Listening to the forest: how acoustic monitoring is transforming mohua conservation in the Southern Lakes
Date: 26th June 2026
How do you monitor a small forest bird across thousands of hectares of rugged mountain terrain?
Traditionally, monitoring forest birds has relied on skilled field teams carrying out methods such as five-minute bird counts, transect surveys, observing birds and tracking breeding activity. These approaches remain the foundation of our monitoring programme, providing information that technology alone cannot. But they capture only a snapshot in time.
Acoustic monitoring adds another powerful tool to the conservation toolbox. By placing automated recorders throughout the forest, we can passively “listen” for weeks at a time without disturbing wildlife, capturing thousands of hours of bird activity that would otherwise go unnoticed. Combined with traditional field observations, it gives us a far more complete picture of where birds are, how they are using the landscape and how populations change over time.
For the past four years, Southern Lakes Sanctuary has been working with Digilab NZ to develop and refine acoustic monitoring techniques for mohua in the Makarora Valley. By combining automated sound recorders, artificial intelligence and traditional field observations, we are gaining a deeper understanding of mohua behaviour and the wider forest ecosystem than ever before.
Turning birdsong into data
During the 2025–26 breeding season, Southern Lakes Sanctuary staff deployed 12 automated recording devices throughout a key mohua breeding zone in the Makarora Valley. Positioned in areas known or suspected to be used by mohua, the recorders captured nearly 1,500 hours of audio between November and January.
Rather than manually listening to every recording, Digilab’s specialised algorithms analysed the audio, identifying and classifying bird calls automatically. This technology can distinguish between different types of mohua vocalisations, including chatter calls, melodic songs and trilly songs.
In total, more than 20,000 mohua vocalisations were detected across the monitoring period.
These detections allow researchers to identify where birds are active, how territories are being used and how behaviour changes throughout the breeding season. By combining acoustic monitoring with our long-term banding programme, researchers are now beginning to identify individual mohua by the unique characteristics of their calls. The technology can also distinguish between male and female vocalisations, offering a powerful new way to monitor breeding pairs, understand population structure and determine ratio of females to males as the population continues to recover.
More than just mohua
While mohua remain a key focus, one of the most exciting developments was the use of Digilab’s multi-species detection algorithm.
The system was trained to identify 18 native bird species from the recordings, including kea, kākā, korimako (bellbird), tītipounamu (rifleman), pīwakawaka (fantail), pipipi (brown creeper) and ruru. The result was a remarkable snapshot of biodiversity across the valley.
From the 1,489 hours of daytime recordings alone, the system identified more than 249,000 bird detections. Miromiro (tomtit) and riroriro (grey warbler) were among the most commonly detected species, while kea, korimako, pīwakawaka, pipipi, tītipounamu and tūī were recorded at every monitoring site.
Even the limited overnight recordings proved valuable, detecting ruru at all twelve recorder locations.
This ability to monitor multiple species simultaneously opens up exciting opportunities for conservation. Rather than collecting information on a single target species, we can now begin building a broader picture of how entire bird communities respond to predator control and habitat management over time.
Confirming what we see in the field
Sites where field teams observed territorial mohua pairs were also the locations with the highest levels of acoustic detections. At some locations, differences in detection rates between nearby recorders helped pinpoint where birds were nesting.
This strong alignment reinforces the value of acoustic monitoring as a complementary tool rather than a replacement for traditional conservation work. Field observations provide context and behavioural insights, while acoustic monitoring delivers a continuous stream of information that would be impossible to collect through human observation alone.
Expanding the technology across the Southern Lakes
The success of acoustic monitoring in Makarora is now helping shape biodiversity monitoring across other Southern Lakes Sanctuary project areas.
In December 2025, Southern Lakes Sanctuary and Digilab also deployed 12 acoustic recorders throughout the West Matukituki Valley alongside our annual bird monitoring programme. Over 1,000 hours of recordings were analysed using the same multi-species algorithms, detecting more than 136,000 bird calls across 17 native species.
The monitoring provided an important snapshot of the valley’s native bird community, with species including miromiro, kea, kākā, korimako, kakaurai, koekoeā, tītipounamu and pīwakawaka recorded across the network of sites. Pekapeka (long tailed bats) were also recorded at 9 out of 12 sites.
Perhaps most excitingly, the recorders also detected mohua at four locations just two months after birds were reintroduced to the valley in October 2025.
Because several recorder locations had also been monitored in 2024, the team was able to compare bird activity between years using the same detection algorithms. This demonstrates another strength of acoustic monitoring: by returning to the same sites over time, we can build long-term datasets that reveal how bird communities change as restoration work progresses.
The future of conservation listening
What began as a project focused on understanding mohua in Makarora has evolved into a powerful tool for monitoring biodiversity across the Southern Lakes Sanctuary. From tracking breeding behaviour and territory use to measuring the response of entire bird communities following predator control and species reintroductions, acoustic monitoring is helping us answer questions that would once have required thousands of hours in the field.
As these long-term datasets continue to grow, they will become increasingly valuable for understanding how our ecosystems are recovering. By combining field observations with artificial intelligence, we’re not just listening to the forest – we’re learning from it.
Posted in: Home Page, News and stories