Australia Built a 44km Fence to Save Extinct Species — Now There’s a New Problem.

How do you catch the last cat inside an area the size of a small city? Not the first cat, not the 40th, the last one, the single animal that has watched every other cat around it die and learned something from each one. Three people solved that problem in Australia's Tanami Desert on the 4th of December 2018. But the reason anybody had to solve it is stranger than the answer and starts with the fact that this cat was locked in. The place is New Haven. A former cattle station 300 kilometers northwest of Alice Springs. Its fence runs 44 kilometers, stands 1.8 meters tall, and cost 5 million dollars, over a million wire clips, fitted by hand, 138 kilometers from the nearest sealed road. What that buys is 9,400 hectares, about one thousandth of one percent of Australia. Nobody claims this saved a country, it was built as a test, because Australia has lost at least 34 mammal species since 1788, around 35 percent of all modern mammal extinctions worldwide, with feral cats implicated in 27. When the gate shut in March 2018, 46 cats were sealed inside. Reaching the last one took nine months, and three Walpiri rangers reading sand at dawn. Four years later, 66 bilbies went through that gate. By 2026, there were around 530. So why does a number as small as 46 justify all of that? Why zero rather than nearly zero? And a harder question waits at the other end. The researchers who spent nearly 30 years perfecting fences like this one now argue that cats should be let back inside. Deliberately. Not because the fences failed, because they worked, and something went wrong in the winning. It starts with what those 46 cats were removed for. A bilby is roughly rabbit-sized, with blue-grey fur, a long pink nose, pink ears, and a black-and-white tail. Being a marsupial, the female carries her young in a pouch that faces backwards, so sand does not pour in while she digs. Seeds, bulbs, termites, and small lizards, hunted at night by smell and hearing rather than sight. One more detail matters more than all of those. A bilby takes almost all its water from its food and never drinks. The feral cat does exactly the same thing. In desert country, most predators are tied to water. Dingoes kill cats, but a dingo has to drink, which chains it to bores and soaks. Since a cat needs no waterhole at all, the deep desert became the one habitat on the continent where it operates unsupervised. None of which means bilbies lived in paradise before 1788. Dingoes hunted them, and owls did too, but both were tied to the same water as everything else. There is also one moment each night when a bilby is defenceless. Digging head down, sand flying, its two best senses are swamped. A 2024 Murdoch University study titled Housework or Vigilance measured exactly that at Western Australian sites. In the days after a cat passed a burrow, bilbies cut their digging sharply, and cat activity around burrows roughly doubled near the full moon. So the animal is forced to choose between maintaining its home and watching for what will kill it. And what will kill it arrives on a schedule set by the moon. Run that across a continent for two centuries, with two to six million feral cats depending on rainfall, each killing around 740 animals a year. The lesser bilby is gone, and so is the So is the pig-footed bandicoot. What goes with them is easy to miss. A single bilby shifts one to two tons of soil a year, digs burrows spiraling three meters deep, and shelters as many as 16 other species in them. Australia was losing the machinery that kept its ground working. So why not simply kill all the cats? Kill 90% in a region, and survivors breed back within two seasons into territory you emptied for them. Control is not removal. And that gap has absorbed most of the money Australia ever spent on this. Hence, the fences. The rabbit-proof fence in Western Australia was finished in 1907. Its main line running 1,833 kilometers and the rabbits were past it before it was done. The dingo fence runs 5,614 kilometers and is still patrolled at around 10 million dollars a year. It stands 1.8 meters high, exactly the height of New Havens. Dingos did stay off the sheep country and two separate ecological worlds were created in the process. The side without dingoes generally faring worse, which suggests, though the debate is live, that removing a top predator costs more than it saves. That is the record New Haven was built against. Now the mala, a rufous hare-wallaby, roughly cat-sized, belonging to a To a Warlpiri songline that runs from the Central Northern Territory into South Australia. There may once have been 10 million. By the 1950s, the species was assumed extinct until two small colonies surfaced northwest of Alice Springs in the 1960s and 70s. That was all of it. 22 animals were taken into care in 1981 and senior Yapa men were invited to the release site first because the mala was their animal and the country was theirs. An electrified fence went up in 1986 around roughly one square kilometer, the mala paddock, and by 1992 it held about 150 animals. So the team tried the step everyone was waiting for, releasing 81 mala into open desert between 1990 and 1992. Some lasted 18 months, and some bred. Everyone was eventually killed or vanished, and the sand after one of those losses pointed to a single fox. Then in October 1991, a wildfire took the last wild colony. The only mainland mala left alive were the captives and the animals inside that square kilometer of wire. Six years later, a single fox got in. Nobody knows how long it was there before they found it, and by the time they caught it, it had killed 70 mala. Close to half of everything the species had left on the mainland. 