Using an innovative combination of advanced chemistry and artificial intelligence, scientists have successfully detected molecular signatures of life in rocks over 3.3 billion years old, pushing back the timeframe for detecting ancient life by roughly one billion years.
The findings, published in the Proceedings of the National Academy of Sciences, demonstrate that even when individual biomolecules have been destroyed by billions of years of geological processes, distinctive patterns can still reveal whether organic matter originated from living organisms.
Led by researchers at the Carnegie Institution for Science, the study analyzed 406 diverse samples ranging from modern organisms and ancient fossils to meteorites and synthetic compounds using pyrolysis gas chromatography-mass spectrometry combined with machine learning. Unlike previous research in this area, the team decided to look beyond preserved biomolecules to the patterns of fragmentation and degradation that organic molecules undergo over time.
The key insight is that biological molecules are functionally selected by life, meaning that even highly degraded molecular assemblages retain patterns distinct from those of non-biological processes. The research team divided their samples into nine categories, including modern animals and plants, fossil microbes, coal and wood, meteorites, and synthetic organic mixtures, then used these to train machine learning models.
According to the study, the machine learning models achieved remarkable accuracy rates in distinguishing different types of organic matter. The system correctly classified biogenic versus non-biological fossil samples at 93 percent, and photosynthetic versus non-photosynthetic samples with the same accuracy. When comparing modern organisms to meteoritic materials, the model achieved 100% accuracy.
“Ancient life leaves more than fossils; it leaves chemical echoes,” said co-author Dr. Robert Hazen in a press statement. “Using machine learning, we can now reliably interpret these echoes for the first time.”
Using conservative criteria requiring high confidence scores from multiple models, the researchers confidently identified biological origins in 11 ancient rock samples. Most significantly, they detected molecular evidence for life in 3.33-billion-year-old sedimentary rocks from the Josefsdal Chert in South Africa’s Barberton Greenstone Belt.
The study also found evidence for photosynthesis, the process that eventually transformed Earth’s atmosphere by producing oxygen, in rocks as old as 2.52 billion years from the Gamohaan Formation in South Africa.
The detection of photosynthetic signatures in these ancient rocks is particularly noteworthy because it provides molecular confirmation of what paleontologists have long suspected based on fossil structures. The Gamohaan Formation contains unique three-dimensionally preserved carbonate structures called microbialites that strongly suggest very ancient photosynthetic microbial communities, and the new molecular evidence supports this interpretation.
In simple terms, life on Earth is a lot older than we previously thought. The study also identified photosynthetic signatures in the 2.30-billion-year-old Gowganda Formation of Ontario, Canada.
The research suggests that life on Earth, particularly life capable of photosynthesis, began roughly a billion years earlier than previously assumed. The evidence of molecules that reliably indicated life had only been found in rocks that were around 1.6 billion years old. This new method doubles the time window we can use.
Moreover, this research also has implications for astrobiology. The ability to discriminate between biological and meteoritic organic matter is particularly relevant for analyzing samples from Mars or icy moons like Europa and Enceladus.
“This innovative technique helps us to read the deep time fossil record in a new way,” explained co-author Katie Maloney. “This could help guide the search for life on other planets.”
MJ Banias covers space, security, and technology with The Debrief. You can email him at mj@thedebrief.org or follow him on Twitter @mjbanias.