News Feature: Probing the limits of “evolutionary rescue”
Could species threatened by climate change and other stresses avoid extinction through rapid evolution?
In the space of five years, the field crickets of Kauai fell silent. The quiet was deafening to evolutionary biologist Marlene Zuk, who had spent a decade crawling through Hawaii’s vacant lots and church lawns, collecting the insects for her research at the University of California, Riverside. When she started her work, Zuk remembered males as always chirping. But beginning in the 1990s, she saw and heard fewer crickets. It seemed Kauai’s population had careened off an ecological cliff toward extinction.
One obvious culprit, Zuk thought, was a small parasitoid fly with remarkable hearing (1). Female flies use their fine-tuned ears to locate a male cricket chirping in the grass and drop their larvae onto his back. The maggots burrow through his carapace and eat his soft insides, bursting out to pupate in the soil about a week later. The drama of the cricket and the fly unfolded nightly in the front yards and hotel lawns of Hawaiian paradise, forcing a big trade-off for male crickets: sing for sex and court a gruesome death.
By the early 2000s, Zuk had all but stopped hearing field crickets on Kauai. The roadsides she frequented to collect insects no longer thrummed with the distinctive, nails-along-a-comb chirping of the males. One night in 2003, she opened her car door to silence on her field site. “I thought ‘that’s that, but you may as well get out of the car’,” Zuk remembers. She stepped out, clicked on her headlamp, “and all of a sudden I started seeing all these crickets.”
“If you’re not a cricket person, you will not fully appreciate the cognitive dissonance this generates,” Zuk says emphatically. Chirping is a cricket’s sexual signal. Losing it means males should not be able to attract females or have offspring. Yet, here Zuk was, seeing crickets, and not hearing a thing. “It was like, what the hell?” she says. On closer inspection, the Kauai males still rubbed their wings together. The crickets were trying to sing; their wings had just stopped making sound.
The silence turned out to be genetic (2). A mutation in a single sex-linked gene had altered wing development for some male field crickets, Zuk’s research group found. Instead of growing the rough file and scraper–like structures males usually rub together to sing, their wings became smooth and soundless. Normally, these flat-wing males would face terrible odds of reproducing because females find males by localizing their calls. But with a sharp-eared fly hunting singing crickets, silent males were much less likely to be eaten inside-out. It seemed that their favorable mutation had rescued the population, as their genes spread.
The case of the quieted crickets offers up an intriguing question: Can evolution act fast enough to save a population plunging toward extinction under the strain of environmental change? Researchers are increasingly considering the possibility of recovery in at least some species, a concept called evolutionary rescue. The crickets’ silent-wing mutation could be one example. It spread like wildfire because staying quiet conferred a big advantage.
And yet, detecting evolutionary rescue in wild populations is still hard to do with any certainty. Other factors can also rescue populations, such as changing behavior or moving to a new habitat. Still, understanding when evolution can arrest and reverse population decline has major implications for the field — and for the future of wildlife conservation policies.
From Theory to Practice
The classic graph in evolutionary rescue is a U-shaped curve representing a population changing in size over time after an abrupt shift in the environment. First, the population plummets, then bellies out, and finally rebounds by evolving a trait that allows it to persist. The first of these curves for evolutionary rescue appeared in a 1995 article by theoretical population geneticist Richard Gomulkiewicz and theoretical ecologist and evolutionary biologist Robert Holt (3). Why do some populations survive environmental change, the two men asked, while others don’t? When does evolution intervene?
Combining fundamental equations from population biology and genetics, Gomulkiewicz and Holt calculated that a population was most likely to obey its U-curve and persist when it was initially large, with a diverse pool of genes for natural selection to act on. And it couldn’t go extinct so fast that evolution had no time to kick in or dip below a critically low population size. One key assumption: the population is closed, meaning no individuals are migrating in or out. In evolutionary rescue, as it was defined in 1995, natural selection acts on the pool of genes already present in the population.
After Gomulkiewicz and Holt’s early work, the field matured slowly. “Evolutionary rescue was a mid ‘90s idea that sat around in the literature without taking off for quite a while,” says ecologist Andrew Gonzalez of McGill University and the Quebec Center for Biodiversity Science in Montreal. He and colleague Graham Bell were the first to demonstrate evolutionary rescue in the lab using yeast. Bell and Gonzalez set up hundreds of brewer’s yeast populations of varying sizes and stressed them with salt (4). Larger populations more readily adapted, they found, following Gomulkiewicz and Holt’s U-curve prediction.
But there were important caveats. Natural selection on existing genes isn’t the only way to save a population. New individuals can migrate into a declining population and keep it from shrinking further just by showing up, even if they don’t breed (a phenomenon known as ecological or demographic rescue), or they can bring in beneficial genes (genetic rescue) by breeding. Genetic rescue can also happen if new genetic material arrives by wind, water, or other means — think pollen floating through the air (5⇓–7). Most of the time, the two concepts go hand in hand, explains evolutionary ecologist Ruth Hufbauer. New individuals migrate into a population and then breed, facilitating gene flow and sometimes genetic rescue.
