Part 3. Seed dispersal. Chapter 17: dispersal by wind
It was mentioned a few chapters ago that wind-dispersed species have tricks for making sure that most seeds abscise when the wind speeds are greater than average. But there are other equally important traits the plant can control: the rate at which the seed falls and the height at which it begins its journey. These two together determine the amount of time that the horizontal wind can move the seed laterally before it hits the ground.
One invariable characteristic of wind-dispersed plants is that they place the seeds as high up as possible. This is most easily seen in wind-dispersed meadow plants that attach the seeds to the top of a vertical shoot rising well above the leaves (consider dandelions in a field; cattails in a marsh). Likewise, wind-dispersed trees tend to place the seeds well up in the crown. (Crown refers to the upper part of the tree where the leaves are.) For example, with pines the female cones are mainly in the upper third of the crown (the male cones are lower: in the middle third). Why place the seeds high? Two reasons. First, as mentioned, the seed will travel farther; double the release height and you double the distance achieved. Second, horizontal speeds are greater as you get further up above the ground. Why? Because the drag of the branches slows down the wind; the higher you get, the fewer drag-promoting elements you have.
As for minimizing the fall rate, this is done via a force: the production of lift or drag. Recall from your high school science class that objects in freefall will accelerate downward at 9.8 meters per second squared; that is, the speed keeps increasing during the descent. Picture a skydiver who has just jumped out of a plane. Immediately, the descent velocity of this individual goes from 0 to about 19 meters per second after 2 seconds of descent (= 9.8 times 2). The force acting on the skydiver is gravity, and if nothing else intervened she would keep accelerating. Indeed, after 35 seconds she would have fallen several kilometers and be poised to break the sound barrier. But this will not happen because of another force—drag—which began operating as soon as she jumped out of the plane. Drag is proportional to two things: (1) the square of the local speed around her body and (2) her cross-sectional area. Given that the descent speed is going up in a linear manner with time due to gravity, then drag for a freefalling object is going up as the square of time. Thus, relatively quickly, drag goes from a force of 0 to one that rivals—indeed it soon cancels out—gravity. When you have no net force—in this case, drag now equals gravity for our skydiver—then it falls at a constant rate, and this is referred to as the terminal velocity. But area affects drag too: picture the skydiver who maximizes his area by immediately extending his arms and legs and keeping his body parallel to the ground. He accelerates rapidly of course (drag is almost nothing when he starts) but the acceleration is lessening as the drag rapidly grows. He ends by falling at a terminal velocity of about 56 meters per second. This will most certainly lead to unpleasantness unless he opens his chute. That parachute, greatly extending his area while adding only minimally to his mass, decelerates him to a new, lower, and much safer terminal velocity of about 4 meters per second depending on the type of parachute. Near the other extreme he can greatly reduce his area by holding his limbs close to the body and refusing to pull the ripcord: this suicidal cannonball descent would have a terminal velocity of 90 meters per second. Returning to the gravitational force, a crucial factor in determining the terminal velocity is the mass (parachute plus body) AND the area. Indeed, the ultimate issue in determining the final rate of fall is the ratio of total mass to total area: the larger this ratio is, the faster the terminal velocity. Small animals have a small mass, and so they do not even need parachutes: it has been well-documented by the SPCA that a cat in New York City can often survive, however dispiritedly, a fall from the top of a skyscraper as it instinctively rights itself and splays its limbs (thus markedly increasing the drag by increasing the area), suffering only sprains or broken bones. (Instinctively spreading the limbs is a product of natural selection.) You or I or a rhinoceros can jump from a 20-story building and stretch out our limbs like a cat but we are beyond that crucial minimum ratio of mass to area: we will hit the pavement at far too great a speed and most assuredly make a mess the city has to clean up. In summary, elephants and humans can use jumping from a precipice as a sure-fire venue for suicide while the mouse regards a fall from that height as merely a heart-stopping journey; the intermediate-sized cat poised at the edge of that sheer granite wall is at the critical point where death may or may not occur.
