Life finds a way. Paleontologists have reiterated that maxim over and over, often drawing from the way organisms seemed to bounce back from the five mass extinctions Earth has suffered. These events winnowed down the number of species alive on the planet, the story goes. And the survivors set the foundation for life to flower again the way a forest can be refreshed by fire. But a new study suggests that paleontologists may be off tempo. Some of the greatest evolutionary radiations of all time happened without a mass extinction to spark them, and not every disaster opened space for new life.

In broad strokes, the connection between mass extinctions and dramatic radiations of new species seems intuitive. Organisms have their particular niches, and extinction creates vacancies. By this logic, “new species can only originate if a new empty niche appears or if the species can outcompete another one,” says biologist Daniele Silvestro of the University of Fribourg in Switzerland, who was not involved in the study.

Conventional wisdom had it that mass extinctions facilitated this zero-sum game on a larger scale. “A classic example is the suggestion that mammals could not radiate until dinosaur competitors were removed in the end-Cretaceous mass extinction event,” when an asteroid impact triggered the world’s fifth massive loss of species, says data scientist Jennifer Hoyal Cuthill of the University of Essex in England. But by analyzing fossil occurrences logged in the collaborative Paleobiology Database using a machine-learning algorithm, Hoyal Cuthill and her colleagues found that life’s most prominent pulses did not usually follow the world’s greatest biological wipeouts.

The diversity of life seems to shift at a range of times, the researchers concluded. Sometimes new species rapidly evolved without a mass extinction to trigger the explosive radiation. At other times, extinction crises cut back life on Earth, and speciation did not immediately follow. In the end-Cretaceous example, some groups of organisms—such as mammals and birds—were diversifying before the mass extinction began. The pattern, visualized as flows of diversity over time, was much more chaotic than expected. “I found it incredibly exciting as the first visible effects of major evolutionary events started to emerge,” Hoyal Cuthill says.

To track biodiversity through time, the researchers turned to the Paleobiology Database. The team selected more than more than 1.27 million fossil data points from the database to determine where and when prehistoric organisms occurred through the ages. The resulting graph told a more complex story than mass extinctions clearing the decks for rapid evolutionary radiations. “The two most extreme examples of radiation we identified were [near] the beginning of the Cambrian and Carboniferous periods,” Hoyal Cuthill says. The first, about 532 million years ago, was when animal life burst on the scene, and the predecessors of today’s arthropods, vertebrates and many other creatures evolved. The second, about 358 million years ago, occurred when some vertebrates evolved to live on land for the first time in their history.

Neither of these events was tied to a mass extinction. Instead, Hoyal Cuthill notes, “these are times when life is thought to have been diversifying into new ecological arenas.” During the Cambrian, animals began to feed on each other in more complex ways. The world developed its first truly complex food webs, from wormlike organisms that slurped up bacteria in the sediment to free-swimming predators festooned with compound eyes and grasping appendages. During the Carboniferous, forests proliferated over the planet and hosted communities of invertebrates, which fed the amphibious vertebrates that were just emerging onto land. Organisms’ ability to proliferate in new ecological roles was the key. As study co-author Nicholas Guttenberg says, “You don’t necessarily have to chip an existing piece off to allow something new to appear.”

The researchers also found that mass extinctions were rarely directly followed by radiations—the Ordovician-Silurian mass extinction happened 440 million years ago, for instance, but the data set suggests biodiversity did not rebound until more than 20 million years later. Of the “big five” mass extinctions over time, four were not immediately followed by evolutionary radiation. The one exception occurred as the Permian period gave way to the Triassic period: After the worst mass extinction of all time, about 252 million years ago, life quickly recovered in the early part of the Triassic. And the predecessors of dinosaurs and other reptiles underwent a major evolutionary pulse.

Taken with the other upticks in biodiversity Hoyal Cuthill and her colleagues detected, the findings indicate that mass extinction and mass radiation are not always coupled together as if life on Earth is maintaining an equilibrium. Mass extinction or radiations can happen with or without the other, constantly shifting depending on the circumstance.

Detecting these patterns is only possible because of decades of previous discoveries. “It’s really super exciting to see such sophisticated computational and data science approaches applied to questions in paleontology,” says data scientist Emma Dunne of the University of Birmingham in England, who was not involved in the new paper. Fieldwork finds new fossils, museum preparation gets specimens ready for study and phylogenetic work places those fossils in context of the tree of life. And then all those data can be placed into the Paleobiology Database to form the basis of new, broad-scale analysis, Dunne adds. “It’s pretty darn amazing,” she says.

These results are not only about the past, the study’s authors note. The world is currently going through a new extinction crisis, where species are disappearing faster than new ones are evolving. Under the old logic, experts might expect life to bounce back in a million years or so after the worst plays out. Judging by the new results, however, life’s diversity may not recover quickly. Or there might be some other circumstance—such as species pioneering new niches—that is needed for the next evolutionary pulse. “While it’s not necessarily entirely gloomy,” Dunne says, “it doesn’t bode well for the future of biodiversity.”