70 dead animals in one incursion looks like proof that fenced sanctuaries do not work. The field drew a narrower conclusion, and that narrowness is why we are still talking about this. That fence was electric and had held for 11 years. What it was not built for was an opponent that could afford to be wrong every night, forever, and only had to be right once. Read that way, the failure looks less like a bad idea than a bad specification, and everything built afterwards was designed around that reading. Arid Recovery, near Roxby Downs, opened in 1997 and worked out the design everything since has copied. It worked and it was tiny, a few thousand hectares in a country of 7.5 million square Square kilometers. The Mala survived only by being shipped out, to an island in 1998 and to an enclosure at Watarka in 2000. Alive and homeless. New Haven exists because of a cattleman who barely ran cattle. Alex Coppock took over the lease in 1959 and his wife named the property. Keeping only enough cattle to live on, rather than a commercial herd, meant that for 40 years, 2,600 square kilometers of desert sat under a lease that was hardly grazed. Conservation groups argue that is precisely why so much of its wildlife survived, and given what heavy grazing did to comparable country, the reading is hard to fault. Birds Australia bought the station in 2000 for its birds. The Australian Wildlife Conservancy took over in December 2005. Then the fence, and the first thing everyone asks about it. A cat can jump higher than 1.8 meters, so on paper this fence is too short. What saves it is that no cat clears a fence in a single leap, they jump at it, grab, and climb the last part. Stopping the jump is impossible. Stopping the climb is not, and the design does it three ways. The floppy top is the clever one. Mesh curves outward, and the top edge is left deliberately loose, so a cat that grabs it and puts weight on it folds the edge backwards and drops where it started. A loose edge cannot be climbed, and it costs almost nothing. Beneath that, a buried skirt of mesh runs down and outward under the sand, so anything digging hits wire and keeps hitting wire. Outside that, two electrified wires. Birds fly over all of it, which is a fair objection. So do bats and insects, and owls hunt across it freely. Four-legged predators walking in at ground level were the only target. Harder is why dingoes are excluded, given that dingoes kill cats. No fence yet built stops a cat and passes a dingo, so the choice was between a place with neither and a place with both. AWC took the first and keeps boars running outside the wire for the dingoes still out there. I would genuinely like to know what you would have chosen. The gate shut in March 2018 and immediately turned one problem into another. Sealed inside were the 46 cats, two foxes and an unknown number of rabbits, and all of them had to go because a sanctuary containing two cats is a feeding pen with good security. Rabbits counted because they compete for the plants the native diggers need and because they feed cats. Clearing a fenced area beats clearing open country for one unglamorous reason, nothing comes back in. The graph flattens rather than falling though. Cats 1 through 40 are hard work. The final six are a different job entirely, since those animals watched the traps go in and lived through the campaign that killed every cat they knew. No trap or toxin finished the job. A WC's feral control officers Murray Schofield and Liam O'Rourke began work two months before the fence was done, alongside the New Haven Warlpiri rangers, and searching that area was never the method. Reading it was. You look at a stretch of sand and know which animal made the marks, which way it went, how long ago, and whether it was hunting or moving through. Then you follow, sometimes for hours, until the cat tires before you do. 21 years earlier, that same skill had read the tracks of the fox inside the Mala paddock, long after it was too late. New Haven sits on Nagalia-Walpiri and Laritja country, and AWC calls the method two-way, Yapa knowledge and Kardia knowledge on the same ground at once. There was a false alarm in September 2018, a sign that turned out to be nothing. Then on the 4th of December, the last one was caught, and for the first time in over a century, a piece of Central Australia held none of them at all. The first species back had been waiting 25 years, and it arrived as an emergency. In late 2017, a wildfire tore through the Mala enclosure at Watarka, leaving survivors in open ground with hawks overhead. 27 animals were evacuated I graduated to New Haven and held in a small pen called Malakulangu. Their lineage traces back to the Mala paddock, making them direct descendants of the animals that Fox did not reach. In July 2019, 30 of them walked out into the full area. After more than 25 years, the Mala was living where it belonged rather than being stored somewhere safe. Others followed. The red-tailed Phascogale in 2020, a carnivorous marsupial barely bigger than a mouse, the Woylie in 2021, the Greater Bilby, Burrowing Bettong and Central Rock Rat in 2022, the Golden Bandicoot in 2023, the Brush-tailed Possum in 2024. The plan calls for at least 10. The Bilby is where this gets hard to believe. 