Can an extinct species be brought back?

Sequencing recovers a genome by assembling short fragments against a reference from a living relative. Some of it never lines up: in a 2026 study of the extinct Christmas Island rat, 3 to 4 per cent of the genome stayed uncovered however deeply it was sequenced, and the gaps held variants that matter. What it produces is a readable sequence and not a molecule: there is no way to turn a list of letters into a chromosome in a cell without building it.

A 2015 commentary by Beth Shapiro set out the state of the field plainly, describing the realistic path as genome engineering rather than resurrection: identify the genetic differences that matter, edit them into a living relative’s cells, and see what results.

Those two genomes came from animals separated by forty thousand years: one from the mainland population and one from the last survivors on Wrangel Island.

The later genome carries the signature of a severe decline: a fifth less diversity than the mainland animal, and long runs of identical sequence, which the authors traced to generations of breeding in a small population rather than to closely related parents.

Later work softened that picture without removing it. A 2024 study of twenty-one Siberian mammoth genomes found that the island population recovered quickly from its bottleneck and stayed stable for six thousand years, while mildly harmful mutations built up. Any genome chosen to copy is one animal’s, carrying its population’s history, which is an awkward detail for any plan that treats a genome as a recipe.

An Asian elephant edited to carry mammoth variants for cold tolerance, hair and fat is an elephant with mammoth traits. It has elephant mitochondria, a different gut microbiome, an elephant upbringing, and no mammoth to learn from.

Whether that counts as de-extinction depends on what a species is taken to be, which is not a question the technology settles.

The strongest argument for the project is functional: restoring a grazing animal to a landscape that evolved with one might restore a process rather than an ornament.

The evidence for that is a model and field sites grazed by living animals such as reindeer, not mammoths, and the habitat, the climate and the other species have all changed in the interval.

Critics argue that a credible promise of reversal would change how people weigh a loss, and the promise is more credible in headlines than in laboratories. In 2025 grey wolves edited at about twenty places in fourteen genes were announced as dire wolves brought back.

The same tools have a quieter use on species that are not yet gone: restoring lost genetic diversity to small populations, the nearer and duller application, and the one Shapiro called potentially the most important. A black-footed ferret cloned in 2020 from cells frozen in 1988 carried a lineage missing from the captive population.