Wind: drag-producing seeds
One class of wind-dispersed seeds imitates paratroopers and the luckier urban cats by minimizing the mass/area ratio. An example of this type of adaptation is the kapok (mass of white fibers) surrounding the seeds of members of the Bombacaceae family (recall Figure 12.4), or the tuft of fibers that lies above the seeds of the aptly-named cottonwoods (or any other member of the poplar family). The ancestors of cotton (taxonomically related to the Bombacaceae, not to the suggestively named cottonwoods) had this same method of dispersal. As with cotton, the fibers in this group are rather chaotically organized; some are pointing almost vertically and are therefore contributing little to the drag production as the dispersal unit (seed plus fibers) descends vertically. The other major parachute-type dispersal unit has the fibers more soberly organized, more or less in the horizontal plane, and so all of them can do some good resisting passage through the air during the descent. This second type is very common among herbaceous plants (e.g., the daisy family: asters, dandelions, and many others) but relatively rare for trees. In both groups, the fibers are hollow (to reduce their mass) and about 15 to 40 microns wide. (Drag production would be far better if for every fiber of 20 microns diameter we could substitute a bunch of much narrower ones, say 1 micron diameter, but no plants have pulled off this trick because a functional cell needs to be a good deal larger than that: around 20 microns is the minimum required to allow the cytoplasm to circulate within the confines of the cell walls.) The two designs function somewhat differently. In the second case, the dandelion-like units, individual fibers each create drag; while in the first type (Bombacaceae), the huge mass of fibers acts as a single ball: the bulk of the air is forced to move around the entire kapok mass. (As an extreme, picture holding up a coarsely-woven rug in the face of a stiff wind: the air is forced to mainly move around it). The problem with the planar array of the dandelion design is that it is limited to small seeds, usually around 1 to 10 mg, because as the seed mass increases, the drag required to balance all that extra mass must also increase. But at too great a drag, the individual fibers would begin to bend upward (thus reducing the effective drag-creating area and therefore increasing the terminal velocity.) By contrast, the kapok design, while messy, has all the fibers shouldering the burden as a mutually supporting ensemble. So, we have seeds as massive as 100 mg using this approach to drag creation. But even here there are limits: with a seed mass about one half of a gram (the size of a small acorn) the entire group of fibers begins to bend, to streamline, thus making the unit fall even faster as area is reduced.
Another way to create drag is to dispense with the plant analogue of a parachute and simply be much smaller than the seeds discussed so far. Recall our counter-intuitive discussion of the Reynolds number in relation to wind pollination: the tiny object moving in air experiences a drag force similar to you moving through a pool of tar… A very small seed will, like a pollen grain, experience an enormous drag as it falls through the air, thus reducing its speed. Even better, its weight (the force that drag must oppose) is correspondingly less because it is so small. Thus, for example, a round seed of 0.1 milligrams with no appendage will fall at a constant rate (terminal velocity) of 2 meters per second (which is not good compared to the dandelion 0.3 meters per second), while an identical smaller seed of 0.01 milligrams falls at 0.9 meters per second. In the limit, as we imagine seeds as small as microscopic pollen grains (a few groups of plants do have seeds this tiny—for example, the orchids), we see terminal velocities as low as 0.03 meters per second. (Remind yourself that, at an impossible limit, 0 meters per second means that the seed would have no tendency to move downward or upward but would instead be happily buffeted by the tiniest updrafts and downdrafts, like an oxygen molecule among the other air molecules.)
In general, though, very few seeds take advantage of this trick of being exquisitely small. Indeed, the seeds of orchids—as small as any seed can possibly be—have been referred to insultingly by botanists as mere sacks of DNA. There is no serious maternal investment of nutrients in the endosperm or cotyledons of orchids, and so the germinating orchid is, initially, utterly dependent on a fortuitous encounter with mutualistic fungi whose hyphae must provide it with the required substitute for the root system it does not have the maternal resources to build for itself. But even somewhat larger seeds than orchids—say, seeds with a mass of 0.01 to 0.1 mg—are rare among plants and essentially non-existent among trees, and the reason remains the rooting problem just after germination—because the initial root permitted by the resources within 0.1 mg seed is so short that it extends only a millimeter into the ground; exactly the thin, uppermost region of the soil that is dry as a bone on a hot summer day only one days after the last rain. (A few millimeters lower down it can still be quite moist of course.) We will pursue this further later on but let us conclude by declaring that while the number of very tiny seeds that can be produced by some abstract tree is, of course, great, the survival rate of the resulting tiny germinants will be terribly low.
