Merlin bird app to feed data into global biodiversity project
Merlin bird app to feed data into global biodiversity project

Merlin app data to boost eBird database

The Merlin bird identification app will soon allow users to automatically feed real-time bird identifications into eBird, one of the world's largest citizen-science biodiversity projects. The update aims to aid conservation of at-risk birds.

Since 2021, the free Merlin app, created by the Cornell Lab for Ornithology, has used machine learning to provide almost instantaneous sound-identification for birdsong, along with an image for each bird identified. Users' detections will now be automatically collected on the global online database eBird, which contains over 2 billion bird observation records.

Declining bird populations in the UK

In the UK, the total bird population has fallen by more than 70 million in the last 50 years, according to the British Trust for Ornithology. The Guardian created an audio soundscape comparing the abundance of birdsong in 1976 with today.

Wide Pickt banner — collaborative shopping lists app for Telegram, phone mockup with grocery list

Almost 2 million people in the UK used Merlin in May this year to identify birds in gardens, woodland, and countryside. Different birdsong makes distinct patterns on spectrograms, and Merlin has been trained to recognise the shapes and attribute them to a species.

Integration with eBird platform

Cornell also runs eBird, created in 2002 to gather millions of bird observations from citizen scientists, building one of the world's largest environmental science platforms. Jessie Barry, one of the leaders of the Merlin project, said: “The eBird mobile app will soon have the ability to upload recordings, which can be recorded in Merlin. Upcoming feature developments will make an even better link to the eBird systems so that we can use the data from what users ‘hear’ with Merlin to monitor bird populations.”

Barry added: “This data helps create tools that can be used to further conservation, inspire support and inform ecological management strategies.”

App reach and expansion

Currently, Merlin can identify 2,066 bird species, including most birds in the US, Canada, and Europe, and common species in India, Central and South America. Barry noted: “It is always an ongoing project to collect additional species. There’s a few we would like to add but we are always adding more and improving the performance of the models.”

The app has been downloaded more than 40 million times in 240 countries, up from 33 million downloads as of December last year. Britain has the second highest total number of users, with almost 2 million on the app in May this year. Canada, Germany, France, Spain, and the Netherlands are in the top 10 countries with the highest usage.

Concerns about misidentification

While Merlin is praised for connecting people to nature and aiding conservation, some concerns exist about misidentification. The European Bird Census Council recommends not using Merlin in official breeding bird surveys and has set up a monitoring group to coordinate acoustic bird monitoring across Europe.

Prof Richard Gregory of the RSPB said: “It is super-positive that it is increasing in popularity and use, and especially to a new, wider and different group of people. Everywhere I go, I see people are using Merlin, connecting with nature and learning their birds, becoming more curious to know more. Fantastic – a revolution.”

However, Gregory warned the app still makes mistakes and had identified his dachshund as a mallard. “If you aren’t an expert, you wouldn’t necessarily know there was an error. It is interesting to hear that Cornell is making changes which mean that recordings from Merlin will go to eBird more easily, so that’s great if the species recognition is correct [but] a problem for conservation when it is not.”

Barry responded that research teams regularly using these data sources would address challenges with data quality. “Our ability to understand changes in bird populations will be better served with more data to work with than if we don’t collect it at all.”

Pickt after-article banner — collaborative shopping lists app with family illustration