Ancient chemical clues reveal Earth’s earliest life 3.3 billion years ago
MSU researcher contributes rare fossils that help train AI to detect life’s oldest molecular signature
Michigan State University
image:
MSU researcher Katie Maloney contributed samples of rare, exceptionally well-preserved seaweed fossils (e.g., macroscopic algae) from Yukon Territory, Canada. These fossils are almost one-billion years old and represent one of the first seaweeds known in the fossil record when most life still needs to viewed through a microscope.
view moreCredit: Katie Maloney
A new study uncovered fresh chemical evidence of life in rocks more than 3.3 billion years old, along with molecular traces showing that oxygen-producing photosynthesis emerged nearly a billion years earlier than previously thought.
The international team, led by researchers at the Carnegie Institution for Science, paired cutting-edge chemistry with artificial intelligence to reveal faint chemical “whispers” of biology locked inside ancient rocks. Using machine learning, the researchers trained computers to recognize subtle molecular fingerprints left behind by living organisms, even when the original biomolecules have long since degraded.
Among the collaborators was Michigan State University’s Katie Maloney, an assistant professor in the Department of Earth and Environmental Sciences, who studies the evolution of early complex life and its impact on ancient ecosystems. Maloney contributed samples of exceptionally well-preserved one-billion-year-old seaweed fossils from Yukon Territory, Canada. These samples represent one of the first seaweeds known in the fossil record, when most life can only be viewed through a microscope.
The study, published in the Proceedings of the National Academy of Sciences, not only deepens understanding of Earth’s earliest biosphere but also has implications for the search for life beyond Earth. The same approach could be used to analyze samples from Mars or other planetary bodies to detect whether they once harbored living organisms.
“Ancient rocks are full of interesting puzzles that tell us the story of life on Earth, but a few of the pieces are always missing,” Maloney said. “Pairing chemical analysis and machine learning has revealed biological clues about ancient life that were previously invisible.”
Earth’s earliest life left behind little in the way of molecular traces. The few fragile remnants such as ancient cells and microbial mats were buried, crushed, heated, and fractured within Earth’s restless crust before being thrust back to the surface. These transformations all but obliterated biosignatures holding vital clues to the origins and early evolution of life.
The new work suggests that the distribution of biomolecular fragments found in old rocks still preserves diagnostic information about the biosphere, even if no original biomolecules remain.
Indeed, this new research shows that life left behind more than anyone ever realized – faint chemical “whispers” locked deep inside ancient rocks.
The team used high-resolution chemical analysis to break down organic and inorganic materials into molecular fragments, then trained an artificial intelligence system to recognize the chemical “fingerprints” left behind by life. Scientists examined more than 400 samples from plants and animals to billion-year-old fossils and meteorites. The AI model distinguished biological from non-biological materials with over 90% accuracy and detected signs of photosynthesis in rocks at least 2.5 billion years old.
Until now, molecular traces that reliably indicated life had only been found in rocks younger than 1.7 billion years. This new method roughly doubles the window of time scientists can study using chemical biosignatures.
“Ancient life leaves more than fossils; it leaves chemical echoes,” said Dr. Robert Hazen, senior staff scientist at Carnegie and a co-lead author. “Using machine learning, we can now reliably interpret these echoes for the first time.”
For Maloney, whose research focuses on how early photosynthetic life transformed the planet, the implications are profound.
“This innovative technique helps us to read the deep time fossil record in a new way,” she said. “This could help guide the search for life on other planets."
