A groundbreaking new study using advanced artificial intelligence (AI) has revealed that the asteroid strike that wiped out the dinosaurs 66 million years ago caused only a modest decline in shark and ray species. The findings challenge previous understandings of how severely this mass extinction affected life in the oceans.
Led by Swansea University, the research brings together deep‑learning AI models with the most comprehensive global fossil dataset ever assembled for sharks and rays. Using these tools, the team traced shark and ray diversity – the number of species – across 145 million years, creating the most detailed long‑term picture of their evolutionary history to date.
The researchers found that:
- Species numbers comparable to modern levels were already reached during the Cretaceous period, more than 100 million years ago.
- The asteroid impact caused a relatively small decline of around 10 per cent in shark and ray species – a stark contrast to the mass extinction suffered by non-avian dinosaurs and many marine predators.
- The number of shark and ray species peaked in the mid-Eocene period, around 50 million years ago, when oceans supported far more species than exist today.
- Since their peak, sharks and rays have lost more than 40 per cent of their species, leaving today’s oceans far less diverse.
These patterns were invisible to previous methods but emerge clearly with advanced AI methods that explicitly correct for spatial and temporal biases in the fossil record, in addition to a new fossil dataset.
Lead author Dr Catalina Pimiento of Swansea University’s Department of Biosciences and of the University of Zurich’s Department of Paleontology explains: “We built a new, carefully curated dataset of fossil shark and ray occurrences, which involved reviewing hundreds of studies, extracting data, checking it and resolving inconsistencies, resulting in a global synthesis of fossil occurrences spanning the past 145 million years.”
Study co-led author Dr Daniele Silvestro of ETH Zurich said: “We then applied a new AI method that is far better at recognising and correcting the uneven and incomplete nature of the fossil record. Previous approaches could account for general differences in sampling, but not for the fact that fossils are unevenly collected across geographic regions or species. Our model can learn these patterns – for example, recognising when fewer fossils in a particular region can reflect limited sampling rather than real biological decline.”
The findings, published in Current Biology, also place the modern conservation crisis facing sharks and rays into a deeper evolutionary context.
Dr Pimiento added: “Today’s sharks and rays are the survivors of a long history of change, including extinction events that have only recently become visible in the fossil record. For the past 40 to 50 million years, their diversity has been trending downwards.
“That long-term decline matters today, because it suggests that modern sharks and rays are already starting from a reduced baseline. In other words, they’re not just facing human pressures such as overfishing and climate change as a healthy, thriving group – they have already lost a lot of evolutionary potential over tens of millions of years. Understanding their past helps us see how important it is to protect the species we still have today.”
Read ‘Deep learning reveals the hidden patterns of shark and ray diversity over the past 145 million years’ in full.