The few tiny-seeded tree species that have taken advantage of this mechanism tend to specialize, not surprisingly, on the wetter habitats. They include Tea tree, an Australian native with a seed mass of about 0.05 mg and no appendages—this would give it a terminal velocity of about 1 meter per second—whose riotous invasion of southern Florida wetlands a few decades back has since cost that state hundreds of millions of dollars expended on eradication programs that achieved, at best, mixed success. It is perhaps more famous globally for its use in skin-care products and as an antisceptic. Another example is the ancestral Metrosideros (seed mass equals 0.1 mg) that has spread across the Hawaiian island chain from Polynesian archipelagos far to the south, presumably by wind rather than by any interest of seafarers migrating in outrigger canoes wanting to bring it to their new home. Another Australian tree (Eucalyptus regnans—very distantly related to Metrosideros) also has no appendage for its seed; nonetheless it can achieve large distances thanks to its status as one of the tallest trees in the world; third-world countries have widely planted it (or close relatives) to establish drought-tolerant fiber plantations, woodlots, and windbreaks.
More generally, Australia, isolated for so long from the rest of the world, tends to have evolved plant species that are not very competitive. For example, the seed dispersal capabilities of Australian trees are generally quite modest, embarrassingly so. Most of them can barely move seeds 30 meters away. And the reason for this is the lack of dispersal appendages. If the seeds were tiny, dispersal would be adequate. But given the aridity of so much of the continent, the seeds need to be larger than that to deal with dry soils at the germination stage. Why are the seeds so inadequately provisioned? Because the millions of years of isolation meant they did not face better competitors; they had traits good enough for them to persist given the equally low competitive ability of the species around them.
Wind: lift-producing seeds
As we saw, the force of gravity can be countered by drag, permitting the seed to move laterally for a longer period; but an alternative upward force is available to natural selection as it seeks to cancel gravity: lift. There are several very different wing designs that generate lift, all of them arrived at independently by evolution.
As with drag, the trick to maintaining a low terminal velocity is to minimize the ratio of total mass (seed plus wing) to wing area. Perhaps the most common type is the asymmetric wing typical of maples, many tropical species (especially members of the pea family), and most conifers. The seed is located at one end; given that the wing is thin, then the span-wise (long axis) center of mass lies just inside the seed, and so the seed autorotates around this mass center, generating lift as it does so. The wing is kept stable in autorotation by a strengthened leading edge that is especially noticeable in larger versions of this design such as maples (Figure 17.1) and mahogany (Figure 17.2) where the thickening is provided by the vascular bundle). That is, the spanwise mass center is about 15% of the distance from the edge, and the pitch (angle to the horizontal) of the wing is relatively constant at about 10 to 20 degrees. Interestingly, all the species that have adopted this design look remarkably similar, each being about 3.5 times longer than they are wide (e.g. the mahogany in the aptly named Figure 3.5). When the same body shape is hit upon independently by different lineages, it is called convergent evolution. (Consider the similarity in the shape of the head for the lion family and the group of marsupial “cats” that filled the same niche in South America and Australia; or, more cogently here, the wings of soaring animals like the albatross (a bird) and the pterodactyl (a familiar Mesozoic reptile).) Invariably, convergent evolution means that, given the physical constraints, we are looking at the only design that works well or, at least, the one that works much better than others at the moment. In the case of the asymmetric wing, the similarity is due to a stability constraint: if the wing is too short relative to its width it cannot turn with enough torque, and it will stall (violently pitch upward and then simply drop nose downward at a fast speed). If the wing is too narrow relative to its length, it is difficult to maintain stability, and again it will pitch rapidly upward, leading to a stall. It turns out that the ratio of wing length to wing width (termed the aspect ratio) needs to be between about 3 to 4 to avoid instability, and this is why all the plant species with this design have independently converged on the same shape (aspect ratio) and thus look similar.
Figure 17.1. Autorotation of sugar maple.
Figure 17.2. Winged seeds of Mahogony (left; within its fruit at top) and Dalbergia (right). Digital.
