## SUMMARY

Most basidiomycete fungi actively eject their spores. The process begins
with the condensation of a water droplet at the base of the spore. The fusion
of the droplet onto the spore creates a momentum that propels the spore
forward. The use of surface tension for spore ejection offers a new paradigm
to perform work at small length scales. However, this mechanism of force
generation remains poorly understood. To elucidate how fungal spores make
effective use of surface tension, we performed a detailed mechanical analysis
of the three stages of spore ejection: the transfer of energy from the drop to
the spore, the work of fracture required to release the spore from its
supporting structure and the kinetic energy of the spore after ejection.
High-speed video imaging of spore ejection in *Auricularia auricula*
and *Sporobolomyces* yeasts revealed that drop coalescence takes place
over a short distance (∼5 μm) and energy transfer is completed in less
than 4 μs. Based on these observations, we developed an explicit relation
for the conversion of surface energy into kinetic energy during the
coalescence process. The relation was validated with a simple artificial
system and shown to predict the initial spore velocity accurately (predicted
velocity: 1.2 m s^{–1}; observed velocity: 0.8 m
s^{–1} for *A. auricula*). Using calibrated
microcantilevers, we also demonstrate that the work required to detach the
spore from the supporting sterigma represents only a small fraction of the
total energy available for spore ejection. Finally, our observations of this
unique discharge mechanism reveal a surprising similarity with the mechanics
of jumping in animals.

## INTRODUCTION

Most basidiomycetes, including many edible mushrooms, actively disperse
their spores through a mechanism known as ballistospory
(Buller, 1909-1950;
Ingold, 1939). The spores, or
ballistospores, are borne on the gills of mushroom caps or equivalent
reproductive structures (Fig.
1A). Each spore develops on an outgrowth known as the sterigma to
which it is attached *via* the hilum – a constriction of the
sterigma that works as an abscission zone
(Fig. 1B,C). Spore ejection is
preceded by the condensation of Buller's drop at the hilar appendix located on
the proximal end of the spore (Fig.
1D,E). Buller's drop is nucleated by the secretion of hygroscopic
substances (such as mannitol) that decrease the vapor pressure of the
incipient droplet (Webster et al.,
1995). In the meantime, a film of water develops on the spore
probably following a similar process. When the drop reaches a critical size,
it touches the water film on the spore surface. At this point, surface tension
quickly pulls the drop onto the spore thus creating the necessary momentum to
detach the spore from the sporogenic surface. The spore can then fall freely
under the action of gravity. Upon emerging from the cap, the spore is carried
away by air currents to a distant location where it can germinate to produce a
new mycelium and, ultimately, new mushrooms.

Surface tension is almost imperceptible at length scales at which humans
operate. However, at microscopic length scales, surface tension forces
dominate over the force of gravity. This fact can be understood from a simple
scaling argument. The force of gravity on an object such as a spore scales as
*F*_{g} ∼ ρ*gR*^{–3}, where ρ
is the density of the object, *g*=9.8 m s^{–2} is the
gravitational acceleration and *R* is the characteristic length of the
object. By contrast, the surface tension force is
*F*_{γ}∼γ*R*, where γ is the
liquid's surface tension (γ=72×10^{–3}
Nm^{–1} for water at room temperature). Considering the ratio of
these forces: *F*_{γ}/*F*_{g} ∼γ
/ρ*gR*^{2}; it can be seen that as *R* gets
small, the surface tension force becomes increasingly important and dominates
the force of gravity for *R* smaller than 1 mm. This simple phenomenon
has profound consequences on the release of spores. The dispersal of most
fungal spores by wind requires that the spores be small thus making the force
of gravity inconsequential compared with adhesion forces. As a result, spores
tend to cling to each other and to the gills of mushroom caps. Active spore
ejection provides a solution to this problem, which explains the great
diversity of mechanisms for spore release in fungi and nonvascular plants
(Straka, 1962). However,
unlike other active dispersal mechanisms, which involve mass release of spores
from specialized launching structures, ballistospores are self-propelled by
water.

