Coronavirus infections are down across much of the United States. Hospitalizations, too. Deaths are finally dropping from their dismaying late-summer peak of more than 2,000 a day. Most people are vaccinated, and booster shots are gaining approval. Officials in the United States are hoping the worst of the pandemic is over.

But so much depends on the virus itself. It is not static. It mutates. Delta, the variant of SARS-CoV-2 now causing virtually all infections in the United States, is more than twice as transmissible as the virus that emerged in Wuhan, China. The possibility of further significant mutations in the virus looms like a giant asterisk over any discussion of the trajectory of the pandemic.

In recent weeks, scientists who closely monitor the virus have said it still appears to have plenty of room to evolve.

“I see nothing that suggests this virus is quieting down,” said Kristian Andersen, an immunologist at Scripps Research. “I don’t think this virus is as transmissible as it can be.”

Scientists are tracking dozens of “sublineages” in the delta line of viruses, each with a slightly different array of mutations. One of those sublineages has spread with unusual speed in the United Kingdom recently and is gaining attention from researchers.

Coronavirus variants like omicron, delta and mu are an expected part of the virus's life cycle, but vaccines can prevent more infectious variants from forming. (John Farrell, Hadley Green/The Washington Post)

So far, there is no compelling evidence that any of the delta offspring have evolved into new, more dangerous variants. The attention-grabbing sublineage in the United Kingdom has taken several months to reach 8 percent of new infections there, so although it may have an advantage, it is not spreading with the kind of explosive speed seen with the ancestral delta strain, noted William Hanage, an epidemiologist at the Harvard T.H. Chan School of Public Health.

“This is nothing like that, but still worth keeping an eye on,” Hanage said in an email. But he stipulated, “We’d have to be idiots to think the virus is done with us, and it will continue to evolve.”

Many scientists suspect that the next “variant of concern,” if and when one does emerge, is most likely to descend from delta. But viral evolution is inherently unpredictable.

“You can’t predict the future — biology is too complicated. No one should even try,” said Joel O. Wertheim, a biologist at the University of California at San Diego who studies how viruses evolve.

‘They didn’t want to believe it’

There was a time, early in the pandemic, when the scientific orthodoxy held that the coronavirus didn’t mutate much, certainly not as promiscuously as influenza. The virus has a proofreading mechanism that limits genetic errors as it replicates.

But the virus surprised the experts. The first significant change in the virus was identified by Bette Korber, a theoretical biologist at the Los Alamos National Laboratory in New Mexico. She had been scrutinizing the genomes of virus samples from around the world and noticed that one mutation, known as D614G, had become common in the virus in dozens of geographic locations. This mutation altered the positioning of the virus’s spike protein — its tool for binding to cells.

Korber, in collaboration with researchers at Duke University and the University of Sheffield in England, concluded that the strain with the mutation was more transmissible than the first strain that circulated in China. They posted their findings online — and slammed into a wall of scientific skepticism.

Maybe, other scientists suggested, it was just a coincidence. Others simply didn’t believe that a single mutation was likely to change the virus dramatically.

“They didn’t want to believe it. They wanted to believe it was holding still,” Korber said, recalling a “rough” period for her in the days after she posted her findings. “People wanted to hope for the best.”

The smoking gun in the case of D614G is that the circulating virus never randomly favored the original strain.

“We could show that within weeks after it entered a new local population, the new form would soon dominate; the reverse, a switching from the mutant D614G to ancestral, just wasn’t happening, and random events don’t all go in one direction,” she said in an email.

No one today doubts that the coronavirus is capable of evolving rapidly — and dangerously — as it spreads through the human population. It is a generalist virus — able to infect many different mammals. It has been known to jump from humans into minks and back into humans. Zookeepers are coping with infections among lions, tigers, gorillas and other captive animals.

Scientists such as Korber and Wertheim are studying the genomic sequences regularly posted on global databases. In a purely scientific sense, this has been a fascinating spectacle. In 1918, when the influenza pandemic killed upward of 50 million people worldwide, scientists had minimal understanding of the pathogen causing so much disease and death. More recent pandemics occurred before the maturation of genomic sequencing technologies. The Internet has enabled rapid data-sharing. Scientists around the planet are watching viral evolution in real time.

“Evolutionary space is vast for this virus,” Korber said.

The coronavirus can change in two fundamental ways. First, it can become more transmissible, by binding better to receptors in the nose, replicating more quickly once it invades the body, or becoming more efficient at aerosol transmission.

Second, it can elude immunity. Many mutations alter the physical shape of the spike protein on the surface of the virus. That shape-shifting can make the virus a more elusive target for antibodies produced through vaccines or prior infection. Those antibodies are hunting for the earlier version of the virus. Some still hit their target, but others miss and fail to neutralize the virus.

So far, evolution has led the virus down the first route, to become more transmissible.

Most mutations are harmful to the virus. A large fraction of mutations have no effect. But a very small fraction of them alter the virus in a way that could prove advantageous and help the emergence of a new variant.

Typically, major functional changes in a virus involve a suite of mutations, as happened with alpha and delta. Despite scrutiny by scientists, some of those mutations remain enigmatic. Experts don’t know what they do, exactly.

In rare cases, single-point mutations, in which one letter of the genetic alphabet is swapped with another, make a difference. This is what happened with D614G. Robert W. Shafer, an infectious-diseases research professor at Stanford University, said experiments suggest the D614G mutation increases the probability that one of the three protruding structures on the spike protein will be in the right shape to more easily bind to human cells.