Journal
Proceedings of the National Academy of Sciences
Article Title
Organic geochemical evidence for life in Archean rocks identified by pyrolysis–GC–MS and supervised machine learning
Article Publication Date
17-Nov-2025
Chemical evidence of ancient life detected in 3.3 billion-year-old rocks: Carnegie Science / PNAS
New method also detects molecular signs of photosynthesis almost 1 billion years earlier than previously documented; Combining chemistry and AI, pioneering method could revolutionize search for extraterrestrial life
image:
Organic matter extracted from samples of 2.5-billion-year-old rock containing fossilized microorganisms like the one in this photomicrograph still contains biomolecular fragments that may have been produced via photosynthesis. Download at https://bit.ly/4nwph2K
view moreCredit: Andrew D. Czaja
Pairing cutting-edge chemistry with artificial intelligence, a multidisciplinary team of scientists today published fresh chemical evidence of Earth’s earliest life – concealed in 3.3-billion-year-old rocks – and molecular evidence that oxygen-producing photosynthesis was occurring over 800 million years earlier than previously documented.
In a groundbreaking study published in the Proceedings of the National Academy of Sciences, scientists from the Carnegie Institution for Science and several partner universities and institutions analyzed over 400 samples, including ancient sediments, fossils, modern plants and animals, and even meteorites, to see if life’s signature still exists in rocks long after the original biomolecules are gone.
Using high-tech chemical analysis to break down both organic and inorganic materials, Michael L. Wong, Anirudh Prabhu, and colleagues trained AI to recognize chemical ‘fingerprints’ left behind by life – signals that can still be detected even after billions of years of geological wear and tear.
The results prove the possibility of distinguishing materials of biological origin (like microbes, plants and animals) from materials of non-living origin (like meteoritic or synthetic carbon) with over 90% accuracy.
Impressively, these methods teased out chemical patterns unique to biology in rocks as old as 3.3 billion years. Previously, no such traces had been found in rocks older than about 1.7 billion years. The results, therefore, roughly double the window of time in which organic molecules preserved in rocks can reveal useful information about the physiology and evolutionary relationships of their original organisms.
The work also provides molecular evidence that oxygen-producing photosynthesis (the process used by plants, algae and many microorganisms to harness sunlight) was at work at least 2.5 billion years ago. This finding extends the chemical record of photosynthesis preserved in carbon molecules by over 800 million years.
Besides helping find evidence of Earth’s earliest life, this work advances a potential way to identify traces of life beyond our planet.
Life’s evidence in ancient cells battered to near obliteration
Earth’s earliest life left behind little in the way of molecular traces. The few fragile remnants such as ancient cells and microbial mats were buried, crushed, heated, and fractured within Earth’s restless crust before being thrust back to the surface. These transformations all but obliterated biosignatures holding vital clues to the origins and early evolution of life.
Paleobiologists who search for signs of Earth’s most ancient life have long relied mainly on fossil organisms, including microscopic fossils of single cells and filaments, and the mineralized remains of cellular structures such as microbial mats and mound-like stromatolites, which provide convincing evidence of life as far back as 3.5 billion years ago. However, such remains are few and far between.
A second line of evidence relies on the preservation of diagnostic biomolecules in ancient rocks. Life’s hardiest organic molecules – those derived from cell membranes or some metabolic processes – have been found in sediments as old as 1.7 billion years, while much older carbon-rich rocks preserve isotopic signatures that hint at a vibrant biosphere 3.5 billion years ago.
However, most ancient rocks preserve neither fossil cells nor any surviving biomolecules. The vast majority of ancient carbon-bearing sediments have been heated and altered in ways that break every diagnostic biomolecule into countless small fragments. Those fragments have proven too small and too generic to provide any clues about ancient life – until now.
The new work is based on the hypothesis that life’s molecules are rigorously selected for their biological functions (in keeping with a new law of nature proposed in 2023). Unlike the random distribution of molecules found in carbon-rich meteorites and other abiotic organic mixtures, life makes a few kinds of molecules in high abundance. Each chemical in a living cell has its own function. The new work suggests that the distribution of biomolecular fragments found in old rocks still preserves diagnostic information about the biosphere, even if no original biomolecules remain.
Indeed, this new research shows that life left behind more than anyone ever realized — faint chemical "whispers" locked deep inside ancient rocks.
