Figure 1

Figure 1. Location of Ibirapuera Central Park where waterfowl were found dead from divergent highly pathogenic avian influenza A(H5N1) clade 2.3.4.4b virus, São Paulo, Brazil, 2025. Map shows location in South America....

Figure 2

Figure 2. Maximum-likelihood phylogenies of NP and PB2 internal genes from divergent highly pathogenic avian influenza A(H5N1) clade 2.3.4.4b virus, São Paulo, Brazil, 2025. A) Alignment of NP internal gene sequences. B)...

In June 2025, we collected cloacal and tracheal swab samples from 2 dead white-faced whistling ducks (Appendix Figure 1) near the lake in Ibirapuera Park, São Paulo, Brazil (Figure 1). We detected AIV and H5N1 from the swab samples by using reverse transcription real-time PCR. We generated complete genome sequences (GenBank accession nos. PX062323–PX062330) (Appendix Table 1) directly from the original swab samples by using the Ion Torrent S5 Platform (Thermo Fisher Scientific, https://www.thermofisher.com). We classified genotypes by using GenoFlu (https://github.com/USDA-VS/GenoFLU). The hemagglutinin (HA), neuraminidase, polymerase acidic, and matrix protein (MP) segments were classified as Eurasian lineages, whereas the nonstructural protein segment was classified as a North American lineage. The remaining internal genes (polymerase basic 1, polymerase basic 2, and nucleoprotein) were unassigned (Appendix Tables 2, 3). We resolved the origins of the unassigned segments by using BLASTn and maximum likelihood (ML) analysis. The polymerase basic 2 and nucleoprotein sequences shared >94% nucleotide identity and clustered tightly with lineages from Colombia and the Brazilian Pantanal (7,8) (Figure 2). That topologic incongruence demonstrates reassortment with regional LPAIV gene pools.

Figure 3

Figure 3. Phylogenetic tree of hemagglutinin gene from divergent highly pathogenic avian influenza A(H5N1) clade 2.3.4.4b virus, São Paulo, Brazil, 2025. Maximum-likelihood phylogenetic tree (left) showing alignment of global hemagglutinin gene sequences....

Our phylogenetic analysis showed that the polymerase acidic and MP segments remain conserved within the Eurasian lineage, clustering with the clade 2.3.4.4b backbone, characteristic of the current panzootic (Appendix Figure 2). However, the nonstructural gene (North America lineage) clustered with AIVs established during the initial introduction of H5N1 viruses into the Americas (2) (Appendix Figure 3). Our genetic analyses of the HA gene sequences showed 97.9% nucleotide identity with the 2023 reference isolate from Brazil (GenBank accession no. OR269887.1) (6), and >98.6% nucleotide identity with recent North American strains (Appendix Table 3), suggesting a divergence from the original lineage established in Brazil. The maximum likelihood phylogenies we generated of the HA and neuraminidase genes also showed the sequence divergence (Figure 3; Appendix Figure 4). The isolates we sequenced form a monophyletic cluster among North American lineages (2022–2025) (9), suggesting a second introduction of the HPAI H5N1 virus in South America (10,11).

Figure 4

Figure 4. Time-scaled Bayesian maximum clade credibility tree of divergent highly pathogenic avian influenza A(H5N1) clade 2.3.4.4b virus, São Paulo, Brazil, 2025. The tree was inferred by using the Gaussian Markov Random...

We defined the timeline for the second introduction of the HPAI H5N1 virus into South America by using Bayesian temporal analysis. The estimated time to the most recent common ancestor (tMRCA) with North American H5 virus lineages was ≈mid-2023 (mean tMRCA May 2023; 95% highest posterior density March 2023–July 2023) (Figure 4). Our tMRCA estimates provide temporal support for the second H5N1 virus introduction from North America and subsequent circulation until it was detected in Ibirapuera Central Park.

We identified a divergent reassortant HPAI H5N1 virus strain in white-faced whistling ducks from Ibirapuera Central Park, São Paulo, Brazil. As one of the city’s largest green, cultural, and recreational areas, attracting thousands of visitors each day (12), the park is a notable human–animal-environment interface. Following laboratory confirmation of H5N1, park authorities immediately implemented prevention, management, and isolation measures to reduce the risk for transmission.

Our phylogenetic and temporal analyses support a second introduction of HPAI H5N1virus into South America from North America, after reassortment with locally circulating LPAIVs in South America. Our findings suggest that reassortment likely occurred before the viruses dispersed through the interior regions of Brazil by regional migratory birds (5), although the precise movement dynamics of those waterfowl require further investigation. Similar exchanges have been described in other H5N1 genomes from South America (5,11), indicating ongoing genetic exchange between introduced HPAI viruses and regional AIV viruses, contributing to the increasing genomic diversity of H5N1 viruses circulating in South America. Together, those reassortment events position South America as a critical region for AIV surveillance.

Our results highlight that HPAIV can be detected in a major urban park that harbors multiple animal species. The coexistence of resident waterfowl, migratory birds, synanthropic mammals, domestic animals, and abundant human activity creates an opportunity for cross-species transmission events. Our observations underscore the necessity of integrating systematic monitoring of urban wildlife into One Health surveillance frameworks to improve early detection and mitigate AIV risks.