Mushrooms use convectively created airflows to disperse their spores

We analyzed spore deposition beneath cultured mushroom (shiitake, Lentinula edodes, and oyster, Pleurotus ostreatus, sourced from CCD Mushroom and Oley Valley Mushroom Farm) as well as wild-collected Agaricus californicus. Pilei were placed on supports to create controllable gap heights beneath the mushroom and placed within boxes to isolate them from external airflows. We measured spore dispersal patterns by allowing spores to fall onto sheets of transparency film and photographing the spore deposit (Fig. 1C). In all experiments, spore deposits extended far beyond the gap beneath the pileus. Spores were deposited in asymmetric patterns, both for cultured and for wild-collected mushrooms (Fig. 1D). Using a laser light sheet and high-speed camera (Materials and Methods), we directly visualized the flow of spores leaving the narrow gap beneath a single mushroom (Fig. 1E): Movie S1 shows that spores continuously flow out from thin gaps, even in the absence of external winds.

What drives the flow of spores from beneath the pileus? Some ascomycete fungi and ferns create dispersive airflows by direct transfer of momentum from the fruiting body to the surrounding air. For example, some ascomycete fungi release all of their spores in a single puff; the momentum of the spores passing through the air sets the air into motion (4). Fern sporangia form overpressured capsules that rupture to create jets of air (16). However, the flux of spores from a basidiomycete pileus is thousands of times smaller than for synchronized ejection by an ascomycete fungus, and pilei have no known mechanism for storing or releasing pressurized air. The only mechanism that we are aware of for creating airflows without momentum transfer is by the manipulation of buoyancy—an effect that underpins many geophysical flows (17) and has recently been tapped to create novel locomotory strategies (18).

Many mushrooms are both cold and wet to the touch (19); the expanding soft tissues of the fruit body are hydraulically inflated, but also lose water quickly (Fig. 2A). We made a comparative measurement of water loss rates from living mushrooms and plants. The rate of water loss from mushrooms greatly exceeds water loss rates from plants, which use stomata and cuticles to limit evaporation (Fig. 2B). For both plants and pilei, rates of evaporation were larger when tissues were able to actively take in water via their root system/mycelium than for cut leaves or pilei. However, the pilei lost water more quickly than all species of plants surveyed under both experimental conditions. In fact, evaporation rates from cut mushrooms were comparable to those from a sample of water agar hydrogel (1.5% wt/vol agar), whereas evaporation rates from mushrooms with intact mycelia were twofold larger (Fig. 2B). Taken together, these data suggest that pilei are not adapted to conserve water as effectively as the plant species analyzed.

The high rates of evaporation lead to cooling of the air near the mushroom and may have an adaptive advantage to the fungus. Previous observations have shown that evaporation cools both the pileus itself and the surrounding air by several degrees Celsius (20). Specifically, because latent heat is required for the change of phase from liquid to vapor, heat must continually be transferred to the mushroom from the surrounding air. We compared the ambient temperature of the air between 20 cm and 1 m away from P. ostreatus pilei with intact mycelia, with temperatures in the narrow gaps between and beneath pilei, using a Traceable liquid/gas probe (Control Company). We found that gap temperatures were consistently 1–2 °C cooler than ambient temperature (Fig. 2C). The surface temperature of the pileus, measured with a Dermatemp infrared thermometer (Exergen), was up to 4 °C cooler than ambient temperature, consistent with previous observations (ref. 20 and Fig. 2C). Evaporation alone can account for these temperature differences: In a typical experimental run, a pileus loses water at a rate of $3\times {10}^{-5}$ kg⋅m⁻²⋅s⁻¹ (comparable with the data of ref. 21). Evaporating this quantity of water requires $E_{\text{vap}}\approx 70$ W/m² of vaporization enthalpy. At steady state the heat flux to the mushroom must equal the enthalpy of vaporization; Newton’s law of cooling gives that the heat flux (energy/area) will be proportional to the temperature difference, $\Delta$ between the surface of the mushroom and the ambient air: $E_{\text{vap}}=h\Delta T$, where $h=10-30$ W/m² °C is a heat transfer coefficient (22). From this formula we predict that $\mathrm{\Delta }T\approx 2.5-7$ °C, in line with our observations.

