A team at the University of Michigan has used artificial intelligence to uncover a striking pattern in the evolution of songbirds. Their analysis, published in Nature Ecology & Evolution, suggests these birds did not evolve at a steady pace. Instead, major changes arrived in rapid bursts that lined up with shifts in Earth's climate.
The researchers studied more than 2,000 species and assembled over 170,000 individual skeletal measurements using Skelevision, an AI tool developed by their lab in collaboration with NYU. The system photographs bird skeletons and accurately measures 12 bones across a bird's skeleton, allowing researchers to digitize entire museum collections far more efficiently than would otherwise be possible.
The analysis revealed a period of especially rapid body-shape evolution around 35 million years ago, coinciding with the Eocene-Oligocene transition — a period marked by intense global cooling. A cluster of evolutionary slowdowns around 15 million years ago also matched major geological events.
Bird communities at more extreme latitudes, where seasonal temperatures vary more sharply, tend to include species that evolve faster than those living closer to the equator. The findings suggest environmental variability may play an important role in driving changes in body shape.
The work highlights how AI is making it possible to extract information from preserved specimens at a scale that would have been difficult to imagine in the past, giving museum collections new scientific value.