A lower limit on seed mass for winged seeds is around half a milligram, conferring a terminal velocity of about 0.5 meters per second. At this point the necessarily small wing becomes too short—cannot generate enough speed as it turns—to push through the air in a tight helix and maintain adequate lift. (We return to the Reynolds number: a really small wing is starting to perceive the air as a tar-like medium.) Is there, likewise, an upper limit for winged seeds? Having a very large seed means you fall faster—just like drag-producing seeds with thin long fibers. But it is not clear that there is an upper biomechanical limit: the pea family member Centrolobium in the New World tropics is the biggest example of this type, weighing several grams (the size of an acorn) and with a seed coat studded with thorns—you do not want to be hit on the head by this as it autorotates down through the branches at 4 meters per second. The real upper limit for this design, one supposes, is that the wing becomes superfluous. Consider a hickory nut (the same size as the Centrolobium) with no adornment vs this huge investment in the thick woody wing by Centrolobium. If the hickory falls from a height of 20 meters it will never have time to attain terminal velocity before it hits the ground; its averaged speed during the acceleration would be about 10 meters per second. That makes the wing investment made by Centrolobium look pretty smart until you remember that the highest wind speeds by far are in the upper part of the forest. During the first 5 meters of descent (when most of the lateral movement by the wind might occur) the hickory nut’s average speed will be about 5 meters per second. That is, during the first part of the descent—the most important part because this is where horizontal wind speeds are greatest—the mean falling rates of Centrolobium and hickory are much the same. One would guess that the massive wing of the Centrolobium is, like your appendix, a vestige of something that once served some useful function: the ancestor likely had a much smaller seed. In any case, we can perceive why Centrolobium represents not a biomechanically imposed upper mass limit but an evolutionary limit.
(As a final note on this type of wing: three decades ago, NASA commissioned a study on the feasibility of using this maple wing design –with the concentrated mass of a camera, appended to a very long and light wing, replacing the equally-concentrated mass of the seed—as a cheap way to obtain closer pictures of the surface of Mars, a 360 degree panorama being obtained with each rotation as the device descended through the atmosphere. The very low density of the Martian air meant that the wing needed to be well over 10 meters long to fall at a suitably slow rate. As you know, NASA ended up choosing the more promising method of a ground-hugging robot—the rover—unpacked from a vehicle that, lacking all the elegance of a pine or even a Centrolobium, was coughed up by the orbiting spacecraft, parachuted toward the red surface, and then bounced for several hundred meters along the ground after an exceedingly hard landing. This is not the approach expected by a generation that grew up on computer-generated science-fiction films.)
A second way to fashion the wing is somewhat less common with the seed again at one end, but now there is no thickening along one side of the wing; unlike the maple or pine, this wing is symmetric (Figure 17.3 shows the winged seed of a velvet ash). In consequence, there is no spanwise stability; in effect, the wing stalls—but in a predictable way. Thus, there is simultaneously a double rotation: as before, we have the wing rotating around the seed in the horizontal plane, but now also the wing itself is rotating rapidly around its central axis. (The maple wing maintained a constant wing angle, cleaving the air like a helicopter rotor.) This bilaterally symmetric design is less efficient than the asymmetric one: for the same mass, the terminal velocity will be about 30% lower for the asymmetric wing. (Some authors have argued that the second rotation makes the bilaterally symmetric wing more stable in turbulent air but there is no compelling evidence—or even any non-compelling evidence—for this statement. As often happens in ecology, things keep getting repeated because they sound neat. Ecology has its own version of urban myths and, as with the popularity of tabloids, it thrives on intellectual lethargy.) A variant on the design of the ash occurs in many tropical species (e.g., Dalbergia: Figure 3.5) as well as the invader Ailanthus in the eastern US: the seed is in the middle of the wing rather than at one end. This means that the first type of autorotation is no longer a tight helix at one end; rather the seed describes a very wide helix, as much as a meter wide, as it falls through the forest.
Figure 17.3. Bilaterally symmetric wing of black ash. Graphite. By V. Crisfield.
A third type of wing is more literally like a helicopter than what we have seen, but very rarely found in North America. Many tropical seeds (the type case is the Dipterocarpaceae family of southeast Asia (Figure 17.4), almost all of them tree species) have a number of wing-like appendages (rotors, to continue the helicopter analogy) attached to the base of a single large seed. A variant is found with the North American basswood: a modified leaf serves as a pair of rotors—the seed hanging down from the middle (Figure 17.5). For equal mass, this design is even worse than that of the ash, perhaps explaining why it is so uncommon. Even with only a single fruit attached (but there is usually more than one), basswood descends at about 1.7 meters per second, one of the worst performances of any North American wind-dispersed species.