Cooling air from the laboratory ambient termperature ($T=18$ °C) down to $T=16$ °C increases the density of the air by $\mathrm{\Delta }{\rho }_T=\alpha \mathrm{\Delta }T=0.008$ kg⋅m⁻³, where α is the coefficient of expansion of air. Cold dense air will tend to spread as a gravity current (23), and an order of magnitude estimate for the spreading velocity of this gravity current from a gap of height $h=1$ cm is given by von Karman’s law: $U_g\sim \sqrt{2\mathrm{\Delta }{\rho }_{T}gh/{\rho }_0}\approx 4$ cm/s. Although the air beneath the pileus is laden with spores, spore weight contributes negligibly to the creation of dispersive air flows: In typical experiments, spores were released from the pileus at a rate of $q=$ 540 $\pm$ 490 spores⋅cm⁻²⋅s⁻¹ (data from 24 P. ostreatus mushrooms). If the mass of a single spore is $m_s=5\times {10}^{-13}$ kg and its sedimentation speed is $v_s={10}^{-3}$ m/s (see discussion preceding Eq. 1), then the contribution of spores to the density of air beneath the pileus is $\Delta {\rho }_S=m_{s}q/v_s=0.003$ kg⋅m⁻³, less than half of the density increase produced by cooling, $\mathrm{\Delta }{\rho }_T$. Indeed, water evaporation, rather than spores and the water droplets that propel them (11), constitutes most of the mass lost by a mushroom. To prevent spore ejection, we applied a thin layer of petroleum jelly to the gill surfaces of cut mushrooms. Treated mushrooms lost mass at a statistically indistinguishable rate from that of cut mushrooms that were allowed to shed spores (Fig. 2B).

Although our experiments were performed in closed containers to exclude external airflows, it is still possible that spore deposit patterns were the result of convective currents created by temperature gradients in the laboratory, rather than airflows created by the mushroom itself. To confirm that spores were truly dispersed by airflows created by the mushroom we rotated the mushroom either 90° or 180° halfway through the experiment and replaced the transparency sheet. Because the box remained in the same orientation and position in the laboratory, we would expect that if spores are dispersed by external airflows, the dispersal pattern would remain the same relative to the laboratory. In fact we found consistently that the direction of the dispersal current rotated along with the mushroom (Fig. S1), indicating that mushroom-generated air flows are dispersing spores.

We explored how the distance over which spores dispersed depended on factors under the control of the parent fungus. The distance spores dispersed from the pileus increased in proportion to the square of the thickness of the gap beneath the pileus ($R^2=0.90$, Fig. 3). However, we found no correlation between spore dispersal distance and the diameter of the pileus or the rate at which spores were produced ($R^2=0.07$ and $R^2=0.17$ respectively, Fig. S2).

Spores were typically deposited around mushrooms in asymmetric patterns, suggesting that one or two tongues of spore-laden air emerge from under the pileus and spores do not disperse symmetrically in all directions (Fig. 1C). These tongues of deposition were seen in wild-collected as well as cultured mushrooms (Fig. 1D). We dissected the dynamics of one of these tongues by building a 2D simulation of the coupled temperature and flow fields around the pileus. Although real dispersal patterns are 3D, these simulations approximate the 2D dynamics along the symmetry plane of a spreading tongue. In our simulations we used a Boussinesq approximation for the equations of fluid motion and model spores as passive tracers, because their mass contributes negligibly to the density of the gravity current. Initially we modeled the pileus by a perfect half-ellipse whose diameter (4 cm) and height (0.8 cm) matched the dimensions of a L. edodes pileus used in our experiments. However, if cooling was applied uniformly over the pileus surface, then spores dispersed weakly (Fig. 4A). Weak symmetric dispersal can be explained by conservation of mass: Cold outward flow of spore-laden air must be continually replenished with fresh air drawn in from outside of the gap. In a symmetric pileus, the cool air spreads along the ground and inflowing air travels along the undersurface of the pileus. So initially on leaving the gills of the mushroom, spores are drawn inward with the layer of inflowing warm air; and only after spores have sedimented through this layer into the cold outflow beneath it do they start to travel outward (Fig. 4A, Top).

