Scientists find the seeds that could survive for decades — and one crop clearly outlasts the others
Some seeds stored for future food security can remain alive for decades in the freezer. Others deteriorate far sooner than scientists once expected. A major Australian study has now put real-world numbers on that difference, identifying mung bean (Vigna radiata) as the most durable of six tropical food crops tested under long-term genebank conditions.
The headline-friendly version is irresistible: scientists have found an “apocalypse seed” capable of surviving for generations in suspended animation. The scientific reality is less theatrical and considerably more important. The research shows that the genetic reserves on which agriculture may one day depend cannot simply be frozen, forgotten and assumed to remain viable for centuries.
Researchers at the Australian Grains Genebank in Horsham, Victoria, analysed four decades of germination records from tropical grain crops stored at −20°C, the temperature commonly used for long-term conservation of so-called orthodox seeds. The team examined sorghum, common bean, soybean, mung bean, pigeon pea and adzuki bean, combining historical records with thousands of fresh germination tests carried out in 2024.
The scale of the archive is substantial. The Australian Grains Genebank holds around 217,000 accessions representing roughly 1,250 species, making it one of the world’s major repositories of crop diversity. Its tropical collection alone contains more than 31,000 active accessions. In 2024, researchers conducted another 3,861 germination tests, adding them to a dataset containing more than 24,000 viability observations accumulated over decades.
Mung bean was the clear survivor
Among the six species, mung bean showed the slowest decline in viability. Using the standard Ellis-Roberts seed-survival model, researchers calculated a longevity parameter of 63.5 years — technically, the time required for viability to decline by one probit unit under the storage conditions used.
That distinction matters. It does not mean that every mung bean seed will suddenly die after 63.5 years. Seed survival follows a curve, and different seed lots start with different quality. What the figure provides is a comparative measure of how slowly viability deteriorates.
The next strongest performers were sorghum and soybean, with longevity estimates close to 49 years, while common bean was around 33 years and pigeon pea approximately 31 years. At the other end of the scale sat adzuki bean (Vigna angularis), whose estimated longevity was just 17.4 years.
The contrast is considerable. Two species that look perfectly manageable when first placed into cold storage can require entirely different conservation schedules decades later.
Even more striking were the direct germination results. In the 2024 tests, both mung bean and sorghum still produced seed lots with at least 90% viability after more than 44 years in storage. Common bean retained similarly high viability in some lots after almost 38 years, while soybean remained above 90% after more than 34 years.
These are impressive figures, but they come with an inconvenient caveat: performance varied markedly between individual accessions of the same species. A crop cannot therefore be assigned one magic expiry date and left untouched until that date arrives.
Why the old assumption of centuries-long security is dangerous
Seed conservation has long benefited from mathematical models predicting that properly dried and frozen seeds can remain viable for very long periods — in some cases theoretical estimates extend into centuries or even millennia.
Those calculations are useful, but they are not substitutes for actual seeds ageing in actual freezers.
The Australian data show that some accessions fall below important viability thresholds within 20 to 40 years, even under carefully controlled long-term conditions. That is a rather uncomfortable result for anyone tempted to imagine a seed vault as a biological time capsule that can simply be locked and ignored.
Seed banks do not merely store packets. They manage living genetic material that is slowly deteriorating.
Under international FAO Genebank Standards, cultivated crop accessions generally enter storage with high germination capacity, and managers are advised to monitor viability before it falls too far. A widely used management threshold is 85% of initial viability. Once an accession approaches unacceptable levels, the seeds must be regenerated: planted, grown, harvested and returned to storage as a fresh generation.
That process sounds straightforward. It is not.
Every regeneration cycle costs money, labour and greenhouse or field space. More importantly, each regeneration introduces an opportunity for genetic drift, accidental selection, contamination or loss of rare genetic variants. Testing too often also consumes the very seeds the bank is attempting to conserve.
The art of genebank management is therefore slightly perverse: test often enough to avoid losing the material, but not so often that the act of protecting it helps destroy it.
Why this matters far beyond one Australian freezer
Crop genebanks are effectively insurance policies for agriculture. They preserve old cultivars, landraces, breeding material and wild relatives containing genes that may become valuable decades later.
Those traits can include resistance to drought, heat, salinity, fungal disease, insects and new pathogens. As climate change alters growing conditions and narrows the safety margins of major crops, this genetic diversity is becoming less of an academic archive and more of an agricultural strategic reserve.
The global conservation system is vast. Around the world, seed collections collectively hold millions of accessions. The Svalbard Global Seed Vault alone now safeguards more than 1.4 million duplicate seed samples from over 6,500 species, stored at approximately −18°C as a security backup for collections held elsewhere.
But Svalbard does not magically maintain those seeds forever. The institutions depositing the material remain responsible for monitoring viability and regenerating their original collections. The Arctic vault is a backup copy, not a biological repair shop.
That makes accurate longevity estimates exceptionally important. If a genebank assumes a collection will remain healthy for 100 years when significant deterioration actually begins after 25 or 30, irreplaceable genetic diversity can quietly disappear before anyone notices.
Mung bean is more important than its modest reputation suggests
Mung bean may lack the political weight of wheat, rice or maize, but it is hardly an agricultural curiosity. Native to South Asia and domesticated thousands of years ago, Vigna radiata is now grown widely across Asia and increasingly elsewhere as a short-duration pulse crop.
Its seeds are rich in protein, while the plant can contribute nitrogen to agricultural soils through biological fixation. It is also relatively suited to warm climates and short rotations, qualities that make it increasingly interesting in discussions around climate-resilient farming.
Its extraordinary performance in cold storage therefore has practical value. Conserving mung bean diversity efficiently could help breeders retain access to traits needed for future heat tolerance, disease resistance and crop adaptation.
Still, describing mung bean as an indestructible “doomsday crop” would be scientifically sloppy. The study evaluated seeds under controlled genebank conditions: dried to low moisture levels, sealed in specialised packaging and maintained around −20°C. A packet of mung beans left in a kitchen cupboard obviously does not acquire the same superpowers.
The uncomfortable lesson from the seed vault
The most important finding is not that one humble legume can survive frozen for decades. It is that seed longevity is far less uniform and predictable than conservation models sometimes assume.
Mung bean performed exceptionally well. Adzuki bean did not. Even within the same species, different seed lots deteriorated at different rates.
That variability forces genebanks to abandon lazy, one-size-fits-all monitoring schedules and move toward evidence-based management tailored to individual crops — and, where possible, individual accessions.
The distinction matters because these collections are not museums. They are repositories of biological options.
If future crops need genes for surviving hotter summers, unfamiliar diseases or radically different rainfall regimes, those genes may already be sitting inside a foil packet in a freezer somewhere.
Keeping that packet frozen is the easy part. Making sure what is inside can still grow when humanity finally needs it is the real job.