Given that a large mushroom can shed spores at the astonishing rate of 40 million spores per hour (Buller, 1909-1950); the release of ballistospores has rightfully attracted some attention (Buller, 1909-1950; Ingold, 1939; Money, 1998). As early as 1939, Ingold determined that the surface energy in Buller's drop is sufficient to account for the kinetic energy of the spore (Ingold, 1939). He, however, concluded his discussion of the topic remarking that `although there appears to be sufficient surface energy to discharge the spore it is not too easy to see how this energy could be mobilized to bring about discharge' (Ingold, 1939). More recently, Turner and Webster (Turner and Webster, 1991) were able to predict the initial spore velocity with respectable accuracy based on a few judicious assumptions. The development of high-speed video cameras and their recent application to visualize ballistospore ejection (Pringle et al., 2005) provide, for the first time, a way to address Ingold's question with direct measurement of all key parameters in the problem.

Here, we present a detailed analysis of how surface tension is used for
spore ejection in *Auricularia auricula* (`tree ears') and
*Sporobolomyces* yeasts. In particular, we quantify the forces and
energies of the three stages of the ejection process: the transfer of surface
energy from the drop to the spore, the work of fracture required to release
the spore from the sterigma and the kinetic energy of the spore after
ejection. Our analysis reveals an exquisite fine-tuning of the different
stages that yields a surprisingly high efficiency for the transfer of energy
from Buller's drop to the spore.

## MATERIALS AND METHODS

### Specimen preparation

To initiate spore development and spore discharge, dehydrated
*Auricularia auricula* (Fr.) J. Schrot fragments were first imbibed on
a wet towel and then kept under humid conditions with the fertile surface
facing downward. After a few hours, spore ejection had begun as indicated by
the presence of white spores on the bottom of the dish. We cut thin vertical
sections (0.5 mm) of the fungus and laid them flat on a microscope slide
covered with a thin (100–200 μm) layer of 2% agar. Sterigmas were now
oriented horizontally so that spores were ejected perpendicular to the optical
axis of the microscope. Spores from yeast-like species were isolated from
leaves. Although the yeasts were not identified to the species level, they are
members of the Urediniomycetes (the rust fungi), likely to be of the genus
*Sporobolomyces*. The yeasts were plated from a primary culture onto a
thin layer of a 2% nutrient agar. After a few days, the spores germinated to
form hyphae, sterigmas and new spores. Our yeast cultures may have included
more than one species but we found little quantitative differences between the
different cultures. Therefore, for simplicity, we are treating all samples as
a single taxon. All experiments were performed on *A. auricula* and the
yeast species, except for the work of fracture of the hilum, which was
performed on *A. auricula* only.

### Microscopy and imaging

All imaging was done in transmitted light with ×20 and ×40
objectives. Images were captured with a Phantom V7.0 (Wayne, NJ, USA) or a
Photron Ultima APX-RS (San Diego, CA, USA) high-speed camera at a frame rate
of up to 250,000 frames s^{–1} and exposure times as short as 1μ
s. The high acquisition rate necessary to capture spore ejection can be
achieved only when image resolution is low (typically 32×128 pixels).
Although our analyses were performed on these raw images, the frames from the
time-lapse sequences are presented in the figures at higher resolution to
improve clarity. We include as supplementary material three movies (AVI
format) for *A. auricula* and one for the *Sporobolomyces*
yeasts (see Movies 1–4 in supplementary material). The frame rates for
Movies 1–4 are, respectively 90,000 frames s^{–1}; 80,000
frames s^{–1}; 250,000 frames s^{–1}; 90,000
frames s^{–1}.

### Spore ballistics

We developed image analysis routines in Matlab (The MathWorks, Natick, MA,
USA) to track the centroid of the spore and the rotation of the spore's major
axis over the entire trajectory. Although spore translation in *A.
auricula* and *Sporobolomyces* yeasts could be tracked reliably in
all time-lapse sequences, only *A. auricula* offered two spores with
rotation confined to the imaging plane that could thus be analyzed for their
angular velocity. The *Sporobolomyces* yeasts could not be positioned
such that the spore trajectory was confined to the focal plane of the
microscope; the spores thus moved quickly out of focus. To compute the spore
velocity, we used a 3-D tracking algorithm that relies on the size of the
out-of-focus spore to infer its vertical position. The calibration for the
vertical position was obtained by imaging particles at known vertical
displacements above or below the focal plane and recording the size of the
out-of-focus particles.