To understand how mushrooms can overcome the constraints associated with needing to maintain both inflow and outflow, we performed a scaling analysis of our experimental and numerical data. Although the buoyancy force associated with the weight of the cooled air draws air downward, warm air must be pulled into the gap by viscous stresses. For fluid entering a gap of thickness h, at speed U, the gradient of viscous stress can be estimated as $\sim \eta U/h^2$, where η is the viscosity of air. We estimate the velocity U by balancing the viscous stress gradient with the buoyancy force, $\rho g\alpha \mathrm{\Delta }T$; i.e., $U\sim \rho g\alpha \mathrm{\Delta }T{h}^2/\eta$. We then adopt the notation that if f is a quantity of interest (e.g., temperature or gap height) that can vary over the pileus, then we write $f_r$ for the value of f on the right edge of the pileus and $f_{\ell }$ for its value on the left edge of the pileus. If $U_r=U_{\ell }$, then there is the same inflow on the left and right edges of the pileus, and dispersal is symmetric and weak.

If there are different inflows on the left or right side of the pileus, then there can be net unidirectional flow beneath the pileus, carrying spores farther. Assuming, without loss of generality, that the net dispersal of spores is rightward, the spreading velocity of the gravity current can be estimated from the difference: $U_g\sim U_{\ell }-U_r$ (right-moving inflow minus left-moving inflow). The farthest-traveling spores originate near the rightward edge of the pileus and fall a distance $h_r$ (the gap width on the rightward edge) before reaching the ground. Because the gravity current spreads predominantly horizontally, the vertical trajectories of spores are the same as in still air; namely they sediment with velocity $v_s$ and take a time $\sim h_r/v_s$ to be deposited. By balancing the weight of a spore against its Stokes drag, we obtain $v_s=2\rho a^{2}g/9\eta$, where $\rho =1.2\times {10}^3$ kg/m³ is the density of the spore, and $a=2-4$ μm is the radius of a sphere of equivalent volume (4). The sedimentation velocity, $v_s$, can vary between species (typically $v_s=1-4$ mm/s), but does not depend on the flow created by the pileus. The maximum spore dispersal distance is thenwhere we use the notation ${\left[f\right]}_r^l$ to denote the difference in the quantity f between the left and right sides of our model mushroom. Unidirectional dispersal therefore requires either that $\mathrm{\Delta }{T}_{\ell }\ne \mathrm{\Delta }{T}_r$ (i.e., there is a temperature gradient between the two sides of the pileus, Fig. 4A, Middle) or that $h_{\ell }\ne h_r$ (i.e., the mushroom is asymmetrically shaped, Fig. 4A, Bottom). These two cases can be distinguished by the dependence of dispersal distance upon the gap height: For a temperature-gradient–induced asymmetry we predict $d_{\text{max}}\propto {\left[\mathrm{\Delta }T\right]}_r^{\ell }{h}_r^3$ (and $U_g\propto {\left[\mathrm{\Delta }T\right]}_r^{\ell }{h}_r^2$), whereas for a shape-induced asymmetry, $d_{\text{max}}\propto \mathrm{\Delta }T{\left[h\right]}_r^{\ell }{h}_r^2$ (and $U_g\propto \mathrm{\Delta }T{\left[h\right]}_r^{\ell }{h}_r$). Both scalings are validated by numerical simulations (Fig. 4B). When rescaled using Eq. 1, data from simulations with different values of $\mathrm{\Delta }{T}_{\ell }$, $\mathrm{\Delta }{T}_r$, $h_{\ell }$, $h_r$, ${\left[\mathrm{\Delta }T\right]}_r^l$, ${\left[h\right]}_r^l$, or $v_s$ all collapsed to two universal lines corresponding to temperature and height asymmetries (Fig. 4 C and D).

