Beth Shapiro on De-Extinction: Dire Wolves, Mammoths, Conservation Biology, and the Ethics of Engineering Life

1. Guest Background

The guest in this Huberman Lab episode is Dr. Beth Shapiro, presented in the episode evidence as an evolutionary biologist and the Chief Scientific Officer of Colossal Biosciences. Huberman adds that she was previously a professor at UC Santa Cruz and an investigator with the Howard Hughes Medical Institute before leaving academia for Colossal. That background matters because the interview is not a generic conversation about spectacular extinct animals. It is an episode analysis of how a scientist working at the center of ancient DNA, genome engineering, and conservation argues for a specific meaning of de-extinction.

Colossal is described as working on de-extincting species such as the woolly mammoth, dodo bird, and dire wolf, while also using genomics for broader species preservation. Shapiro’s own origin story in the interview adds useful context. She says she began at the University of Georgia in broadcast journalism, worked in local television and radio, and then shifted after a nine-week geology and archaeology field course made her want to tell stories about landscapes, people, and ecosystems. She later studied parasitoid wasps in Panama in an ecology and population-ecology context, received a Rhodes Scholarship, went to Oxford, and joined Alan Cooper’s ancient DNA lab. Ancient DNA appealed to her because it combined geology, paleontology, and storytelling: sequences from vanished ecosystems could be used to explain how environments changed when humans arrived or when the world warmed after the last ice age.

That mixed background helps explain the interview’s texture. Shapiro speaks as a company scientist defending ambitious projects, but also as an evolutionary biologist who keeps returning to evidence, biological limits, animal welfare, ecological function, and governance. Huberman’s role is to press the questions many listeners would ask: What is a species? Why not dinosaurs? Is a dire wolf really a dire wolf? Would mammoths simply be released in Alaska? If genetic editing is acceptable for animals, what happens when similar logic reaches human embryos? The episode covers de-extinction, human genetic engineering, ethical implications, and species preservation through that back-and-forth.

2. What the Episode Covers

The episode’s main subject is Shapiro’s attempt to move de-extinction away from the pop-culture image of resurrecting the past and toward a more technical, ecological, and governance-focused account. The discussion begins with species concepts. Shapiro argues that “species” is a human concept: biology does not care which category humans choose, but humans need labels to communicate, conserve, regulate, and reason. She distinguishes biological, genetic, and geographic species concepts, not to dismiss classification, but to show that different questions need different classificatory tools. Her buffalo and bison example makes the point accessible: animals may share a common name because of appearance or use, while DNA shows that they are not close relatives.

The Neanderthal discussion turns the same point into an evolutionary lesson. Shapiro says modern human ancestors who left Africa bred with Neanderthals, and many people today carry about 2% to 5% Neanderthal DNA. Those fragments differ person by person; if the fragments in living people are pooled, she says they could reconstruct more than 90%, possibly more than 95%, of the Neanderthal genome. This is not treated as a trivia point. It becomes an example of how ancient DNA changes the scale of comparison. Before the Neanderthal genome, asking what makes humans human meant comparing humans with chimpanzees or bonobos across 3 million to 5 million years of divergence. With Neanderthal DNA, that branch is shortened to roughly 300,000 to 500,000 years, narrowing the candidate list of relevant variants.

The de-extinction discussion then moves through technical feasibility. Shapiro says dinosaurs cannot be brought back because there is no recoverable dinosaur DNA: the oldest DNA recovered from bone comes from a mammoth bone roughly 1 million to 2 million years old, while dinosaurs disappeared more than 66 million years ago. She explains that DNA breaks down after death through UV damage, freezing and thawing, and microbial decay. Cold Arctic conditions can preserve DNA far longer than hot, wet places. The dodo example is especially instructive: Shapiro says she tried hundreds of bones from Mauritius, but the useful high-quality dodo genome comes from a bird that was taken alive to Europe and preserved in the Danish Museum of Natural History.

From there, the episode asks how a target species should be chosen. Shapiro says the choice is driven by motivation and includes technical, ethical, ecological, and social reasons. A species’ extinction cause, ecological role, remaining niche, risk of re-extinction, and possible harm to other species all matter. The final third of the conversation expands the frame: the same tools used for de-extinction, including multiplex genome engineering, cellular rejuvenation, wild-animal iPS cell technologies, cloning, genotype-to-phenotype mapping, and ecological modeling, may also serve living endangered species. Huberman then pushes the ethical extension to humans by raising IVF embryo screening, the Chinese CRISPR baby case, and embryo sequencing companies that claim to screen for disease or optimize traits. The episode therefore becomes less a story about spectacular animals than a broad analysis of when genome engineering is evidence-guided repair and when it becomes ethically unstable selection.

