De-extinction

De-extinction is the attempt to bring back an extinct organism. In practice, no project restores an extinct species. What the working programmes produce is a proxy: a living species edited to carry some of the extinct animal's genetic variants and, if the edits work, some of its traits. Whether that constitutes de-extinction is a question of definition, and the definition has become the main point of public dispute.

§Three routes, only one of which is genomic

Back-breeding selectively breeds living descendants for traits resembling the ancestral form. The Heck cattle of the 1920s and the later Tauros programme aimed at something aurochs-like; the Quagga Project in South Africa breeds plains zebras for reduced striping. These produce animals that look the part and are genomically ordinary.

Cloning by somatic cell nuclear transfer requires intact cells, which extinction usually precludes. The one success remains the bucardo, a Pyrenean ibex whose last individual was tissue-sampled before her death in 2000; a cloned kid was born in 2003 and died within minutes from a lung malformation.¹ The same technique is used productively for species that are endangered rather than extinct, as in the cloning of a black-footed ferret and a Przewalski's horse from cell lines cryopreserved decades earlier. That is genetic rescue, and it works. The mechanics are covered in Human cloning.

Genome editing of a close relative is the route every current headline project takes. It exists because ancient DNA is fragmentary. Even exceptionally preserved permafrost specimens yield short, chemically damaged fragments rather than chromosomes, and there are no living cells to clone from.² Sequencing a mammoth is therefore possible; growing one from mammoth cells is not. The workaround is to sequence the extinct genome, compare it with a living relative, identify variants underlying traits of interest, and install those variants in the relative's cells with CRISPR–Cas9 or its derivatives.

§What has actually been done

1. 2003
   Bucardo cloned
   A cloned Pyrenean ibex is born and dies within minutes, the only extinct animal ever brought to term and still the only one.
2. 2013
   The idea goes public
   A TEDx conference organised by Revive & Restore and National Geographic gives de-extinction its name and its research agenda, including passenger pigeon and heath hen projects.
3. 2016
   Conservation guidance
   The IUCN Species Survival Commission issues guiding principles treating de-extinction products as proxy species, to be judged by conservation benefit rather than by resemblance.
4. 2021
   Colossal founded
   Ben Lamm and George Church start a company to pursue mammoth, and later thylacine and dodo, with venture funding on a scale the field had not previously seen.
5. 2025
   Edited proxies announced
   Colossal publicises mice edited for cold-adapted coat and metabolic traits and three grey wolves carrying, on the company's account, about twenty edits at fourteen genes and described as dire wolves; taxonomists and conservation biologists reject the second characterisation.

The 2025 canid announcement is the clearest case study. Three grey wolf pups were produced carrying, on the company's account, about twenty edits across fourteen genes, selected from a dire wolf genome assembled from ancient specimens, and presented as the return of a species that has been extinct for roughly ten thousand years. Specialists pointed out that dire wolves are not closely related enough to grey wolves for a small edit set to reconstitute them, that the edited traits were largely morphological, and that the resulting animals are grey wolves with alterations. The company's own scientific framing has been more careful than its announcements, using a functional definition under which an animal counts if it performs the extinct species' ecological role; the exchange is set out under Colossal Biosciences.

Mammoth work is further from an animal and technically more interesting. It requires elephant pluripotent stem cells, which were reported only recently and remain hard to work with; multiplex editing of many loci; and gestation. The last is the binding problem. An elephant pregnancy runs close to two years, the only available surrogate species is itself endangered, and the alternative is an Artificial womb for a several-hundred-kilogram fetus, which is a much harder problem than the neonatal work described in Ectogenesis.

Marsupials invert the difficulty. The thylacine programme uses the fat-tailed dunnart, a mouse-sized relative, because marsupial gestation is very short and development continues in the pouch. The size difference between host and target is extreme, and the germline route requires deriving and editing marsupial stem cells with tools that were developed for mice. Birds are harder still: avian de-extinction depends on culturing and editing primordial germ cells, which works reliably in chickens and poorly in pigeons, the group containing the dodo's relatives. All three programmes are gated on reproductive biology, not on editing, and the field's dependence on Induced pluripotent stem cells and on In vitro gametogenesis-adjacent techniques is more decisive than its dependence on nucleases.

§What "de-extinct" would have to mean

A species is not a list of variants. It is a genome, a developmental environment, a gut microbiome, a set of learned behaviours transmitted between generations, and an ecological context. An edited elephant would have elephant mitochondria unless those were replaced, elephant epigenetic inheritance, an elephant microbiome and no mammoths to learn from. Beth Shapiro, an ancient-DNA specialist who later became chief science officer at Colossal Biosciences, argued before joining the company that the honest target is a functional proxy rather than a resurrection.³

The IUCN's guidance takes the same line, treating any such organism as a proxy to be evaluated on whether it delivers conservation benefit and on what risks its release carries.⁴ Under that standard the interesting question about a cold-adapted elephant is not whether it is a mammoth but whether releasing it into Arctic tundra does what proponents claim, which is to convert moss and shrub tundra to grassland, raise surface albedo and slow permafrost thaw. That hypothesis is contested among ecologists and has never been tested at scale.

Restoration or distraction

Proponents argue that de-extinction develops genetic-rescue tools with immediate use for endangered species, and that the publicity funds conservation science that would otherwise go unfunded. Critics reply that resources spent on proxies could conserve more biodiversity if spent on habitat, and that the promise of reversal weakens the case for prevention.⁵

§Risks and governance

Releasing an edited organism into the wild is a deliberate ecological intervention, and it inherits the governance problems of Gene drives without the self-propagating mechanism. Existing frameworks fit awkwardly. A proxy is a genetically modified organism under most national biosafety law, a candidate for reintroduction under conservation law, and in some jurisdictions not obviously a protected species at all, since the extinct taxon it names is not on any list. Whether a proxy could be granted the legal protections of the species it imitates is unsettled and matters commercially.

Animal welfare is the more immediate concern. Cloning and interspecies gestation have high failure rates and produce sick neonates, and the surrogate species used are often themselves threatened. The bucardo died of a developmental defect within minutes of birth, which is the modal outcome of nuclear transfer in a novel species rather than an unlucky one.

The strongest technical caution is unrelated to ecology. Multiplex editing of many loci compounds the chance of the chromosomal losses and rearrangements described in Off-target effects in genome editing, and a programme that plans to make dozens or hundreds of changes in one genome is operating well beyond the edit counts validated in therapeutic work such as Human germline editing or in the engineered pigs of Xenotransplantation.

§Outlook

The near-term output of these programmes is likely to be tools rather than animals: better stem-cell derivation in non-model species, higher-multiplex editing, cryobanking of cell lines from living endangered populations, and improved ancient-genome assembly. Those have conservation value independent of whether a mammoth is ever born, and the same infrastructure supports work in Synthetic genomes and Genetic code expansion and recoding.

The unresolved question is what standard should govern the claim. If a proxy counts as de-extinction when it fills an ecological role, the burden is ecological evidence that it does, which no project has yet supplied. If it counts when it resembles the extinct animal, the field is doing selective breeding with better instruments, and the Precautionary principle arguments raised against it have little to attach to. The companies have so far been allowed to choose the standard and announce the result against it.

§See also

• Colossal Biosciences
• Human cloning
• CRISPR–Cas9
• Gene drives
• Artificial womb
• Xenotransplantation
• Induced pluripotent stem cells
• Precautionary principle

§References