Skip to main content

Deleting a Species

We are on the brink of being able to genetically engineer an extinction. Should we?

In a windowless London basement, behind three sets of locked steel doors and a wall of glass, thousands of Anopheles gambiae mosquitoes cling like Marvel supervillains to the sides of white mesh cubes. The room is negatively pressurized, so air is constantly sucked inward to ensure that the mosquitoes, which have been subjected to a new and astonishingly powerful kind of genetic engineering, never escape.

If the modifications to these whining mosquitoes were perfected, and they were somehow able to make their way to sub-Saharan Africa, they would have an effect on their kin unlike any animal that has ever existed. The Anopheles are equipped with a genetic tool that ensures that they are either sterile—they can't produce viable eggs—or, if fertile, that they will pass that sterility gene on to nearly every offspring. And the same would be true for their descendants, which would continue to spread the genetic sabotage into future generations.

If some future version of the mosquitoes were released, these deadly modifications could spread through the African tropics, crashing the population as they went. And because Anopheles is the primary African vector for the parasite that causes malaria, its collapse would likely take down malaria with it. Within a few years, the last great scourge of humanity, which kills upwards of half a million people per year, would be vanquished on the African continent. It would be one of the greatest health achievements of all time. And yet the intentional eradication of a species is not something we should pursue without a lot of foresight, and the release of highly invasive genetically modified organisms (GMOs) into the wild is itself deeply disturbing.

Known as a gene drive, the ability to force particular genes into future generations of an entire species only became available to humans with the development of CRISPR, the gene-editing tool that has enabled us to make precise changes to an organism's DNA. Kevin Esvelt was a fellow at Harvard University's Wyss Institute for Biologically Inspired Engineering in 2013 when he figured out how to build a gene drive. In a 2014 paper, he proposed several applications for his invention, including hobbling weeds that had become resistant to herbicides, reducing malaria-carrying mosquitoes, and eliminating invasive rodents on islands, where they wreak havoc on indigenous birds and plants.

A version of this story originally appeared in the June/July 2018 issue of Pacific Standard. Subscribe now and get eight issues/year or purchase a single copy of the magazine.

A version of this story originally appeared in the June/July 2018 issue of Pacific Standard. Subscribe now and get eight issues/year or purchase a single copy of the magazine.

Many traditional conservationists were horrified by the prospect, yet other groups embraced it. The Gates Foundation made gene drive a centerpiece of its anti-malaria efforts, and the eco-warriors at Island Conservation, who have long used poison to combat invasive mice and rats, seized on gene drive as a more precise weapon in their war to save native species. New Zealand is considering using a gene drive in its push to eliminate invasive rodents, weasels, and possums by 2050. Kevin Esvelt wants to engineer mice that are immune to the bacterium that causes Lyme disease, whose cycle of transmission goes from mice to ticks to people. Dengue, Zika, and several other mosquito-borne diseases are promising gene-drive targets. A lab in California is working to limit the damage caused by an invasive species of fruit fly, and labs in Australia and Texas are developing "daughterless mice" (capable of conceiving only male offspring). The first gene-drive field trials are anticipated within the next decade.

With earlier-generation GMOs, such as Monsanto's Roundup Ready crops, arguments often hinged on the potential for those genes to escape into the environment. Conservationists believed escape was inevitable, while corporations downplayed the risk, but nobody was suggesting that GMOs be let loose in nature—until now.

When I first heard about gene drive, I thought of "ice-nine," the form of water in Kurt Vonnegut's 1963 novel Cat's Cradle that is solid at room temperature and acts as a seed crystal for adjacent water molecules, turning them solid. At the end of Cat's Cradle, the frozen body of a man who has committed suicide by drinking ice-nine falls into the sea, and all the world's oceans and rivers are forever frozen, extinguishing most life on Earth. Gene drives have similar dystopian potential. In theory, a single lab could alter the entire planet. And the technology has arrived far quicker than our ability to grapple with its staggering implications.


Gene drives work by gaming inheritance, forcing their way into the genetic make-up of future generations. Sexually reproducing species usually have two versions of each of their genes, one inherited from each parent, and they randomly pass one of those to each offspring. Individuals that inherit more useful genes thrive, and are therefore more likely to reproduce and pass on those good genes, while individuals that inherit disadvantageous genes are less likely to get the chance to reproduce. In this way, evolution causes detrimental genes to disappear from the gene pool.

