The tweet double that saves birds

Who could recognise a bird species by its song? Enthusiasts try, but not everyone has the same level of knowledge. This is why, in Finland, artificial intelligence is used to organise amateur sightings and scientific monitoring of migrations. The result is a digital twin – a virtual replica that enhances biodiversity protection and brings people closer to listening to the sounds around them. Here is how

The climate crisis anticipates migratory events, creating ecological mismatches between breeding times and resource availability, especially for birds migrating in Europe. The climate crisis anticipates migratory events, creating ecological mismatches between breeding times and resource availability, especially for birds migrating in Europe. These creatures, more than others, suffer the impacts of these changes, and their rapid response can help us detect, mitigate, and prevent them. «The environment is now changing extremely fast, and we need to find ways to generate and use scientific knowledge much more quickly than we do», explains Otso Ovaskainen, Professor of Mathematical Ecology and Statistics at the University of Jyväskylä (Finland).

These are not just words. With his team: Ovaskainen has launched a project that combines citizen science with AI-based bird sound classification to better understand how they move and how we can act to mitigate the global and local effects of the climate crisis. The project is called Biodiversity Digital Twin¹, a digital copy of bird biodiversity capable of evolving with it.

The dual risk of citizen science

The idea arose from necessity but also from the immediate availability of vast amounts of data, provided by a systematic bird presence-counting programme active in Finland for several decades. «This is valuable information for assessing the influence of climate change on – Ovaskainen explains – The large number of bird enthusiasts holds enormous potential for science, as it allows for much larger datasets than those collected solely by professional researchers. However, there are many issues that can compromise the value of this data». By definition, for a project to be truly considered citizen science, scientists and researchers must also be directly involved, often on a voluntary basis.

«Some citizens are very knowledgeable about birds, while others know them less well. It can be very difficult to assess from the data which observations are reliable – the researcher notes – An even more serious problem is that most of the data is “presence only”, meaning it reports when citizens have observed certain species but not when they have tried to observe them and found none».

The risk is twofold: you can encounter false positives if someone claims to have seen a species that wasn’t really there, or false negatives if the observer fails to report a species that was actually present.

There is also the issue of spatio-temporal distortion, as citizen science projects are generally not based on systematic or randomised sampling schemes, but rather on opportunistic sampling. Not everything reported should be taken as gospel, but at the same time, it should not be discarded. Innovative solutions are needed to optimise the use of this data. A double-check is needed, and an application with a dual approach, like the one designed by Ovaskainen.

The dual solution of the digital twin

A digital twin is a virtual copy of a physical object used to simulate real-world changes. In his digital twin dedicated to migratory birds, Ovaskainen combines official long-term bird monitoring programmes with citizen science input get the best of both worlds.

«Our aim is to link the data citizens send via smartphones to our databases and use both methods simultaneously on a subset of information to improve calibration» he says.

In practical terms, citizens can “double” birdsong digitally by recording it and uploading it to the application. While, in the traditional process, they were asked to enter details of the song in specific captions, now the classification of species and the calibration of vocalisations is handled and refined by artificial intelligence. «By storing raw audio, we can also later reclassify the data by exploiting new, better performing AI models», Ovaskainen adds, describing how the app he created works.

It is called MK, an acronym for “Muuttolintujen kevät”, which means “Spring of Migratory Birds”, and as an output, it provides the most probable species classifications along with their reliability rate. Recently, citizens have also been given the option to contribute systematic five-minute recordings in predefined locations, earning their own “field scientist” scores.  It’s a way to engage them, while also generating continuously updated predictions on the spatio-temporal distributions of birds and their singing activity.

Widespread awareness and deep scientific knowledge

The dual track of digital-physical and citizen science-formal science offers double benefits. For scientists, the app helps better understand the factors behind changes in bird distribution, primarily those linked to climate change and habitat shifts. For users, it allows them to relate their own observations to the advancing spring migration.

«We are also noticing a significant impact in terms of awareness – adds Ovaskainen – Digital twins like this are making people more aware of the environment and nature».

Our project quickly gathered 250,000 users, many of whom had never actively observed birds before. Many have told us that it has brought them closer to this world, allowing them to experience new and interesting things».

Active and appreciated, this Biodiversity Digital Twin prototype is just the beginning of an ambitious project. Having found a way to obtain reliable data from citizen science and adequate real-time scientific bird monitoring, the team that created it now wants to further develop the platform. «We will develop an internet portal that displays data and predictions providing a proof-of-concept of a real-time biodiversity digital twin», says Ovaskainen, convinced that this kind of technology has the potential to revolutionise biodiversity research.

The future of research is double and digital

Whether it is migratory birds or other animal or plant species, a tool capable of converging the efforts of citizens and scientists to create a “digital duplicate” for studies and research is also a political tool, according to Ovaskainen. «By “playing” with the digital twin, it is possible to contribute to improving strategies and plans for biodiversity conservation and the economic mechanisms that guide them. Greater public awareness of science can also encourage more people to take part in environmental initiatives – he says – Monitoring ecological communities in real time through digital technologies makes it possible to scale data from local to global levels, facilitating conservation actions based on information more quickly than in the past». Ovaskainen speaks of a new era of automated monitoring systems that can accelerate the European Green Deal² and facilitate more sustainable land-use decisions. He mentions examples such as ARISE³ and BioAcousticsAI⁴, and looks to the future of research, hoping that the prototype he is working on will contribute to greater accountability of those who shape medium and long-term strategies, and lead to greater investment in Digital Twin technologies.

To study the impacts of the climate crisis, to limit them, to test solutions more courageously, knowing that you are doing so on a digital double without disturbing a nature and a planet that urgently need targeted and effective solutions, not hasty or, worse, token efforts. The birds migrating to Europe are showing us this – and singing it to us.  

---

1. For an explanation of the prototype app created by the Finnish research team, seeOvaskainen, O., Lauha, P., Gordillo, J. L., Nokelainen, O., Rahman, A., Souza, A., Talaskivi, J., Tikhonov, G., Vancraeyenest, A., & Lehtiö, A. (2024). Prototype Biodiversity Digital Twin: Real-time bird monitoring with citizen-science data. Research Ideas and Outcomes, 10. https://riojournal.com/article/125523/element/8/220330// ↩︎
2. For further details on the Green Deal, see The European Green Deal. (2021, July 14). European Commission. https://commission.europa.eu/strategy-and-policy/priorities-2019-2024/european-green-deal_en ↩︎
3. The ARISE project dedicated to monitoring biodiversity through digital twin technologies implemented in the Netherlands https://www.arise-biodiversity.nl/ ↩︎
4. Details of the BioAcousticsAI project, a successful example of AI monitoring and citizen science https://bioacousticai.eu/ ↩︎