66 went in across May and June 2022, drawn from two separate populations on purpose, since a group founded on one gene pool is a problem waiting to surface. A WC's first estimate, three and a half years later, came back at around 530, a modelled figure rather than a headcount, and arid zone numbers swing with rainfall, so treat it as indicative. The growth itself is real, and the reason is the best single fact in this story. A bilby is pregnant for 14 days, the shortest gestation of any mammal on earth. Young are born about the size of a jellybean, finish developing in the pouch over roughly 75 days, and females can breed from 6 months old and raise up to 4 litters in a good year. All of which looks like an animal built to sit out a bad desert and expand the moment conditions turn, and 2024 handed them exactly that. New Haven took 637mm of rain, the wettest year since 2001, and breeding success ran between 85% and 100% of adult females caught. Rain did the work, not the fence, is the objection you would expect. Except that the same rain fell across the unfenced desert and brought back nothing, which suggests rain does not create a population so much as give an existing one room to expand. Inside the wire was the only place anything was left to expand. Golden bandicoots tripled in two and a half years. Why only 9,400 hectares when the property is 30 times that? Because you must clear whatever you enclose and prove it stays clear, indefinitely. Construction was the short part of this job. If you stand at the fence at dawn and the line is visible, though not in the plants, spinifex and mulga look much the same either side. The difference is in the sand. Cat tracks, fox tracks, and rabbit scrapes outside, thousands of small diggings across the dunes inside. Certainty that there are zero cats rather than one clever survivor is impossible, so they check relentlessly 44,000 camera trap nights in 2024 alone, plus strips of sand smoothed so anything crossing has to sign its name. The whole achievement is a negative result that has to be reproduced forever. To Warlpiri people, the bilby is ninu, and the burrowing bettong is pudaya. These animals were missing from country that still held the stories about them. Alice Henwood, one of the trackers who found the last cats, has described it plainly. For a long time, they did not see any mala. One photograph from the 2019 release carries the whole thing. Christine Ellis, ranger and tracker, one of the people who emptied that fence of cats, stands at the release site holding a baby named Jeffrey Jongala Wilson, while 30 mala go out onto country his family has stories about. Schoolchildren from nearby Niripi have helped release bilbies, putting their hands on an animal their grandparents could describe from memory and their parents never saw. Australian shops sell chocolate bilbies at Easter rather than rabbits, and this is the reason. Almost all of it runs on donations, which points at the honest weakness of the whole model. A fence nobody is paid to check is a fence with a hole in it. So where does this actually stand? In favor, eight species returned, several breeding faster than the recovery plans allowed for, and the pattern repeating elsewhere. At Mallee Cliffs in New South Wales, 107 bilbies released between 2021 and 2023 produced a first estimate of 1,840 against and all of it from inside the field. Not everything worked. The red-tailed Phascogale went in during 2020 and has not been detected since June 2022. So that reintroduction failed and nobody is certain why. Success carries its own risk, too, since populations with nothing hunting them can climb past what the land feeds and crash in the next drought. Deepest of all, and still unresolved, inside that fence are hundreds of animals descended from dozens that have never met a predator. Catherine Mosby and colleagues at Arid Recovery have shown anti-predator wariness fades within a few generations, leaving them poor stock for release into country that still has cats. The safer you keep them, the less capable they become of being safe. Which brings us to the question from the beginning. Mosby's answer is to let a small number of cats back in on purpose. In Arid Recovery's trials, Bilby and Betong numbers rose through the first two years alongside 1 to 4 cats, and held with as many as 10 to 12. It sounds like sabotage, and it may be the only version of this project with an ending. I would like to know whether you would do it. Stage 2 would expand the predator-free area to 100,000 hectares, needing a fence of around 180 kilometres, and would be the largest cat clearance by area ever attempted. In 2026, it still reads as a plan rather than a construction site, and earlier target dates have slipped. New Zealand is attempting the opposite under predator-free. 2050, clearing rats, stoats, and possums from an entire country. Different tools, same conclusion. A fence is a starting position, and the finish line is the day the animals inside no longer need one. Nobody has reached it, which leaves a question I cannot answer cleanly. Is an animal living behind a fence still a wild animal? New Haven proved something narrower than the headline. Australia is not fixed, and this was never going to fix it. What those hectares show is that, gone from this region, described a pressure, not the soil. Two great fences tried to hold back a continent, and a continent was too big to hold. This one held a shape. In 1997, one fox found one gap and killed 70 mala. In 2018, three Warlpiri trackers found 46 cats and left none. Australia fenced the problem out. America is trying the opposite, putting something back in. If you have not seen how a dead Oregon creek was rebuilt by a scientist with a stack of fence posts and a rodent that works for free, that one is on the channel. Same idea, run backwards. Subscribe, and I will keep finding them.