Oxygen defines complex life on Earth: animals breathe it, plants make it as a byproduct of photosynthesis, and it’s impossible for most of us to imagine our planet without it. But in fact, Earth spent nearly half of its 4.5-billion-year history with almost no atmospheric oxygen. That all changed around 2.4 billion years ago, when oxygen produced by tiny organisms started to accumulate in the atmosphere and seas, kicking off something known as the Great Oxidation Event (GOE). Over a few hundred million years, oxygen levels increased from practically zero to about 0.1-1% of current levels.
For a long time, many scientists assumed that oxygen-breathing organisms could only have evolved after oxygen became widely available. However, new CUHK research now challenges this view. It has long been known that scattered “oxygen oases” existed in the ancient world. CUHK researchers have now found that it was within these microscopic environments that the earliest aerobic bacteria emerged, around 2.7 billion years ago, roughly 200 to 400 million years before the Great Oxidation Event.
The research team, led by Professor Luo Haiwei, Associate Professor of the School of Life Sciences and Director of the Simon F.S. Li Marine Science Laboratory, used AI to analyse 80,000 bacterial genomes from the public Genome Taxonomy Database. Rather than instructing the AI to focus on genes known to mediate oxygen utilisation, the research team allowed the machine learning algorithm to autonomously determine which genomic characteristics most effectively distinguish oxygen-respiring bacteria from their anaerobic relatives. The model ultimately selected 40 genes predominantly linked to core cellular operations, such as energy generation, stress management, and the biosynthesis of essential molecular components, which Professor Luo describes as “fundamental cellular housekeeping processes”.
The model proved remarkably robust despite using only 40 genes. Because these genes are part of an integrated cellular system, the model can still accurately predict an organism’s oxygen phenotype even if some genes are missing from the genomic data. By contrast, traditional methods rely on specific oxygen-related genes. If those genes are missing from the data, scientists cannot tell whether the organism is truly anaerobic or the genes were simply missing due to incomplete genome assembly. This is the fundamental limitation of traditional approaches.
The oxygen oases in which these bacteria lived were themselves created by bacteria, the ancestors of today’s cyanobacteria, which began producing oxygen as a byproduct of photosynthesis around 2.7 to 3.0 billion years ago.
“These aerobic bacteria existed long before oxygen accumulated in the atmosphere”, says Professor Luo. “This shows that life was adapting to use oxygen locally, before Earth’s atmosphere was transformed and oxygen began to accumulate globally. You don’t need detectable oxygen across the entire planet; just small, persistent, and localised pockets are enough to drive a major metabolic breakthrough”.
You might expect that if oxygen developed in small pockets, it would simply disperse evenly within the atmosphere. That didn’t happen, says the professor, because the early Earth was chemically hungry for it. “The oceans were rich in dissolved ferrous iron and other minerals that rapidly reacted with oxygen, consuming it almost as quickly as it was produced. Volcanic gases also acted like a chemical sponge, soaking up free oxygen. This meant oxygen could only persist in protected micro-environments, such as within slimy bacterial mats or at the surface of sediments”. These would have mostly been small and scattered, at a scale from metres down to millimetres, but potentially fairly widespread, he adds.
Adapting to breathe oxygen would have been an evolutionary advantage for the bacteria, allowing them to generate energy more efficiently and giving them a competitive edge in nutrient-limited environments. In addition to the bacteria, the pockets of oxygen would also have hosted oxygen-producing photosynthesisers and other anaerobic microbes, forming simple microbial communities. “They were the pioneers of oxygen-based life”, says Professor Luo. “Their existence shows that life was already preparing for an oxygenated world long before the atmosphere changed. All later aerobic organisms, including animals and plants, inherited this metabolic innovation”.
This has important contemporary implications, he adds: “By showing how life adapted to local oxygen pockets long before oxygen accumulated in the atmosphere, we gain insights into the profound, long-term interplay between organisms and their environment. This perspective helps us understand how life might respond to current climate change, such as rising carbon dioxide levels affecting ocean chemistry and oxygen levels”.
Building on their findings, the team plans to develop a model that also includes partly oxygen-breathing microbes, allowing for a more sophisticated understanding of oxygen adaptation. “Such a tool is highly relevant to modern ecosystem health, as many critical environments are experiencing oxygen stress, such as coastal dead zones with seasonal oxygen loss, or coral reefs where warming and nutrient pollution drive oxygen fluctuations”, he says. “A more precise genomic predictor for these subtle metabolic strategies could help us monitor microbial community responses to such stresses and more accurately anticipate ecosystem resilience in our changing world”.