As we shall show in the Results section, the Reynolds number (*Re*)
for spore ejection is small. Therefore Stokes' law provides a good description
of the drag force acting on the spore
(Happel and Brenner, 1983).
Assuming a spherical spore, the ballistic trajectory of the spore will thus be
governed by the following force balance:
*D*=6πμ*Rv*=*ma*, where *D* is the drag
force, *R, m, v* and *a* are, respectively, the mean radius,
mass, velocity and acceleration of the spore (including the fused drop), andμ
=1.84×10^{–5} Pa s is the dynamic viscosity of air.
The force balance equation can be rearranged to give:
(1)
Integrating gives
*v*(*t*)=*V*_{0}exp(–*t*/τ_{T}),
where the characteristic decay time associated with the translational velocityτ
_{T}=*m*/6πμ*R* and *V*_{0} is
the initial velocity of the spore. We can integrate again to find the spore
position along the axis of discharge (*x*) assuming that
*x*(0)=0:
(2)
A similar equation can be derived for the viscous dissipation associated with
the rotation of the spore:
(3)
where α is the angular position of the spore, Ω_{0} is the
initial angular velocity and τ_{R}=*m/*20πμ*R*
is the characteristic decay time for the rotation of the spore
(Happel and Brenner, 1983).
Eqns 2 and
3 were used to fit the observed
spore trajectories and infer the parameters *V*_{0},Ω
_{0}, τ_{T} and τ_{R}.

### Measurement of rupture force

The rupture force of the hilum in *A. auricula* was measured with
custom-made micropipettes calibrated on an analytical balance (0.1 μN
precision). Using a micromanipulator, a micropipette was brought into contact
with the top of the spore, perpendicular to the sterigma. A water film
provided adhesion between the spore and the glass micropipette. In some
experiments, we also used poly-l-lysine-coated micropipettes to
enhance adhesion. The micropipette was then displaced slowly until the spore
detached from the sterigma or until the adhesion between the spore and pipette
failed. The force was calculated from the deflection of the micropipette with
an error of ±5%. To infer the spring constant of the sterigma, we
measured its elongation δ just prior rupture (error of 10%).

### Surface energy available for spore ejection

The energy available to eject the spore comes from the surface energy
stored in Buller's drop. For *A. auricula*, the surface energy freed
during the fusion process (Δ*E*_{p}) can be calculated
from the coalescence of a spherical drop onto a plane
(Fig. 2A). The energy is equal
to the difference in surface area of the spore–drop system before and
after coalescence, i.e.:
(4)
where γ_{SV}, γ_{SL} and γ are the energies
associated with the spore–vapor, spore–liquid and
liquid–vapor interfaces, respectively. *R*_{D} is the
radius of the drop before fusion, *A*_{S} is the area of the
spore covered by the drop after fusion and *A*_{D} is the drop
surface area after fusion.

Using Young's law for the contact angle
(γ_{SV}=γ_{SL}+γcosθ)
(de Gennes et al., 2003), we
have:
(5)
The coalesced drop is a spherical cap of radius
and contact angle θ for
which the area is
,
the volume is
and the projected area onto the spore is
.
Then:
(6)
From the conservation of the volume, one can write:
(7)
Therefore, the surface energy available for spore ejection is:
(8)
As would be expected, Δ*E*_{p} is proportional to the
total surface area of Buller's drop
() times a factor that accounts
for the degree of spreading of the drop onto the spore
().
The surface energy for the nearly spherical spores of the
*Sporobolomyces* is easy to derive assuming that Buller's drop envelops
the spore (Fig. 2B).

### Error analysis

The main error in our experimental observations comes from the length
measurements made on video images. These measurements are used to assess the
spore and drop radii and for calculating their volumes. The length
measurements were precise to ±0.5 pixels whereas the diameter of the
drop was <7 pixels and the spore's dimensions were ∼8×15 pixels.
As seen in Eqn 8, the prediction
of the freed surface energy depends on two length measurements:
*R*_{D} and .
Prediction of the initial spore velocity *V*_{0} requires in
addition the width (*W*_{S}) and length
(*L*_{S}) of the spore. The absolute error on the velocity
estimate (Δ*V*_{0}) is given by the following equation:
(9)
where
pixels are the absolute errors for the length measurements. Using
Eqn 9, we find that the relative
error on the predicted velocity isΔ
*V*_{0}/*V*_{0}=22%.

## RESULTS

Spore ejection is best described by first analyzing the spore ballistics to infer the spore initial velocity and kinetic energy. We then proceed to a mechanical analysis of the stages that precede ejection.