Hufbauer teased all three kinds of rescue apart in experiments with red flour beetles in her lab at Colorado State University in Fort Collins (8). Tiny denizens of grain silos, the beetles live their lives immersed in wheat flour: they eat it, live in it, and breed in it. Hufbauer raised hundreds of beetle populations in wheat flour enriched with nutritious yeast and then dumped them into clear plastic boxes with corn flour and a lower percentage of yeast, a less-nutritious environment. If the beetles didn’t adapt to their newfound meal, they would die. Then Hufbauer encouraged them to survive. To simulate demographic rescue, she added extra beetles from the same stock to some of the populations. For other populations, she swapped out just one beetle with an individual of a different genetic background, simulating genetic rescue. Sometimes she did both. Sometimes she did neither: her control populations didn’t receive any extra help. If they survived, it would be through evolutionary rescue.
After six generations in the corn, across both the experimental and control groups, some populations had evolved and rebounded. Their bodies grew smaller, and were likely to use fewer resources in a resource-poor environment. Genetically rescued populations — the ones with extra genes from one beetle — had the largest population sizes at the end of the experiment, compared with demographic rescue and control populations. But surprisingly, Hufbauer says, even some of the control populations survived. “We fully expected,” she says, “that they would really go extinct,” but they “were able to adapt and rescue themselves, essentially.” Natural selection acted on the beetles’ existing genes, it seemed, yielding the same U-curve predicted in 1995. It was the telltale signature of evolutionary rescue.
Over the last 25 years, studies such as this one have taken evolutionary rescue from the realm of purely theoretical to experiments with actual populations of multicellular organisms. “Now people have confidence it’s not just in mathematicians’ brains and Petri dishes,” Gonzalez says. But making the leap from yeasts and beetles in the lab to organisms in the wild has been much harder, researchers acknowledge. Even working with small laboratory critters means monitoring hundreds of replicate populations evolving over generations — a feat of tracking that’s much harder in the bush. What can rapid evolution really do to prevent extinction in the wild, Gonzalez asks? “That turns out to be a question of enormous applied value.”
Rescue favors the easily overlooked, smaller creatures. Organisms that swarm in large numbers, reproduce quickly, and have many young, studies suggest, are most likely to evolve their way out of extinction. New field studies hint at evolutionary rescue in wild populations of rats, rabbits, phytoplankton, and minnows called Atlantic killifish (9⇓–11). A 2016 study, for example, found that killifish populations from filthy urban estuaries tolerate industrial chemical concentrations hundreds to thousands of times higher than populations from cleaner sites, thanks to rapid selection on a handful of genes (12). Such examples suggest evolutionary rescue could be relevant to the real world — and that evolution may occasionally work fast enough in environments rapidly being degraded by people.
But wild cases are hard to verify. Take Kauai’s field crickets. Even such a suggestive case — with an identified mutation, that’s beneficial and widespread — isn’t definitively evolutionary rescue. Crickets and flies coexist on other Hawaiian islands too, where flat-wing males are much rarer, suggesting Kauai’s population might not have needed the mutation to avoid going extinct. If the crickets weren’t headed for oblivion, then their rebound wouldn’t qualify as rescue. “There’s always some uncertainty,” Gonzalez says.
Real-world populations don’t live in the isolation of a Petri dish, and evolutionary adaptation isn’t their only tool to deal with environmental change. New behaviors and migration can also help a population survive stressful situations.
In the cricket’s case, it seems a combination of genetic change over time across the population, as well as behavior, helped their populations rebound. A silent male might be safe from the fly, but staying quiet presents mating challenges. “How does a female find you?” says Zuk, who’s now at the University of Minnesota in St. Paul. “And even if she finds you, what’s going to make her mate without a song?” A behavior of the silent males may have been key. They hang around the few singing males in the grass and intercept females headed the same way. All crickets will sometimes carry out this so-called satellite behavior, Zuk says, but it “seems to be more pronounced in places with the flatwings.” Zuk thinks the mutation found a toehold because of satellite behavior (13). Evolution alone didn’t save the crickets; behavior helped it along.
This sort of behavioral flexibility in a changing environment is one example of phenotypic plasticity — the ability to display different traits under different circumstances. It can look a lot like evolution, but it’s not. Ants in the genus Pheidole, for example, carry genes for huge heads and bodies, which most species normally don’t express. The genes can be expressed, however, in larvae exposed to a juvenile hormone, according to a 2012 study in Science (14). Ants born after exposure to the hormone grow into super-soldier–like adults with massive heads. But the ants aren’t evolving. Huge-head genes already existed in the population, sleeping in the genome.
Adaptation — becoming better suited to the environment — can happen by evolution (as in genetic change over time) or by changing gene expression so the same genotype shows a new phenotype (as in the ants). One reason that wild cases of evolutionary rescue are so hard to prove, Gonzalez says, is because phenotypic plasticity and evolutionary adaptation can look so alike. Pure plasticity, as in the ants’ case, isn’t rescue. But when plasticity and genetic change are combined, as in the crickets, evolutionary rescue can occur. Zuk’s case seems to be rapid evolution made possible by phenotypic plasticity; the silent-wing gene wouldn’t have spread without a way for males and females to find each other and mate.
