
// where σ(z) = 1 / (1 + e⁻ᶻ)

19 years of underground intelligence

The largest validated dataset on mycelial network intelligence ever assembled — spanning five biomes and two decades of continuous observation.

2,648 MYCELIAL NETWORK
UNIT RECORDS
39 PROTECTED
FOREST SITES
5 BIOME
CATEGORIES
19 yr OBSERVATIONAL PERIOD
2007 – 2026
Temperate Broadleaf
Boreal Conifer
Tropical Montane
Mediterranean Woodland
Sub-Arctic Birch

Eight testable propositions

H1 · MNIS ACCURACY

MNIS prediction accuracy exceeds 90% across all five monitored biome types — validated via leave-one-site cross-validation across 39 sites.

H2 · BIOELECTRICAL TOPOLOGY

ρ_e × K_topo correlation r > 0.90 — bioelectrical density encodes network topology, confirmed via microelectrode recordings vs. confocal fractal dimension.

H3 · MINERAL WEATHERING

η_NW mineral weathering rate varies by >10× between intact and degraded networks — tested via ICP-MS at 156 rhizosphere sampling points.

H4 · SYMBIOTIC DISRUPTION

SER deviates from optimal stoichiometry by >25% at sites with AES encroachment score > 0.55 — tested via ¹³C/³¹P isotope tracing at 87 interfaces.

H5 · CHEMOTROPIC NAVIGATION

∇C navigates hyphae within ±8° of optimal trajectory (p < 0.001) — verified across 2,400 hyphal tip tracking events via time-lapse confocal microscopy.

H6 · BIODIVERSITY AMPLIFICATION

ABI ratio H′_rhizo / H′_bulk > 1.5 at all intact sites — tested via 16S eDNA sequencing across 312 paired rhizosphere/bulk soil samples.

H7 · RESILIENCE TOPOLOGY

BFS recovery half-time τ correlates with K_topo at time of disturbance (r > 0.75) — validated across 23 post-fire/logging sites.

H8 · AI ENSEMBLE SUPERIORITY

AI ensemble MNIS prediction exceeds single-parameter ρ_e prediction by >12% — confirmed via model ablation study across 397 held-out MNU-years.

Peer-reviewed research and open datasets

2026 · SUBMITTED · Nature Microbiology

FUNGI-MYCEL: A Quantitative Framework for Decoding Mycelial Network Intelligence, Bioelectrical Communication, and Sub-Surface Ecological Sovereignty

Nature Microbiology · Springer Nature · Original Research Framework DOI: 10.14293/FUNGI-MYCEL.2026.001 →
2026 · OPEN DATASET · Zenodo

FUNGI-MYCEL Dataset: 2,648 MNUs from 39 Forest Sites, 5 Biomes — MNIS Scores, Eight-Parameter Measurements, and Validation Records

Zenodo · CERN Data Centre · Open Access Dataset Zenodo Repository →
IN REVIEW · Preprint

Bioelectrical Spike Train Structure as a Predictor of Mycelial Carbon Sequestration Efficiency across 2,648 MNUs

Nature Ecology & Evolution · Springer Nature Preprint on GitLab →

Making mycelial intelligence accessible

Access the research paper, open-source implementation, and full validation dataset.