Why Trees Are Tall

There are three main reasons why trees are tall:

1. To get more sunlight for their leaves.


2. To protect their leaves from ground-based leaf eaters.


3. To increase the distance that the wind carries their seeds, for trees that use wind-based seed dispersal.

Of these, #1 is the most important, because it affects almost all trees and it affects them more strongly than #2 or #3. Redwood trees can tower 100 meters over the ground; I'm not aware of any ground-based leaf eater taller than the giraffe, which tops out at around 6 meters.

As for #3, there are diminishing returns to growing taller just to send your seeds farther away. If you are trying to avoid competing with your progeny for light and soil nutrients, you only need to get them past your canopy and root system. If you are trying to prevent seed eaters from getting too many seeds in each bite, then the average seed displacement is proportional to height (see equation 1 of Nathan et al.'s "Mechanistic models for tree seed dispersal by wind in dense forests and open landscapes"), making the seed density inversely proportional to height squared. There's also the fact that as your seeds get more distant, the probability increases that they land in a biome not friendly to trees, like a lake.

Why does being taller help trees get more sunlight? Because they are in competition with other trees for that sunlight. To quote Kevin Simler's superb Parable of the Redwoods:

[T]he redwood is locked in an evolutionary arms race ("height race") with itself. It grows tall because other redwoods are tall, and if it doesn't throw most of its effort into growing upward as fast as possible, it will literally wither and die in the shadows of its rivals.

This competition has real costs for the trees; every bit of energy they use to grow upwards is energy they can't for example use to fight disease. (For trees, the race to the top can be a race to the bottom, so to speak.) In fact, foresters will sometimes thin forests for the express purpose of making the remaining trees healthier, though in that case the only reduced competition is for water and soil nutrients; the trees still grow tall as fast as they can.

And why is that? Why can't trees detect when there are no other trees around and start slacking in the height department? You might think that the answer to that is easy: trees can't do that because they don't have eyes to see with, and even if they did, they don't have brains to make decisions with.

Ten or twenty years ago, that might have been the end of the story, but lately scientists have been discovering that plants can (a) detect pretty subtle properties of the light falling on their leaves, (b) communicate this information to other parts of the plant, and (c) continue to act on this information even after the light stimulus is removed. There has even been a bit of a backlash, not about any of those experimental results, but rather about whether words like "memory", "nervous system", or "think" should be used to describe them.

I have not specifically seen a study showing that a tree can detect another nearby tree via the shade pattern on its leaves, but there are studies showing that other, non-tree plants can do this by measuring the "decreased ratio of red to far-red light (R:FR) characteristic of foliage shade". There are also experiments with trees detecting when nearby trees have been attacked by insects. So I think enough of the pieces are there that if "not growing when all alone" was a good strategy for trees to have, it is close enough in gene-space that 360 million years of tree evolution could have found it.

So why might it not be a good strategy? My guess is that it is because even if a tree is alone right now, it probably won't stay alone for very long. And when the neighboring trees do come, the original tree would have wished it had had an even bigger head start.

We can state the hypothesis more formally: let d be the duration before a neighboring tree is detected, Es be the energy saved per unit time by not growing while alone, and Ea be the sunlight energy received by the tree during its entire lifetime had it stopped growing during d, and Eg be the sunlight energy received by the tree had it grown nonstop. Then the "slacker tree" strategy is a bad one if and only if

d * Es < Eg - Ea

To recap, we have adopted the methodology of a three-year-old: we started with a why question, and we kept on asking why for each answer that we got. We end with one last why question: if trees are tall because they are competing for sunlight, why are they competing for sunlight? Why don't they cooperate instead?

Let's consider a specific proposal: all trees should grow to 20 meters tall, then stop. That's tall enough to annoy the giraffes and ensure good seed dispersal. What tree could object to such a plan?

The problem is that this strategy isn't evolutionarily stable. A mutation comes along that increases the 20 meters to 21 meters, and in just a few generations, all the trees are now 21 meters tall. And then 22 meters. And then the mutation arises that removes the "then stop" part entirely. It takes over, and we are back where we started.

To make the plan work, we need a way to punish the trees that don't follow the plan. So maybe trees are tall because they can't punish other trees, or can't detect the trees that need to be punished?

