People around the world are helping save Galapagos iguanas

The “Iguanas from Above” project brought together nearly 14,000 volunteers to count endangered Galapagos marine iguanas. These reptiles, found only in the Galapagos Islands, are facing many threats. Counting them from the ground is difficult due to the inaccessible terrain.

To overcome this, researchers at Leipzig University launched a drone-based survey. They then invited the public to analyze these aerial images online.

“This expands the role of citizen science, provides significant support for researchers, and offers a valuable opportunity to actively engage people in conservation topics,” noted Dr. Amy MacLeod, who leads the campaign.

This effort resulted in more than 1.3 million classifications across nearly 58,000 images. Each image was reviewed up to 30 times. The goal was not just participation but reliable population estimates.

Best methods for counting iguanas

To ensure reliability, scientists compared volunteer results with expert evaluations. A gold standard dataset of 4,345 images served as the benchmark.

Volunteers achieved 68-94% detection accuracy. They were more likely to miss iguanas than to add extra individuals by mistake.

Rather than use a majority vote, the team applied a minimum threshold method. If five or more volunteers agreed on iguana presence, the count was accepted. This method improved accuracy, especially when images were complex or crowded.

Clustering methods, especially HDBSCAN, performed best for counting iguanas. It grouped volunteer marks to identify individual iguanas more reliably than other approaches.

Accuracy of volunteer counts

The success of volunteer counts depended heavily on image quality. During the first project phase, drone images had poor focus and lighting.

These issues led to frequent undercounting. Later phases used clearer images, which resulted in higher accuracy.

Volunteers had trouble distinguishing iguanas from dark volcanic rocks. This was especially true for small or well-camouflaged iguanas. When image quality was high, count differences between volunteers and experts narrowed considerably.

Even with better images, undercounting remained common. Researchers found that when many iguanas were in a single frame, volunteers often reported fewer than were actually there.

What made people join or quit

Not all volunteers contributed equally. Most submitted fewer than 50 classifications. But a small group, known as “super volunteers,” completed thousands. Some volunteers preferred staying anonymous, yet still showed high dedication.

Feedback from surveys revealed what discouraged participants. Blank images – photos without any iguanas – reduced motivation. Blurred images and difficult terrain also caused fatigue. Repetition made the task less engaging for some users.

Still, the volunteers’ dedication proved strong. One contributor analyzed every image in all three phases. Many were driven by the desire to help science rather than by any external rewards.

Are the iguana counts reliable?

The research team tested whether removing inputs from anonymous or inexperienced users would improve results. They found it reduced the number of useful classifications and sometimes harmed accuracy.

While using more classifiers per image might seem helpful, it did not always increase reliability. The key was balancing quantity with quality and giving more weight to the counts of skilled participants.

The HDBSCAN method gave the most consistent results. It required fewer assumptions and handled spatial data better. It also offered a base for future machine learning models that could automate parts of the work.

Toward a smarter monitoring system

This project confirms that volunteers can help count iguanas, but still tend to undercount. The team plans to apply correction factors for future population estimates. They aim to use the data to update the IUCN Red List and improve conservation plans.

They are also testing whether volunteers can identify reproductive behaviors, such as males in breeding colors or lekking sites. These insights could enrich understanding of marine iguana ecology.

Moving forward, the researchers will integrate machine learning into the workflow. AI models will be trained using volunteer data. These tools may help detect iguanas and filter out empty images. The hope is to create a semi-automated pipeline for faster and cheaper monitoring.

Wider impacts and future vision

This method could apply to other hard-to-monitor species in remote areas. Online platforms like Zooniverse show that people worldwide are eager to help conservation science.

The combination of drone surveys and crowdsourced analysis can reshape how we approach large-scale wildlife monitoring. According to the researchers, if experts alone had analyzed the images, it would have taken decades.

Now, with volunteer help and upcoming AI tools, they aim to complete a full population survey of the Galapagos marine iguana by 2026. The project is still live, and anyone can join to help protect this unique species.

The study is published in the journal Scientific Reports.

—–

Like what you read? Subscribe to our newsletter for engaging articles, exclusive content, and the latest updates. 

Check us out on EarthSnap, a free app brought to you by Eric Ralls and Earth.com.

—–