Figure 17.4. Here we see 2-, 3-, and 5-winged Dipterocarp species autorotating in a Malaysian forest. Digital.
Figure 17.5. Basswood, a terribly inefficient design. V. Crisfield. Graphite.
Note that for all these winged seeds, the “rotors”—whose number varies among these types from one to five—spin the entire ensemble as it falls. Indeed, the behavior is much like a helicopter when power is lost. A helicopter with a dying engine is designed to shift the pitch of the rotors as it descends so that they passively—like basswood—spin, reaching a terminal velocity sufficiently low that the pilot has some chance of climbing out of the wreckage alive.
A fourth design for wings makes no attempt to be stable. A species like a birch or the tropical Tabebuia has two wings at either end of a relatively small seed. As it tries to autorotate, it succeeds for a few revolutions but then pitches upward, and then dives, and starts the process over again. Nonetheless, the upward movements certainly contribute to a slowing of its overall descent rate. With a similar design, paper birch falls at about 0.6 meters per second, while redwood (Figure 17.6), with its useless, teasing fringe of wings, descends at about 1.3 meter per second. In short, this design, lacking the careful construction of the asymmetric wing, is sensible only if you add very large wings. With all these species, the wings are usually membranous (soft, translucent) rather than woody because they do not need to rotate under great force (high speed) in a tight helix, and the seeds small.
Figure 17.6. Winged seeds of paper birch (top) and coast redwood (left). The dime reminds you these are small seeds. Digital.
Finally, two other wing types are found: both quite rare and restricted to the tropics. There is one type like a paddlewheel with the central axis formed by an elongate seed. The other type is a pure glider: as might be designed by an engaged high school student folding a sheet of paper. Both designs are poor at translating wing mass into a reduction in terminal velocity.
Let us summarize what we have learned about design principles at the intersection of natural selection and aerodynamics. At very small seed masses, no appendage is required. But subsequent germination requirements make such tiny seeds unusual among plants and especially among trees. In the range of seed sizes 0.1 to 10 mg, the most efficient design is the drag-producing seed with a planar array of straight fibers. These can reliably create terminal velocities of 0.2 to 0.5 meters per second. Nonetheless many species within this size range do not use this method: consider the winged seeds of many conifers. At somewhat higher masses, the most efficient design is the asymmetric wing. As we approach seed masses of a gram or more, any lift or drag promoting appendage is leading to such a high terminal velocity as to be useless; evolution will favor dispersal by animals rather than waste energy constructing woody wings.
Consideration of this range from good designs to dreadful designs can be refreshing given the constant “gee whiz” stories in newspapers and science magazines, all of them stressing how this or that trait is marvelously adapted to its present environment. But as Darwin pointed out, the surest reason to believe in evolution via natural selection rather than in an artificer God with a fondness for the perfect anatomic trait, is the presence of jury-rigged contrivances—the legacy of natural selection having to work with the baggage that the lineage brings with it as it shuffles forward in time—and, as well, sub-optimal traits. To persist for millions of years, a plant species need not be optimal in all, or even any, of the traits it uses to gather light and nutrients, disperse seeds, ward off predators, etc. It needs only, in total, to be better than its competitors some of the time. And thus, lodgepole pine in the Rockies would be better off if it had the planar fibers of a dandelion rather than an asymmetric wing. But consider how radically the female cone would have to be altered to accommodate this evolutionary change. Shortly, we will see a way in which a pine, with slight modifications of the ancestral design, can become animal dispersed. But even then, the changes are quite modest (loosen the grip of the wing on the seed; or don’t let the scales flex open after seed maturity is reached; or make the seed much bigger). Generally, species are prisoners of their evolutionary descent, their possibilities greatly circumscribed by what happened before. Thus, the newest legal assault on Darwinism by creationists (although it is a venerable argument dating back several centuries)—intelligent design—is very odd, because the great majority of what goes on in nature is not especially well-designed, much of it is merely good enough, and some is downright embarrassing. Coming at the issue from a very different perspective, the creationists emphasize the same ornamentations as journalists—utterly focused on the really sweet examples of what natural selection (or God) might do. The eyes of a mammal: golly! But when you look closely, biological progress looks exactly how we might expect it to look: ragged, duct-taped, unfinished, and yet modestly functional. It appears that God muddles through just like everybody else.