Experimental data from real mushrooms fits well with $d_{\text{max}}\sim h_r^2$ (Fig. 3). Additionally the scaling [1] accords with experimental observations that dispersal distance does not depend on pileus diameter or on the rate of spore release (Fig. S2). We cannot directly confirm that there are not temperature gradients over the pileus surface. But these data are consistent with shape asymmetry, that is, variation in the thickness of the pileus and height of the gap playing a dominant role in asymmetric spore dispersal from real mushrooms. Additionally, we made another direct test of our scaling law [1] by measuring $U_g$ (Fig. 4E and Materials and Methods). When the dimensionless prefactors in [1] were kept constant by using the same pileus, but the height of the gap, $h_r$, was varied, we found that $U_g\sim h_r$, consistent with the scaling for shape asymmetry-driven flow (Fig. 4F).

In nature, pilei may grow crowded together or under plant litter or close to the host or substrate containing the parental mycelium (Fig. 1 A and B); thus, in addition to needing to disperse from the narrow gap beneath the pileus, spores may potentially also need to climb over barriers to reach external airflows. Although the cold, spore-laden air is denser than the surrounding air, to replace the cold air that continuously flows from beneath the pileus, warm air must be drawn in to the pileus. Our simulations showed that when the gravity current met a solid barrier, the warm inflow and cold outflow could link to form a convective eddy (Fig. S3). We studied whether this eddy could lift spores into the air and what effect this might have upon spore dispersal. We found that when mushrooms were surrounded by a vertical barrier (Materials and Methods), spores were dispersed over the barrier (Fig. 5A) provided that their horizontal range, predicted using Eq. 1, exceeded the height of the barrier (Fig. 5B).

Surprisingly, spores climbing over the barrier were dispersed apparently symmetrically in all directions from the top of the barrier (compare Fig. 5A with Fig. 1C) over the entire area of the box containing the mushroom and barrier (Fig. 5A). In particular, the total horizontal extent of the spore deposit (the size of the box) typically greatly exceeded the horizontal range predicted by Eq. 1, even without accounting for the distance traveled by the gravity current in crossing over the barrier.

The enhancement of spore dispersal by a barrier can be attributed to the action of the recirculating eddy: Spores that climb the barrier may enter the current of air that is pulled down to the mushroom to replace the cold spore-laden air (Fig. 5C and Movie S2). This eddy can carry spores up and farther away from the barrier, and it is likely that no longer being constrained to travel along the ground surrounding the mushroom contributes to their increased range.

Why does the height that gravity currents need to climb up the barrier not reduce their dispersal distance? In a climbing gravity current spores continue to sediment vertically downward. Because the vertical velocity of a climbing gravity current ($U_g\sim 4$ cm/s) is typically much larger than the sedimentation speed ($v_s\sim 1$ mm/s), spores do not sediment out of the vertical gravity current as it climbs. To test this idea quantitatively, we numerically simulated spore dispersal by a gravity current that encounters a barrier inclined at an angle θ to the horizontal. Because the velocity of the gravity current is directed parallel to the barrier, spores can still sediment toward the barrier, but at a reduced sedimentation velocity $v_s\cos \theta$. (Sedimentation parallel to the barrier, with velocity $v_s\sin \theta$, can be neglected.) Revisiting Eq. 1 we therefore predict that spores will disperse a distance $U_g{h}_r/\left(v_s\cos \theta \right)$; i.e., they will travel a factor of $1/\cos \theta$ farther up the slope than they would travel horizontally, and this prediction agrees quantitatively with numerical simulations (Fig. 5C). In particular, if $\theta =90°$ (a vertical wall), we predict no sedimentation onto the wall, consistent with experimental observations that the distance that a gravity current climbs up a vertical wall does not reduce its horizontal spreading.

