AI Measured 170,000 Bird Bones and Found Songbird Evolution Ran in Violent Bursts, Not a Steady Climb

Museum drawers hold millions of bird skeletons that nobody has time to measure. University of Michigan researchers pointed an artificial intelligence tool at them instead, and the resulting dataset has overturned a comfortable assumption about how songbirds took over the planet.

The team built a system called Skelevision to pull more than 170,000 individual skeletal measurements from over 15,000 museum specimens covering upwards of 2,000 bird species. Fed into evolutionary models spanning roughly 45 million years of passerine history, the measurements showed that body-shape change did not accumulate at a steady drip. It came in rare, violent bursts separated by long stretches in which rates of change actually slowed.

The largest burst lands about 35 million years ago, at the Eocene-Oligocene transition — one of the sharpest cooling events in the Cenozoic, when ice began building on Antarctica and forests across the northern hemisphere gave way to more open habitat. Passerines, the perching birds that today account for more than half of all bird species, were reorganizing their skeletons precisely as the world's habitats were being rearranged.

"This is really important for evolutionary theory because there's a long history, going back 100 years, that predicts the emergence of new groups called evolutionary radiations is often associated with an explosive burst of diversification," said Jake Berv, the lead author and a postdoctoral fellow at Michigan. The prediction is old; what has been missing is data broad enough to test it across an entire order of birds rather than a handful of well-studied lineages.

Senior author Brian Weeks, an associate professor at Michigan, said the shape of the signal is as informative as its timing. "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," he said. In other words, the bursts look like colonization events — birds moving into ways of living nothing else was using yet.

Geography turned out to matter too. Weeks noted that "it looks like there's a connection between latitudinal gradients and rates of morphological evolution that has been underappreciated," pointing toward a link between where a lineage lives and how fast its body plan changes. The tropics have long been recognized as holding more species; this suggests they may also run evolution at a different speed.

The study, "Rates of passerine body plan evolution in time and space," was published in Nature Ecology & Evolution by Berv, Charlotte M. Probst, Santiago Claramunt, J. Ryan Shipley, Matt Friedman, Stephen A. Smith, David F. Fouhey and Weeks, with support from Schmidt Sciences and the David and Lucile Packard Foundation.

The broader implication is about method. Natural history collections have been accumulating specimens for two centuries, and the limiting factor on using them has always been human hands with calipers. Machine measurement removes that ceiling, and questions that required decades of manual work are suddenly answerable in a single study.