Trail braiding

Why does a footpath across a wet meadow end up ten times wider than the line anyone actually walked?

Walk a popular route across boggy ground and you meet something odd: a scar twenty metres wide where a path half a metre wide would do. Nobody wanted that. Every walker who contributed to it was trying to keep their boots dry and stay off the worst of the ground, which sounds like restraint rather than damage.

So the puzzle is not why people damage paths. It is why a crowd of people each avoiding damage produces far more of it than a crowd walking straight down the middle would have done.

Start with one walker at one moment. The tread ahead is a churned puddle. The walker steps onto the vegetated verge beside it — a small, sensible act with an obvious private benefit and no visible cost.

Now ask what that step does to the verge. A grass sward on wet ground resists a certain amount of traffic and then stops resisting. Boots shear the leaves, break the surface mat, and press the soil grains together. Compacted soil takes water in far more slowly than the loose, root-riddled soil beside it, so rain that used to soak in now stands on the surface. And the roots that were binding the top few centimetres are dying.

Notice where that leaves us: the verge the walker stepped onto is now on its way to becoming exactly what they stepped off. Wetter, softer, less vegetated. In a few weeks it is a puddle too, and the next walker steps one pace further out.

That is a positive feedback, and it is worth being precise about what closes the loop. Use degrades the tread. Degradation makes the tread unattractive. Unattractiveness pushes use sideways onto ground that has not been degraded yet. And that sideways use starts the same process one lane over. The path does not widen because more people came. It widens because the mechanism that repels walkers is the same mechanism their walking creates.

Two properties turn that loop from a nuisance into the twenty-metre scar.

The first is that the damage curve is steeply curvilinear. Most of the vegetation loss on a given lane happens within its first modest number of passes — the sward goes from intact to largely destroyed early, and further traffic does comparatively little extra. Recreation ecologists have measured this repeatedly, though the exact number of passes depends entirely on the vegetation type, wetness, and soil, so I will not attach a figure to it. The consequence is what matters: a brand-new braid lane carrying a small fraction of the traffic still loses most of its vegetation.

The second is the asymmetry in timescales. Destroying the sward takes a season of light use. Rebuilding it on wet, compacted ground takes several growing seasons of no use at all — and on a popular route there is no such interval. So the loop only ratchets outward.

The instinctive fix is to spread walkers out so no single line takes the punishment. The curvilinear damage curve says this is precisely wrong on a durable, already-hardened surface: dispersing traffic across ten lanes does not give each lane a tenth of the damage, it gives each lane most of the damage. Concentrating traffic on one sacrificial line, and making that line able to take it, is the counter-intuitive but well-supported management response — and it is why boardwalks, stone pitching, and drainage work when signs asking people to stay on the path do not.

That also identifies which claim in the account is load-bearing, and how it could be refuted. The whole argument turns on damage being front-loaded rather than proportional to use. If a lightly used braid lane carrying, say, a tenth of the traffic were found to have lost about a tenth of its vegetation cover, the curvilinear claim would be false, dispersal would genuinely dilute impact, and braiding would be self-limiting rather than self-reinforcing. That is a straightforward field measurement — count passes onto marked lanes and record cover — and it is the experiment the recommendation rests on.

Two honesty notes. Avoidance of wet ground is the dominant driver on soft, poorly drained terrain, but it is not the only cause of widening: walking side by side to talk, passing oncoming groups, and shortcutting the inside of bends all widen a tread without any mud involved, and on dry hard ground those are the main mechanisms. And I have described this as though the walker's motive were dry boots, which is the usual case, but the loop runs identically on a motive of avoiding a rough or eroded surface — what the feedback needs is only that degradation makes a lane less attractive than its neighbour.

Finally, the comparison to a braided river is a resemblance, not a shared mechanism. A river braids because a sediment-laden flow cannot maintain one channel. A path braids because each user makes an independent avoidance decision. The pattern rhymes, the physics does not.

• Why some vegetation types recover from trampling in a season and others effectively never do.
• How drainage rather than surfacing is usually the real fix, since standing water is what triggers the avoidance.
• Whether the same avoidance loop explains desire lines across a lawn, where there is no mud and the trigger is distance rather than wetness.

Every term the collection defines is gathered in the glossary.