Conventional genetic engineering is limited by the rules of reproduction. Most engineered traits have a 50/50 chance of being passed down, and unless a trait confers some advantage to the organism, it should eventually disappear. Since most genetic engineering to date has bred traits that benefit people, not the organisms themselves, so far no GMOs have made significant inroads into nature. But a gene drive can practically guarantee inheritance. And since beneficial genes are favored by natural selection anyway, the unique value of engineering a gene drive lies in propagating a detrimental trait, possibly even all the way up to extinction.

To make a gene drive, you start with the gene-editing tool CRISPR, which consists of two parts: a gene-slicing enzyme and a string of genetic code that tells the enzyme where to cut. CRISPR is shockingly easy to use. You don't need a world-class lab, and you don't have to be a genius. I've created antibiotic-resistant bacteria in a friend's kitchen. You just order your CRISPR from a DNA-synthesis company (the going rate is $65 plus shipping), specifying the exact 20-letter sequence of DNA you want it to target. It arrives as a few drops of liquid in a test tube. You add that liquid to another test tube containing cells of the organism you want to modify, along with any new DNA you want inserted, then heat it up. The CRISPR finds the spot, makes the cut, and the new DNA gets stitched in place.

Kevin Esvelt was part of the team at Harvard that helped develop CRISPR, and he was the first to realize that the CRISPR mechanism itself could be inserted directly into an organism's genome to create a gene drive. Once there, the CRISPR would eliminate the natural counterpart of the gene it is attached to, and the cell would copy the functioning, genetically engineered version of the gene (containing the CRISPR) in its place. The organism would then have two working copies of the CRISPR gene, one of which would be guaranteed to be passed down to each of its offspring, where the process would repeat, until virtually every individual in a population carried the engineered trait.

With conventional genetic engineering, there is usually a 50/50 chance of an engineered trait being passed down, but a gene drive can practically guarantee inheritance, allowing an altered gene to spread through a population.

With conventional genetic engineering (top), there is usually a 50/50 chance of an engineered trait being passed down, but a gene drive (bottom) can practically guarantee inheritance, allowing an altered gene to spread through a population.

It was a brilliant insight, with enormous implications. According to the unwritten rules of science, Esvelt's next move should have been to quietly create a gene drive in his lab and then publish a paper announcing the achievement to the world and staking his claim to it. Instead, he paused to consider the consequences.

When I first met Esvelt in 2017 at Editing Nature, a summit convened at Yale University's Institute for Biospheric Studies to weigh the ramifications of engineering the wild, I was struck by his demeanor. He seemed haunted and tightly wound, as if he'd just come from a dark future he was hoping to save us from. His boyish smile and wispy blond hair reminded me of Tintin, but his gravelly, leading-man's voice vibrated in an unusual timbre. Like the long dungchen horns of Tibetan monks, it seemed to resonate with both awe for the world and sorrow for its eventual passing.

As soon as Esvelt realized how easy it would be to build a gene drive, he knew he had a potential ice-nine on his hands. "This thing self-scales," he told the biologists, conservationists, and ethicists gathered at Yale that day. "You can't run a field trial. You can't introduce it anywhere in the endemic environment without having it spread probably to every population."

After his 2013 discovery, Esvelt knew others would soon hit upon the same insight, and he felt that the runaway nature of gene drive was not something that could be trusted to biotech specialists working in isolation. "Your decision to go ahead and build it in the lab means that you are performing an experiment that could affect other people," he said. "And if you don't tell them that you're doing it in advance, you're actively denying them a voice in the decision. And frankly, that's wrong."

Esvelt pictured the headline sure to follow an accidental gene-drive release: "Scientists Convert Entire Species to GMOs. Is CRISPR to Blame?" He feared that a botched trial could turn the public against the technology and destroy its vast potential. So shortly after their breakthrough, he and his colleagues at the Wyss Institute called a meeting of top ecologists, biologists, ethicists, and national security experts. They explained the technology to the group, and discussed the best plan of action. And their remarkable conclusion was that the only way to ethically explore the potential of gene drive was to change the culture of science. "We need to at least tell other people what we are thinking of doing before we even begin experiments," he explained. "This is difficult, because every incentive in science points against it. If you share your brilliant idea, you're inviting some larger, better-funded lab with spare hands to steal it, get it working first, publish, and get the credit."

