On an outing, a couple of companions mentioned that they were reading Beaver Land by Leila Philip. They remarked about several curious facts about beavers, including a claim I found dubious: that beavers could direct the trees they felled. That is, the beavers could decide and then execute a plan that would make the tree fall in a direction they chose. Of course, to anyone who does this, as I do, that’s obviously crap.

Think about how beavers chew away at a tree—they chew away at the tree until the top portion of the tree either falls over or is connected by a dull point of the connected fibers of the heartwood, right at or near the center of the tree. The weight distribution of the tree can take the tree any way it wants, and the wind can take the tree opposite of its center of gravity. These beavers aren’t directing the fall at all; they are merely taking what comes to them.

Along with this claim was that beavers have some sort of special hearing that allows them to hear the tree fibers cracking so they know when it’s going to fall, and they can escape safely. Anyone who has been in this situation knows that you don’t need special hearing, and you don’t even need to be particularly close. The “hold wood” fibers snapping is loud, even with a chainsaw running.

So, I went to look for the research, and I found the distressing article: The Orientation of Beavers (Castor canadensis) when Cutting Trees (OHIO J SCI 103 (5):143-146, 2003).

This entry isn’t about beavers; it’s about the idea that you can have an idea of beaver behavior without ever looking at one, or putting forth a way to predict what would happen. How do you know what beavers do if you never saw one?

Again, science is the art of prediction. Can I observe the world, detect patterns, and make correct predictions? In this case, given a tree, can I predict to a high degree where a beaver will stand, its general approach to the tree, and his execution of its plan?

I suspect you can’t, and this research doesn’t provide such a rubric. I suspect the beaver stands wherever he was when he encountered the tree, and the direction of travel is much more predictive than anything else. I also expect that beavers don’t really have a plan; they act compulsively and randomly. But it’s up to someone else to prove through a mountain of evidence that that’s not true.

How would you judge that? Trail cameras might work, but more likely there would be a graduate student in a tent watching a piece of land for days on end and taking notes every 15 minutes. Why? Because cameras are a capital expense, and graduate students pay their university to do this. Later, I’ll discuss why that isn’t what happened.

Beavers die

One of this group’s assertions was that beavers had a special power to avoid being hurt by trees. Of course, that’s stupid and false:

The idea that beavers can control the direction of a tree’s fall is now out of favor based on two types of observations: 1) many trees get caught in the foliage of other trees, and 2) falling trees sometimes kill beavers

But this statement is similarly stupid. That a tree gets caught up doesn’t mean that the beaver didn’t control the direction. In a dense forest, any direction is likely to cause a hang-up, especially in the US Northeast (and this research is from Ohio). And, that falling trees sometimes kill beavers doesn’t mean beavers didn’t control the direction. That can be “shit happens”, that there are stupid beavers (choosing a bad direction is still choosing), or nothing special at all.

There’s nothing here that comes from actual observations of beavers, and as such, saying anything about how beavers orient themselves to the trees is just the bunk.

The elephant and the three blind men

In the parable of the three blind men and the elephant, three different people feel different, select parts of an elephant then tell others what an elephant looks like, having never touched all of an elephant. And, as the parable’s situation forces, having never seen an elephant.

So, let’s suppose that we want to discover if beavers employ a plan and execute it with skill to repeatedly and consistently succeed. What’s the first thing we’d want to do?

If I were designing this experiment, I’d want a bunch of graduate students, volunteers, and field workers to observe beavers and record what they do. I’d interview as many people as I could about beaver behavior. If I could, I’d ask beavers what they are up to.

This paper goes in a different direction. It ignores the beavers and looks at the tree stumps.

Wait, what? How is a tree stump going to tell them anything? Here’s the glaring red flag that shows that the authors not only don’t know anything about beavers, they don’t know anything about trees.