3. Core Views: Reasoning, Examples, and Limits

The episode’s central view is that de-extinction should be understood as functional reconstruction, not the literal copying of an ancient individual. Shapiro uses the mammoth project to correct the Jurassic Park template. Colossal is not growing ancient DNA into an animal or filling dinosaur gaps; for mammoths, it compares mammoth genomes with those of Asian elephants, which she says are about 99% similar depending on the calculation. The target is not the complete genome of one ancient mammoth. The target is the set of changes shared by mammoths and different from elephants, especially changes that plausibly matter for appearance, physiology, and ecological function. In her framing, the result would be an elephant capable of living in habitats that mammoths occupied, not a time-machine duplicate.

The dire wolf example makes this reasoning concrete. Shapiro says Colossal sequenced fossil dire wolf genomes, learned from them which genetic changes were associated with larger, more robust, light-coated animals, selected 20 edits, and engineered those changes into a gray wolf genome. Yet the light-coat trait also shows why de-extinction is not simply obedience to ancient DNA. Shapiro says variants near the fossil dire wolf light-coat genes raised animal-welfare concerns in a dog or gray-wolf background, including ocular cutaneous albinism, blindness, or deafness. The team therefore used a safer light-coat route known from living dogs and gray wolves. This is one of the episode’s strongest examples because it contains both ambition and restraint: ancient DNA guides the target, but animal welfare can override a more literal reconstruction.

Shapiro’s species argument supports the same view. If someone insists on a strict genetic species concept, engineered mammoths or dire wolves will always be vulnerable to the claim that they are not “really” the extinct animal. Shapiro’s reply is not that names never matter, but that name-gatekeeping can crowd out the more important questions: Is the animal healthy? Are the traits biologically grounded? Can it fill an ecological role? Does the technology help living species? The Neanderthal and bear examples show why a simple label can mislead. Modern humans and Neanderthals interbred; brown bears and polar bears diverged about half a million years ago and can still interbreed when habitats overlap, but gene flow may persist in one direction and not the other because adaptation matters. The limitation is that classification cannot be abandoned in practice. Regulation, public communication, animal management, and ecological release still need operational definitions. Shapiro’s argument works best against the claim that incomplete genetic identity makes the project worthless; it does not remove the need for careful naming and governance.

A second major view is that ecological legitimacy cannot be inferred from technical success. Making an engineered animal is one question; releasing it is another. Shapiro rejects the idea of simply making a mammoth and releasing it into Alaska, citing the difficulty of creating the first animal, the need for safety and care, IACUC protocols, and the authority of multiple regulatory agencies. She is even more restrictive about dire wolves: she says they should not be released because they would compete with gray wolves, which already face survival pressures, and because there is no ecological need. This distinction is essential. The episode does not argue that every de-extinct animal belongs in the wild. It argues that any release must pass a separate ecological and regulatory test.

At the same time, Shapiro does use ecological examples to explain why some projects are worth exploring. Pleistocene Park is presented as a case where large herbivores such as bison, wild horses, deer, and muskox change snow cover, surface insulation, plant recovery, and landscape diversity. Yellowstone gray wolf reintroduction is used to illustrate cascading effects of restoring a top predator: prey populations changed, riparian plants recovered, and river flow was affected. The thylacine discussion is more speculative: Shapiro suggests that Tasmania’s missing top predator may relate to current ecological problems and that sick Tasmanian devils might have been removed more readily if top predators remained. These examples should not be given equal evidentiary weight. Yellowstone is a historical reintroduction example; Pleistocene Park is an experimental analogue; the thylacine claim is a plausible ecological hypothesis. The episode is strongest when it keeps those levels distinct.

A third core view is that de-extinction and living-species conservation are not a zero-sum pair. Shapiro’s repeated answer is that both rely on the same tools, technologies, and needs. Charismatic extinct animals draw public engagement, enthusiasm, student attention, and investment, but the same technical stack can be redirected toward urgent conservation. The northern quoll example shows a possible functional edit: Colossal Australia partners made a single-letter genome change that changes one amino acid, and dish-based measurement suggests it may let quolls tolerate cane toad toxin. The black-footed ferret story shows a different tool: cloning from Frozen Zoo tissue can reintroduce genetic diversity into a population bottlenecked by earlier near-extinction. Shapiro also says Colossal announced red wolf cloning alongside the dire wolf work, using the same toolkit to introduce red wolf ancestry into the existing red wolf population.

This view has power, but also needs boundaries. Public enthusiasm and new investment can expand conservation capacity, but they do not automatically guarantee good priorities, independent oversight, or long-term stewardship. Shapiro partly answers that concern by describing local advisory panels, the Tasmanian advisory group, Maori stewardship of the moa project through the Naitahu Research Centre, care reports, and public communication. That is not a “trust us” model; it is a staged model involving local people, conservation biologists, regulators, and public education. Its limitation is that the model depends on transparency, outside scrutiny, stable funding, and the willingness to slow down when evidence is weak.