A Natural Ally
So what can rapid evolution really do in the wild, and what are its limits? Scott Mills chuckled at that question, on the phone from his office at the University of Montana in Missoula. “That’s it,” he says. “We don’t know.” Mills and other wildlife biologists want to make evolution an ally in the race to conserve disappearing species. Montana’s winter mountains give them a unique vantage to ask how.
On the hillsides there, a long list of predators prey on snowshoe hares — “the candy bar of the forest,” Mills says. Camouflage is a hare’s best defense. The animals blend in with the landscape by growing a brown coat in spring, which turns snowy white as the days grow short in fall. But as Montana’s climate changes, snow is falling later, and melting earlier in the season, leaving hares mismatched with their environment and very visible to predators. Snowpack is expected to decrease by roughly 40 to 69 days in western Montana this century (15). “White animals on brown ground stick out,” Mills says. “Our hares in Montana get clobbered in weeks where they’re white on brown background.”
Mismatched hares can’t keep pace with warmer winters and decreasing snow because their trigger to molt and shed isn’t temperature; it’s day length. Mills has found that hares don’t have much phenotypic plasticity to change their coats, overriding day length for another seasonal cue. “So then we have to ask,” he says, “is there a possibility to adapt fast enough, via natural selection?”
The answer is: maybe. In more southerly parts of the snowshoe hare range, such as coastal Oregon and Washington, snow is unpredictable and rarely sticks. Hares there keep a brown coat year-round, molting and shedding from brown to brown. A single gene is responsible, which came from mating with black-tailed jackrabbits, and spread through snowshoe hare populations living in low-snow conditions, Mills reported last June (16).
Liaisons with another species can accelerate evolution, but unless they coincide with population declines
“The promise of evolutionary rescue, is that maybe some fraction will recover, maybe there is some hope.”
and high-speed environmental change, they don’t qualify as rescue. In this case, the winter brown coats probably spread through Pacific Northwestern hares between 3,000 and 15,000 years ago, so it’s hard to say whether it initiated rescue or not in those populations. But the adaptive brown gene showed Mills that climate can shape coat color. “Not many traits are as definitively shaped by climate as this one,” Mills says. “Because whether you’re mismatched is 100% determined by the average persistence of snow.”
When could a trait shaped by climate help species survive the kind of rapid change Mills is seeing in Montana? He figured that polymorphic populations — where winter white and winter brown hares coexist — would offer the richest palette for natural selection to act on and, therefore, the highest odds of evolutionary rescue. In another 2018 article, Mills showed, using data from natural history collections, that polymorphic populations of hares and other seasonally coat-changing species pop up across the Northern Hemisphere (17). In places such as Washington’s Cascade Mountains, both hare color morphs hop between patches of snow and towering red cedars. Hares aren’t endangered, but they illustrate how conservation might embrace polymorphic areas, such as the Cascades, where evolutionary rescue is most likely.
Although Mills isn’t certain rescue can happen in this case, he sees the hare’s story as a metaphor for the conservation community because evolutionary rescue is “nowhere on the radar of reserve design.” It’s been clear since the first theory article in 1995 that large populations are more likely to rescue with a manageable extent of environmental change. Subsequent studies showed connected populations, with migration, gene flow, and some history of similar stress may be the most likely to adapt and survive. But how exactly humans might foster rapid evolution is the next unanswered question, Mills says — one that “goes to the heart of climate resilience for wild species.”
How effective reserves could be depends heavily on the rate of climate change, Gonzalez adds. Whether Earth sees 2 °C or 4 °C of warming and whether that’s in 50 years or 100 or 200 will decide which populations are even candidates. Polar bears and other charismatic mammals aren’t likely contenders because their generation times are long. Evolutionary rescue takes 10 to 100 generations, he says, meaning hundreds of years for large mammals. Rapid change will outpace them before rescue kicks in. Faster-breeding creatures, such as insects, are the better bet. Indeed, Kauai’s field crickets shifted from chirping to 90% silent males in fewer than 20 generations, or about a decade. Even so, Gonzalez would still choose policies that slow down climate change and keep populations big and connected, he says, to “allow evolutionary rescue to be a possibility, even if it’s not likely.”
The next frontier for the field may be studying it at community levels. Individual populations are woven into communities, so when one group rescues, there may be domino effects for the species it interacts with, Gonzalez explains. Stressing whole ecosystems — such as small ponds teeming with bacteria, water bugs, and fish — and then watching as adaptation unfolds (or doesn’t) at multiple trophic levels could help clarify community evolutionary rescue’s role in the fate of ecosystems themselves.
Understanding rapid evolution may not stop many extinctions, but it could lead to conservation policies that maximize the potential for rescue. Considering how bleak the story of man’s impact on wildlife can be, “the promise of evolutionary rescue,” Gonzalez says, “is that maybe some fraction will recover, maybe there is some hope.”
Published under the PNAS license.