The transmission of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) from humans to other animal hosts — known as zooanthroponotic transmission — continues worldwide and has been described for several companion, captive and wild animal species1,2,3. Indeed, evidence indicates that SARS-CoV-2 is a generalist pathogen that can infect at least one nonhuman animal species from nearly every mammalian order (Fig. 1). Abundant, sustained and protracted circulation of SARS-CoV-2 in the human population is likely to promote the risk of establishment of secondary animal reservoirs for otherwise-extinct variants and tangential paths of viral evolution. In addition, the circulation of SARS-CoV-2 in animals poses the risk of virus transmission back to humans, complicating variant-specific vaccination strategies. Cases of animal-to-human transmission of SARS-CoV-2 have been documented, and involve infected mink in several countries, white-tailed deer (Odocoileus virginianus) in North America4, pet hamsters in Hong Kong5 and cats in Thailand6. Multihost transmission of SARS-CoV-2 has resulted in novel SARS-CoV-2 clades4, and variants with enhanced binding affinity to the angiotensin-converting enzyme 2 (ACE2) receptor7, which is exploited by the virus for cell entry. Here, we describe these risks in more detail and argue that enhanced surveillance of SARS-CoV-2 presence in nonhuman species and of cross-species transmission is an essential element of the ongoing global response to this new virus.

Fig. 1: The breadth of hosts known to be susceptible to SARS-CoV-2 as of April 2022.
figure 1

Consensus maximum likelihood tree of mammals (4,098 species; 100 bootstraps) obtained from data hosted on VertLife27, as outlined in the Supplementary Information. Mammalian hosts with either documented experimental or natural infections with SARS-CoV-2 are annotated. The full list of susceptible host species and associated references are provided in Supplementary Table 1.

Full size image

Persistence, host adaptation and recombination risks of spillover into animals

A reservoir is an epidemiologically connected animal population or environmental niche in which a pathogen such as SARS-CoV-2 can be maintained and transmitted to humans8. White-tailed deer potentially form one such reservoir for SARS-CoV-2 and have been suggested to perpetuate lineages that may be extinct in humans9. Additionally, a cluster of infections detected in white-tailed deer from Ontario, Canada (assigned to PANGO lineage B.1.641) has probably been circulating outside of the human population in wildlife since 2020 or early 2021 (ref. 4). The persistence of SARS-CoV-2 in zoonotic reservoirs is of particular concern as it facilitates host-specific adaptation and potential recombination with other viral lineages or species, threatening the health of wildlife, domestic animals and people.

Putative host-adaptive mutations in SARS-CoV-2 following spillover events into farmed mink and wild white-tailed deer have been identified10. Although the functional role of these mutations has not yet been determined, one deer-specific mutation was found in the PLpro domain of nonstructural protein 3 (Nsp3). Nsp3 is known to interfere with host innate immunity, and disrupts the type I interferon pathway by direct cleavage of host IRF3 (ref. 11). This is relevant to the key question of whether SARS-CoV-2 may become more pathogenic in humans after acquiring animal-specific adaptations. A common evolutionary trade-off when a virus adapts to novel hosts is reduced virulence in the original host. This feature has been known since the late 19th century and led to the development of the first attenuated vaccines, in which a viral strain is passaged in animal hosts. The establishment of SARS-CoV-2 in a new animal host can be independent of its effect on human health if transmission to and from humans is not essential for the virus to persist in, and adapt to, its nonhuman animal host12. If a secondary reservoir for SARS-CoV-2 can persist without the need for constant reseeding through repeated spillovers from humans, animal-adapted viral lineages may adapt and diverge from strains in circulation in humans, irrespective of their transmissibility in the former12.

Another concern regarding the circulation of SARS-CoV-2 in animals is the potential for recombination with other lineages of SARS-CoV-2 or other animal-borne viruses that may allow them to emerge as novel zoonotic viruses. A key feature of coronavirus replication is the occurrence of homologous RNA recombination13,14. Coronaviruses can easily jump between nonhuman animal species or between animals and humans15. Recombination events followed by host-specific adaptation have contributed to the emergence of several new coronaviruses that have successfully jumped into novel host species14, resulting in established diseases such as bovine coronavirus disease, canine coronavirus disease and the two coronaviruses (HCoVs) that are endemic in the human population, HCoV-OC43 and HCoV-229E. Recombination involving the spike gene amino-terminal and receptor-binding domains and some accessory proteins between clades of SARS-CoV-related coronaviruses circulating in bats are also believed to have contributed to the emergence of SARS-CoV16.

