The research team, led by Ying Shi of the Shanghai Academy of Agricultural Sciences together with colleagues including Fengwu Zhou of Fudan University and corresponding author Huili Liu, set out to test the conventional view that stray animals necessarily harbor greater viral richness than owned pets. Urbanization reshapes the environments in which companion animals live, and these changes may alter their viromes in ways that matter for zoonotic disease surveillance. Yet the dynamics of the viral microbiome in urban household and stray pets, and the forces that drive the assembly of these viral communities, had remained poorly characterized. Rather than assuming that street life means more viral exposure, the researchers measured it directly.

To do so, the team collected a total of 245 fecal or rectal swabs from household and stray cats and dogs across Shanghai. The laboratory workflow was deliberately comprehensive. Each sample underwent enrichment for virus-like particles, a step that concentrates viral genetic material while filtering out the overwhelming bulk of bacterial and host DNA. The researchers then co-extracted both viral DNA and RNA, recognizing that viruses spanning these two broad genetic classes could differ in their epidemiological relevance. After amplification, the material was sequenced on an Illumina paired-end platform, generating the short-read data from which the viral communities would be reconstructed computationally.

Identifying genuine viral sequences in metagenomic data is a notoriously difficult problem, and the team approached it with redundancy. Viral contigs were identified using four complementary tools: VirSorter2, DeepVirFinder, VirFinder and geNomad. Only sequences passing through this multi-tool gauntlet, and then further validated with the quality-assessment program CheckV, were retained for analysis. The resulting set of viral contigs was classified taxonomically, placed into phylogenetic context, and subjected to standard ecological statistics, including alpha and beta diversity measures, redundancy analysis and network co-occurrence modeling. This combination allowed the researchers to ask not only which viruses were present, but how the communities were structured and what factors best explained their composition.

The sequencing effort yielded 123 vertebrate viral operational taxonomic units, or vOTUs, drawn from virus families that include Coronaviridae, Poxviridae and Parvoviridae. These are groups of considerable public-health interest: coronaviruses in particular have repeatedly demonstrated the capacity to move between animal and human hosts, while parvoviruses are important pathogens of domestic carnivores. Cataloguing such viruses in animals that share close, daily contact with humans is a foundational step for any urban zoonotic surveillance program, and the Shanghai dataset provides one of the most detailed pictures to date of what those animals carry.

The headline result was a surprise. Household dogs exhibited the highest Chao1 richness, an estimator of total species richness, among all four host groups examined, and their viral richness was significantly higher than that of household cats, stray cats and stray dogs, according to Wilcoxon rank-sum tests corrected for multiple comparisons using the Benjamini-Hochberg false discovery rate procedure, with adjusted p-values below 0.05. Household dogs also harbored 73.7 percent of all host-specific vOTUs found in the study, fourteen out of nineteen, and every one of those unique viral units belonged to the Coronaviridae. In addition, household dogs showed the highest richness of vertebrate-associated viruses from families with recognized zoonotic potential. For dogs at least, life indoors appears to come with a more varied viral repertoire, not a narrower one.

Intriguingly, the excess richness in household dogs did not translate into a fundamentally different viral community. When the researchers compared overall community composition between household and stray dogs using PERMANOVA, a permutation-based test of differences between groups, the result was not statistically significant, with an effect size of R-squared equal to 0.07 and a p-value of 0.115. In other words, the household-stray contrast in dogs was richness-specific rather than reflecting a broad compositional shift. Household dogs carried more viral types, but the overall character of their viral communities resembled that of stray dogs. This distinction between the number of viral lineages present and the identity of those lineages is one of the study’s central conceptual contributions, separating elevated viral richness from genuine compositional differentiation.

Beyond the household-versus-stray comparison, the data revealed striking host specificity in urban pet viromes. Only about 15 percent of vOTUs were shared between cats and dogs, and network analysis of viral co-occurrence showed clear modular separation between the two host species. This modularity suggests that host identity, rather than shared urban space alone, strongly constrains which viruses circulate in which animals. The team also identified 59 putative novel viral lineages, defined as sequences showing less than 80 percent amino acid similarity to known lineages. Such a substantial reservoir of uncharacterized diversity in familiar companion animals underscores how much of the urban viral landscape remains unmapped, even in species that have lived alongside humans for thousands of years.

Perhaps the most policy-relevant finding came from the multivariate modeling of viral community assembly. Community composition was significantly influenced by socio-environmental factors, specifically human population density and green space coverage. This points to the roles of host density and urban greening as drivers of viral community structure: places where animals and people are packed more tightly together, or where green corridors alter the ecology of urban wildlife and pets, appear to shape which viruses pets carry. For surveillance planners, the message is that viral risk is not distributed uniformly across a city but varies with the urban fabric itself.

The authors conclude that both host identity and urban environmental context need to be incorporated into One Health surveillance frameworks, the integrated approach that recognizes the interconnected health of people, animals and ecosystems. The study was supported by the SAAS Program for Excellent Research Team 2025 and the Shanghai Agriculture Applied Technology Development Program, and all animal procedures were approved by the Animal Care and Use Committee of the Shanghai Academy of Agricultural Science, with owner consent obtained for household sampling and stray animals sampled humanely after capture by local authorities. As cities grow denser and greener in parallel, understanding how those design choices sculpt the viral communities of our closest animal companions may prove to be an essential component of preparing for the next viral emergence.

Subject of Research: Viral microbiome diversity and community assembly in urban household and stray cats and dogs

Article Title: Higher viral microbiome diversity in household than stray pets: insights into urban community assembly

Article References: Shi, Y., Zhou, F., Tao, J., Li, B., Cheng, J., Tang, P., Li, Q., & Liu, H. (2026). Higher viral microbiome diversity in household than stray pets: insights into urban community assembly. Microbiome. https://doi.org/10.1186/s40168-026-02525-3

Image Credits: AI Generated

DOI: 10.1186/s40168-026-02525-3

Keywords: viral microbiome, virome, household pets, stray animals, cats, dogs, metagenomics, vOTUs, Coronaviridae, zoonotic surveillance, urbanization, One Health