Banded vegetation in drylands
A Socratic walk-through of banded vegetation in drylands — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why does dryland scrub arrange itself into even stripes that nobody planted?
Fly low over parts of Niger, or Somalia, or the Australian mulga country, and the ground below looks ploughed. Dark stripes of thorny scrub, tens of metres wide, alternate with pale bare ground, and the stripes run for kilometres in near-parallel arcs. In West Africa it has a name — brousse tigree, tiger bush — because from the air it reads as an animal's markings.
The first instinct is to look for the ruler. Old field boundaries, a buried geology, a fire history, someone's planting scheme. But the pattern appears where no one has farmed, on ground that is geologically uniform, and it appears independently on three continents in unrelated floras. So the ruler is not out there. Something about the way water and plants interact must be generating regular spacing from nothing but local rules — and that is a much more interesting claim than it first sounds, so it is worth being careful about whether it holds.
Reasoning it through
REASONING #Start with the one resource that is genuinely scarce. These are places receiving a few hundred millimetres of rain a year, in hard bursts. Ask what the soil surface does with a hard burst, and the pattern begins to come apart into pieces.
Bare dryland soil does not simply drink. Raindrop impact and drying seal the top few millimetres into a crust, and a crusted surface sheds most of what lands on it. So bare ground is not passive — it is a producer of runoff. Where does that runoff go? Downslope, and these landscapes are almost always gently sloping, often well under two percent. Almost flat, but not flat.
Now ask what a patch of vegetation does to the same burst. Litter, roots and the burrowing of termites and other soil fauna keep the surface open, so a vegetated patch takes water in far faster than the crust around it. Follow the consequence: a patch does not merely survive on its own rainfall. It intercepts and absorbs the runoff generated by the bare ground upslope of it. The patch is effectively harvesting a catchment much larger than itself.
Here is the step worth pausing on. A plant helps its immediate neighbours — shade, litter, an open surface, water held in place. But it harms things further away, because it drinks the water that would otherwise have travelled on. Facilitation at short range; competition at long range. Does that combination on its own explain a stripe?
Work it through. Suppose the vegetation is initially scattered at random. Any slightly denser clump absorbs slightly more runoff, which lets it grow denser still — a positive feedback that concentrates the vegetation. But it also strips the water from the ground downslope of it, which cannot then support growth, so a bare gap opens below each clump. And that bare gap, being crusted, generates runoff for the next clump down. The system cannot settle into an even carpet, because an even carpet leaves no bare ground to harvest from. It cannot settle into isolated dots either, because the slope gives the interaction a direction and clumps merge sideways along the contour. What is left is bands running along the contour, separated by roughly the distance over which runoff can travel — which is why the spacing is regular. Nobody set the wavelength; the hydrology did.
This is a recognised class of mechanism — a scale-dependent feedback, mathematically the same family as the reaction-diffusion instabilities Turing described, and captured in dryland form by Klausmeier's 1999 model. The pattern is not an accident on top of the ecology. It is the only stable arrangement.
Two honest caveats. First, at several sites the bands have been observed to creep slowly upslope, because the upslope edge gets the water first and colonises while the starved downslope edge dies back; but this is not measured everywhere and the rates reported vary, so treat migration as a common behaviour rather than a universal law. Second, on genuinely flat ground the same feedback produces spots, labyrinths or holes rather than stripes — and the popular claim that the sequence between those forms gives an early warning of desertification is a live argument in the literature, not settled fact.
The analogy
THE ANALOGY #Think of a queue of people sharing an umbrella walk in heavy rain, strung out along a corridor with a single leaking gutter overhead. The person in front catches the drips; the one immediately beside them is sheltered and does well; the one a few paces behind gets nothing at all and drops out. Nobody organises the queue, yet it settles into clusters with gaps — because each cluster is fed by the dry stretch behind it, and a continuous crowd would leave no stretch to feed from.
people in a queue can see each other and step aside, whereas the plants have no perception of the pattern whatever — each one only ever experiences the water arriving at its own roots, which is precisely why the regularity is so surprising.
Clarifying the model
THE MODEL #Three refinements hold the picture together.
First, the bare ground is not wasted land. It is functional; it is the catchment. A band of tiger bush may be receiving two or three times its own direct rainfall, which is why woody vegetation persists at rainfall totals that could not support a continuous cover. Clear the interbands and plant them, and you can destroy the bands by removing what fed them.
Second, "emergent" here has a precise meaning and not a mystical one. Every rule in play is local — a crust sheds, roots absorb, water runs downhill. The stripes are a property of the whole slope that no single rule contains, but they are entirely derivable from those rules. Emergence is a statement about levels of description, not about extra forces.
Third, the direction matters. The pattern is oriented by the slope, so bands run along the contour and the whole structure has an upslope and a downslope face that behave differently. That asymmetry is the fingerprint that distinguishes a runoff-driven band from a pattern produced by some other mechanism.
A picture of it
THE PICTURE #How to readEnter at the angled node — rain arrives everywhere equally, which is the point. The hexagon is the bare interband doing its real job of shedding water rather than absorbing it. Follow the runoff into the band's upslope edge and round the loop: absorption feeds growth, growth feeds absorption. The diamond is where the pattern is actually set, because the answer decides whether the far edge lives. Taking the "no" branch reaches the hazard node, and note that it feeds two ways — back to fresh bare crust, closing the cycle, and forward to the rounded outcome, which is the slow upslope creep.
What became clearer
WHAT CLEARED #The stripes are not a design imposed on the landscape; they are what a landscape looks like when help is local and harm is not. Once a plant improves the ground beneath it while draining the ground beyond it, an even cover becomes impossible and a regular spacing becomes inevitable — and the bare ground between the bands turns out to be the thing keeping the bands alive.
Where to go next
ONWARD #- Why the same feedback yields spots, labyrinths and gaps on flat ground, and what decides which.
- Whether pattern transitions really are an early warning of dryland collapse, or an artefact of the models.
- How termite activity contributes to infiltration inside the bands.
Key terms
TERMS #| Term | What it means |
|---|---|
| Tiger bush (brousse tigree) | banded dryland vegetation alternating with bare ground, first described in West Africa. |
| Scale-dependent feedback | facilitation acting over a short distance combined with inhibition acting over a longer one; the general engine of self-organised spatial pattern. |
| Soil crust | a sealed surface layer formed by raindrop impact and drying, which sharply reduces infiltration and generates runoff. |
| Klausmeier model | a 1999 water-and-plant model that reproduces dryland banding from local rules and a gentle slope. |
Every term the collection defines is gathered in the glossary.