The distance (x) any one seed is transported by the wind is straightforward to estimate (Figure 17.7).
x = uh/f
where u is the mean horizontal wind speed, h is the release height of the seed, and f is its fall rate (terminal velocity). What about the entire crop of seeds on the tree? Some will experience downdrafts and low horizontal speeds and not go far at all. A few brave seeds will catch a strong updraft and travel a few football fields away. Dispersal of a single tree’s crop by the wind tends to have a characteristic curve. There is a rapid rise in seeds per m2 with distance traveled up to a peak at about 5-15 meters from the tree. Then there is a gradual decline out to greater distances. We have measured dispersal curves (using seed traps) out to distances as great as one and a half kilometers and while the proportion of seeds making it this far is small (<5%) there is little doubt that this fraction is instrumental is how our wind-dispersed species manage to keep pace with rapid climate change.
Figure 17.7. The basic equation for seed dispersal by wind. Bristlecone pine. Digital.
Wind: snow and other low-friction surfaces
If seeds fall on a snow surface they are, at least until the next snowfall buries them, available for re-entrainment by the wind—this time as a secondary dispersal event. Three things are required to take advantage of this mode of dispersal. First, of course, the species must be in a place that reliably gets snow in the winter—far enough north or high enough in the mountains—and then the seed must be abscised during this cold season so that it is on top of rather than under the snow layer. Oddly, almost no tree species take advantage of this potentially brilliant means of seed dispersal. As discussed in a much earlier chapter, most North American species north of Mexico abscise seeds in the autumn, with perhaps only 5 to 10% of northern conifer seeds falling on winter snow in an average year. Essentially none of the seed crop abscises during winter for maples, ironwoods, and most other hardwoods. The big exceptions are yellow birch (Figure 17.8) and black birch (eastern North America) which abscise around 30% (that estimate is quite rough) of their crop in the winter.
Figure 17.8. Yellow birch seeds on snow. Digital.
The second requirement for successful dispersal on snow involves our favorite ratio—total seed mass to total area. For the wind to pick up the seed from the surface, the seed must generate enough lift to counteract both gravity (pushing the seed downward) and friction (resisting a lateral push). Thus, the same trait that makes a seed well-dispersed by wind as a primary event (and, totally coincidentally, float on a water surface as a secondary event) will also, as a second coincidence, aid it in this secondary event on a cold terrestrial surface. A half-second gust of air scoops up the seed for a moment and then drops it back onto the snow further along. This saltation (from the Latin word for jumping) is the same form of locomotion that small pebbles use as they bounce downstream along riverbeds, or that sand grains in the wind employ to extend Saharan dunes; and the process will continue for seeds as long as the wind remains gusty, and the seed avoids holes and other sheltered (low wind speed) spots in the snow. These dead-air spaces—picture the lee side of a fallen log—effectively trap the seeds because the wind will now be too feeble to pick them up again. As an example: when you are cross-country skiing in an open area in a yellow birch stand in Michigan or among black birches in New England on a windy day, you will note the huge numbers of seeds that have been stranded in the shallow ruts made by skis earlier that morning.
The third necessity is that the primary event must get the seed out into an open area (field, snow-covered river or lake). The reason is that the wind speeds near the snow surface in a forest (especially in a conifer stand, but even in a leafless hardwood stand) are too low. The trunks and the branches of shrubs exert too much drag on the wind and sufficiently slow it so that the odds of re-entrainment have become tiny. Further, the very rough surface (snow-covered fallen logs, the bases of living trees, etc.) in a forest makes for lots of those dead-air places that will trap a seed before its journey has seriously begun.
But for those seeds that can initially be dispersed into an open area there is essentially no limit on how far they can go on snow. We can expect them to travel all the way to the next stand of trees (which should trap them: uneven log-strewn surface; low wind speed), or, if on a snow-draped, iced-over river, then as far as the next major curve in the stream. While distances of many kilometers can be presumed to be routine, we actually know very little concretely about what happens to seeds on snow or ice. The only study so far ever done was based on my students in Quebec watching seeds, spray-painted in gay colors, that moved at least 100 m. They undoubtedly went farther than that. . . but consider that if the seeds are moving in a gust, then tiny bullets of hardened snow are simultaneously flying into the face of the complaining undergraduate, and the wind-chill is so great that you cannot sufficiently pay even someone as poor as a student to continue the field study beyond that first numbing day. Thus, we know little about secondary dispersal on snow.