### Spore ballistics – the `sporabola'

To quantify the initial velocity and kinetic energy present in the spore at
the moment of ejection, we analyzed the ballistic trajectory of the spore
(Figs 3 and
4; Movies 1–4 in
supplementary material). Buller coined the word `sporabola' to describe the
particular trajectory followed by the spore
(Buller, 1909-1950). The shape
of the sporabola results from the interplay of gravity and viscous forces
acting on the spore. The Reynolds number at ejection is
*Re*=*V*_{0}*L*_{S}/ν≈0.5, whereν
is the kinematic viscosity of air (1.4×10^{–3}
m^{2} s^{–1}). Given the small Reynolds number, Stokes'
law provides a good description of the drag force acting on the spore
(Happel and Brenner, 1983). The
spore position along the axis of discharge *x* is thus given by (see
Materials and methods section):
(10)
The spore rotation, clearly seen in Fig.
3A,D, is also damped by air viscosity. The angular position is:
(11)
As shown in Fig. 3, these
relationships fit the data very well and yield, for the spore shown in
Fig. 3A, an initial velocity of
*V*_{0}=0.8 m s^{–1}, an angular velocity ofΩ
_{0}=9×10^{4} rad s^{–1} and decay
times of τ_{T}=184 μs and τ_{R}=66 μs (see
Table 1 for a summary of the
data). Fig. 4 shows spore
ejection in a *Sporobolomyces* yeast. For this sequence, the spore
velocity is *V*_{0}=1.6 m s^{–1}.

According to Stokes' law, the decay times areτ
_{T}=*m*_{SD}/6πμ*R* andτ
_{R}=*m*_{SD} */*20πμ*R*,
where *R* and *m*_{SD} are, respectively, the mean
radius and mass of the spore–drop complex (see Materials and methods)
(Happel and Brenner, 1983). We
can therefore compare the decay times inferred from the fitted spore
displacement in Fig. 3 with
those predicted by the theory. For the decay time associated with the
translational velocity (τ_{T}), the mean ratio of the measured
over the predicted decay times is 0.91 (standard deviation: σ=0.08, for
*N*=4 measurements). For the decay time associated with the angular
velocity (τ_{R}), the average ratio is 1.08 (σ=0.19,
*N*=2). The measurements are therefore in surprising close agreement
with the theory.

Finally, we can look at the kinetic energy of the spore. The translational
energy is
and the rotational energy is
,
where *r*_{g} is the spore's radius of gyration. The radius of
gyration for a prolate spore rotating about its short axis is
*r*_{g}=(*a*^{2}*+b*^{2})*/*5,
where *a* and *b* are the minor and major semi-axes of the
spheroid. Substituting values for the sequences shown in
Fig. 3, we find
*E*_{K}=2.3×10^{–13} J and
*E*_{R}=6.3×10^{–15} J. Therefore, the
amount of energy transferred into translation of the spore is at least 30
times greater than the energy associated with the spore's rotation.

### Ejection model

We now address the most fundamental question of the ejection mechanism– how the surface energy stored in Buller's drop is transformed into kinetic energy. To answer this question, we need a proper understanding of the fusion process. Fusion takes place over a time interval of less than 4 μs and is therefore just below the temporal resolution of most high-speed cameras currently available. Using a frame rate of 250,000 images per second and a shutter speed of 1 μs, we obtained some new and critical information about the early stages of spore ejection (Fig. 5). The first frame in Fig. 5A shows the drop that has condensed at the base of the spore. On the second frame, the drop has touched the spore and coalesced. The drop has not spread over the spore completely because its outline can still be discerned. In the third frame, the spore has been ejected while the top border of the drop is still visible on the spore. Finally, the last frame shows the spore rotation in and out of the image plane. This sequence of images establishes that the drop travels only a short distance on the spore and does not spread over the entire surface. Fig. 5B,C offers additional evidence of the partial fusion of the drop, which, as we will show, has some important implications for the amount of surface energy available to release the spore.