Finally we note that, although our experiments were designed to exclude external winds, in nature wind speed tends to increase with height above the ground (14), so spores that travel upward and then away from the vertical wall may be more likely to reach dispersive winds.

We compared rates of water loss between L. edodes and P. ostreatus mushrooms and plants [one potted specimen of each of Tagetes erecta, Ocimum basilicum, Salvia elegans, Chrysanthemum sp., and Dahlia variabilis, acquired from Home Depot, and cut leaves of Cerces canadensis, Pittosporum tobira, T. erecta, and Trachelospermum jasminoides, collected on the University of California, Los Angeles (UCLA) campus]. To measure the rate of evaporation from mushrooms with intact mycelia or from whole plants we wrapped the pot or the log on which the mushroom grew in plastic, so that water was lost only through the pileus/leaves. We also compared with water agar (1.5% wt/vol) samples that were poured into Petri dishes and left uncovered. All samples were allowed to dry in a laboratory ambient temperature (19 °C, 60% relative humidity). We measured surface area of samples by cutting the leaves and pilei into pieces and photographing the pieces.

For spore dispersal measurements mushroom caps were placed in closed containers to isolate them from surrounding airflows. Depending on the gap height and length of gravity current, we used 15-cm diameter suspension dishes, food storage containers to which we added cardboard drop ceilings, or file boxes. To prevent condensation from forming under the mushroom pileus we added a small quantity ($\sim$ 0.2 g) of desiccant (DampCheck) to each container. Mushroom gap height was controlled by balancing the mushrooms on skewers spanning between two cardboard trestles, short lengths of tubing, or wire connectors or by suspending them from the ceiling of the container. We measured both the distance dispersed by the spores and the asymmetry of the spore print. To calculate the print asymmetry in Fig. 1D, we traced, using ImageJ, the coordinates of the boundary of the spore print. We then calculated asymmetry moments $c_n={\displaystyle \int e^{ni\varphi }\hspace{0.17em}dA}$, where the integral is carried out over the area of the spore print, and ϕ is the angle made between the line joining each point to the centroid of the originating pileus and an arbitrary orientation axis. We characterized the asymmetry using a parameter $\text{Asym}=\left(\left|c_1\right|+\left|c_2\right|\right)/c_0$.

Spore dispersal was directly observed by illuminating the pileus with a laser sheet created by expanding a vertical laser beam from a 0.5-W Hercules-450 laser (LaserGlow) with a plano-concave lens (f = −7.7 mm; ThorLabs). The mushroom pileus was placed inside an 18 $\times$ 22 $\times$ 22-mm box: The ceiling and three walls of the box were lined with matte black aluminum foil (ThorLabs), the fourth wall was constructed of transparent acrylic, and the floor of the container was covered with photography velvet to minimize scatter from the laser. The laser light sheet passed through a slit in the ceiling of the box to illuminate the rim of the mushroom cap. Spores traveling through the plane of the light sheet acted as bright tracer particles of the flow, and we filmed their movement using a FASTCAM SA3 high-speed camera (Photron). After 2–4 s the laser heating created a thermal plume that obscured the airflows created by the mushroom. Accordingly, we analyzed only the first 2 s of movie captured from each experiment and waited 5 min to allow the surface to return to temperature equilibrium between runs of the experiment. Although our observation time was necessarily short, the time between observations was long enough for even slow overturning eddies to emerge. Spore velocities were measured by using a hybrid of particle imaging velocimetry (based on the code of ref. 26) to measure the velocities of groups of spores and individual particle tracking (27).