Esvelt decided to make an example of himself. He published his paper before doing any experiments, with the hope that all gene-drive research would follow the precautions and protocols he laid out, the most important of which was pre-registration of all experiments so they could be vetted by all potential stakeholders. Since then, he has spent as much time lobbying against the unwise use of gene drives as he has advocating for them, sometimes using language that distresses his fellow scientists. "We are walking forwards blind," he said in a 2016 interview that is frequently cited by gene-drive opponents. "We are opening boxes without thinking about consequences. We are going to fall off the tightrope and lose the trust of [the] public."

Not since Robert Oppenheimer has a scientist worked so hard against the proliferation of his own creation. "When you see something that is technically sweet, you go ahead and do it and you argue about what to do about it only after you have had your technical success," Oppenheimer said in 1954. "That is the way it was with the atomic bomb."

And that is how it has been with gene drive. Esvelt now runs something called the Sculpting Evolution group at the Massachusetts Institute of Technology. When I sat down in his office and asked him if he had convinced many scientists to forego the technical sweets, he shrugged. "It will never happen unless we change the incentives," he said. "Most scientists, however supportive in theory, say they just can't take the risk." The allure of scientific immortality—or at least tenured professorship—is simply too strong, and while those working with gene drives claim to follow rigorous safety protocols, few are willing to openly share what they are inventing behind the closed doors of their labs.


We are on the cusp of a gene-drive explosion. Many agricultural pests are potential targets, as are weeds that have evolved resistance to Roundup. California's cherry growers are funding gene-drive research to eliminate the spotted-wing fruit fly, which lays its eggs in soft fruits. Tata Trusts of India recently gave the University of California–San Diego $70 million to train a new generation of Indian scientists to use gene drives for agriculture and disease control. And in the fall of 2017, the biotech firm Oxitec released genetically engineered diamondback moths (which infest broccoli, cabbage, and other brassicas) in a field trial in upstate New York. The moths carry a gene that kills females in the larval stage, and though there is no gene drive involved at present, it would be a logical next move.

The most vocal critics of gene drives have been two conservation organizations, Friends of the Earth and the ETC Group. Jim Thomas, co-executive director of ETC, told me that, for all the emphasis on curing disease and saving endangered species, he sees "Big Ag" lurking in the background. "Ultimately, I think that's where this technology lands," he said. "It becomes a kind of insecticide. If there's money to be made here, that's what’s going to drive it." Thomas sees potential for abuse in the developing world. "How does a powerful technology shift power relations? And what does that mean for those that are marginalized and vulnerable?"

In September of 2016, 30 environmental luminaries, including Jane Goodall, David Suzuki, and Vandana Shiva, joined with ETC to publish an open letter calling for a moratorium. "We believe that a powerful and potentially dangerous technology such as gene drives, which has not been tested for unintended consequences nor fully evaluated for its ethical and social impacts, should not be promoted as a conservation tool," they wrote. "Given the obvious dangers of irretrievably releasing genocidal genes into the natural world, and the moral implications of taking such action, we call for a halt to all proposals for the use of gene drive technologies."

Friends of the Earth joined ETC in bringing the call for a moratorium to the December 2016 meeting of the United Nations' Convention on Biological Diversity, which covers the equitable use and regulation of biological resources, including genetically modified organisms. The Convention has previously halted controversial technologies such as ocean fertilization and sterile-seed crops by establishing moratoria, but with gene drive it merely called for better risk-assessment. Friends of the Earth and ETC vowed to continue to rally support for a moratorium, which will be debated in Montreal this July and then voted on at the next meeting in Egypt this December.


Most gene-drive scientists accuse these groups of exaggerating the risks of genetic engineering and playing to the public's fears, but Natalie Kofler, who founded Yale's Editing Nature initiative to facilitate public deliberation around gene editing, thinks it's vital to take their point of view seriously. "The followers of those groups share a worldview with many people that I discuss this with on a daily basis," she told me. "They feel deeply that it is wrong to tamper with the DNA of wild things. There's a sacredness to it that we shouldn't mess with. And that is a worldview that is very quickly dismissed by scientists and technologists. And because it's not being acknowledged as something valid for discussion, I think it's creating a huge polarization."