Consideration of the forces acting during the coalescence of Buller's drop reveals that ballistospore ejection is the fungal equivalent of jumping (Fig. 6). The same three ingredients are present – a lowering of the center of mass, a quick release of energy and an interaction with a rigid support. Growth of Buller's drop at the proximal end of the spore lowers the spore's center of mass (i.e. it brings it closer to the sterigma) as well as provides the energy to be used during ejection. This step is the ballistospore's way of bending its `legs' in preparation for jumping. As soon as fusion begins, the drop exerts on the spore a surface tension force directed towards itself and the spore exerts on the drop a force of the same magnitude but of opposite direction (Fig. 6A). With no external interaction (isolated system), the drop and spore would move towards each other, and the global center of mass would remain immobile. Thus, there would be no ejection. In the case of ballistospores, the sterigma plays the role of the rigid support. Its presence prevents the spore from moving towards the drop by exerting a reaction force opposing the surface tension force applied by the drop. The sterigma force is the external force acting on the spore–drop complex that leads to the motion of the center of mass. The same requirement for interaction with a rigid support is found in jumping. There, the moments applied at the leg joints must be resisted by the ground to generate the impulse that will accelerate the center of mass. Ballistospore ejection, however, differs from jumping in one important way. The spore is not resting on the sterigma but is attached to it. Therefore, as the spore launches forward, it will put the sterigma under tension. The latter must break easily to release the spore.

This scenario emphasizes the critical role played by Buller's drop and the
sterigma during spore ejection. We can subdivide the ejection process into
four stages (Fig. 6). During
the first stage, Buller's drop grows thus lowering the center of mass of the
spore and storing the energy that will be used during ejection. The second
stage encompasses the early coalescence during which the sterigma is under
compression and provides the counter-acting force necessary to move the global
center of mass of the spore–drop complex. It is this force that allows
Buller's drop to be accelerated up to a characteristic speed
*V*_{D}. In the third phase, the drop decelerates as it
transfers its momentum to the spore. The sterigma is now under tension and
needs to break easily to release the spore without dissipating its kinetic
energy. Finally, the fourth stage is the release of the spore.

The simplest model for energy transfer suggests that the kinetic energy of
the drop is equal to the difference in surface energy,Δ
*E*_{p}, between the initial state just before
coalescence and the final state just after coalescence (energy loss will be
considered in the last section). The validity of this assumption can be
ascertained by estimating the Reynolds number for the drop motion. We find
*Re*=*V*_{D}*R*_{D}*/*ν≈50.
The relatively large value for the Reynolds number confirms that viscous
effects are small compared with inertial effects, leading to an efficient
transfer of surface energy into kinetic energy. Therefore, we can write
,
where *m*_{D} and *V*_{D} are the mass and
velocity of the drop, respectively. The latter phase of the coalescence is an
inelastic shock between the drop and the spore. Although the energy is not
conserved, the linear momentum is conserved which implies that
*V*_{0}=*m*_{D}*V*_{D}/*m*_{SD}.
This model answers Ingold's question of how the surface energy stored in
Buller's drop is transformed into kinetic energy of the spore.

### Rupture force of the sterigma

The strength of the chitinous wall of the sterigma could easily exceed the
force created by the fusion of Buller's drop. Therefore, to predict the
initial velocity of the spore at ejection, the energy required to break the
hilum must be known. We measured the rupture force (*F*_{B}) by
pulling on spores with calibrated glass microcantilevers. The microcantilever
was brought in contact with the distal end of the spore and gradually pulled
away (Fig. 7, insets). Surface
tension between the cantilever and the spore allowed us to put the spore and
sterigma under tension. Our measurements reveal two spore classes
(Fig. 7). Some spores are
weakly attached to the sterigma and are removed with a force between 0.08 and
0.3 μN (mean *F*_{B}=0.15 μN). Other spores are strongly
attached to the sterigma and cannot be removed with forces up to 1.2 μN
(the maximal force that could be applied with the experimental set-up). For
these spores, the force required to fracture the hilum is higher than the
adhesion force between the cantilever and the spore. Most attempts to increase
the adhesion between the spore and the cantilever, and thus apply higher
forces on the hilum, failed; probably because the wet spore surface does not
allow strong bonding. However, numerous trials with cantilevers coated with
poly-l-lysine yielded a few strongly bonded cantilevers. For these
experiments, the hilum either ruptures for forces in the low range observed
before or for large forces above 1 μN and up to 4.8 μN
(Fig. 7). The two spore classes
provide direct evidence for the development of an abscission zone at spore
maturity to allow easy release of the spore
(van Neil et al., 1972;
McLaughlin et al., 1985). The
upper force range gives an estimate of the force required to rupture the hilum
before the abscission zone has fully developed.