Although there is no evidence to date for recombination between SARS-CoV-2 and any of the endemic HCoVs, it remains possible that SARS-CoV-2 may recombine with coronaviruses that are circulating in animals. Such recombinants could lead to the secondary introduction of novel variants with increased cross-transmissibility or virulence in humans. Furthermore, SARS-CoV-2 variants that emerge from recombination in an animal host could be antigenically different, largely escaping the protection afforded by SARS-CoV-2 vaccines and antibody-based therapeutic agents that target spike antigens. This possibility is underscored by an excess of recombination in the spike gene17. Indeed, evidence of recombination events occurring between endemic coronaviruses and other clade A betacoronaviruses exists for the dromedary camel, the reservoir of MERS-CoV18. Here, the endemic camel virus HKU23 (DcCoV-HKU23) was found to have genomic segments of a rabbit virus (RbCoV-HKU14) in the haemagglutinin esterase gene as well as homologue fragments of a rodent betacoronavirus in the spike protein18.

What surveillance of SARS-CoV-2 in humans and animals can teach us about pandemic emergence

From a macroecological perspective, humans are only a single node in a complex network of host species that pathogens continually jump between. Although there is some evidence that features associated with the virus, host and environment contribute to the cross-species transmission of pathogens, the drivers of host jumps are not fully understood19. The ongoing animal multihost transmission of SARS-CoV-2 necessitates a focus on the drivers of interspecies viral spillover, maintenance and adaptation. Analysis of the ecological and genetic features of both host and virus through complex modelling approaches such as machine learning may yield valuable insights into the drivers of SARS-CoV-2 exchange (direct or indirect) across vertebrate hosts. More broadly, these modelling approaches enable the rapid triaging of resource-intensive surveillance efforts and in vitro and in vivo characterization of pathogens that are predicted in silico to be zoonotic. Most importantly, understanding the ecological and molecular drivers of virus transmission among hosts may eventually allow us to preempt the emergence of novel zoonotic diseases. However, such complex modelling approaches require large datasets with comprehensive metadata including host health status and immunological profiles, and these are currently not available. Such datasets can only be generated through coordinated surveillance efforts in both humans and animals.

Challenges and opportunities in the coordination of multipillar disease surveillance

Given the risks associated with secondary spillovers into zoonotic reservoirs, there is an urgent need to implement integrated surveillance programmes to coordinate the monitoring of SARS-CoV-2 in humans, the environment, and domestic and wild animals. However, there are several challenges to implementing integrated disease surveillance. In most countries where surveillance systems exist, separate agencies manage surveillance in humans, animals and the environment, often with no intersectoral communication or coordination. In high-income countries, disease surveillance systems in humans and animals are governed by entirely different sets of regulations, policies, standards and funding sources, which makes intersectoral cooperation challenging. By contrast, in most low- and middle-income countries, resource limitations and poor infrastructure are substantial barriers. Regional and national governments and international agencies should implement policies that ensure intersectoral cooperation and provide dedicated, long-term and international funding for sustained and coordinated pathogen surveillance.

Additionally, many livestock producers lack the financial means or support to test their animals for target pathogens outside of funded surveillance programmes, and may fear economic losses and a threat to their livelihoods if diseases are detected. Also, there are few mandated wildlife surveillance programmes other than those for specific pathogens such as rabies and chronic wasting disease. Therefore, comprehensive education and compensation programmes will be essential to improve overall knowledge of and response to zoonotic diseases in livestock and wild animals, including transmission routes, symptoms, economic effects and potential health risks. Such initiatives require support from the government, coordinated partnerships and collaboration with all stakeholders in food animal production, wildlife services and conservation, and the large corporate and private sectors.

