Why Demanding 100% Species Accuracy Misses the Conservation Point Entirely
A growing strand of scientific purism argues that de-extinction efforts must achieve perfect genetic fidelity before they can be taken seriously. The position—advanced by researchers including Philip Seddon, Pontus Skoglund, evolutionary biologist Nic Rawlence and ancient-DNA specialist Jeremy Austin—rests on the belief that anything short of an exact genomic match fails both scientifically and ethically. Austin, based at the Australian Centre for Ancient DNA, has famously dismissed de-extinction as “fairytale science,” questioning whether mammoth or moa projects represent serious conservation at all. But this view sits uneasily with how evolution works and how conservation is actually practiced.
Seddon’s assertion that “extinction really is forever” leans on the idea that species are discrete, fixed entities with clear boundaries. Yet evolutionary biology tells us the opposite: populations shift continuously over time, adapting to new pressures and diverging from their historical baselines. Modern conservation routinely works with populations that are genetically distinct from their pre-decline ancestors, and it does so without hesitation.
Skoglund’s critique of genome-edited wolves as inadequate proxies for extinct dire wolves reflects a similar line of thinking. It assumes ancient DNA offers an idealized benchmark that should be re-created exactly. In reality, variation across millennia and geography is the rule, not the exception. Conservation translocations often introduce animals into habitats they have been absent from for generations, with the goal of restoring ecological processes, not historical precision.
Rawlence’s description of edited organisms as “designer GM animals rather than true de-extinction” highlights a further conceptual disconnect. The argument focuses on definitions rather than outcomes. If an engineered organism can occupy an ecological niche, influence vegetation, or reshape a landscape in ways the extinct species once did, then its exact genomic composition becomes less relevant than its functional performance.
It is notable that the world’s primary conservation authority takes precisely this view. The International Union for Conservation of Nature (IUCN)—the organization responsible for the Red List and global conservation standards—defines de-extinction in strictly functional terms. In its Guiding Principles on Creating Proxies of Extinct Species for Conservation Benefit, the IUCN states the aim is to produce:
“a suitable proxy able to restore ecological functions or processes that might have been lost as a result of the extinction of the original species.”
The IUCN does not require perfect genetic replication. In fact, it explicitly frames de-extinction around ecological functionality, not molecular precision.
History provides a further reality check. Early captive breeding programmes for the California condor, black-footed ferret and Arabian oryx faced similar criticisms—accusations that artificially bred or genetically narrow populations were too compromised to merit recovery. Those views did not hold. Each species exists today because conservationists prioritised survival and ecological viability over concerns about purity.
Austin’s view that functional de-extinction is inadequate reflects an academic stance at some distance from day-to-day conservation decision-making. Wildlife managers consistently face trade-offs: genetic composition, habitat constraints, and the urgency of mitigating ecosystem decline. Their priority is function—whether an organism can help stabilise or rebuild a system—not whether its genome aligns with an extinct predecessor.
Treating species as artifacts that must be restored exactly as they once were overlooks a basic point: ecosystems respond to what organisms do, not to how precisely their DNA mirrors the past. The purist framework would disqualify many of conservation’s successes and risk immobilising emerging technologies that could help address biodiversity loss at a scale traditional methods cannot meet.