A dormant mycelium sheet that moves when the air turns damp

A membrane that forms where broth meets air

Nature Communications published a study online on 12 August 2026 describing flexible sheets made from fungal mycelium, and it is unusually frank about what they cannot do. Rui Nie, Baoyuan Li, Luxuan Li and colleagues at Harbin Institute of Technology, with Xiaoman Liu and Xin Huang as corresponding authors, grew Trametes versicolor in a liquid medium and let the mycelium knit itself into a continuous membrane where the broth meets the air. Fermentation ran for twenty days at 28 °C and 85 % relative humidity, and a fresh membrane held more than 90 % water whatever its age or treatment. Scaling the system up to a 185 × 235 mm container produced full-size membranes; that measurement is the vessel, not the sheet. The original paper carries two dates, online 12 August 2026 and version of record 10 September 2026, the second also recorded in the Crossref metadata.

The two-facedness comes from the way the fungus grows: layer A on top is aerial hyphae spreading horizontally above the medium, layer B beneath is nutritive hyphae pushing down into it. The membrane is then steeped in poly(vinyl alcohol) and glycerol and dried. The result, which the authors call a macerate, is a supple leather-like sheet that keeps its Janus structure. Dried without glycerol it is brittle: the PVA acts as a crosslinking binder, the glycerol as plasticiser.

Which face does what

The asymmetry is measured rather than asserted. Layer A is the more porous of the two, 37.31 ± 2.16 % against 32.42 ± 1.33 %, takes up 1.24 times more moisture at saturation and adsorbs it faster; layer B has the denser skeleton but a moisture diffusion coefficient 1.70 times higher. Both share a critical relative humidity of roughly 60 %. What decides the bending is swelling of the skeleton itself, put by micro-computed tomography at 9.27 % for layer A against 3.38 % for layer B, rather than any change in the pores.

Which fungus owns which number

Six fungi went into the study. Trametes versicolor and Phanerochaete chrysosporium BKM-F-1767 were bought from a Shanghai culture collection; Coprinus comatus, Pleurotus ostreatus and Agaricus bisporus came from a research institute in Wuhan; and a black yeast-like Cladophialophora sp. was isolated from a wild Trametes versicolor growing on the Harbin campus. Across that family the abstract reports elongation from 11 to 349 % and toughness — the area under the stress–strain curve — from 0.1 to 18.0 MJ·m⁻³, which the authors call the highest values yet reported for flexible mycelium materials. That is their own comparison against the published literature, not an independently certified record.

Keeping each number with its own fungus matters. The model Trametes versicolor sample treated with 10 % PVA and 10 % glycerol has a Young's modulus of 23.7 MPa, 36 % elongation and 1.0 MJ·m⁻³ toughness — respectable, and nowhere near the top of the range. The extremes belong to the Cladophialophora macerate under the same treatment: 18.0 MJ·m⁻³ and an elongation the Results section puts at 350 %, while the abstract closes the range at 349 %. Both figures are in the paper and are best left as they stand. Nor is any of this plastic-free. Poly(vinyl alcohol) is a synthetic polymer, and when the team tried natural polysaccharides in its place — carboxymethyl cellulose, Trametes polysaccharides, pullulan, pectin — the mechanical performance suffered.

A bonded strain gauge, not a nervous system

In a humidity chamber a 5 × 25 mm strip clamped at one end was taken from 20 % to 85 % relative humidity over eight minutes, then back down across the following twelve. Dynamic vapour sorption gave an average uptake of 374.9 mg·g⁻¹ over five consecutive cycles at 25 °C and 85 % humidity. The library of bending angles runs from (−19.0 ± 10.3)° to (43.2 ± 10.7)°; standard deviations of about ten degrees on angles that reach forty-three mean the motion is modest and only broadly repeatable, and the photograph of an opening paper flower in Figure 2 is more persuasive than the error bars beside it warrant.

The direction of bending is not fixed either. An untethered sheet drying from its fresh state curls towards layer B; the same material gaining moisture in the chamber deforms reversibly towards layer A; a strip held at one end bends consistently towards layer B. The Pleurotus sheet deformed in the opposite direction to the Trametes one.

In the fan demonstration the electrical signal is not the fungus talking. A commercial BF350 strain gauge was glued to the macerate, its output amplified and fed to an Arduino UNO-R3 and a relay. Once ambient humidity passed a critical value of about 60 %, the output voltage reached the relay trigger of roughly 1.7 V and switched on a fan; the airflow dried the strip, the strip straightened and the circuit opened again. The macerate does have an electrical response of its own — its resistance falls as it takes up moisture, the drop 25.51 % steeper through face A than face B — but that is a passive resistance change in a dried composite, not an electrophysiological signal from a living fungus.

The authors head off the obvious misreading themselves: the device works on a minute-scale timescale by design and suits long exposure rather than real-time control. Recovery lags uptake because desorption and adsorption proceed at different rates. In a demonstration of crawling along a ratcheted surface, a 5 × 30 mm strip was humidified for five minutes and took an hour to regain its shape.

Dormant is not the same as growing

Nothing here is alive in the sense of being at work. The cells sit metabolically quiet but viable: propidium iodide staining found membrane damage in both hyphae and spores, yet MTT assays produced the characteristic purple formazan and Mito-Tracker Red CMXRos labelled intact mitochondria in the hyphae. After three months of dry storage, samples transferred to agar regrew — visible from the fifth day, covering the plate by the tenth.

Three months is the measured span of survival in dry dormancy. The one-year figure does different work: the authors give it as the longest dry storage that should be allowed if the spores are to stay alive, and separately report that a used macerate left outside sterile conditions for a year could still be rescued on an antibiotic medium and grown back into Trametes versicolor. Neither is a demonstrated service life for a finished object.

Regeneration is likewise not self-healing in service. A cut sheet only knitted together after being laid on nutrient medium with the cut faces touching, at 28 °C and 85 % humidity. Fragments dropped into liquid medium with antibiotics rebuilt a fresh membrane in twenty days. The first regrown generation actually outperformed the original sheet; the decline became clear only by the third, and even that one could be recycled into a sheet exceeding 200 % elongation.

Where the paper stops

There is no product here, and the authors do not pretend otherwise. There is a life-cycle assessment — cradle to gate, to ISO 14040 and 14044, for one kilogram of material — which makes it worth saying what it leaves out: transport, use and disposal. It locates the carbon hotspots in the process and suggests that a lower PVA loading and bio-based binders would bring them down; it does not settle whether such a sheet beats anything it might replace. There is no washing, abrasion or in-service durability data, and no manufacture beyond a single vessel. What there is instead is a measured mechanism, with its limits printed alongside it, in a freely readable full text.

The underlying behaviour needs no laboratory to observe. A dried turkey tail bracket picked up on a walk stiffens in a heated room and softens after rain, and thin polypore fruit bodies curl and flatten with the weather. What is new is that the asymmetry has been designed deliberately, quantified, and published together with everything it fails to do.