Only two-and-a-half kilometres from the Beehive, one of the most interesting conservation projects in the country is breaking ground. The Karori Wildlife Sanctuary Trust, a private community group, is setting up a "mainland island" in the former Karori Reservoir, a valley nestled between the two populous suburbs of Karori and Highbury/Brooklyn, and bordered at the top by reserves, lifestyle blocks and farms. Mainland islands have been established in a number of locations around the country by the Department of Conservation over the past decade, following success in increasing the numbers of kokako in the original Mapara "island" south of Te Kuiti. All these mainland islands rely on ongoing pest control and natural physical boundaries to keep predator numbers inside the island low. This approach wasn't possible in an urban area. The trust decided instead to build a fence—one that would exclude 14 undesirable species of mammal: possums, stoats, ferrets, cats, dogs, rabbits, hedgehogs, weasels, two species of rat, mice, deer, goats and pigs. Nobody had previously constructed such a fence. What was more, there was virtually no research into the physical capabilities of New Zealand's key predators. "For example, we didn't know how high feral cats could jump, or how well possums could climb, or how deep rats could burrow," said Stephen Fuller, the island's general manager. Fuller, an ecological consultant, coordinated a team of engineers, scientists and environmental consultants, fencing contractors and metal and wire fabricators to trial fence designs starting in early 1996. To enclose the 250 ha reserve, 8.6 kilometres of fence was going to be required—not an undertaking to be approached haphazardly. The team erected a wooden building with two compartments, separated by the test fences under trial. Animals were put on one side, food on the other. At first the animals were aided to traverse the barrier between the compartments to reach their food. Then the aids were removed, and the animals left to their own devices. One possum was videoed making 100 unsuccessful attempts to scale the fence in a single night. Over 18 months, 100 possums, nine feral cats, seven stoats, 10 Norway rats, 20 ship rats and 55 house mice pitted their strength and cunning against various designs. "The trials put the fences under more strain than they would get in the wild," noted Fuller. "After three days without food, the animals were highly motivated to beat the fence. Normally they wouldn't try so hard." The final fence design specifically avoided using electrification, the most common method used for keeping climbing animals off pest-control fences. "Our proximity to the city, and often extreme weather conditions, posed too many maintenance and operational problems for electricity," said Fuller. The fence design chosen, and now under construction, is 2.2 m high (cats can jump up to 1.85 m), and made of a mesh tightly woven from 2.5 mm high-tensile galvanised wire. The gaps in the mesh are only 6 mm by 50 mm-narrower than the head of a mouse at weaning. The fence has a shallowly buried underground skirt of the same mesh extending 400 mm out from the bottom of the fence at right angles, tests having suggested that when some animals find that they cannot get through the fence, they start digging against it. A modular design enables damaged portions to be easily replaced. The crowning glory is a distinctive "top hat," a smooth, rounded hood of galvanised zincalume which extends 250 mm out from the fence and has been shown to effectively repel possums and other climbers. The hood has also been tested to withstand Wellington's wind gusts. Secured by wooden posts at three-metre intervals, this elaborate fence costs $180 per metre in materials alone. Adding the other costs associated with construction—clearing trees, surveying, putting in drains and gates, and erection—swell the cost to $240 per metre. In preparation for construction, a four-metrewide lane was cleared along the line of the fence—three metres to be outside the wire and one metre inside—so animals don't have any climbing aids. This lane is to be kept permanently bare. Most of the fence follows the natural contour lines of the valley ridge. The steepest section is where it plunges to the valley floor, opposite the main road into the sanctuary. Staff gathered to watch one bulldozer being anchored to another and dropped down a 45-degree slope to clear the vegetation from this section. "By the time it reached the bottom, the driver had sore legs from standing on the dashboard," said Fuller. "We all cheered." The last of the perimeter was cleared in February 1999, and fence building started immediately. Construction is expected to be complete by August. The trust has launched a public campaign encouraging Wellingtonians to buy a fencepost to support the sanctuary. By June, 750 posts (each bearing the donor's name) had been purchased. So far, $6.5 million has been raised towards the $9 million total cost of the project. This fence is expected to have broader applications, and the design has been patented. "Ours is the Rolls-Royce of fences," remarked Fuller. "Other projects may need to exclude fewer types of predators, and could use our research to develop simpler fence designs." Possible uses extend beyond conservation, and the sanctuary is talking with a power station and food manufacturer, both of whom want to exclude mammals without relying on electric fencing. After the fence is completed, all mammals inside the area will be eradicated—hopefully by January 2000—and then the real excitement will begin, as native birds and reptiles are reintroduced to the valley. Wellingtonians should be hearing kiwi calling in the night within two years.