TL;DR
Researchers used genomic epidemiology to reconstruct the transmission history of SARS-CoV-2 between humans and white-tailed deer in Iowa. The study confirms multiple human-to-deer "spillover" events followed by sustained deer-to-deer transmission, and while no direct "spillback" to humans was detected in this dataset, an alarmingly low case-finding rate (4.2%) suggests that such events could easily go unnoticed.
What Problem Does This Paper Address?
Since the start of the pandemic, SARS-CoV-2 has proven to be a generalist virus, infecting various animal species. White-tailed deer (Odocoileus virginianus) are of particular concern because they are highly susceptible, live in large populations across North America, and often interact with humans.
The central challenge is determining the "directionality" of transmission. If deer act as a persistent reservoir, the virus could evolve independently in deer populations and eventually "spill back" into humans as a new, potentially vaccine-evasive variant. This paper seeks to quantify these transmission dynamics: How often does the virus jump from humans to deer, how well does it spread among deer, and is there evidence that it is already jumping back?
Methods
The authors employed genomic epidemiology, a method that uses the mutations in viral genomes to reconstruct "who infected whom."
- Phylogenetic Reconstruction: They used BEAST 1.10.4 to build dated phylogenetic trees. By comparing the genetic distance between viral samples and their collection dates, they estimated the timing of the "Most Recent Common Ancestor" (MRCA) for different clusters.
- Transmission Inference: The core of the analysis used the R package TransPhylo. Unlike standard phylogenetics, TransPhylo can infer the number of unsampled cases in a transmission chain. This is critical because we only ever sequence a tiny fraction of total infections.
- Sensitivity Analysis: Because deer biology differs from humans, the authors tested whether different assumptions about "generation time" (the time between being infected and infecting others) or "sampling time" (the time between infection and being tested) would change the results.
Datasets
The study analyzed a specific cluster of genomic data from Iowa:
- Deer Sequences: 52 whole-genome sequences collected from retropharyngeal lymph nodes of hunted or roadkill deer (Sept 2020 – Jan 2021).
- Human Sequences: 141 human SARS-CoV-2 sequences from the same region and timeframe, sourced from the GISAID database.
- Reference Data: The deer sequences were part of a larger, previously published dataset of 283 deer samples from Iowa.
Key Insights and Ideas
- Deer as a Sustained Reservoir: The estimated common ancestor for the deer sequences dates back to March 2020—months before the high levels of infection were actually observed during the fall hunting season. This suggests the virus was circulating in deer populations long before it was detected.
- The "Unsampled" Iceberg: The model inferred a median of 4,394 unsampled cases within the transmission tree. This leads to a case-finding rate of just 4.2%. In other words, for every deer or human case sequenced in this study, there were likely 24 others that went undetected.
- Multiple Spillovers: The phylogenetic tree showed deer sequences "nested" within human clusters multiple times. This confirms that the virus didn't just jump to deer once; it jumped repeatedly in independent events.
- Biological Parity: Despite the differences in species, the estimated generation time for both humans and deer was roughly the same: a median of 5 days.
Key Results
- Reproduction Number ($R$): The inferred $R$ value was 1.05, indicating a slow but steady exponential growth of the outbreak across both species during the study period.
- Directionality: The study found clear evidence of human-to-deer transmission and deer-to-deer transmission.
- Spillback Evidence: No supported transmission pairs were identified moving from deer to humans. However, the authors emphasize that because the case-finding rate is so low (4.2%), spillback cannot be ruled out—it simply wasn't captured in this specific, limited sample.
- Geographic Connectivity: Transmission links were found between deer in geographically distant counties, suggesting that either deer move further than expected or, more likely, human travel is facilitating the spread of the virus to different deer populations simultaneously.
Limitations and Future Work
The primary limitation is sampling bias. The deer samples came from hunted animals or roadkill, which is not a random or representative sample of the population. Furthermore, the human genomic data available in public databases is often incomplete or delayed.
The authors argue that we currently lack "active surveillance" at the human-animal interface. To truly understand the risk of spillback, we need:
- Targeted Surveillance: Testing individuals with high deer contact (hunters, hikers, and wildlife officers).
- Longitudinal Studies: Tracking how the virus persists in deer populations outside of the hunting season.
- Independent Biological Data: More research is needed on how long SARS-CoV-2 RNA persists in deer lymph nodes compared to human nasal passages to refine transmission models.
Summary updated