Still, Kofler finds the idea of a ban on research "totally ridiculous." This is a brand-new technology, she said. "Right now we don't know nearly enough about how it works, how the public perceives it, or how it will impact the environment to take stances of opposition or support. Right now, we need to be comfortable to stay in the gray zone—to comprehensively explore this issue with the degree of openness and transparency that it deserves. So, if anything, more research—scientific and sociological alike—needs to take place."

Jim Thomas points out that there's a difference between a moratorium and a perpetual ban: "There's a feeling that taking a judicious pause and taking the time to think carefully means nothing is ever going to move forward. But that's not what a moratorium is."

When the stakes are as high as they are with gene drive, who could argue with a judicious pause? People in Africa, Esvelt says. Every year you delay work on gene drives, another half-million people die. "Who am I to tell somebody who's lost children to malaria, and has more children at risk, that they can't do it because somebody else doesn't agree? Why should some people get veto power over a technology that could save the lives of other people's children?"


And yet, despite that sentiment, Esvelt keeps making things more difficult for his colleagues. Last November, I along with several other journalists received an unusual email from him. "I'm writing because we have a couple of papers coming out next week that are personally embarrassing for me, but are likely consequential enough for gene drive, conservation, and science policy that you might find them interesting," he wrote. What followed was a surprising statement: "My decision to list invasive species control as a potential application of gene drive in our original 2014 eLife paper was an embarrassing mistake.... It was profoundly wrong of me to even suggest it." Additional modeling, he explained, showed that gene drives were even riskier than he'd thought. For that reason, one of his new papers concluded, with the possible exception of malaria, "we should not even consider building drive systems likely to spread indefinitely beyond the target area."

The new papers triggered a wave of fresh panic in the media. "'Gene Drives' Are Too Risky for Field Trials, Scientists Say" reported the New York Times. Most of the coverage focused on Esvelt's mea culpa, and when we met in his MIT office shortly after, I asked him if that was the reaction he'd expected.

"Of course!" he responded. "I'm not totally naive. 'Inventor tries to stuff genie back in bottle'—that's a story. It doesn't happen very often that a scientist says, 'I was wrong.' Maybe it should happen more often."

Esvelt believed that other researchers were underestimating the risk of engineered organisms escaping a field test, even on an isolated site, in part because of a wild card beyond the scope of any mathematical model—human nature. "You build it, you try it anywhere, and someone who has an interest is going to move it illegally to take advantage. It would be totally cost-effective for someone to hire mercenaries to fly in, capture mice, and fly out again. But that's not the sort of thing most scientists think about."

I was reminded of Jeff Goldblum's chaos mathematician in Jurassic Park. "If there's one thing the history of evolution has taught us," he warns the park's designers, "it's that life will not be contained. Life breaks free. It expands to new territories and crashes through barriers painfully, maybe even dangerously.... Life finds a way."

Some of Esvelt's colleagues saw the move as a publicity stunt: Instead of drives "likely to spread indefinitely," Esvelt was recommending a new, self-limiting type called Daisy Drive that he had recently designed. In Daisy Drive, multiple drives are linked in an organism's genome in a kind of daisy chain. Drive A drives Drive B, and B drives C, and C drives D, and so on. But because nothing drives A, it follows normal inheritance patterns and gets quickly diluted in the gene pool. Those individuals who don't inherit A have nothing to drive B, which then gets diluted in subsequent generations. Like the stages of a rocket, the drives continue to fall away until the whole system stops working after a set number of generations. In theory, Daisy Drive allows you to affect a local population for a set amount of time.

Esvelt now hopes to use a self-limiting drive such as Daisy to combat Lyme disease in the northeastern United States, where it has become so prevalent that many people no longer risk walking in the woods and fields. Almost 40 percent of Nantucket residents have reportedly contracted Lyme disease, and that is where Esvelt has proposed to begin his "Mice Against Ticks" experiment, as well as on neighboring Martha's Vineyard. To make sure the local stakeholders understand the implications, Esvelt has been holding community forums on the islands since 2016, and most residents seem open to the idea. After an initial field test on an isolated and uninhabited island, he would release thousands of Lyme-resistant mice on Nantucket and Martha's Vineyard. If all went well, the eventual goal would be to release Daisy Drive mice on the mainland. The Lyme infection cycle would then be broken, and eventually the Daisy Drive would disappear as well. After a few generations, the mice would revert to normal.

