University of Michigan Scientists Use AI to Find Bursts in Bird Evolution

Evolutionary theory has long posited that life doesn’t change in a linear crawl, but in pulses. While the fossil record has frequently hinted at this pattern, proving it requires a scale of data that has traditionally been impossible for humans to process. That changed with the application of a specialized AI tool and a massive dataset of skeletal measurements.

Skelevision and the 170,000 Measurement Dataset

The breakthrough relied on Skelevision, an AI tool developed by Brian Weeks’ laboratory in collaboration with New York University’s David Fouhey. The system allows researchers to digitize museum collections with speed, scanning individual specimens in roughly 45 seconds by photographing them against a consistent measurement grid.

The team used this technology to analyze more than 15,000 museum specimens, primarily from the U-M Museum of Zoology. This effort resulted in a dataset of over 170,000 individual skeletal measurements across more than 2,000 species of passerines. To make sense of this volume, lead author Jake Berv created a statistical method called bifrost, which treats the entire skeleton as an integrated unit rather than analyzing bones in isolation.

Climate Shifts and the Eocene-Oligocene Transition

The AI analysis revealed that bird body shapes didn’t shift uniformly. Instead, the researchers identified a massive burst of evolution approximately 35 million years ago. This spike coincided with the Eocene-Oligocene Transition, a geological era where Earth rapidly shifted from a warm “greenhousestate into a coolericehouse” world.

Flowering Plants Face Extinction Risk for 21.2% of Evolutionary History

According to the researchers, these climate upheavals created new ecological niches, forcing natural selection into overdrive. However, this acceleration eventually stalled. The data shows a cluster of evolutionary slowdowns around 15 million years ago, which aligned with another period of global cooling. By this point, most available ecological niches had already been filled.

“This pattern we found with rare, big increases in the rates of evolution and lots of small decreases in the rate of evolution is really consistent with a pattern where lineages are exploring new ecological space and changing rapidly to take advantage of that opportunity.”

Brian Weeks

Geography as a Predictor of Evolutionary Speed

Beyond climate history, the team tested whether current geography influences the rate of change. They found a correlation between latitude and morphological evolution.

This finding supports the broader theory of adaptive radiations—the idea that the emergence of new groups is often tied to explosive bursts of diversification. Whether triggered by a new continent or a shift in temperature, these pulses account for a significant portion of life’s diversity on Earth.

Broader Implications for Evolutionary Theory

The use of AI to corroborate century-old theories suggests a shift in how paleontologists and biologists approach morphology. By automating the measurement of thousands of specimens, researchers can now ask questions about the sequence of changes required to explain modern variation.

The University of Michigan study was supported by the David and Lucile Packard Foundation and Schmidt Sciences. As AI continues to refine the ability to scan and measure physical specimens, the remaining uncertainty lies in identifying exactly which ecological triggers—be they geographic or climatic—consistently spark these evolutionary accelerations.