There are biological constraints to how efficient a virus can become at transmission. Where those limits are, though, is not fully understood.

“This has been an old question with flu: Is flu going to run out of options?” said Sarah Cobey, a computational biologist at the University of Chicago who studies the ecology and evolution of viruses. She notes that influenza has never run out of ways to evade immunity. H3N2, a subtype of influenza observed since the 1968 pandemic, continues to develop new methods to dodge the immune system’s antibodies.

Scientists studying the evolution of the coronavirus say their research reinforces the need to vaccinate widely and rapidly. There is too much virus in circulation. Mutation is a numbers game. The more chances a virus has to mutate, the more likely it is that a fitter variant will gain traction.

President Biden’s science adviser, Eric Lander, said driving down the amount of virus in circulation is critical to reducing “the number of shots on goal this virus has.”

Lander answered quickly when asked what the virus is going to do next: “Nature is very creative. I think the key for us is to be prepared for whatever happens.”

Preparedness requires situational awareness, one part of which is genomic surveillance — studying samples of the virus to see how it is mutating. In December, as the alpha variant was beginning to infiltrate the United States, researchers were producing genomic analysis on only a small fraction of virus samples in the country. Many experts worried the country was blind to new mutations.

The tracking has improved dramatically since then. Scientists in the United States have sequenced 2.3 percent of 40.5 million reported cases, according to a Washington Post analysis that includes data from The Post’s case tracker, Johns Hopkins University and GISAID, an initiative sharing information on the coronavirus and flu virus. That’s seven times the surveillance rate in January, when the United States ranked 38th in the world for genomic surveillance. Nine months later, the nation had risen to 28th.

Yet genomic surveillance remains patchy in some U.S. communities, and genetic surveillance is limited globally.

“It is very likely that the next variant will be a daughter of delta, and it may emerge anywhere,” said Vaughn S. Cooper, who studies the evolution of microbes at the University of Pittsburgh.

‘Respect for this virus’

At one point early this year, the alpha variant was spreading so rapidly that it became responsible for roughly 70 percent of all new infections. Alpha, like the D614G variant, caught the world by surprise, turning up in genomic sequences in England in September 2020 with an extensive package of mutations already in place.

That sudden appearance of a remarkably mutated variant led researchers to speculate that it could have originated in an unknown immunocompromised patient who had been treated for months. This was a sobering conjecture.

Alpha was one and a half times more transmissible than earlier forms of the virus. Then came delta. First identified in India, it initially made little impact in the United States early this year when it first showed up in genomic sampling. By early May, delta still accounted for only about 1 percent of new U.S. infections.

Then, it exploded. By July it was dominant, and by August, it had nearly wiped out all competitors. Anyone getting sick with the coronavirus in the United States today is almost certainly dealing with delta.

Other variants — such as mu, which achieved some media attention over the summer — are still in the mix, but just barely, and they continue to be outcompeted by delta.

A virus can become more lethal as it evolves, and there is some evidence that delta is more likely to cause severe illness. But the opposite is also true: Viruses can mellow over time. Pathogens that incapacitate their victims will not be as successful, typically, as those that are less “virulent” and allow the infected person to keep moving around and spreading the virus.

“Natural selection favors pathogens that transmit. It doesn’t really care about whether it kills the host or makes it sick,” Wertheim said.

The most important consequence of new variants, Shafer and seven other experts wrote in a recent Nature Reviews Genetics report, is how they will affect vaccines. So far, there is no evidence the delta variant is evolving into a vaccine-evading form, scientists said.

Cooper, the University of Pittsburgh scientist, said new variants are emerging because of the sheer number of cases, as the virus jumps from person to person hundreds of thousands of times a day.

Concerns over viral evolution should accelerate vaccination, not hold it back, Cobey and three co-authors argued in a review paper in April. Consider a breakthrough infection: Most people who have a breakthrough infection don’t have sufficient antibodies to ward off the virus. Or their antibodies aren’t targeting the right sites on the virus — or both. Whatever the case, Cobey said, the virus isn’t under pressure to escape immunity because there’s just not a lot of immunity in play.

“There’s very little evolution that happens in the course of an infection in most people, really,” Cobey said.

Some vaccine-makers are preparing customized, variant-specific formulas, but none appears close to public use. Pfizer-BioNTech has tailored a form of its vaccine to target the delta variant’s spike protein, though this is not yet being tested in people. Moderna, too, has a candidate vaccine that’s specific to delta. But the company is not testing a delta-centric version in people, either, in part because data on people who have had a third dose of the original vaccine shows boosted antibodies that neutralize delta.

Johnson & Johnson has said a Phase 3 trial of its original vaccine showed protection against delta and other circulating variants, but the company is “constantly assessing the need for an update, and we are looking at how to do so if the need emerges.”

The development of effective vaccines has given the world a tool to end the emergency phase of the pandemic. But vaccine inequity is a global problem. While the richest countries have vaccinated more than half their population, the vaccination rate in the poorest countries is minuscule. Variants can arise anywhere, and the virus has plenty of people in which “to try out its moves,” as Korber puts it.

Maria Van Kerkhove, a World Health Organization epidemiologist, said past pandemics offer guidance about how to respond to this one. But every pandemic is different, she said, because every virus is different.

“I really feel that we need to keep the respect for this virus,” she said. “And while we want it to get to some predictability, this virus still has quite a lot left in it.”

Dan Keating and Harry Stevens contributed to this report.