For a finite rupture force, the spore velocity is reduced by an amountΔ
*V* that depends on the work done to fracture the hilum. During
the late phase of the coalescence process, the sterigma is stretched until the
hilar region is fractured. Given a stiffness *k* and an elongationδ
for the sterigma, the elastic force acting on the sterigma is
*F*_{E}=*k*δ. When *F*_{E} reaches
*F*_{B}, the hilum breaks. The energy needed to sever the
attachment is equal to the work done by the elastic deformation:
. We
measured *k* of the sterigma for different spores with the force
experiment reported in Fig. 7
and found values between 0.45 N m^{–1} and 1.5 N
m^{–1} (mean of 0.72 N m^{–1}). Using our
measurements of rupture force and stiffness, we can compute the energy
required to liberate the spores in *A. auricula*. The energy of
fracture is *E*_{B}=1.6×10^{–14} J and
corresponds to a velocity reduction of 3.4%. Therefore, the work of fracture
dissipates only a small fraction of the kinematic energy of the spore.
However, Buller's drop does not contain enough energy to rupture the hilum
before the abscission zone has been weakened. This observation may explain
why, on some occasion, fusion of Buller's drop fails to release the spore
(Buller, 1909-1950).

### Transfer of surface energy

The central component of our model is the calculation of the surface energy
available to accelerate the drop. This energy is equal to the difference in
surface energy between the initial state just before coalescence and the final
state just after fusion. The exact expression for the difference in energy
depends on spore geometry and final drop geometry. For *A. auricula*,
our observations of the coalescence process
(Fig. 5) reveal that the fused
drop adopts a geometry close to a spherical cap. The difference in surface
energy is (see Materials and methods):
(12)
This equation gives a measure of the energy available to accelerate the drop.
Setting the drop kinetic energy
() equal
to the freed surface energy and using
,
we find for the drop velocity:
(13)
The expression for *V*_{D} takes into account the spore
geometry and wettability through
. Using the conservation of
momentum between the drop and spore, it is possible to predict the initial
spore velocity *V*_{0}. The prediction for the ejection shown
in Fig. 3A is 1.2 m
s^{–1} whereas the observed velocity is 0.8 m
s^{–1}. The ratios of the predicted and observed velocities for
the entire set of experiments are listed in
Table 2. The predicted velocity
is surprisingly accurate given that energy loss, either to break the hilum or
through dissipation during the fusion process, has not been taken into
account.

It is also possible to predict the angular velocity of the spore. A torque
is exerted on the spore because the surface tension force is applied at some
distance from the point of contact between the spore and the sterigma
(Fig. 6A). This torque explains
the rotation of the ejected spore. Using the conservation of angular momentum
(Happel and Brenner, 1983), we
have
,
where *l* is the distance between the global center of mass and the
point of drop fusion. For the discharge shown in
Fig. 3A, *l*=3 μm,
giving a calculated angular velocity ofΩ
_{0}<8×10^{4} rad s^{–1}, in
good agreement with the measured value of 9×10^{4} rad
s^{–1}. We have not been able to investigate this aspect of the
discharge further because the rotation of most spores was not confined to the
imaging plane and thus could not be measured.

In the *Sporobolomyces* yeasts, the spore is nearly spherical and is
covered by a film of water (Fig.
4). The fusion is thus close to the coalescence of a drop of
radius *R*_{D} onto a perfectly wetting spherical spore of
radius *R*_{S}. The difference in surface energy is then:
(14)
where
.
Equating the surface energy and the drop kinetic energy and solving for the
drop velocity, we find:
(15)
Using this equation and the conservation of momentum, we predict a velocity
*V*_{0}=3.4 m s^{–1} whereas the observed
velocity is 2.3 m s^{–1}. Given that some energy is necessarily
lost in the coalescence process and in breaking the hilum, the agreement is
again very good.

### A test of the model

The key assumption of our model is that the drop reaches a characteristic
velocity *V*_{D} that can be predicted from the change in
surface energy of the system. To test the validity of this assumption, we
performed experiments on an artificial system that mimics the fusion of
Buller's drop. A drop was placed on a highly hydrophobic plate while another
plate, this one wettable, was approached slowly from above until it touched
the drop. Contact with the wettable plate induced a fast upward motion of the
drop (Fig. 8). The contrast of
wettability between the two substrates was such that the drop moved in its
entirety from the lower surface to the upper one. This coalescence process is
very similar to what happens when a drop wets the spore. Biance et al.
performed a similar experiment but with plates of similar wettability, leading
to a final state where the drop is split between the two surfaces
(Biance et al., 2004). They
provided a scaling relationship for the horizontal growth dynamics of the
neck. Here, we complement their analysis with a study of the vertical motion
of the center of mass.