Finally, the episode shows that the ethical burden becomes heavier when similar technologies move from animal conservation to human selection. Huberman raises IVF embryo screening, the CRISPR baby case in China, and companies sequencing embryos for disease exclusion or trait optimization. Shapiro distinguishes height from IQ, saying height is relatively measurable and heritable while IQ is difficult to define, measure, and separate from culture. Baby KJ provides the positive counterexample: a child with a urea cycle deficiency received a bespoke CRISPR base-editor treatment developed through academic, industry, NIH, and Children’s Hospital of Philadelphia collaboration. That case supports the medical value of genetic engineering, but it does not settle reproductive enhancement. The episode’s most careful ethical lesson is that shared tools do not imply shared permission. Conservation rescue, disease treatment, and embryo selection require different thresholds.

4. Learning and Application

A practical way to use this episode is to treat any de-extinction claim as a layered claim. Ask what is being restored: a genome, a visible trait, a physiological capacity, an ecological role, or a public story. Shapiro’s mammoth and dire wolf explanations show that serious de-extinction is usually not ancient DNA being grown into an animal. It is comparative genomics plus engineering in a living relative’s genome. That means the most useful questions are not only “Is it really the extinct species?” but also “Which edits were chosen, what evidence links them to traits, what background genome are they entering, and what welfare risks were avoided?” The dire wolf coat-color decision is a good model: the safer route mattered more than literal copying.

A second application is to separate creation from release. An engineered animal born in a controlled setting has not thereby earned a place in the wild. Shapiro’s own boundaries are clear. She says dire wolves should not be released because they would compete with gray wolves and have no present ecological necessity. She also says a first mammoth would require care, safety, resources, and regulation rather than immediate release. For any comparable project, the release checklist should include extinction cause, available niche, disease risks, competition, possible hybridization, food-web effects, stakeholder consent, regulatory review, long-term funding, and an exit plan. Without those conditions, “ecosystem restoration” remains an aspiration rather than an established result.

A third application is to look beyond dramatic appearances. Conservation genome engineering may matter most when it changes survival-relevant functions: toxin tolerance, disease resistance, climate adaptation, or genetic diversity. The northern quoll example points toward a narrowly targeted anti-toxin edit. The black-footed ferret example points toward genetic rescue by cloning older tissue from the Frozen Zoo, followed by possible future edits for plague resistance if the genetic mechanism can be identified. The red wolf example shows how a toolkit publicized through a dire wolf story might also serve a living endangered wolf. The tradeoff is that genetic tools do not replace habitat protection, disease management, local stewardship, or captive-breeding expertise. They add options; they do not remove ecological work.

A fourth lesson is that public communication is part of governance. Shapiro says Colossal talks publicly through social media, podcasts, documentaries, papers, preprints, talks, and National Geographic Live. She also argues that the dire wolf story made people say “extinction” and “synthetic biology” for the first time and discuss their agency in the future world. For high-risk science, communication should not only celebrate breakthroughs. It should explain uncertainty, failed paths, animal-welfare constraints, regulatory steps, dissent, and what would cause a project to stop. Otherwise, public fear that scientists are doing things merely because they can will become a rational response to missing information.

A fifth application is to evaluate inaction as a choice. Shapiro’s ecological ethic is not that every intervention is justified. It is that humans have already altered ecosystems through hunting, domestication, species movement, vaccination, predator protection, agriculture, and habitat change. In fast-changing habitats, refusing translocation, assisted reproduction, genetic modification, synthetic biology, or de-extinction may mean accepting a less biodiverse future. The boundary is equally important: recognizing the cost of inaction does not approve every intervention. It simply forces the comparison to include both intervention risk and default decline.

The human genetics discussion requires a stricter boundary. Baby KJ’s bespoke CRISPR treatment illustrates a medical case where a severe disease, a specific mechanism, direct patient benefit, and safety testing can make gene editing ethically compelling. Embryo selection for IQ, height, or other preferred traits is different. Huberman’s examples of IVF screening, the Chinese CRISPR baby case, and embryo sequencing companies raise consent, inequality, measurement validity, and social-pressure problems. Shapiro’s distinction between height and IQ is a useful warning: even a heritable trait can be easier or harder to define and measure. A practical ethical rule from the episode is to avoid treating disease rescue, conservation rescue, and reproductive optimization as if they belonged to the same permission category.

Source

• Original episode: Bringing Extinct Species Back to Life | Dr. Beth Shapiro