Free-thinking computers reverse-engineered the fossil record to identify the causes of a cataclysm.
To settle the long-standing debate about whether a massive asteroid impact or volcanic activity caused the extinction of the dinosaurs and more species 66 million years ago, a team at Dartmouth College took an innovative approach – they removed scientists from the debate and let computers decide.
Researchers report in the journal Science A new modeling method powered by interconnected processors that can work through reams of geographic and climate data without human input. They used approximately 130 processors to reverse-analyze the fossil record to identify events and conditions. Cretaceous-Paleogene (K-Pg) extinction event, which cleared the way for the rise of mammals, including mammals, leading to early humans.
A new perspective on historical events
“Part of our motivation was to assess this question without preconceived hypothesis or bias,” said Alex Cox, the study’s first author and a graduate student in Dartmouth’s Department of Earth Sciences. “Most models move forward. We adapted a carbon-cycle model to run the other way, using effect to find cause through statistics, providing only minimal prior information when working towards a particular outcome.
“In the end, it doesn’t matter what we think or what we thought before—the model shows us what we see in the geologic record,” he said.
The model crunched more than 300,000 possible scenarios of carbon dioxide emissions, sulfur dioxide emissions, and biological productivity over the 1 million years before and after the K-Pg extinction. By a category Machine learning Known as Markov chain Monte Carlo — it’s not like a smartphone predicts what you’ll type next — processors worked together independently to compare, iterate, and recalculate their results until they reached a scenario that matched outcomes preserved in the fossil record. .
Identifying the causes of destruction
The geochemical and organic remains in the fossil record clearly capture the catastrophic conditions during the K-Pg extinction. Animals and plants around the world suffered massive extinctions as food webs collapsed under an unstable atmosphere—rich in sun-sucking sulfur, airborne minerals, and heat-blocking carbon dioxide—shifting en masse from frigid conditions to scorching conditions.
Although the effect is clear, the cause of the destruction is not resolved. Early theories of the cause of volcanic eruptions were eclipsed by the discovery of an impact crater in Mexico called Chicxulub, caused by a mile-wide asteroid now thought to be the primary cause of the extinction event. However, theories are beginning to coalesce as fossil evidence points to a one-two punch not seen in Earth’s history: An asteroid may have slammed into a planet already reeling from massive, highly violent eruptions of volcanoes in western India’s Deccan Traps.
But scientists still don’t know — or agree — to what extent each event contributed to the mass extinction. So Cox and his adviser, Brenn Keller, a Dartmouth assistant professor of geosciences and study co-author, decided to “see what you get if you let the code decide.”
Modeling results and volcanic impact
Their model suggested that the release of climate-changing gases from the Deccan traps alone would have been enough to trigger the global extinction. The traps were exploding about 300,000 years before the Chicxulub asteroid appeared. During the roughly 1 million years of eruptions, the Deccan vents are estimated to have released 10.4 trillion tons of carbon dioxide and 9.3 trillion tons of sulfur into the atmosphere.
“Historically we’ve known that volcanoes cause mass destruction, but this is the first independent assessment of volatile emissions taken from sources of their environmental impacts,” said Keller, who last year published a paper linking four of Earth’s five mass extinctions. Volcano.
“Our model worked through the data independently and without human bias to determine the amount of carbon dioxide and sulfur dioxide needed to create the climate and carbon cycle perturbations we see in the geologic record. These amounts are consistent with what we expect in emissions from the Deccan traps,” said Keller, adding that the Deccan volcanic K- who worked extensively to study the relationship between Pg extinction.
Asteroid impact and the modern environment
The model revealed a steep drop in organic carbon accumulation in the deep ocean during the Chicxulub impact, which may have been responsible for the extinction of numerous animal and plant species due to the asteroid. The record also contains traces of a temperature drop at the same time that the mammoth meteorite would have been thrown into the air when it hit a surface rich in sulphur—a short-lived cooling agent—caused by high levels of sulfur. In that part of the planet.
The asteroid impact would have ejected both carbon and sulfur dioxide. However, the model found no spike in the emissions of either gas at that time, suggesting that the asteroid’s contribution to the extinction did not control gas emissions.
Conclusion: Methodological innovation and future applications
In the modern environment, Cox said, the burning of fossil fuels has pumped about 16 billion tons of carbon dioxide into the atmosphere annually from 2000 to 2023. That’s 100 times higher than what scientists had projected for the highest annual emission rate from the Deccan Traps. While alarming on its own, Cox said it will take a few thousand years for current carbon dioxide emissions to match the total amount released from ancient volcanoes.
“What’s really exciting is that the results we’ve achieved are broadly physically plausible, making it interesting that the model can technically run completely wild without strong frontal barriers,” he said.
Interconnecting processors has shortened the time it takes to sample and analyze such large data sets from months or years to hours, Cox said. His and Keller’s method can be used to invert models of other Earth systems, such as climate or the carbon cycle—the outcomes of geologic events are well known, but not the factors that lead to them.
“This kind of parallel inversion has never been done before in Earth science models. Our method can scale to include thousands of processors, which provides a much wider solution space to explore, and is more resistant to human bias,” Cox said.
“So far, people in our field have been more impressed by the novelty of the method than the result we’ve achieved,” he laughs. “Any Earth system is ripe for reverse, even though we know its effect, not its cause. The better we know the output, the better we can characterize the input that caused it.
Note: Alexander A. Cox and C. Brenhin Keller, 28 September 2023, “Bayesian inversion of emissions and export productivity across the Late Cretaceous boundary”. Science.