We found that after a brief acceleration, the drop's center of mass moves
upward at a constant speed (Fig.
8H). Therefore, the fusion process is associated with a
characteristic velocity of the center of mass. We measured this characteristic
velocity in a series of experiments and plotted it as a function of the
theoretical velocity predicted from Eqn
13 (Fig. 8I). The
observed drop velocity is proportional to the predicted velocity, with a
proportionality constant β=0.28. The parameter β is a measure of the
efficiency of the transfer of surface energy to kinetic energy. The value ofβ
below one indicates that a fraction of the surface energy is lost in
the coalescence process and therefore not available to accelerate the drop. By
the same token, we can interpret the velocity ratios listed in
Table 2 as a measure of the
efficiency of the energy transfer in the ballistospores. *Auricularia
auricula* and the *Sporobolomyces* yeasts show a similar efficiency
with more than two third of the surface energy liberated contributing to the
kinetic energy of the spore.

## DISCUSSION

The use of surface tension by ballistosporic fungi offers a new paradigm for performing work at the micron scale. One clear advantage of this mechanism is that work, being performed by the fusion of a water droplet, comes virtually for free. It is only under this condition that the innumerable spores contained in a mushroom cap can all be equipped with their own discharge apparatus. The ballistosporic mode of dispersal is in sharp contrast with the mass release of spores or propagules by specialized launching structures found in other taxa (Straka, 1962). It is therefore of great interest to uncover the design principles that make surface tension an effective source of energy.

As first stated by Ingold (Ingold,
1939), the key step for ballistospore release is the transfer of
surface energy stored in Buller's drop to the spore. Our analysis emphasizes
the critical role played by the sterigma. First, during the early phase of the
coalescence process, the sterigma provides the external force that prevents
the spore from moving toward the drop. The global center of mass of the
spore–drop complex is thus projected forward leading to ejection. In the
late phase of the coalescence process, the sterigma is now put under tension
and should fracture easily to prevent dissipation of the spore energy. Our
measurements of the force required to release the spore from the sterigma show
that an active weakening of the hilum takes place before ejection. The
characteristic rupture force of 0.15 μN (*N*=15) recorded for a
weakened hilum is comparable with the rupture force of 0.1 μN reported for
wind-dispersed fungal conidiospores (Aylor,
1975). This value is large compared with the gravitational force
acting on the spore (*F*_{g}<2×10^{–6}μ
N) but small compared with the surface tension force that can be exerted
by a drop at this scale
(*F*_{γ}=γ2π*R*_{D}<1.4 μN,
where *R*_{D}=2.25 μm is the drop radius). However, a force
of up to 4.8 μN is necessary to detach an unweakened spore
(Fig. 7), i.e. three times the
surface tension force. Therefore, without an active weakening mechanism, spore
ejection would be impossible.

To predict the initial velocity of the spore, we developed a model that
focuses on the surface energy freed during the coalescence process. This model
predicts with surprising accuracy the initial translational and angular
velocity of the spore, particularly if one makes allowance for energy
dissipation during fusion. A prediction of the model is that the geometry of
the fused drop affects the amount of energy available to eject the spore.
Consequently, spore morphology and the wetting properties of the spore surface
can play an important role in the transfer of surface energy to kinetic
energy. The low efficiency of energy transfer in our artificial system when
compared with ballistospores (Table
2) also emphasizes the challenges associated with the fine-tuning
of such a mechanism. It is likely that the difference in scale between the two
systems explains the higher efficiency for ballistospore ejection. It is also
noteworthy that our model predicts similar efficiency of energy transfer for
the two species studied despite differences in spore geometry and a threefold
difference in the initial velocity between the *Sporobolomyces* and
*A. auricula* spores.

A way to evaluate viscous loss is to calculate the energy loss in volume
during the fusion process. This energy is
*E*_{V}=*T*_{D}μ∫ξ^{2}*dV*,
where *T*_{D} ∼
*R*_{D}/*V*_{D} is the characteristic time for
the drop merging process and ξ is the shear rate. By taking a
characteristic shear rate ξ ∼
*V*_{D}/*R*_{D} due to the small deformation of
the drop, the integration gives
.
Hence, the ratio between viscous energy loss and the surface energy of the
drop is:
.
The energy ratio corresponds to the capillary number (*Ca*). Here, this
ratio is much smaller than one, indicating small viscous loss.

