Artificial intelligence closes the gaps in the fossil archive

- EN - DE- FR- IT
 (Image: Pixabay CC0)
(Image: Pixabay CC0)

The patchy fossil record makes it difficult for paleontologists to draw an accurate picture of the extent of past biodiversity and to understand how it has changed over time. A study led by Rebecca Cooper and Daniele Silvestro from the University of Fribourg shows how artificial intelligence (AI) can make this task easier.

Scientists estimate that there are currently more than eight million animal and plant species. In order to understand this incredible diversity, it is essential to trace the processes that have influenced global biodiversity over time - including mass extinctions - as accurately as possible. To this end, paleontologists make use of the fossil record, which unfortunately only represents a small, often patchy sample of the species that have ever lived on our planet. In fact, only a tiny fraction of plants and animals fossilize, between 0.01 and 0.1 % of organisms.

AI as a savior in times of need

PhD student Rebecca Cooper, member of the Swiss Institute of Bioinformatics (SIB) and lead author of the study, and Assistant Professor Daniele Silvestro, head of the Computational Evolutionary Palaeobiology group at the SIB, in collaboration with Joseph Flannery-Sutherland from the University of Birmingham, have shown that despite scattered fossil data, it is possible to assess this diversity over time - thanks to the power of artificial intelligence. The study, published in Nature Communications on May 17, 2024, introduces a new software called DeepDive, which is able to reconstruct the evolution of species richness over time.

’At the end of the Permian, more than 251 million years ago, massive volcanic activity led to the worst known mass extinction,’ explains Rebecca Cooper. ’Using DeepDive, we were able to detect severe losses in marine animal diversity and discovered that it took millions of years for diversity to recover.

Revolution for paleontologists

The new program generates hundreds of thousands of synthetic data sets that replicate the fossil record. From these, an AI model learns how the number and location of fossils can inform us about the true extent of hidden biodiversity. ’For over 50 years, paleontologists have struggled with traditional statistical techniques to overcome the uncertainties and limitations of the fossil record,’ says Joseph Flannery-Sutherland. ’Now artificial intelligence provides a powerful tool to solve many of these problems and allows us to understand how biodiversity has changed over geological time. DeepDive opens up really exciting possibilities!

Initial results speak for themselves

The authors also used their software to analyze the fossil record of elephants and their extinct relatives such as mammoths and mastodons. The result: with the help of DeepDive, they were able to discover that more than 35 species existed until recently, before many of them rapidly became extinct. ’Our study shows that this biodiversity, which evolves over millions of years, can be wiped out very suddenly, as hundreds of extinction events in recent centuries have shown,’ says Daniele Silvestro, ’which forces us to appreciate the irreplaceable nature of today’s biodiversity.

At a time when AI is permeating most new technologies - and increasingly our lives - the risks of misuse and the dangers of these ever more powerful models are constantly being debated. However, the study shows that many scientific advances can result from AI. ’It is undeniable that this approach has helped us to better understand the nature and mysteries of the evolution of life on our planet,’ concludes Rebecca Cooper.

Original publication and software

DeepDive: estimating global biodiversity patterns through time using deep learning , Nature communications, 17.05.2024

The DeepDive program is free of charge and can be viewed here:­ve-project


About the SIB

The (Swiss Institute of Bioinformatics) is an internationally recognized non-profit organization dedicated to the science of biological and biomedical data. Its scientists are passionate about creating knowledge and solving complex questions in many fields, from biodiversity to evolution and medicine. They provide key databases and software platforms as well as bioinformatics expertise and services to academic, clinical and industrial groups. The SIB unites the Swiss bioinformatics community, which consists of around 900 scientists, and promotes collaboration and knowledge exchange. The institute helps to keep Switzerland at the forefront of innovation by promoting advances in biological research and improving health.