Directional felling

In the world of forestry, “directional” felling refers specifically to making the tree lay over in a direction against its prevailing forces. For example, suppose the tree in front of me leans to the left. Can I fall it to the right instead? In general, yes I can (and I mean me, standing next to the tree with a chainsaw, edges, and an axe).

To accomplish a directional fell (and I write “fell” to mean the process of cutting down a tree instead of “fall”, which I reserve for the actual movement of the tree to end up on the ground), you can’t do what beavers do: chew away the tree until the remaining fibers fail. Most forestry directional felling is going to maintain some of the tree’s fibers and use those fibers to constrain the tree from falling in the disfavored directions. This is the “hinge”, and typically it’s about 80% of the diameter of the tree and slightly in front of the center of the tree, but not the center of gravity, which is likely behind or to the side for directional felling. If the center of gravity is in front of the hinge, it isn’t really directional felling.

Even during the fall, the hinge is (you hope) intact until the face-cut aspects meet, stopping the free fall of the tree and causing the momentum of the falling weight to redirect and to rip the hinge apart. The bole often “jumps” off the stump, although there are various ways to control that, such as with the face-cut angle (bigger angle, more time before the aspects meet). This typically happens when the angle of the bole (the main part of the tree) is around 45 degrees to 30 degrees to the ground, in which case the direction of fall is a fait accompli. Indeed, in some falls, these fibers never completely fail because the tree’s branches hold the bole off the ground, or the face-cut angle was large enough to control the bole longer, which makes sense when you don’t want to hit something else. As a fun fact, this is an example of a classic simple machine: a device that redirects a force. And, the wedge you used is another simple machine.

But, I don’t think the people involved in this paper have ever had to cut down a tree, or at least one large enough they might lose their life if they did it incorrectly.

What did the beaver want?

This study’s idea of intent has no basis in fact. They look at the stump, long after the beaver and the bole are gone, and assign some mark on that tree as “most recent”, and then use the position of that mark to infer something about the plan and execution of the fell.

First, how does anyone who didn’t witness the beaver chewing on the tree know what the last mark is? Understand that a beaver doesn’t start chewing on a tree and proceed to completion. This could happen over days or weeks, and fresh marks could look much brighter than older, weathered marks. But, in a particular session, where the beaver gnaws on the tree and all the marks are made close in time, which one is last? Do the marks on the left of the tree degrade or weather the same as the ones on the right? Given a couple of days, do marks made at the same time look the same? Nothing in this paper addresses that.

Second, is the direct action of the beaver the main factor in the actual fall? Is there some point during the beaver’s gnawing that the tree is going to fall despite further gnawing? For example, the beaver gnaws the tree to the point where the next gust of wind will start the irreversible fall of the tree, but the beaver continues to work past that point. Which mark is the one that sufficiently compromised the tree? Most likely, that is not the last mark that beaver made.

Last, were all marks on the tree made before the fall or the inevitability of the fall? Were any marks made after the tree was on the ground? I think that beavers likely don’t gnaw on trees that are on the ground, but what if the tree hung up in another, had effectively fallen, meaning it was no longer supported by the stump? Does the beaver know that it’s pointless to gnaw and continue until he gives up and does something else?

It’s not important that these questions prove or disprove the feasibility of the study, but that the authors don’t have answers and as such can’t be used as the basis for an inference. There’s no cited research that establishes this basis. Indeed, much of the cited literature is popular science periodicals instead of formal studies.

The stump is not the tree

Earlier, I wrote about indirect measures and that most science is not actually a serious attempt to add to human knowledge. That is the case here as well. There is nothing in this paper that increases our understanding of beavers. We already know the paper did not observe any beavers. They also didn’t observe any trees that were still standing. They had to find something they could measure with minimal effort, and then extrapolate (rather than interpolate) ideas from that. This is rife with “begging the question”, where you take as true some unproven assumptions that would falsify your conclusion if they were actually false.

