In the desert of northwest Australia, about 10 miles east of the small mining town of Newman, lies a natural wonder. If you fly overhead, you’ll see vast carpets of green spinifex grass, pockmarked by barren red circles, as if some deity had repeatedly stubbed out a cosmic cigar on the parched landscape.

These disks of bare soil are called “fairy circles,” and they’re not unique to Australia—they also exist 6,000 miles away in Namibia. There, the circles number in the millions, and extend over some 1,500 miles of desert. They comprise different grasses but their patterns are the same: low-lying vegetation freckled by circles of empty soil. They almost seem alive, growing and shrinking with a lifespan of 30 to 60 years.

Local people believe them to be the work (or footprints) of deities and spirits. Scientists have tried to come up with more grounded explanations since they first started writing about the circles in the 1920s. Some suggested that they’re the work of grazing ants, or radioactive gases leaking from underground, or poisonous plants that kill off their competitors.

Over the last century, two main hypotheses for the cause of these circles have fought their way to the top of the scrum. The first is from Norbert Juergens of the University of Hamburg, who says that the circles are the work of sand termites. To store water, he argues, these insects eat the roots of grasses within a circular patch, allowing the underlying grains of sand to catch and absorb the falling rain. The result is Namibia’s version of a beaver dam—an engineered reservoir. And according to Juergens, the fierce competition between neighboring termite colonies causes the regular spacing of the circles.

Many other researchers, including entomologists and botanists, aren’t convinced. They think the circles occur because plants engage in a tug-of-war for water and other scarce nutrients. Due to their battles, the landscape “self-organizes” into rings of deep-rooted grasses, draining water from a central reservoir where no other plants can thrive. This explains why, as the researchers Michael Cramer and Nichole Barger found in 2013, the fairy circles are restricted to places with low rainfall, and why they grow after dry years and shrink after wet ones.

Namibian fairy circles (Stephan Getzin)

Stephan Getzin from the University of Goettingen started off as a fan of the termite hypothesis when he began studying the fairy circles in 1999. But he defected to the self-organization camp after studying aerial images of the fairy circles, and seeing just how regular they are. “They have an extremely regular hexagonal spacing, like a honeycomb,” he says. “That pattern persists throughout the landscape for hundreds of thousands of meters. Termites and ants are not known to cause such strictly ordered patterns.” In May 2014, he published a paper outlining his evidence for the self-organization hypothesis.

Three days later, he got an email from Bronwyn Bell, an environmental manager at an iron-ore mine in Newman. We have something similar here, she said. To prove her point, she attached an aerial photo. “When I saw it, it looked really convincing,” says Getzin. Seven months later, he was on a plane bound for Newman.

Getzin and his colleagues found that the Australian circles exist in the same orderly honeycomb as their Namibian counterparts, with almost exactly the same spatial traits. And by measuring temperature and analyzing soil samples, they worked out how the circles might form.

The critical point is that the area around Newman doesn’t get enough rain to sustain an even carpet of plants, so there’s competition for water. Plants that grow a little bit bigger than their neighbors draw in more water: Their deeper roots loosen the soil around them, allowing more water to seep in. Nearby plants benefit, while those further away die of drought. They leave patches of bare earth that are too hard, compact, and hot for seeds to germinate. These empty circles act as rain collectors: any water that falls on them runs off to the side, where it nourishes the encircling plants.

When Getzin simulated all of this on a computer, he produced virtual patterns that are almost indistinguishable from the actual fairy circles.

He thinks that the Namibian circles form in a slightly different way. There, the fight for water mostly occurs underground; in Australia, it happens on the surface. But the basic idea is the same: Water conflicts that play out over meters create patterns that pockmark the land for kilometers. “The new paper moves us closer toward a unifying theory of fairy-circle formation,” says Barger.

It also weakens the termite hypothesis. Getzin found that most Australian fairy circles show no signs of these insects. He also mapped the locations of termite nests in one particular site and found that they’re randomly distributed, clustered in some areas and absent from others. That’s very different to the even hexagonal spacing of the fairy circles themselves.

There’s something compelling about the termite hypothesis, which involves creatures actively shaping their world. It plays to our appreciation of agency, and perhaps our love for underdogs (or under-insects, as it were). By contrast, the self-organization concept is less intuitive. It’s much harder to imagine how thirsty plants, just by sitting there, could produce these beautiful, kilometer-wide patterns. Indeed, Juergens once described the idea to me as “just a synonym for fairies.”

And yet, the natural world is full of examples of beautiful patterns that have deceptively simple origins.

In 1952, Alan Turing, the English mathematician and code-breaker, suggested that the stripes and spots of many animals are produced by two molecules: an activator that produces the pattern, and an inhibitor that blocks it. These diffuse through the skin and interact with each other. Depending on how strongly they interact and how fast they spread, Turing predicted that they’d result in everything from zebra-like stripes to cheetah-esque spots. (Try it out for yourself.)

Many studies have since found many examples of these “Turing patterns.” They’re evident in the pigments of animal skins and seashells, and even the cells of your fingers—examples of local interactions between spreading entities that create complicated, repeating patterns over vast scales. Scientists have seen such patterns in the distributions of parasites and hosts, and the location of plants in deserts. Getzin sees them in the fairy circles.

“Fairy circles have certainly jumped from an obscure phenomenon known to only a few, to a celebrated but mysterious landform,” says Walter Tschinkel, an entomologist from Florida State University. But “correlation between a computed model and nature is still not proof of causation, no matter how nicely the output mimics nature.”

It will take an experiment to truly solve the fairy-circle mystery. Scientists will need to show that they can artificially create or close up the circles by manipulating water levels, soil quality, or termite numbers. That’s a tall order for a phenomenon that exists across miles and decades.

Absent such experiments, it would certainly help to find more fairy circles. Barger thinks the odds are good. “With the increasing availability of satellite images globally it’s likely that we will continue to discover more instances of these circular gap patterns in arid environments in the future.”