Methods
Study sites and study plants
The herbaceous dogwoods (Cornus canadensis, C. suecica, and their tetraploid
hybrid, C. unalaschkensis) are common herbs in boreal forests and grow
circumboreally in Europe, North America, and Asia. We studied Cornus canadensis
in the relatively undisturbed forest at northeastern end of Isle Royale National Park,
Lake Superior, Michigan, USA, in June and July of 2002-2004. C. canadensis often
carpets the forest floor with each flowering shoot producing inflorescences of many
(21-39) small flowers subtended by four showy white bracts. All flowers used in the
experiments were collected by permit from Edwards Island or on the main island of
Isle Royale in the proximity of Monument Rock. Impatiens pallida (Balsaminaceae)
capsules were collected in Williamstown, MA in July 2004.
High-speed videos of Cornus canadensis
Individual bunchberry flowers were secured on a stand with a fine pin and
photographed using a MotionXtra HG-100K camera (DEL Imaging Systems,
Cheshire, CT, USA). Floral opening was initiated by touching the trigger, which
extends from a petal, with a fine wire. A xenon light and two 250 W incandescent
lamps provided the necessary high intensity lighting. Individual frames from the
high-speed videos were imported into Adobe Photoshop and a single point on the
anther, petal, or filament was marked. The x and y coordinates of these points were
measured on each frame and used to determine velocities and accelerations. In
frames where the stamens were partially obscured by the petals, the positions of the
points were estimated by lining up an image of the whole stamen with the visible
sections.
The velocities were calculated from the change in position of the points in
subsequent frames. The uncertainty in velocity, vi, was calculated to be the
quadrature sum of position uncertainties, xi, divided by the time interval, T:
v i 
2
x i1
 x i2
T
. This procedure assumes that the position uncertainties are
random errors that are not correlated and that the uncertainty in the time interval is

small compared to the position uncertainty. Acceleration values, ai, and their
uncertainties, ai, were obtained similarly from velocity data.
Terminal velocity measurement
Videos taken at 1,000 fps of exploding flowers were used to measure the terminal
velocity of the pollen in air (supplementary movie 2). Successive vertical positions
for 7 separate falling pollen grains were recorded from two different explosion events.
The slope of the height of the pollen grain plotted as a function of time was used to
determine terminal velocity. The number quoted is the mean velocity of all 7 grains
and the uncertainty is the standard error of the mean.
High-speed videos of Impatiens pallida
To determine the time required for an Impatiens pallida (Balsaminaceae) capsule to
explode open, we recorded fruit opening at 5,000 fps. We determined the time
required for separation of the middle seed from the centre tissue of the capsule (3.25.4 ms) and the time required for the 5 outer sections of the capsule to separate from
the central tissue of the capsule (2.8-5.8 ms) (n = 3).
Sodium azide study
To determine whether physiological processes are required for explosive flower
opening, 13 closed flowers were incubated in the metabolic inhibitor 0.5% sodium
azide1 for 3.5 h and then triggered; all opened fully. The viability of the filaments was
tested using the vital stain flourescein diacetate, which fluoresces when cleaved by
cytoplasmic esterases2. Following the sodium azide treatment, flowers were
submerged for 10-15 min in a 1:1000 dilution of 1 mg/ml fluorescein diacetate stock,
and then vigorously washed in three changes of distilled water. Individual filaments
were removed and mounted on slides for observation under a fluorescence
microscope using an FITC filter cube. Control (water-treated) filaments fluoresce
strongly in a pattern that follows cell boundaries. None of the 24 azide-treated
filaments examined showed bright fluorescence that followed the control cell pattern;
8 showed no fluorescence at all.
Sucrose experiments
To determine the effect of turgor on floral opening, mature flower buds were
incubated for 4 h in solutions ranging from pure water to 2 M sucrose (0, 0.2, 0.4, 0.6,
0.8, 1, 1.2, and 2 M). Buds were then removed from the solutions, briefly blotted to
remove large water drops, and triggered open. The angle that petals bend from
vertical (n = 36) and the distance between opposing anthers (n = 9) was measured for
each treatment. Both declined as a function of increasing sucrose concentration. For
buds soaked in water, petals bent back a full 180 degrees, and anthers separated 3
mm. Incubation in 1.2 or 2 M sucrose resulted in petals moving less than 45 degrees
and filaments separating by no more than 1 mm.
Force measurements
To measure the force required to open bunchberry flowers, we triggered flowers with
a 0.13 mm diameter wire while recording from above with a 1,000 fps camera. Using
the still frames we measured the maximum distance the wire was deflected beyond
the stigma (centre of the flower). The deflection was converted to Newtons by using
an electronic balance to measure the forces required to deflect the wire known
distances.
Pollinator Observations
We observed pollinators visiting Cornus canadensis at Isle Royale National Park
noting visitation rates and whether or not visitors were able to trigger open flowers.
Samples of each insect type were collected and then weighed just after they were
killed. Large insects, like bumblebees (~100mg) and long-horn beetles (~25mg),
moved rapidly between inflorescences and regularly triggered open flowers. Small
insects (<5mg), like ants and small syrphid flies, were unable to trigger open flowers
and also were poor pollinators as they rarely moved between inflorescences.
Pollen Movement Experiments
To determine the distance pollen can be carried after an explosion, we measured the
distance pollen traveled both indoors and outside. Indoors, we placed an individual
flower in a small hole in the center of a 22 x 22 cm board covered with black plastic.
The flower was then triggered open; pollen exploded from the flower was clearly
visible on the black plastic. We measured the maximum distance pollen traveled
from the center of the flower to the furthest visible pollen grain. Pollen from flowers
exploded indoors traveled horizontally from 2.5 to 22 cm (mean = 8.8 cm, n = 7) with
the distribution of pollen following the weak air currents present in the room. Outside
we used a 1 x 1 m board covered with black plastic. In the presence of a steady wind,
pollen was rapidly carried more than 1 m, exceeding our ability to measure it.
Supplementary References
1.
Vasilyeva, E. A., Minkov, I. B., Fitin, A. F. & Vinogradov, A. D. Kinetic
mechanism of mitochondrial adenosine triphosphatase. Inhibition by azide
and activation by sulphite. Biochem. J. 202, 15-23 (1982).
2.
Rotman, B. & Papermaster, B. W. Membrane properties of living mammalian
cells as studied by enzymatic hydrolysis of fluorogenic esters. Proc. Nat.
Acad. Sci., 55, 134-141 (1966).