The 406 measured samples came from seven major groups:
Modern animals: vertebrates (e.g. fish) and invertebrates (e.g. insects).
Modern plants: including both their photosynthetic parts (e.g. leaves) and non-photosynthetic parts (e.g. roots and sap).
Fungi: including mushrooms and yeast.
Fossil materials: e.g. coal, ancient wood, and shale rich in preserved algae.
Meteorites: carbon-rich space rocks that could resemble prebiotic material.
Synthetic organic materials: made in labs to simulate early-Earth chemistry.
Ancient sediments: ranging from hundreds of millions to over 3 billion years old, with uncertain origins.
The team used pyrolysis–gas chromatography–mass spectrometry (Py-GC-MS) to release trapped chemical fragments from each sample. They then used a specific type of machine learning model called “random forest,” which builds hundreds of decision trees to classify data and to extract latent ecological and taxonomic patterns. This is the first study to combine Py-GC-MS data with supervised machine learning to identify biosignatures in multi-billion-year old rocks.
Says team member Dr. Robert Hazen, Senior Staff Scientist at the Carnegie Institution for Science: “Think of it like showing thousands of jigsaw puzzle pieces to a computer and asking whether the original scene was a flower or a meteorite.”
“Rather than focus on individual molecules, we looked for chemical patterns, and those patterns could be true elsewhere in the universe,” Dr. Hazen added.
“Our results show that ancient life leaves behind more than fossils; it leaves chemical ‘echoes.’ Using machine learning, we can now reliably interpret these echoes for the first time.”
The paper concludes: “Information-rich attributes of ancient organic matter, even though highly degraded and with few if any surviving biomolecules, have much to reveal about the nature and evolution of life.”
A pioneering model
The model’s performance was tested in three main ways:
1. Modern living animals and plants vs non-life samples
Could the model distinguish life-based organic matter from non-living origins (like meteorites or synthetic chemistry)?
Yes, with up to 98% accuracy on known samples.
When applied to ancient rock samples, the model found strong evidence for life in multiple 3.3-billion-year-old formations.
2. Photosynthetic vs Non-photosynthetic
Could the model detect signs that an organism once used sunlight for energy?
Yes, with 93% accuracy.
The method identified photosynthetic signatures in rocks as old as 2.52 billion years.
3. Plant vs Animal
Could it distinguish plant-based life from animal-based life?
Yes again, with 95% correct classification in modern samples.
This type of classification is harder in ancient rocks due to the scarcity of animal fossils in the model’s training set. This is a point of improvement for future work.
Seeing through the fog of time
One key insight was that age makes detection harder. Younger samples from the last 500 million years retained strong biotic signals. For rocks 500 million to 2.5 billion years old, about two-thirds still showed life signatures. But in rocks older than 2.5 billion years, just 47% retained detectable evidence of life.
For each sample, the model didn’t just report “life” or “non-life,” it gave a probability score. If a sample scored above 60% for “biotic,” it was considered a strong hit.
This probability-based approach allows for nuance. For example, a coal sample that had been heated to over 400°C might have lost most of its biological markers and landed in the “uncertain” range. But well-preserved ancient samples—especially those that hadn’t been exposed to intense heat or pressure—still scored confidently in the “biotic” zone.
The authors were also careful not to claim a sample was biotic unless it truly stood apart from abiotic materials, reducing the risk of false positives.
Among the ancient samples that stood out as clear positives:
Biotic material in 3.33-billion-year-old sediments from e.g. South Africa’s Josefsdal Chert
Photosynthetic life in 2.52-billion-year-old rocks from e.g. South Africa’s Gamohaan Formation
Why this matters for science, and space exploration
The results suggest that machine learning applied to degraded organic matter can help resolve long-standing debates about the evolution of life on Earth in deep time.
This method could also assist in the search for signs of extraterrestrial life. If AI can detect biotic “fingerprints” on Earth that survived billions of years, the same technique might work on Martian rocks or even samples from Jupiter’s icy moon Europa.
