A study published in May in Nature Astronomy proposes a way to separate biological chemistry from non-biological chemistry that does not depend on finding any particular molecule. The signal is not which compounds are present. It is how evenly they are spread across the sample.
The paper, titled “Molecular diversity as a biosignature,” has Gideon Yoffe, a postdoctoral researcher at the Weizmann Institute of Science in Israel, as first author, with Fabian Klenner, an assistant professor of planetary sciences at the University of California, Riverside, among the co-authors. The work applies a measure familiar to ecologists, not chemists, and tests it against roughly 100 existing datasets rather than any new sample return.
This is one study, not settled consensus. It is worth taking seriously, and the authors themselves are careful about what it does and does not establish.
What the method actually measures
Ecologists describe biodiversity using two properties. Richness is how many distinct species are present. Evenness is how uniformly the population is split between them. A forest with one dominant tree and a scattering of others is rich but uneven. A forest where a dozen species each hold a similar share is both rich and even.
Yoffe and colleagues applied that same accounting to molecules. They drew on around 100 published datasets covering amino acids and fatty acids from microbes, soils, fossils, meteorites, asteroids, and synthetic laboratory mixtures, then asked whether the diversity statistics alone could sort the biological samples from the abiotic ones.
According to the paper, they could. Amino acids in samples produced by living things were consistently more diverse and more evenly distributed than amino acids formed by non-biological processes. The separation held across a wide range of sample types, which is the part the researchers describe as surprising given how simple the underlying measure is.
The amino acid result is only half the finding
It would be easy to read this as a single clean rule: life spreads its molecules evenly, non-life does not. That is not what the paper shows.
For fatty acids, the pattern runs the other way. In the datasets analysed, abiotically produced fatty acids were distributed more evenly than those made by biological processes. The diagnostic direction depends on which class of molecule you are looking at. The method works by reading the contrast in the right direction for the right compound, not by assuming that evenness always means life.
That detail matters for anyone tempted to compress the result into a slogan. The useful claim is narrower and more interesting: the way a biological process apportions its molecules leaves a statistical signature, and the signature differs between amino acids and fatty acids in a way that itself carries information.
Why the use of existing data is the real draw
The appeal of the approach is that it does not call for a bespoke instrument. The diversity measure runs on tables of abundances, which is the kind of output many mass spectrometers and chromatography systems already produce. In the team’s framing, it may be possible to look for this pattern in measurements that current and planned missions are gathering anyway.
That is a possibility raised by the authors, not a demonstrated capability. The paper analyses laboratory and field datasets, not live mission data from Mars, Europa, or Enceladus. The claim is that the statistical signal survives in the kind of data those missions return, which is a reason to test the method against real mission measurements, not a result already in hand.
The framework also picked up something the researchers had not gone looking for. Biological samples formed a continuum from well preserved to heavily degraded, and even badly altered material kept a trace of the organisation. Fossilised dinosaur eggshell in the dataset still carried a detectable statistical signature shaped by ancient life. The method appears to register not only the difference between life and non-life, but degrees of preservation.
What the study does not establish
Two limits are worth stating plainly. The datasets are drawn from terrestrial biology and from meteoritic and laboratory chemistry, so the method has been trained and tested on the one example of life we have. Whether an independently originated biology would distribute its molecules the same way is an open question the paper cannot answer.
The idea of reading distributions rather than hunting for marker molecules is also not new. Agnostic biosignature work, including a machine-learning approach published by Cleaves and colleagues in the Proceedings of the National Academy of Sciences in 2023, has pursued the same instinct from a different direction. What this paper adds is a simple, interpretable diversity statistic borrowed from ecology, and evidence that it separates biotic from abiotic samples across a broad spread of existing data.
The authors are explicit that no single test settles the question. Klenner notes that any future claim of detecting life would need multiple independent lines of evidence read within the geological and chemical context of the world in question. Yoffe describes astrobiology as a forensic science, an exercise in inferring process from incomplete clues. The diversity measure is offered as one more line of evidence, valuable mainly when it agrees with others.
What to watch
The next test is whether the signal holds when the method meets real mission data rather than curated datasets, where contamination, instrument limits, and sparse sampling all complicate the picture. If the diversity statistics can be extracted reliably from measurements already returning from Mars and the icy moons, the method becomes a low-cost addition to the existing toolkit. If they cannot, it remains a clean result about Earth’s chemistry waiting for a harder trial. UC Riverside’s account of the work is here.
