Researchers from U-M, NYU and others collaborated to analyze thousands of museum specimens, confirming centuries-old ecological rule applies to avian wings
Study: Longer Wing Bones in Warmer Climates Suggest a Role of Thermoregulation in Bird Wing Evolution (DOI: 10.1111/geb.70033)
ANN ARBOR-For centuries, scientists have observed that animals in warmer climates have longer limbs, a pattern known as Allen's Rule. Long attributed to the need to maintain body temperature, the precise mechanism that gave rise to this pattern has remained poorly understood.
A new computer vision system has now confirmed this principle applies to bird wings, too. The finding reshapes our understanding of the evolution of bird wings to include the demands of temperature regulation in addition to flight mechanics. This finding could help scientists better understand how birds may be adapting to climate change.
"Collecting skeletal measurements on a large scale lets us answer big questions about how species evolve and interact with their environments," said Brian Weeks, lead author of the new study and assistant professor at the University of Michigan's School for Environment and Sustainability.
Published in the journal Global Ecology and Biogeography, the study represents the culmination of a six-year collaboration between ecologists and computer scientists at the University of Michigan and New York University.
The team's "Skelevision" system uses artificial intelligence to automatically identify and measure bird bones from photographs.
"We use a deep neural network to detect individual bones in specimen images, identify their type, and create a precise digital outline of each one," said David Fouhey, one of the study's senior authors and assistant professor at NYU Tandon School of Engineering and NYU's Courant Institute.
"Along with the co-designed hardware, we're able to reduce 3D measurement to a 2D task, in which current computer vision systems excel."
Before this technology, researchers tended to study skeletal traits in relatively small sample sizes. The laborious process required manually handling fragile bones and measuring each element with calipers, creating a bias toward more easily measured external traits.
The researchers designed and built a complete end-to-end system for analyzing bird skeletons-developing both the physical imaging hardware with a high-resolution camera positioned above a surface where bird bones are arranged and the sophisticated AI software that analyzes these photographs to identify and measure individual bones.
This integrated hardware-software approach reduces specimen handling time from about 15 minutes to about one minute each. The methodology was established in a 2023 report in Methods in Ecology and Evolution, demonstrating Skelevision's accuracy across 12,450 bird specimens.
This efficiency allowed researchers to analyze wing-bone measurements from 1,520 species of passerine birds across 80 families from all continents except Antarctica. The specimens came primarily from the University of Michigan Museum of Zoology, and the dataset has since been supplemented with specimens from Chicago's Field Museum of Natural History.
"Wing bones play a unique role in thermoregulation," Weeks said. "When birds fly, the muscles supported by these bones become crucial for dissipating the enormous heat generated by flight muscles. This suggests the pattern we're seeing-longer wing bones in warmer climates-is driven primarily by the need for efficient cooling rather than heat conservation.
"Even traits as critical as wings, which we've traditionally studied only for flight mechanics, are being shaped by thermoregulation demands. This has important implications for how birds might respond to climate change."
The technology is now being expanded with an advanced 3D scanning system to measure additional properties like volume and surface area. The researchers have also released their dataset and open-source code.
In addition to Weeks and Fouhey, the study's authors are Christina Harvey of the University of California Davis; Joseph Tobias of Imperial College London; Catherine Sheard of the University of Bristol; and Zhizhuo Zhou of the University of Michigan and Carnegie Mellon University.
The David and Lucile Packard Foundation provided funding for the research.
