Juliana Telles-de-Deus, Ingra M. Claro, Mayara Bertanhe, Charles Whittaker, Márcio Port-Carvalho, Esmenia C. Rocha, Thaís Moura Coletti, Camila A. M. da Silva, Ian Nunes Valença, Tamara Nunes de Lima-Camara, Márcia Bicudo de Paula, Mariana S. Cunha, Jaqueline Góes de Jesus, Pâmela dos Santos Andrade, Victoria Cox, Natalia C. C. F. de Azevedo, Juliana Mariotti Guerra, Juliana Laurito Summa, Ana Paula P. Teixeira, Eduardo S. Bergo, Mariza Pereira, Filipe R. R. Moreira, Alvina Clara Félix, Anderson V. de Paula, Raissa Heloísa de Araújo Eliodoro, Marissa da Silva Lima, Franciane Mendes de Oliveira, Valquíria Reis de Souza, Lucas A. M. Franco, Marcelo S. Nardi, Thaís Caroline Sanches, Eric T. B. C. da Silva, Alice Coimbra, Paulo R. dos Santos, Katherine Lima de Gouveia, Francisco E. S. P. Vilela, Sarah C. Hill, Dilmar A. G. Oliveira, Hélia Maria Piedade, Thais Guimarães-Luiz, Camila M. G. Abreu, Guilherme Casoni da Rocha, Leandro Abade, William M. de Souza, Ben Lambert, Renato Pereira de Souza, Adriano Pintér, Ester C. Sabino, Luís Filipe Mucci, Nuno R. Faria
Abstract Yellow fever virus (YFV) continues to threaten human and wildlife populations in the Americas, yet its transmission at the forest–urban interface remains unclear. Here we integrate ground- and canopy-level mosquito surveillance, systematic monitoring of non-human primate carcasses and viral metagenomics to describe the dynamics of a sylvatic YFV outbreak in a 186-hectare Atlantic Forest fragment embedded within metropolitan São Paulo, Brazil, between 2017 and 2018. Our analyses reveal that transmission was primarily driven by a single genetic cluster introduced during a period of high abundance of the main vector, Haemagogus leucocelaenus mosquitoes. A near-complete hepatitis A virus genome was detected in a YFV-infected howler monkey, suggesting potential co-infections at the human–wildlife interface. Phylogenetic and epidemiological modelling estimated a basic reproduction number, R 0 , for sylvatic yellow fever of 8.2 (95% CI 5.1–12.2), substantially higher than previous estimates for urban outbreaks. Our findings demonstrate that multisource surveillance could provide actionable early warnings in regions at risk for zoonotic spillover.