To measure the ability of mushroom-created airflows to disperse spores over barriers, we surrounded mushrooms by cylindrical walls, formed by taping a strip of transparency sheet inside of a 15-cm diameter Petri dish (Fig. 5A). By varying the width of the strip, we could vary the height of the barrier. To measure whether spores could cross this barrier, we placed a transparency sheet on the floor of the Petri dish (inside the barrier) and on the bench top outside of the barrier. To measure spore deposition just inside and just outside of the barrier, we cut annuli of diameter 15 mm from the two transparency sheets. The two annuli were cut into smaller pieces and then separately vortexed in 10 mL deionized water to wash off spores. The concentration of spores from inside and outside the barrier was counted automatically by the following method: Spore suspensions were added to a hemocytometer to create samples with known volume, which were photographed at 260$\times$ magnification, using a Zeiss AxioZoom microscope. Spores appear as dark ellipses with bright outlines. Images were first denoised using median filtering and contrast enhanced using a Laplacian of Gaussian filter. Images were thresholded to identify bright edges and dark spore interior regions. Spores were segmented morphologically, by first removing all edges shorter than 30 pixels and then all dark regions with diameter smaller than 20 pixels or larger than 80 pixels. Spores were identified by dilating the dark regions, using a disk element with a radius of 5 pixels, and keeping only those regions that overlapped with a detected bright edge. We then subtracted the bright edges from the dark regions, to create one disconnected region per spore. The number of disconnected regions in each image was then counted. The ratio of the number of spores in the outer annulus to the number in the inner annulus gives a measure of the fraction of spores dispersed over the barrier. Because there is a much larger total area of spore fall outside of the barrier than inside, the barrier-crossing rate is proportional to, but not equal to, the fraction of spores that crossed the barrier.

To simulate the trajectories of spores in convective currents, we used the Boussinesq approximation. Specifically to calculate the inertia of the gravity current we take the density of air to be constant ${\rho }_0$, but when calculating the buoyancy force, we use a linear model, $\rho \left(T\right)={\rho }_0-{\rho }_0\alpha T$, where T is the temperature measured relative to the ambient temperature far from the mushroom (so that $T\to 0$ far from the pileus), and α is a coefficient of expansion: $\alpha =3.42\times {10}^{-3}$/K. We nondimensionalize our equations by scaling all distances by L, the half-width of the mushroom; all temperature differences by $\Delta T$, the maximum temperature drop at the mushroom surface; velocities by $U^{\ast }={\left(\alpha gL\Delta T\right)}^{1/2}$; and pressure by ${\rho }_0{U}^{\ast 2}$. The temperature and velocity fields around the pileus satisfy coupled partial differential equations (PDEs):

Spore trajectories were calculated by solving the ordinary differential equations $\dot{x}=u$, $\dot{y}=v-v_s/U^{\ast }$, where $v_s$ is the spore sedimentation speed. Eq. 2 contain two dimensionless numbers: the Reynolds number $\text{Re}\equiv {\rho }_0{U}^{\ast }L/\eta$, formed from our velocity and length scales and the viscosity η of air, and the Péclet number $\text{P}é\equiv U^{\ast }L/\kappa$, which depends on the thermal diffusivity, κ. In a typical simulation $\text{Re}=60$ and $\text{P}é=40$. We used Comsol Multiphysics (COMSOL) to set up and solve the PDEs in a 2D domain, in which the pileus was modeled by a semiellipse with a no-slip, constant temperature boundary condition on its surface. The pileus was set a small distance above a no-slip floor, and the domain was closed by a semicircle on which the no-slip boundary condition was imposed and whose diameter is 40$\times$ larger than the cap diameter. Constant temperature boundary conditions were applied on all surfaces, with $T=0$ on the walls of the domain, and cooling was applied on the pileus surface (see main text for the two principal boundary conditions used). To avoid creating convective overturning of fluid in the gap beneath the pileus, we isolated a section of the floor directly beneath the mushroom and applied a zero-flux boundary condition there. To model the effect of nearby walls, we replaced the semicircular external boundary by boundaries that were set a distance 1.5 L from the midline of the mushroom and oriented at an angle θ to the horizontal, as described in the main text.