From bytes to biodiversity

Wildlife recognition technology:

Although camera traps have revolutionised biodiversity monitoring, a single camera can collect thousands of images and ecologists spend days, sometimes months, manually reviewing them. Artificial Intelligence is improving image processing speed and cost-effectiveness. Our system has processed ~23 million images since we launched, and staff are now regularly processing single camera trap surveys with more than 1.5 million images. The efficiency gains are huge: the AI system is saving two people at least three months’ work each year, just from the AI filtering out images that have no animal in them. AWC aims to develop species recognisers for up to 120 mammals and reptiles.

All ears for conservation:

Environmental sound, often referred to as eco- or bio-acoustics, is a powerful data source for investigating ecosystem health by captured sound. The non-intrusive technology is particularly useful for detecting animals that are hard to find, like Koalas, frogs, and some species of bird, and can help monitor the recovery of threatened species after events like bushfires, or map migratory patterns throughout a season. Current advancements have seen the development of solar-powered units capable of recording continuously (or scheduled) for years with only minimal service visits to change the memory cards. This is delivering efficiencies for staff and benefits for monitoring shy wildlife. AWC is now using such units to listen to arid-dwelling species such as the critically endangered Plains Wanderer.

A bird’s-eye view:

Efficiently tracking reintroduced species and producing reliable estimates of population trends over time is fundamental to the planning, delivery and ongoing evaluation of AWC’s ambitious reintroduction program. Drones equipped with advanced radio-telemetry systems are now being piloted to monitor reintroduced species at Mt Gibson. These drones can detect multiple animals simultaneously, enabling ecologists to monitor movements across large areas. “In the past, tracking translocated animals has involved hours of walking and searching for collar signals,” says AWC Senior Wildlife Ecologist Sophia Callander. “The drone, on the other hand, has proven to be a great tool for detecting multiple signals and finding animals in a much shorter time.” Since first acquiring the drone in 2022, AWC has conducted nearly 200 flights to track radio-collared Brushtail Possums, Chuditch and Numbats at Mt Gibson.

Taking aim at feral cats:

Felixers are box-like units, which use LiDAR, cameras and AI to distinguish feral cats from native animals. Once a feral cat is recognised, a toxic gel is sprayed and ingested when the cat grooms itself, making the system a more humane and less time-consuming model for feral animal control. The units also act as a camera trap, collecting data on prey species and cat distribution across an area. Following the 2019–20 ‘megafires’, Felixers were deployed on Kangaroo Island in South Australia to reduce the impact of cats on populations of displaced species such as the Kangaroo Island Dunnart. AWC is also trialling their use at Mt Gibson as part of an ‘outside the fence’ research project.