We already talked about plants that can detect the shade of nearby plants. What more is needed to detect their height? I'm sure there are lots of different ways it could be done, but here's one: if the tree knows the angle of the sun at two times of the day, and can measure where on its canopy the top of the other tree is casting its shadow, then the height of the other tree is a simple linear equation [1]. Plants are known to solve more difficult equations than that just to figure out when to open and close their stomatal apertures.

So detecting the defecting trees is on the table. What about punishing them? Surely trees can't attack other trees, except in really bad Ent fanfiction?

Actually, there is some evidence that they can, by releasing biochemical toxins. It's called negative allelopathy, and it is a little bit controversial, especially when applied intraspecies, because (a) how do you distinguish it from normal competition for resources and (b) how does the releasing organism avoid being harmed by the toxins it is handling?

But nonetheless, here's an example of same species negative allelopathy, taken from the delightfully named "Chemical Warfare in Plants":

Proebsting and Gilmore (1940) related the problem of re-establishing peach trees in old peach orchards to the presence of toxic substances in the soil. Nutrient deficiency and pathogenic organisms were eliminated as causes after extensive analysis. Peach roots, root bark, and alcoholic extracts of peach roots were all found to be toxic to the growth of peach seedlings when added to virgin soil in which the seedlings were planted.

So, just to be clear, I'm not saying that we ever see trees selectively releasing toxins at other trees based on how tall the other trees are. Similarly to before, I am saying that all the individual components appear to be at least biologically possible, so it is worth asking the question of why we don't see it.

One simple answer might be that toxins aren't strong enough to have any deterrence effect, even when it is a bunch of trees all attacking a single tree. But here's a more fundamental reason: "grow to 20 meters, then stop, and punish any trees taller than 20 meters" is still not an evolutionarily stable strategy. Why? Because it loses to "grow to 20 meters, then stop". Trees following that strategy don't get punished, and don't have to spend any energy creating nasty biochemical weapons.

So "punisher" trees lose to "don't punish" trees, and then "don't punish" trees lose to "don't stop growing" trees. But wait, the "punisher" trees were designed to beat the "don't stop growing" trees. Does that mean we have a rock-paper-scissors situation?

No. A single "don't stop growing" tree will (by assumption) lose when placed in a forest of "punisher" trees. But a single "punisher" tree will not thrive in a forest of "don't stop growing" (aka normal) trees. It will attack all the trees around it, but it will use energy to do so, and the attacks won't be effective enough to keep the surrounding trees from being taller. (If the toxins were that effective, then even normal trees would use them to squash the competition and you would never see two trees next to each other.)

By contrast, a single "don't punish" tree will thrive in a forest of "punisher" trees, and a single "don't stop growing" tree will thrive in a forest of "don't punish" trees. This creates a ratchet effect, in that as long as there are "don't punish" trees, the "punisher" trees are doomed, and as long as there are "don't stop growing" trees, the "don't punish" trees are ultimately doomed as well.

Finally, we might consider the strategy "grow to 20 meters, then stop, and punish any taller trees, and punish any trees that don't punish other trees". This plan, if adopted by enough of the forest, is stable against the other strategies. It is reminiscent of the strategies that humans sometimes use to ensure cooperation: we punish people who break the law, but we also punish people who don't pay the taxes that pay for the police and jails used to punish people.

Sadly though, for all of its theoretical virtues, this final strategy utterly lacks biological viability. It's one thing to punish trees based on their height, which can in principle be assessed by shade patterns on leaves. But if tree A is releasing toxins from its root system into the root system of tree B, I'm hard pressed to come up with any mechanism by which tree C could reliably determine it.

To summarize, trees don't cooperate either because they can't punish defectors or because they can't punish the freeloaders who don't punish the defectors.

And that is why trees have to compete for sunlight, and that is why trees are tall.

[1] Let tree 1 sit at the origin and have height h1. Let tree 2 sit distance d away at (d, 0) and have height h2. Let the relevant part of tree 1's canopy lie on the line y = h1 - alpha x. Let the ray from the sun that touches the top of tree 2 at time t be given by y = h2 + beta_t (x - d), and let x_t be the intersection of that line with y = h1 - alpha x. Then

h2 = h1 + alpha (beta_1 x_2 - beta_2 x_1) / (beta_2 - beta_1)