A program that can accurately predict the next destination of a migrating bird, a feat that is considered one of the toughest in biology, has been developed.
People have been studying bird migration for a very long time.
Getting accurate, up-to-date information on bird locations and migration patterns is extremely challenging.
A novel predictive model that can reliably anticipate where a migrating bird will travel next—one of the most challenging challenges in biology—was recently unveiled by computer scientists at the University of Massachusetts Amherst in partnership with biologists at the Cornell Lab of Ornithology.
While still being refined, the model, known as BirdFlow, should be made accessible to scientists within the next year and will ultimately be made available to the general public.
There have been and are still many efforts to tag and track individual birds, which have given us a lot of useful information. However, it is challenging to physically tag enough birds—not to mention the cost of doing so—to provide a thorough enough image to forecast bird movements.
“It’s really hard to understand how an entire species moves across the continent with tracking approaches,” adds Sheldon, “because they tell you the routes that some birds caught in specific locations followed, but not how birds in completely different locations might move.”
The number of citizen scientists who watch and report migratory bird occurrences has increased dramatically in recent years. Through eBird, a project run by the Cornell Lab of Ornithology and worldwide partners, birders from all over the globe submits more than 200 million bird sightings each year. With hundreds of thousands of users, it is one of the biggest biodiversity-related research initiatives ever created, enabling cutting-edge species distribution modeling via the Lab’s eBird Status & Trends project.
“eBird data is amazing because it shows where birds of a given species are every week across their entire range,” adds Sheldon, “but it doesn’t track individuals, so we need to infer what routes individual birds follow to best explain the species-level patterns.”
BirdFlow uses the estimates of relative bird abundance from eBird’s Status & Trends database and puts that data through a probabilistic machine-learning algorithm. To “learn” to anticipate where individual birds will travel next during migration, this model is fine-tuned utilizing real-time GPS and satellite tracking data.
Using data from 11 different North American bird species (including the American Woodcock, Wood Thrush, and Swainson’s Hawk), the researchers found that BirdFlow not only outperformed other models for tracking bird migration, but that it could also accurately predict migration flows without the real-time GPS and satellite tracking data.
According to co-author and postdoctoral scholar at the Cornell Lab of Ornithology Benjamin Van Doren, “birds today are experiencing rapid environmental change, and many species are declining.”
Van Doren continues, “Using BirdFlow, we can unite different data sources and paint a more complete picture of bird movements with exciting applications for guiding conservation action.”
Image Credit: guizmo_68, CC by 2.0