We have found it useful to compare ballistospore release with jumping in
animals. We first note that the take-off velocity of the spore (1–2 m
s^{–1}) falls precisely within the narrow range of take-off
velocities (1–4 m s^{–1}) reported for good jumpers from
insects to mammals (Vogel,
2005a; Vogel,
2005b). This striking observation suggests that a take-off
velocity on the order of 1 m s^{–1} is a fundamental limit for
jumpers whether they achieve this velocity through muscle work or surface
tension. Vogel (Vogel, 2005b)
posited that the strength of biomaterials may impose limits on the stress that
can be applied to accelerate jumpers and thus may set the maximal take-off
velocity. However, it is doubtful that the same argument would apply to
ballistospores. As we have shown, the surface tension force exerted by
Buller's drop is *F*_{γ}=1.4 μN and is applied on a
cross-section of 5 μm^{2}, which is a level of stress that most
biomaterials can sustain.

In both insects and vertebrates, the velocity of the center of mass is known to increase monotonically during the active part of the jump up to the take-off velocity (Burrows, 2006; Burrows, 2008; Marsh and Johnalder, 1994). The evolution of the center of mass velocity can be accounted for if a finite force is applied during the entire hind limb deployment. By analogy, it would be tempting to assume that the velocity of Buller's drop in ballistospores follows a similar evolution with surface tension, instead of muscle work, providing a roughly constant force over the entire distance traveled by the drop. However, as can be seen in our artificial system (Fig. 8H), this approach would lead to a gross overestimate of the drop velocity. The drop is in fact accelerated over a very short distance and then displaced at a constant characteristic velocity. We have argued that understanding what sets this characteristic velocity is the key to predicting the spore velocity at ejection. Our results show that the characteristic velocity scales with the surface energy freed during the coalescence (Fig. 8I) and thus highlight the importance, in ballistospores, of the final geometry of Buller's drop in determining the energy available for discharge and the take-off velocity of the spore.

**LIST OF SYMBOLS**

*A*_{D}- drop surface area after fusion
*A*_{S}- area of the spore covered by the drop after fusion
- a
- acceleration of the spore
- a, b
- minor and major semi-axes of the spheroid
- Ca
- capillary number
- D
- drag force
*E*_{B}- energy of fracture
*E*_{K}- translational kinetic energy
*E*_{R}- rotational kinetic energy
*E*_{V}- energy loss in volume
*F*_{B}- rupture force of the hilum
*F*_{D}- force applied by the drop
*F*_{E}- elastic force
*F*_{g}- force of gravity
*F*_{γ}- surface tension force
*F*_{R}- ground reaction force
*F*_{S}- force applied by the spore
*F*_{SD}- force applied by the spore–drop complex
*F*_{St}- force applied by the sterigma
- g
- gravitational acceleration
- k
- stiffness
*L*_{S}- length of the spore
- l
- distance between the global center of mass and the point of drop fusion
- M
- moment at the joints
- m
- mass
*m*_{D}- mass of the drop
*m*_{SD}- mass of the spore–drop complex
- R
- mean radius
*R*_{D}- radius of the drop before fusion
- radius of the drop after fusion
- Re
- Reynolds number
*r*_{g}- spore's radius of gyration
*R*_{S}- spherical spore radius
- t
- time
*T*_{D}- characteristic time for drop coalescence
*V*_{0}- initial spore velocity
*V*_{D}- velocity of the drop
*W*_{S}- width of the spore
- v
- velocity
- x
- axis of discharge
- γ
- surface tension at liquid–vapor interface
- γSL
- surface tension at spore–liquid interface
- γSV
- surface tension at spore–vapor interface
- Δ
*E*_{p} - surface energy freed during the fusion process
- δ
- elongation
- θ
- angular position of the spore
- θ
- contact angle
- μ
- dynamic viscosity of air
- ν
- kinematic viscosity of air
- ξ
- shear rate
- ρ
- density
- τR
- characteristic decay time associated with angular velocity
- τT
- characteristic decay time associated with the translational velocity
- Ω0
- initial angular velocity

## FOOTNOTES

Supplementary material available online at http://jeb.biologists.org/cgi/content/full/212/17/2835/DC1

We would like to thank the MRSEC at Harvard University and D.A. Weitz for funding; B. Roman, L. Mahadevan, D. Quéré, and H. A. Stone for useful comments on the manuscript; as well as D. H. Pfister, N. M. Holbrook and M. A. Zwieniecki for help with the experiments.