Many in the disease surveillance field argue that a lack of precise guidelines for data sharing and usage is the primary reason for the mistrust between agencies and the absence of interagency cooperation in disease surveillance20. In addition, data collection standards are very different among agencies, which makes data integration almost impossible. A potential solution is the development of data repositories, universal guidelines for data collection, analysis and sharing, interoperable data standards and protocols among agencies. The utility of data repositories is further underscored by their potential use for viral phenotyping and assessment of risk determinants, such as host cell receptor binding, viral replication, transmission, immune escape and antiviral resistance. This information transforms genomic data into health intelligence, particularly when coupled with key ecological, behavioural and epidemiological elements (such as R0, a measure of transmissibility). Standardization around these risk elements is also critical to knowledge synthesis and decision-making.

Testing all animals everywhere is not practical, feasible or necessary to gain health intelligence for mitigation and control. Instead, a rational approach to targeted surveillance, coupled with laboratory experiments to validate zoonotic potential, is essential. To design this, we can incorporate computational models to predict and shortlist high-risk host species and locations. Molecular simulations have extensively been used to model interactions between ACE2 and the receptor-binding domain to predict the susceptibility of different species to wild-type SARS-CoV-2 and its variants2. Additionally, ecological models have been developed to predict locations where frequent contact between humans and animals, and between animals, increase transmission risk21,22. Such models do not generally include, but can benefit from, incorporating existing knowledge of animal behaviour and ecology, requiring the input of expertise from wildlife ecology and animal husbandry. Also, linkages with human and environmental (wastewater) data enable geotemporal prioritization based on areas of high viral activity among humans and risk of animal exposure. Finally, key epidemiological parameters can inform the prioritization of surveillance by focusing on high-risk transboundary and human–animal interfaces (for example, urban wildlife and peri-domestic animals), as well as endangered animals, such as some species of primates at risk of high mortality.

A robust ‘One Health’ surveillance system for infectious diseases requires four key programmatic pillars to be in place (Fig. 2). Surveillance of environmental samples (mainly involving wastewater) has previously been used to monitor antimicrobial resistance23 and is becoming established for monitoring SARS-CoV-2 (ref. 24). Such surveillance can act as an early detection system; for example, identifying poliovirus in sewage in the UK and New York in 2022 (ref. 25). There has also been massive genomic surveillance for SARS-CoV-2: over 15 million human-derived genomes for SARS-CoV-2 strains have been deposited on the public repository GISAID (current to April 2023)26. However, surveillance programmes in humans are largely limited to patients who are seeking healthcare and often target specific subsets of human pathogens, precluding the ability to detect novel infectious agents. A SARS-CoV-2 surveillance scheme that relies on untargeted RNA metagenomic sequencing would additionally inform on the incidence and prevalence of other known zoonotic pathogens such as filoviruses or highly pathogenic avian influenza viruses (such as H5N1 viruses), and facilitate the early identification of potentially zoonotic pathogens that are yet to be characterized. Expanding this pillar to include routine surveillance of apparently healthy individuals and individuals with severe febrile respiratory syndromes of unknown aetiology may help us to better understand the epidemiology of known or novel infectious agents that are circulating in the population and their potential outbreak risk.

Fig. 2: Four pillars of genomic surveillance.
figure 2

Coordinated SARS-CoV-2 genomic surveillance in humans, the environment, and domestic and wild animals is critical to inform policy development towards an early warning system for detecting outbreaks in humans and animals. AMR, antimicrobial resistance.

Full size image

There is a bigger gap in surveillance among the other two pillars (domestic animals and wildlife), despite the far-reaching effects of animal reservoirs that we have described, for both SARS-CoV-2 and other pathogens at risk of zoonotic emergence. It is crucial that we build on and integrate existing surveillance networks and bolster surveillance pillars that are less established. Such integration requires the global adoption of standardized frameworks such as the Tripartite Zoonotics Guide and the recently released Quadripartite One Health Joint Plan of Action, engagement of stakeholders, and linkages with other One Health preparedness and response activities. Emphasis should be placed on the prioritization of pathogens and populations, biocontrol strategies, early warning systems, and the development of decision-support tools and triggers for action. Clear guidance provides the basis for sustainable and versatile surveillance programmes that require strong, harmonized support at local, national, regional and global levels.

A key lesson we should all learn from the COVID-19 pandemic is that investing in infectious disease surveillance and risk prevention strategies can minimize the devastating global economic and social damages of disease outbreaks. Most importantly, only with the global coordination of existing surveillance efforts can we better manage the aftermath of the current pandemic and predict and thwart future ones.