Air sampling accurately captures circulating zoonotic viral diversity emerging from poultry live-animal markets

Abstract

Environmental surveillance has emerged as a pivotal strategy for early detection of pathogens that pose threats to humans (1) but has not been utilized for zoonotic agents. In Asia, live-bird markets (LBMs) are key human-animal interfaces for zoonotic virus transmission (2). Traditional sampling strategies are time-consuming, expensive, threaten animal welfare and have significant occupational biosafety risks. In this study, we assessed the performance of metagenomics on environmental samples (ES) compared to traditional poultry swabs for detecting avian viral pathogens in LBMs in Cambodia. ES, including air, cage swabs, and carcass wash water, were collected alongside throat and cloacal swabs from domestic chickens and ducks across twelve sampling visits in two LBMs over a 15-month period. Viral nucleic acids were extracted and sequenced using a capture probe-based metagenomics approach. Our results show that metagenomics on ES outperformed traditional poultry samples in detecting the highly pathogenic Influenza A/H5N1, including circulating clades 2.3.4.4b and 2.3.2.1c, which were found in the environment but missed by poultry swabs on multiple occasions. Environmental metagenomics was also highly sensitive in the detection of over 40 other viruses from key pathogen families such as Astroviridae, Coronaviridae, Picornaviridae, and Retroviridae. Viral contigs from ES showed high similarity to those from poultry swabs further highlighting the accuracy of this approach. Our findings highlight metagenomics on ES can precisely and effectively replicate metagenomic results from traditional surveillance samples, offering broader coverage and enhanced detection of avian pathogens. This robust approach could be pivotal for mitigating zoonotic spillover, controlling pathogen transmission at LBMs, and enhancing pandemic preparedness strategies.

Figure 1

Environmental metagenomics accurately detects circulating poultry pathogens. A) Circular heatmap depicts virus detection between poultry swabs and the environment. Whereby a given virus was found in at least poultry sample a blue cell is shown and whereby that same virus was found in at least one environmental sample at the same time and location an adjacent pink cell is shown. Each track of the heatmap represents one visit to an LBM. B) Boxplot showing the difference in virus abundance between taxa that were successfully detected or undetected from panel A. Statistics were calculated using a Wilcox test. C) Bar plot showing the total number of unique viruses detected in poultry swabs compared to each individual environmental sample at each sampling date. Statistics were calculated using Fishers exact test best all groups. Based on PCoA in Supplementary Figure 5, we calculated the distance of all samples from the median centroid coordinates of D) cloacal and E) oropharyngeal swabs obtained from domestic chickens as well as F) cloacal and G) throat swabs obtained from domestic ducks. Statistics were calculated using a Kruskall-Wallis with Dunns post-hoc test. H) Ridge plot showing the probability of each poultry swab being the source of pathogenic viruses for all environmental samples. Statistics were calculated using a Kruskall-Wallis with Dunns post-hoc test. All P-values obtained were corrected for false-discovery rate (FDR) using the Benjamini-Hochberg method. P values are annotated as follows: P < 0.05 *; P < 0.01 **; P < 0.001***.

Figure 2

Most avian viral contigs align to environmental contigs. A) Circular heatmap showing the percentage (%) of all avian viral contigs mapping to those assembled from the environment. B) Summary of panel A showing the percentage alignment rate of avian contigs to those de novo assembled from environmental samples. C) The total number of contigs assembled for each type of poultry swab collected. Alignment rate for D) cloacal and E) throat swabs collected from chickens is shown for each individual environmental sample. Statistics were calculated using a Kruskall-Wallis with Dunns post-hoc test. All P-values obtained were corrected for false-discovery rate (FDR) using the Benjamini-Hochberg method. P values are annotated as follows: P < 0.05 *; P < 0.01 **; P < 0.001***.

Figure 3

Environmental sampling detected highly pathogenic Influenza A contigs more often than poultry swabs. A-C) Maximum likelihood phylogenies of the HA subtypes A) H5, B) H6 and C) H9. D) Heatmap showing the detection rate of different HA and NA subtypes across all twelve visits to the respective markets. Ward.d2 hierarchal clustering reveals HA and NA subtype combinations which co-circulated. E-G) Maximum likelihood phylogenies NA Influenza subtypes D) N1, E) N2, F) N6. Triangle shaped tips show new sequences generated in the current study from one of two LBMs in Cambodia (which are also highlighted with a red arrow). Circle shaped tips indicate reference sequences. Each tip is coloured based on clade designation shown in the legend accompanying each phylogenetic tree.

Figure 4

Environmental derived pathogen contigs are a short phylogenetic distance from those identified in poultry swabs. Phylogenetic tree reconstruction of nine avian pathogens successfully detected using environmental metagenomic surveillance inferred using the maximum likelihood method in IQ-TREE. Sequence type is determined by colour and shape of the node. Novel sequences assembled from the environment are shown in green (triangle) while sequences derived from poultry swabs are shown in red (circle). Reference sequences download from NCBI Virus Database are shown in blue (square). All trees represent full length genomes from the respective virus.