Instead, their indirect measure was the disconnected stump left over and the orientation of surrounding trees. From that, and only that, they try to work backward to what the beaver was thinking and what the beaver did. They use a plumb line to measure the angle of the trunk compared to straight up, and assume that they know something based on that. If you’ve ever worked in a forest, you know this is complete bullshit. Trees do all sorts of crazy things, and knowing what the stump looks like doesn’t tell you about the bole. The angle at knee height is useless. But it is something that you can measure.

Consider, for example, how the weight is distributed in the canopy. A tree leaning to the right might have its center of gravity to the left because there are more or heavier branches on the left. These things matter, but aren’t part of the data.

Early in college, I took a very intense Botany class. The final was a walk in a redwood forest with a list of species but also a list of phenotypes or situations. Not only did you have to know what something was, but how to find it. In that class’s midterm, I answered a question with “phototaxis”, meaning that the plant’s growth is predicted by the source of light. That was the right answer almost, because the answer was “positive phototaxis”, meaning that the plant grows toward light. And almost is not correct. Yeah, tough class, and I think about that a lot. Accuracy matters.

The next time you are in a forest, look at the way the trunks point and where the trees actually grew. Maybe they started in one direction but then switched. You don’t know the distribution of weight based on the trunk. Maybe there was a crowded canopy when the tree was a sapling, so it grew to the side looking for the light from a far-away hole in the canopy. Then a storm clears out some of the canopy, making the dominant light straight up. In the extreme case, these are “pistol butt” trees.

One curious thing about this paper is that it doesn’t mention the tree type. But why would you if a tree is just a tree? Some trees, like birch, are thin and tall without branching, so canopy asymmetry doesn’t matter as much as it does with a pine or oak. The Ohio Department of Natural Resources says the forest is a mix of beech and maple.

With all of this, the paper lacks any science (predictive power) and any investigation of the beaver’s intent. If all the trees are leaning toward that marshland or pond, where there is no canopy and the environment favors a positive phototaxis (which aligns with a favorable geotaxis (negative!)), the beavers might not have any agency at all in the direction that the tree falls because the beaver’s advantageous direction is accidentally also the tree’s natural lean and no matter what they do, it all works out.

But there’s this nonsense:

Far from the shore, the trees tend to be fairly symmetrical and vertical (Loehle 1986). Thus, trees should fall in the direction from which they are cut.

This is stupidly beyond belief; it’s unsupported, has no observational evidence, and is a completely broken syllogism.

It is there only to lead to the next statement:

Therefore, beavers trying to control the direction trees fall should cut from the side nearest the water to minimize the distance the trunk or branches need to be dragged.

But this isn’t how beavers chew on trees. They gnaw on it until the bole is balancing on a point. There’s no steering ability there.

How they could have done

They could have recorded beavers chewing on trees and watching them fall. They wouldn’t have to measure 452 stumps, make large unsupported logical leaps, or entertain fantasies of unobserved beaver behavior.

But the acknowledgments give it away:

This research was funded by the Howard Hughes Medical Institute Undergraduate Biological Sciences Education Program grant to Ohio Wesleyan University.

This is a vanity project to keep students, who are short-term workers, busy. They didn’t care to get to the truth; they cared about a short-term project that could satisfy a grant and be completed by temporary workers in the time available. None of these students likely had the time to conduct an actual study because they have other classes. So how much time did these students spend looking at 432 trees? A weekend? Or, were there multiple cohorts of students who each spent minimal time mildly participating in something that would look sufficiently interesting on their graduate school application?

This was never about science because it was more about science-based entertainment. There’s nothing here that advances human knowledge or increases the predictive probability. Worse yet, it solves no problem that anyone cares about.

But even then, let’s suppose everything in this paper is true? Does it matter? Do the beavers anywhere else act that same? This isn’t about beavers, it’s about the beavers at this site with these trees. What does this tell me about any other group of beavers at any other places? Remember, these lightweight explorations are simply using the most convenient sources of information for the least amount of effort/