Urban Acoustics Project: Wildlife Monitoring in your Back Garden

BY: .

People assume wildlife monitoring and conservation research is done in big nature reserves and game parks - places of wild nature, unfamiliar and foreign to our everyday urban and suburban lives.

However, an innovative study: Urban Acoustics Project from the University of KwaZulu-Natal, led by Professor Colleen Downs, is looking at which species persist in urban environments using acoustic monitors running in near real-time.

Although disconnected from traditionally “wild” areas, backyards, small streams, and public parks (ie, natural or managed green spaces in the urban mosaic landscape) can contain a surprising amount of wildlife, including birds, insects, frogs and small mammals. Interestingly, the presence of these animals provides opportunities to study how they persist in living near humans, how the environment supports animal diversity, and if endangered and vulnerable species can endure in these environments.

One of the biggest obstacles to conservation and diversity work is a lack of time and data: both are crucial to understanding how animals fully interact within landscapes. Some forms of data collection, like surveys based on human observation, can only collect information for a few hours at a time, which leaves a lot of potential animal activity unaccounted for. Part of understanding the full diversity of an area means recognising that animals are active at different times of the day, which means constant monitoring to get comprehensive biodiversity data. “Dedicated sampling surveys” using camera traps or acoustic monitors are a passive means to collect data for longer periods. That is why the Urban Acoustics Project runs its monitors for 24-hours in 10-day cycles, with battery changes on the 1^st and 15^th of each month.

The Urban Acoustics Project uses recording devices (acoustic receivers/ monitors) as its means of data collection. They hope their research may help the monitoring of sounds become a mainstream method for collecting large amounts of biodiversity data. The project plans to use audio recordings to determine the presence of vertebrate species - birds, fruit bats and frogs - in urban areas, and compare species across geographical locations and time. This data would also allow the researchers to determine which species are most active during specific hours of the day, how weather or seasonal shifts affect them and how species are distributed across the study areas.

Globally, there has been an increase in the number of publications researching urban wildlife and conservation. These studies suggest that urban environments support ecological diversity more than previously thought, especially in mosaic urban landscapes. New data gathered in the Urban Acoustic Project can inform urban planners and conservation practitioners about increasing and managing biodiversity while minimising the likelihood of negative human-wildlife interactions. This data is very important as it contributes to improving the knowledge gaps in Africa, South America and Asia compared to the global north.

The data generated by the Urban Acoustic Project is staggering, including the species’ calls and background noise. One of the many reasons that acoustic projects generally do not collect data at the 24-hour scale is that it would take too long for a person to listen and process the recordings into usable data. Here, the Urban Acoustic Project uses new technology, such as machine learning and Artificial Intelligence, to create and train algorithms that can identify audio calls, allowing data to be processed effectively and in a fraction of the time it would take one person to do so.

The technology is still new, and it takes time to train the algorithms accurately. One of the biggest issues is the need to have accurately labelled animal sounds - in this case, bird calls - to train your algorithm so it can pick them out of the collected data. However, if the AI is trained on clean sounds -Vanessa Süßle, one of the AI team working on the Urban Acoustics Project, helped explain - with no background noise, it would not be able to accurately pick out bird calls from the data which have background noise. Here, Dr Matthew Burnett, who is working on labelling data, says that even trained citizen scientists can assist in the labelling process. For example, those with monitors could keep monthly bird lists that can be compared to the acoustic data. Researchers are working on an innovative way to train AI to assist in solving these problems. The team extracts labelled data from databases like Xeno-Canto or Macaulay. They then embed these files, which are already labelled, into the recordings from the gardens involved in the project. Then, they can train the AI models to pick out the bird calls despite the background noise.

AI has been a controversial topic recently as it has become more prevalent in our daily lives, especially on social media. This research shows that AI can have positive applications worldwide and help researchers and scientists analyse valuable data from nature. This could help preserve nature for future generations.

Humans have become increasingly disconnected from wildlife and nature in recent decades. However, this distancing is not total. The research on urban wildlife shows that animals and nature find ways to persist around us even as we distance ourselves from it. This new area of urban wildlife research can help people find ways to bring nature back into our everyday lives in ways that can benefit wildlife, humans and, hopefully, our planet… So what is calling in your backyard?

If you would like to assist as a citizen scientist on the Urban Acoustic Project, please contact Ms Preshnee Singh on 033 260 5127.

Words: Beth Meyer

Photograph and image: Supplied