The source of the ongoing Ebola epidemic in West Africa may have been a hollow tree where children played—and bats roosted. In a visit to Meliandou, the Guinean village where the outbreak apparently started, researchers learned of the tree and linked it to one of the outbreak’s first victims. But, in a frustrating twist, the tree had burned to a stump just before they arrived, thwarting their search for evidence that might confirm the scenario.

A year ago, a toddler in Meliandou died of a mysterious disease; soon, his sister, mother, and grandmother were infected as well. As far as epidemiologists can tell, the family members were the first people to die in West Africa’s Ebola epidemic, which according to the latest statistics from the World Health Organization has now sickened more than 20,000 people and killed at least 7842. But how the toddler caught the virus has been a puzzle. Before the current outbreak, there was only a single known human case of Ebola in West Africa.

Human Ebola outbreaks elsewhere have been linked to outbreaks in wildlife—including duikers (small antelope), gorillas, and chimpanzees—or traced to hunters who butchered animals found dead in the forest. No one knows which animals provide the natural reservoir for Ebola, but bats are leading suspects. Several types of bats can survive experimental infection with the virus, and researchers have found Ebola virus RNA in at least three species of fruit bats. That made the animals—commonly hunted and eaten in Guinea—a top contender as the source of the outbreak.

Soon after the outbreak was identified as Ebola in March 2014, wildlife epidemiologist Fabian Leendertz of the Robert Koch Institute in Berlin went to southeastern Guinea to look for signs of an outbreak in wildlife. Leendertz, with three more German veterinarians and eight Guinean biosurvey experts from the nonprofit Wild Chimpanzee Foundation, spent 4 weeks in the region, capturing bats from four sites and surveying two protected forest areas.

The researchers found no evidence that wild animals were dying of Ebola, they report in a paper published online today in EMBO Molecular Medicine. The populations of chimpanzees, duikers, and other large mammals were at about the same levels they had been in the previous surveys in the region, conducted in 2010 and 2011—“good news for conservation,” Leendertz says. They also found no direct evidence of Ebola virus infections in any of the 169 bats (from at least 13 species) that they captured and tested. But their visit to Meliandou yielded intriguing clues.

There, they found a large tree stump near a well-traveled path to a small river where villagers washed their clothes. The hollow tree was only 50 meters from the house where the toddler lived; children used to play in it, residents told the researchers. But on 24 March, the tree had burned, Leendertz says—and all that was left were the stump, fallen branches, and ashes. (It’s not clear whether someone set it on fire deliberately because of the Ebola outbreak. “There are different stories about why it burned,”Leendertz says.)

When the tree started burning, there was a “rain of bats,” villagers told Leendertz—a small, smelly species with a long tail locally called lolibelo and sometimes “mice that can fly.” In the ash surrounding the tree, the researchers found DNA fragments that match the Angolan free-tailed bat Mops condylurus, an insect-eating species that is widespread across Central and West Africa and that fits the villagers’ description. Other studies have found that the species can survive experimental infections with Ebola.

The toddler could have picked up the virus while playing in the tree, Leendertz says; he might have caught and played with an infected bat or ingested a small quantity of infected bat droppings. “There must have been thousands of bats in there. That is more exposure than you can get by hunting individual bats,” Leendertz notes. He’s disappointed, but not particularly surprised, that they found no infected bats in their samples. “The virus must be extremely rare” in bat populations, he says. Because bats are hunted so much, if the Ebola virus were widespread, “we’d see infections all the time.”

Peter Walsh of the University of Cambridge in the United Kingdom, who studies Ebola in wildlife in Central Africa, says the evidence linking the outbreak to the bat tree is frustratingly indirect. “It is suggestive, but it certainly doesn’t rise to a ‘smoking gun’ level,” he says. The wildlife surveys were too small to convincingly rule out an outbreak in wild animals, he adds.

Leendertz agrees, but notes that there is little intact forest around Meliandou, which would limit contact with wildlife. And the fact that children and women, rather than hunters, were the first people known to be infected also suggests that eating bush meat or contact with forest-dwelling animals is relatively unlikely to have sparked this outbreak. He and his colleagues are continuing to sample bats and other wildlife in the region—most recently from Ivory Coast, close to the Guinean border. In Guinea, he says, surveys aren’t possible at the moment. With the Ebola epidemic still raging there, “people are very suspicious,” he says—especially of anyone who wants to capture bats.

Given the current Ebola outbreak, unprecedented in terms of number of people killed and rapid geographic spread, Science and Science Translational Medicinehave made a collection of research and news articles on the viral disease freely available to researchers and the general public.