Eventually, if the seed has not been trapped by surface elements, then a new snowfall will end its journey. Subsequent melting, months later, will not free it because the moisture around the seed (and wing) makes the dispersal unit now adhere to the wetted bits of snow.
Thoreau, 150 years ago, points out an example with birch he once witnessed where there were three dispersal mechanisms operating. First, there was a primary event on a windy day that placed the seed on snow some distance away from the tree. Then there was a secondary event where wind raced the seed along a snow-covered stream, finally depositing it just above the water-line on the bank. Finally, in the spring, the water, fed by snowmelt, rose high enough to capture the seeds and take them, presumably, he of course did not try to monitor the subsequent diaspora, a long distance downriver.
What about a forest—or field for that matter—without snow? There are too many dead-air spaces at the surface due to the marked unevenness, the wind speeds are therefore too low, and so a seed cannot be re-entrained. Once seeds are deposited on leaf litter in a forest in autumn, they might as well, from the point of view of the wind, have fallen on fly paper. The great thing about snow is that it evens out much of this characteristic unevenness, providing a low-friction surface. Ice is an even better surface for skidding objects. On large lakes and wide rivers, it is normal for snow to be blown off towards the shore, exposing ice Seeds landing on lake ice should enjoy an unimpeded journey to the forest at the far shore. (But there are no studies of this; only Thoreau’s anecdotes).
Another excellent surface for secondary movement by wind is a salt flat. During the maximum continental ice extant 20,000 years ago the jet stream was pushed far to the south so that the presently arid region of Nevada, Arizona, and Utah was covered with forests and, because the jet stream was steering rain-swollen low-pressure centers toward this region, very large lakes. During the previous 70 million years, this region had been broadly domed upward by the eastward motion of the Pacific Plate under the westward-moving North American Plate—and, oddly, much of the central area (Nevada, western Utah) had no outlet to the sea, and so huge inland lakes could form in the lowest-lying parts. When the jet stream shifted back north as climate warmed, the region became far drier, and the forests disappeared except for the highest slopes of the mountains. The lakes—large in area but quite shallow—became ever smaller and more saline as evaporation continued, concentrating the salts, and of course the influx of fresh water from nearby rivers lessened. The Great Salt Lake is a remnant of a much more extensive (and far less brackish) body of water whose wave-cut beaches can still be seen etched into the mountainside overlooking Salt Lake City. So, given the flat surface of these relic lake bottoms, and only an intermittent cover of shrubs that need to be simultaneously drought-tolerant and salt-tolerant, the wind speed near the surface is great and it is not surprising that the seeds of those shrubs can be tumbled long distances across the salt-flats. Other flat desert regions are different in their genesis; they tend to be an erosional surface covered with small stones, and so wind speeds at the surface should not be as great as on the very low-friction salt flats.
Another smooth, low-friction surface is the beach. Recent experiments in South Africa on beach sand show that there is ample dispersal by wind on these surfaces, the seeds easily reaching the netting set up at 25 meters from the release point, so long as the plant cover is sparse (i.e. no dead air spaces).
Now, neither salt flats nor rock-strewn deserts have trees (although the edges of some beaches do). But rivers that cut through these deserts certainly are lined with trees (most of whom have long vertical roots that probe for the water-table), and these ought to be able to move their seeds via secondary dispersal from one river to the next, cross-country across the intervening arid areas. Many herbs and shrubs in deserts appear to have their seeds moved considerable distances in this manner. Again, we have no surety of these statements.
Nonetheless, we should not make too much of the flatter parts of deserts, or the beaches rimming the continents and lakes, and the sparkling snow surfaces with newly-deposited winter-abscised seeds. The great majority of the Earth’s tree seeds initially abscised by wind will subsequently fall among the dead spaces in the dried leaves or foot-high green grass blades, and stay there, never again to feel the fingers of the wind, the invitation to explore the world. Only the appetite of an animal could cause them to be lifted from the ground one last time.