The authors are careful not to overstate their conclusions. They acknowledge:
The need for larger, more balanced sample sets, especially more fossil animals and diverse abiotic materials
Some samples still fall into a gray zone, with mid-range probability scores that don’t allow firm conclusions.
The method is complementary, not a replacement, for traditional techniques like isotope analysis or fossil morphology.
The team plans to refine their models, explore different types of machine learning, and test their approach on rocks from Earth’s Mars-like deserts.
“This study represents a major leap forward in our ability to decode Earth’s oldest biological signatures,” says Dr. Hazen. “By pairing powerful chemical analysis with machine learning, we have a way to read molecular ‘ghosts’ left behind by early life that still whisper their secrets after billions of years. Earth’s oldest rocks have stories to tell and we’re just beginning to hear them.”
Adds Dr. Wong: “Understanding when photosynthesis emerged helps explain how Earth’s atmosphere became oxygen-rich, a key milestone that allowed complex life, including humans, to evolve.”
“This represents an inspiring example of how modern technology can shine a light on the planet’s most ancient stories and could reshape how we search for ancient life on Earth and other worlds. In future, we plan to test materials like anoxygenic photosynthetic bacteria — possible analogs for extraterrestrial organisms. This is a powerful new tool for astrobiology.”
Says co-first author Dr. Anirudh Prabhu of Carnegie Science: “These samples and the spectral signatures they produce have been studied for decades, but AI offers a powerful new lens that allows us to extract critical information and better understand their nature. Even when degradation makes it difficult to spot signs of life, our machine learning models can still detect the subtle traces left behind by ancient biological processes.”
“What’s exciting is that this approach doesn’t rely on finding recognizable fossils or intact biomolecules. AI didn’t just help us analyze data faster, it allowed us to make sense of messy, degraded chemical data. It opens the door to exploring ancient and alien environments with a fresh lens, guided by patterns we might not even know to look for ourselves.”
********
Further comments
“For decades, we’ve searched ancient rocks for traces of life using a limited set of tools. What’s remarkable about this study is that it adds whole new dimensions – not just better instruments, but better questions. Machine learning helps us uncover biological signals that were effectively invisible before. It’s a leap forward in our ability to read the deep-time record of life on Earth.”
Co-author and paleobiologist Andrew H. Knoll, Harvard University
“For decades, organic geochemists have been examining the rock record looking for the diagnostic molecules that could tell us something about the nature of life at that time. These new techniques allow the data to speak for themselves in new ways, and for scientists to find new patterns faster than ever before.”
Co-author H. James Cleaves II, Howard University, Washington DC
* * * * *
Fact box
Technique used: Pyrolysis Gas Chromatography-Mass Spectrometry (Py-GC-MS)
Samples analyzed: Over 400 (modern, fossil, meteorite, and synthetic)
Machine learning success rates:
98% accuracy distinguishing modern life from non-life
95% accuracy distinguishing plants from animals
93% accuracy distinguishing photosynthetic organisms
Oldest signs detected:
Life: 3.33 billion-year-old rocks (Josefsdal Chert, South Africa)
Photosynthesis: 2.5 billion-year-old rocks (Gamohaan Formation, South Africa)
Potential future applications:
Searching for life on Mars, Europa, or other worlds
Improving understanding of early Earth ecosystems
********
The full dataset and code are publicly available through the Open Science Framework and github, inviting further research and exploration into ancient biosignatures. Open data repository: 10.17605/OSF.IO/G93CS; Github: https://github.com/PrabhuLab/PyGCMS-Biosign-ML
