TOWARD THE END of last year, doctors in Nelson Mandela Bay, a city of about 1 million people in the Eastern Cape of South Africa, started to see something alarming. The city had been hit by a tsunami of Covid-19 cases in June and July, swamping hospitals and leading to thousands of deaths. That wave began to subside as winter turned to spring in the southern hemisphere. But starting in November, hospitals in the city and its surrounding province began to fill up with Covid-19 patients again—this time twice as fast they had during the first surge.

To figure out what was going on with the steep uptick in new cases, doctors at those hospitals enlisted the help of Tulio de Oliveira, a geneticist and bioinformatician at the University of KwaZulu-Natal in Durban who leads a national network of sequencing labs. His team began piecing together the genomes of the coronavirus that had caused each person’s infection. For months, these researchers had been periodically doing similar genomic surveillance work to keep tabs on the dozens of strains of SARS-CoV-2 that were circulating around the country, looking for any problematic mutations in the virus’s spike protein. Eight months into the pandemic, in 99 percent of the more than 1,500 genomes they’d sequenced, they’d only found one such mutation. De Oliveira was in the process of submitting those findings to a journal.

Then, on December 1, the first results came back from Nelson Mandela Bay. In each of the 16 samples gathered from 15 clinics around the city, the viruses all possessed a near identical constellation of mutations unlike any that had ever before been seen in South Africa. And eight of those mutations were in the spike protein. “Literally the day before I had written, ‘The spike genome in South Africa is very stable,’” de Oliveira told WIRED in an interview. “Then I saw this new cluster and I thought, ‘Wow, that has changed.’”