A number of self-limiting drives have now been proposed by Esvelt and other researchers, but so far they exist only on paper, which makes Jim Thomas skeptical. "Precision in biology and ecosystems is a bit of a pipe dream," he told me. Ecosystems are remarkably complex, and viruses and parasites have tremendous capacities to evolve.

When I mentioned this critique to Esvelt, he gave me a knowing nod. "The thing everyone is overlooking is, how do you know your gene drive is going to behave over time the way you intend? We've never before engineered something that we anticipate to evolve out of our control. Perfect prediction is impossible." But unlike the skeptics, he believes you can get close enough to proceed with confidence. "You need to model very large populations over multiple generations. We can't do that in mice or mosquitoes, but we can in worms."

And they are. This winter, on the sixth floor of a nondescript MIT office building, behind a locked door with a black-and-orange Biosafety Level 2 warning sign, I held up dozens of petri dishes filled with what looked like twitching, emaciated commas. These were roundworms, C. elegans, also known as nematodes, and there were 5,000 to 10,000 of them per dish, reproducing every three days. "We can do 100 generations in a year with a population of 100 million," Esvelt told me. "If we really wanted to push it, we could probably do a population of a billion. I can't think of another organism that would let us do that."


One of Esvelt's postdocs placed a dish of worms under a microscope and turned on a black light. Through the lens, I could see the silvery squiggles snaking through the agar, eating bacteria. Each had a glowing red esophagus thanks to a fluorescent gene (originally from a jellyfish) that made it easier to track which ones had received the genetic modifications.

These worms will be the first organisms on Earth to harbor a Daisy Drive. Their lives will be confined to thousands of test tubes managed by a liquid-handling robot that can be programmed to move precise amounts of liquid between tubes. Each test tube will harbor an isolated population of worms, so the Sculpting Evolution team can test what happens when Daisy Drive worms invade a new, unmodified population. They can also test whether an engineered drive evolves into something unexpected, given enough time and population growth, and whether an "immunizing reversal drive" can be built that will target such a runaway drive and reset it.

Eventually, the worms could have enough genetic diversity to serve as a decent stand-in for any wild population, and all experiments on them will be pre-registered for feedback from the scientific community. To keep life from finding a way, Esvelt told me the project has five layers of safety containment: physical (the roundworms are kept in a locked lab, and they aren't nearly as mobile as mice, mosquitoes, or fruit flies), ecological (there are no wild C. elegans to breed with on the mean streets of Cambridge), reproductive (most wild C. elegans are hermaphrodites and aren't interested in sex anyway), molecular (the self-limiting Daisy system), and more molecular (the gene drive targets a unique DNA sequence that has been engineered into the Sculpting Evolution worms but isn't found in wild worms).

If all gene-drive research hewed to these standards, I'd sleep better at night. But despite the recommendations from Esvelt, as well as the National Academy of Sciences, there are currently no binding rules in place. And even if everyone currently working on gene drives behaves responsibly—and they seem to be—it's easy to see how, eventually, as the technology spreads, someone, somewhere along the way, will get sloppy.


Public alarm grew louder in December of 2017, with the release of a cache of 1,200 emails between scientists and other gene-drive proponents that had been obtained through the Freedom of Information Act. "Gene Drive Files Expose Leading Role of U.S. Military in Gene Drive Development," announced a press release, which noted that most gene-drive projects—including the London mosquitoes, Texas mice, and MIT roundworms—were being funded by the Department of Defense's Advanced Research Projects Agency (DARPA) as part of its Safe Genes program. Although DARPA had publicly announced it was funding the projects months earlier, this was not well known to the general public, and a number of news outlets ran with the story. The Guardian's headline read, "U.S. Military Agency Invests $100m in Genetic Extinction Technologies."

In its response, DARPA pointed out that its goals were defensive: "Our feeling is that the science of gene editing, including gene drive technology, has been advancing at a rapid pace in the laboratory," wrote the agency's chief of communications. "These leaps forward in potential capability, however, have not been matched by advances in the biosafety and biosecurity tools needed to protect against potential harm if such technologies were accidentally or intentionally misused."