* * * * *
Organic geochemical evidence for life in Archean rocks identified by pyrolysis-GC-MS and supervised machine learning,” by Michael L. Wong, Anirudh Prabhu, et al
Published by The Proceedings of the National Academy of Sciences, Washington DC, USA
Authors:
Michael L. Wong 1,2*, Anirudh Prabhu1*, Conel O’D. Alexander1, H. James Cleaves II 1,3, George D. Cody1, Grethe Hystad4, Marko Bermanec5, Wouter Bleeker6, C. Kevin Boyce7, Andrea Corpolongo8, Andrew Czaja8, Souvik Das9, Robert R. Gaines10, Dan Gregory11, John Jaszczak12, Emmanuelle Javaux13, Jaganmoy Jodder14, Andrew H. Knoll15, Martin Van Kranendonk16, Katie M. Maloney17, Nora Noffke18, Robert Rainbird19, Emersyn Slaughter20, Roger Summons21, Frances Westall22, Jasmina Wiemann23, Shuhai Xiao24, and Robert M. Hazen1**
1 Earth and Planets Laboratory, Carnegie Institution for Science, Washington DC 20015, USA
2 NHFP Sagan Fellow, NASA Hubble Fellowship Program, Space Telescope Science Institute, Baltimore MD 21218, USA
3 Department of Chemistry, Howard University, Washington DC USA
4 Mathematics and Statistics, Purdue University Northwest, Hammond IN 46323, USA
5 Department of Earth Sciences, University of Graz, 8010 Graz, Universitätsplatz 2/II, Austria
6 Natural Resources Canada, 601 Booth Street, Ottawa, Ontario K1A 0G1, Canada
7 Department of Geological Sciences, Stanford University, Stanford, CA 94305, USA
8 Department of Geology, University of Cincinnati, Cincinnati, OH 45221, USA
9 State Key Laboratory of Critical Earth Material Cycling and Mineral Deposits, Nanjing University, Nanjing 210023, China
10 Office of the President, Pomona College, Claremont, CA 91711, USA
11 Department of Earth Sciences, University of Toronto, Toronto, Ontario M5S 3B1, Canada
12 A. E. Seaman Mineral Museum, Michigan Tech, Houghton, MI 49931, USA
13 Early Life Traces & Evolution-Astrobiology, University of Liège, 4000 Liège, Belgium
14 Department of Geosciences, University of Oslo, 0316 Oslo, Norway
15 Department of Earth and Planetary Sciences, Harvard University, Cambridge, MA 02138, USA
16 School of Biological, Earth & Environmental Sciences, University of New South Wales, Sydney, NSW 2052, Australia
17 Department of Earth and Environmental Sciences, Michigan State University, East Lansing, MI 48824, USA
18 Department of Ocean and Earth Sciences, Old Dominion University, Norfolk, VA 23529, USA
19 Geological Survey of Canada, Ottawa, Ontario K1S 5B6, Canada
20 Department of Geological Sciences, University of Florida, Gainesville, FL 32611, USA
21 Department of Earth, Atmospheric and Planetary Sciences, MIT, Cambridge, MA 02139, USA
22 Centre de Biophysique Moléculaire, CNRS-UPR4301, Orléans, France
23 Department of Earth & Planetary Sciences, Johns Hopkins University, Baltimore, MD 21218, USA
24 Department of Geology, Virginia Tech, Blacksburg, VA 24060, USA
* Co-First Authors
* * * * *
About Carnegie Science: https://carnegiescience.edu/about
* * * * *
Journal
Proceedings of the National Academy of Sciences
Method of Research
Data/statistical analysis
Subject of Research
Cells
Article Title
Organic geochemical evidence for life in Archean rocks identified by pyrolysis-GC-MS and supervised machine learning
Article Publication Date
17-Nov-2025
The black features within this thin slice of rock are 2.5-billion-year-old microbial structures. This study suggests that the organic matter preserved within this complex microbial community may have been produced by photosynthetic microorganisms. Download at https://bit.ly/4nwph2K
Credit
Andrea Corpolongo
Ancient sediments reveal Earth’s hidden wildfire past
Heriot-Watt University
image:
Dr Clayton Magill
view moreCredit: Heriot-Watt University
An international team of scientists, including a senior researcher at Heriot-Watt University in Edinburgh, Scotland, has uncovered new evidence of ancient wildfires that reshapes our understanding of Earth’s turbulent Early Triassic epoch, about 250 million years ago.
