Human civilisation needs a global genetic archive, not just species lists: the practical target is to preserve viable cells, tissues, gametes, embryos, seeds, and genome-level reference data for as many taxa as possible, because restoration after collapse depends on both genetic information and living reproductive material.
1) How many species are at risk
The latest IUCN Red List update reports 175,909 assessed species, of which 49,505 are threatened with extinction[1]. That is roughly 28% of all assessed species[1]. A separate IUCN statement in 2026 also highlighted that 20% of assessed soil species are at risk, equal to at least 1,758 species in that group alone[1].
The correct strategic conclusion is not “a few endangered species”; it is that the planet is already managing a large-scale archive failure across entire branches of the tree of life.
2) Genome banking versus seed banking
A seed bank stores plant seeds under controlled dry, cold conditions so the plant can be regrown later. Its strength is cheap long-term preservation of many plant species; its weakness is that it covers only seed-producing plants, and seeds are not usable for many organisms, especially animals, fungi, and many plants with recalcitrant seeds.
A genome bank stores genetic material from any species: DNA, tissue, cells, gametes, embryos, and sometimes full cryopreserved reproductive lines. The Frozen Ark explicitly frames its mission this way: preserving DNA, tissue, and viable cells from endangered animals so material is not lost and can support future conservation and breeding decisions[2][3].
The operational difference is decisive:
- Seed bank = preserves plant reproduction units
- Genome bank = preserves biodiversity information and, where possible, living cellular potential across the whole biosphere
For civilisation continuity, a seed bank is necessary but insufficient.
3) DNA storage media: longevity and limits
### Silicon chips
Silicon-based DNA storage is attractive because silicon is mature, scalable, and can last for decades in controlled conditions. However, the limiting factor is usually the DNA synthesis/reading chemistry and packaging, not the chip substrate itself. In survival planning terms, silicon is best treated as a data-carrier layer, not as the primary preservation medium for biological restoration.
### Synthetic DNA storage
Synthetic DNA is the most information-dense archival medium known. It can theoretically store enormous amounts of data in tiny volumes, but long-term survival depends on dry, cold, dark storage and protection from hydrolysis, oxidation, and radiation. Current practical deployments are still constrained by cost, error rates, and retrieval complexity. The advantage is density; the weakness is that reading and writing are not yet civilization-cheap.
### Crystal storage
Crystal storage is the best answer for extreme longevity. A landmark demonstration stored human genome data in 5D optical glass using femtosecond lasers; the material is designed to survive very long timescales under harsh conditions, far beyond ordinary digital media. For a lunar archive, crystalline media are the strongest candidate for ultra-long-term inert storage because they tolerate heat, time, and radiation better than biological media, though they still store information, not living cells.
### Bottom line
- Best for living recovery: cryopreserved cells, gametes, embryos, tissue
- Best for dense information archive: synthetic DNA and crystal media
- Best for operational data systems: silicon chips
- No single medium is enough; a resilient civilisation archive needs all three layers
4) Frozen Ark project
The Frozen Ark Project is a collaborative international effort to safeguard the DNA, tissue, and viable cells of endangered animals[2][3]. It was established in 2004 as a consortium archiving endangered species samples for future conservation use[4].
Its purpose is not only post-extinction remembrance. It also supports present-day conservation by helping researchers:
- assess genetic diversity
- avoid inbreeding in captive breeding programs
- identify closely related individuals
- preserve material before wild populations vanish[2][3]
Public reporting on the project notes collections on the order of 48,000 samples representing about 5,000 to 5,500 species[5][6]. That is useful scale, but still far below what is needed for full biospheric continuity.
5) De-extinction feasibility
De-extinction is possible in limited cases, but it is not a universal restoration method.
### What is feasible
- Species with very recent extinction
- Species with good frozen cells or high-quality DNA
- Species with close living relatives that can act as surrogates
- Traits that can be re-created by genome editing or back-breeding
### What is not feasible today
- Reconstructing a species from fragmentary DNA alone
- Restoring complex extinct ecosystems from a single genome
- Recovering extinct species whose developmental biology, epigenetics, microbiome, and maternal environment are gone
### Strategic judgment
De-extinction is best treated as a targeted rescue technology, not a civilisation recovery plan. It may restore selected species or proxies, but it cannot rebuild the original biodiversity baseline without preserved living material and ecological context.
6) Minimum viable genetic diversity requirements
A species archive is not sufficient if it stores only one genome. The minimum requirement is population-level diversity.
### Practical conservation thresholds
- 50/500 rule: historically used rule of thumb suggesting roughly 50 effective breeders to limit short-term inbreeding and 500 effective breeders for long-term evolutionary potential.
- Modern revision: many conservation geneticists now argue the long-term target should be 1,000 or more effective breeders to better preserve adaptability.
### Archive design implications
For a viable reconstruction buffer, the archive should aim for:
- multiple unrelated founders
- samples from many geographic subpopulations
- representation of rare alleles
- both sexes, across multiple age classes
- repeated sampling across time, not a one-time snapshot
### Minimum archive target
For high-value species, the floor should be