Hunting for a phantom: how bioacoustic surveys helped find the elusive night parrot

Photo Credit: S.Murphy, Conservation Partners

An Enigmatic Bird

The endemic night parrot (Pezoporus occidentalis) is one of Australia’s most elusive bird species. In fact, it was considered extinct by many for almost a century before being rediscovered in 1990. Today, the night parrot is listed as critically endangered, but conservation work targeting it is burdened by a lack of knowledge about this elusive species. Its behaviour adds to the problem, as this small parrot inhabits extremely remote areas and is nocturnal, amplifying the difficulties to observe it.

Apart from a few sparing sightings throughout the years, two dead specimens were collected in 1990 and 2006 respectively, giving researchers valuable DNA-samples and physiological information. The birds seem to be adapted to the harsh environmental conditions of the outback, needing little water and hiding in spinifex thickets during the heat of the day, only coming out to forage after sunset.

Protection for the Unknown

One team conducting dedicated Night Parrot surveys in Queensland is Conservation Partners (CP). As visual methods are not effective tools to detect night parrots, the CP team is using passive acoustic recorders. Recorders are typically deployed for six months or more, often via helicopter due to access constraints. Around 145 autonomous sound recorders, including 37 Solar BARs, have been deployed across a vast area of outback QLD during 2024 and 2025. Deployment sites are usually chosen on the basis of high-resolution aerial imagery, showing the occurrence of roosting habitat, dominated by long-unburnt spinifex, and feeding habitat, which is characterised by floristically diverse run-on systems. A final decision on where exactly a recorder is to be placed is then made from the air or on ground.

The main aim of this long-term study is to locate previously unknown night parrot populations across Queensland’s Channel Country. This landscape is enormous, and remote, and any remaining populations are likely to be extremely small and highly localised, persisting in places with some inherent resilience to threats such as feral cats. Finding these populations is a prerequisite to any meaningful threat reduction or recovery effort.

Fig. 1: Solar BARs lined up to be deployed by helicopter. Metal grates are used to set up protective boxes around the recorders to shield them from cattle. Photo Credit: Steve Murphy

Reliabilty is Key

The nature of the deployment makes every field trip a high-cost operation, making reliable, low maintenance equipment invaluable. The Solar BAR caters to these needs, as once deployed, it weathers the harsh conditions of the Queensland outback well for several years, only requiring a maintenance trip every 6 months to change SD-cards and collect data. Set-up is made even easier through the option of mounting the recorder on a star picket. These can be short and placed inside protective mesh cages or on longer pickets, which also provides a level of protection. However, sometimes, especially with helicopter deployments where space is limited, units are deployed without such protection.

Being mostly deployed on cattle stations, damage by inquisitive livestock sometimes happens. Microphones are often chewed off, and so far, five units have been knocked over and trampled. In the future, an electric fence energizer attached to the aluminium housing may help with this issue.

Fig. 2: High in the air, Murphy and his team make the final decision on the deployment sites based on visual examination. Each deployment is planned extensively beforehand; failure cannot be afforded. Photo Credit: Steve Murphy

Fig. 3: Steve Murphy next to a fully set-up Solar BAR. The device will stay mounted to this star picket for 6 months before its first maintenance service, powered autonomously by its integrated solar panel. Photo Credit: Steve Murphy

Mountains of Data

Throughout the Channel Country, the CP team have accumulated recordings from 13 properties, totalling roughly 250,000 individual audio files and more than 31 terabytes of data. This volume of data is both the strength of the approach and its greatest analytical challenge.

One of the central problems is that night parrot calls are extremely rare within these datasets. Even a single confirmed call can be hugely informative, but detecting it requires systems that can search enormous volumes of audio without missing true events or overwhelming researchers with false positives.

Another central challenge is that night parrot vocalisations vary substantially across space and time. The species has a diverse vocal repertoire, and so the team is working collaboratively with Bush Heritage Australia, Ngururrpa IPA, Fortescue Metals and Rio Tinto to amass a large training dataset from both QLD and WA. This satisfies a key requirement to develop analytical approaches that are generalisable across landscapes.

The original detection pipeline, which the team has been using for many years, was very good at finding potential night parrot calls, but it produced large numbers of false positives that required time-consuming manual validation. To address this, the CP team is developing a new AI system that explicitly balances recall and precision, which are the two parameters that traditionally work against each other; typically, a system might be very good at finding all genuine Night Parrot calls (high recall), but it can only do so by returning a large number of false positives (low precision). The holy grail is building a system that returns high recall and high precision, and this is what CP’s new AI system is promising to do.

Striking Gold

All of this meticulous work finally paid off, when in April 2026, news broke that about 10 night parrot calls had been detected on a single recorder at an undisclosed location during September 2025. The pattern of calling suggests that the recorder was located within the range of the parrot population, but not at the exact roosting location. Calling patterns at long-term stable roosts are characterised by reliable calling practically every single dusk and dawn.

Fig. 4: The known range of the night parrot. Photo Credit: Conservation Partners

The unit where the September detections were made contained no calls during July and August. So the CP team knows it’s in the right neighbourhood, but the task now is to identify exactly where the parrots are returning every single day. Ten years of monitoring at Pullen Pullen shows that populations are typically extremely site faithful.

Once the reliable, long-term roosts are found, conservation efforts will switch to intensive threat abatement, primarily focussed on controlling feral cats. While it could be argued that night parrots at the site have managed to persist despite such efforts, the CP team says that we can always do better at protecting such a critically endangered species. The focus will be on ensuring that the population isn’t just maintained, but increases, so that dispersing young can recolonise surrounding areas.

In early 2026, heavy rains were recorded in south-west Queensland which has triggered breeding activity at Pullen Pullen. Previous research shows that breeding coincides with increased calling and so the CP team is urgently trying to get back in the field to deploy more recorders in the hope of pinpointing the stable roosting sites.

The parrots may be critically endangered and very elusive, but they are not without support. Through the work of the CP team, we have a chance to turn around what has been a precipitous decline in one of Australia’s most unusual species.

Fig. 5: A Solar BAR set up in the middle of a Spinifex growth, the preferred habitat of the night parrot. Photo Credit: Steve Murphy