The findings, reported in Communications Earth & Environment, published by Nature Portfolio under the title Wildfire, ecosystem and climate interactions in the Early Triassic, challenge the long-standing belief in a global “charcoal gap”, a time interval with little or no evidence of fire following the world’s greatest mass extinction.
Traces in the dirt
For decades, the absence of charcoal in the geologic record led scientists to assume that wildfires had all but disappeared after the Permian–Triassic extinction, also known as the "Great Dying”. This was the most severe mass extinction in Earth's history, resulting in the loss of up to 96% of marine species and 70% of terrestrial vertebrate species, primarily caused by massive volcanic eruptions.
This latest study sheds new light on this period, revealing microscopic chemical traces of charred vegetation preserved in sediments.
The team tested 30 sediment samples retrieved from Svalbard, the Norwegian Arctic archipelago better known today as home to the Global Seed Vault. Despite the harsh conditions, the island’s ancient rocks offered pristine material that had remained undisturbed for hundreds of millions of years.
Fire without charcoal
Instead of relying on visible pieces of charcoal, the team searched for molecular fingerprints of combustion known as polyaromatic hydrocarbons (PAHs). These compounds form during the incomplete burning of plant matter and can persist in sediments long after more visible evidence disappears.
Dr Clayton Magill is Associate Professor of Biogeochemistry at the Lyell Centre at Heriot-Watt University and a senior author of the study.
“A lot of folks have not found the normal evidence of fire such as charcoal, ash, burnt fossils so the consensus was that fire wasn’t happening,” he said.
“What our colleague Dr Franziska Blattmann’s work showed is that even without the big pieces of evidence, the microscopic signals are still there. You just need to know where to look.”
The analysis revealed widespread PAHs consistent with burning fresh plant matter rather than volcanic coal deposits or contamination. This strongly suggests that wildfires were, in fact, shaping ecosystems during the Early Triassic, even when the fossil charcoal record seemed to say otherwise.
Modelling fire in deep time
The project, funded by the Swiss National Science Foundation, combined sediment analysis with cutting-edge climate and vegetation modelling. Using an open-source model by Massachusetts Institute of Technology (MIT) named the General Circulation Model (MITgcm), the team successfully reconstructed how shifting climates, ecosystems, and fire regimes interacted in the aftermath of the mass extinction.
“It’s very easy to say, ‘If A occurs, then B will happen,’ but that can be ambiguous,” Dr Magill said. “By using models, we can run our data through theory and test whether it holds up. It doesn’t just say, ‘trust me’ - it shows you the evidence.”
The use of open-source models was especially important, Dr Magill added: “That’s a powerful tool in a world where not everyone has equal access to scientific resources and funding. Open science allows everyone to compete at the highest level.”
The 10-strong team of sedimentologists, palynologists, palaeontologists, physicists and geochemists was led by Dr Franziska Blattmann at the Faculty of Geoscience and Environment at the University of Lausanne in Switzerland. She and her colleagues had worked on the groundbreaking research since 2018 and said: "This study came together through the collaboration of a multidisciplinary team of scientists, working together even amid the challenges of the COVID-19 pandemic. The research highlights how longstanding scientific questions can be advanced and how unexpected discoveries can emerge when collaboration is open, creative and supportive."
Beyond filling in a 250-million-year-old puzzle, the research carries urgent lessons for the present. The Early Triassic was a time of extreme climate swings, ecosystem recovery, and environmental stress, all themes with echoes in today’s warming world.
Journal
Communications Earth & Environment
Article Title
Wildfire, ecosystem, and climate interactions in the Early Triassic
A total of 30 sediment samples were retrieved from Svalbard, the Norwegian Arctic archipelago.
Credit
Dr Franziska Blattmann
No comments:
Post a Comment