The Safe Genes projects focus on learning to limit the reach of gene drives and on ways to detect and disable them, but none of that comforts Jim Thomas. "This has been the history of bioweapons research," he told me. "It's always presented as supposedly defensive: 'We have to develop these tools so we can respond in case someone else develops them.'" Thomas fears the agency's agenda may be much broader. "They're putting a finger in every single major gene-drive project so they can be close to them. So they can understand how these things work." Thomas worries that Daisy Drive is the equivalent of small-scale, tactical nukes. "Once you have this illusion that you can locally control a gene drive, then that opens the door for using it in agriculture or as a weapon." But few experts believe gene drives could make an effective weapon against other people—they are just too slow and obvious. There are easier ways to wage war.

During my most recent visit with Esvelt, I asked if he could imagine some situations where the technologies were too risky to pursue, even in a confined environment. Easily, he said. "There are areas where I would say, no research. And have!" It was after hours on a cold winter night in Cambridge, and Esvelt was looking even more pale and ragged than usual. Still, I pressed him for details. What kind of technology would be too dangerous for research? He shook his head and said, "There are some things I’ve thought of that I'm never going to tell another living soul."


When any new technology arrives, the debate veers toward the best- and worst-case scenarios, the big dreams and the big fears. Gene drives are going to cure malaria. Gene drives are going to become bioweapons. That's our nature. But it's easy to forget how rarely the extremes come to be.

The real test will be after we have a few minor successes controlling diseases or agricultural pests with gene drive. Suddenly we will have one of the greatest hammers ever invented, and we will go looking for nails. Every fast-reproducing plant or animal whose behavior we don't like will be a candidate for redesign. Cockroaches that hate the scent of garbage. Poison ivy that doesn't cause a rash. Fire ants with no fire. There are loose nails everywhere that just need a few whacks to make our lives more comfortable.

"Why not?" goes the counterargument. We've been hammering nature for years. Pollution, habitat destruction, pesticides, insecticides, greenhouse gases. Yale doesn't convene an ethics panel every time somebody clear-cuts a forest or dynamites a reef to harvest the fish. Why is it different once genes are involved?

And yet it is.

Anyone who's ever taken the time to hike to the pristine valley or paddle to the uninhabited island knows the sublimity of finding oneself in a place where the agenda is non-human. It's a reminder that there are ways of being in the world that have little or nothing to do with human ways, patterns of existence that get us out of our own heads and expand the conversation of what it means to be a quivering coil of DNA on the third planet from the sun. It's a form of diversity, and every species is a kind of culture, a cohesive and elegant web of quirks, predilections, and traditions.

We've dammed Glen Canyon. We've littered Everest with ropes and oxygen tanks. Our pawmarks are all over even the wildest places. But we have yet to conquer the DNA of wild things. For the time being, that frontier has been visited by only a handful of early explorers.

In deciding if we have the right to drive a gene through a species, we might think of each genome as a national park, an untrammeled space in a non-geographical dimension. A refuge from an increasingly humanized world. I can hate the whine of a mosquito in my tent and still revere the pristine landscape of its genome. Engineering that genome would be like putting a road system through the Gates of the Arctic. There would be some obvious benefits—and something less obvious would be lost forever.

With every new technology, we tend to shoot first and ask questions later. It's a dynamic built into the DNA of our culture, which rewards the intrepid individuals who plant their flag on the virgin coast. Those ice-nine mosquitoes in their negative-pressure vault may end up being hugely important. They may, in fact, be a gift. A living metaphor of interconnectedness and of consequences, they may force us to consider if the time has come to throw out the Age of Exploration model and create a new system of science that rewards wisdom over cleverness.

That's a big ask, and it may seem absurd right now, as we survey the vast genetic frontier stretching away before us. How could we not poke around just a little? But we have a lot of experience with lost frontiers at this point, so perhaps there's still time to ask what we ought to do with this one. What if, after gazing from the decks of their caravels at the towering forests and teeming estuaries of the New World, the early explorers went back to their funders in Europe and said: Sure, we could make the place safe and productive. We could fill it with cities and farms and factories. But here's the thing: It isn't bad the way it is. It's full of mysteries and other ways of being. So ... this is going to sound crazy, but what if we just left it alone?

A version of this story originally appeared in the June/July 2018 issue of Pacific Standard. Subscribe now and get eight issues/year or purchase a single copy of the magazine.