Deer DNA Pinpoints Where Animals Came From in Chronic Wasting Disease Country

When a wild white-tailed deer tests positive for chronic wasting disease far from any known infected zone, wildlife managers face an urgent question: did the animal walk there, or did it arrive by some other route? A new study published in Ecology and Evolution shows that thousands of genetic markers scattered across the deer genome can answer that question with remarkable precision, placing most animals within 50 kilometers of where they were actually sampled. The work, led by researchers affiliated with Pennsylvania State University and the U.S. Geological Survey, offers wildlife agencies a genomic tool for tracing the origins of disease-positive deer across a five-state region of the American mid-Atlantic and Midwest.

Chronic wasting disease, a invariably fatal prion disorder first detected nearly 60 years ago in a captive Colorado facility, has now reached free-ranging cervids in 37 U.S. states, four Canadian provinces, and three Scandinavian countries. It remains the only self-sustaining prion disease known in wildlife, and while the risk it poses to human health remains speculative, its effects on infected deer populations can be devastating, driving measurable population declines in affected herds. Because white-tailed deer are the most abundant and most heavily harvested deer species in North America, and the primary species affected by the disease, their movements are considered the main driver of natural disease spread into new areas.

Management of the disease in wild populations relies on population reduction through agency culling and public hunting, intensive surveillance, and restrictions on moving deer carcass parts out of affected zones. All of these tools depend on knowing where deer come from and where they can travel. Yet the management units used to regulate hunting and surveillance are frequently drawn along administrative boundaries, roads, or topographic features rather than biological realities, meaning a single genetically coherent deer population may be split across several regulatory zones. When animals cross those artificial lines, monitoring assessments can go badly wrong.

The research team genotyped 1,920 wild white-tailed deer samples collected between 2014 and 2022 in Maryland, New York, Ohio, Pennsylvania, and Virginia, drawing on hunter-harvested animals, roadkill, research captures, and routine disease surveillance. After rigorous quality control that removed markers with more than 10 percent missing data and individuals with poor genotypes, the final dataset comprised 1,749 deer genotyped at 63,834 single nucleotide polymorphisms, or SNPs. That marker count dwarfs the fewer than 20 microsatellite loci typically used in previous deer studies, and the authors hypothesized that this genomic resolution would reveal finer genetic structure and enable more accurate origin inference.

To map that structure, the researchers deployed a battery of complementary methods, each sensitive to differentiation at different hierarchical levels. A principal component analysis showed samples loosely sorting by geography, with New York deer occupying one corner of the ordination space and Ohio deer another. A discriminant analysis of principal components, combined with a k-means clustering step, identified five genetic clusters whose distributions echoed state-level geography. A STRUCTURE-like Bayesian clustering algorithm using the sNMF method went further, resolving ten clusters, while fastStructure with a simple prior supported four. A spatially explicit approach, TESS3R, which incorporates sampling coordinates as priors, continued gaining support up to the maximum number of clusters tested, but individual membership values fell below 0.5 beyond five clusters, leading the authors to regard those results as inconclusive rather than evidence of ever-finer subdivision.

Crucially, the higher-numbered cluster solutions represented subdivisions of the lower-numbered ones, with the locations of genetic barriers remaining consistent across methods. Those barriers appear to correspond to real landscape features. In Pennsylvania, a cluster boundary tracks the Allegheny Front, the escarpment marking the western edge of the Valley and Ridge physiographic province. In eastern New York, a division aligns with the Hudson River and Interstate 87. In western Ohio, a barrier coincides with Interstate 75, and a cluster in southern Pennsylvania and northern Maryland is bounded by Interstates 70, 81, and 83, while the Susquehanna River may partly explain another border. Notably, some barriers weaken along their extent: the division between two clusters in New York is sharp in the central part of the state but dissolves in the north, where admixed individuals predominate.

These patterns carry direct implications for predicting disease spread. Recent chronic wasting disease detections in Ohio between 2020 and 2023 all occurred east of Interstate 75, which appears to constrain deer gene flow and, by extension, movement, suggesting natural westward expansion from those infections is unlikely. In New York, an introduction from Pennsylvania could plausibly spread through the area of one cluster, but further eastward movement seems less probable unless it follows the weaker northern corridor. The finding is timely, given a 2025 detection in a captive facility in Wayne County, Pennsylvania, bordering New York, that prompted creation of a new disease management area. In Pennsylvania itself, the strong genetic structuring means that spatially disparate detections of positive wild deer are unlikely to reflect natural expansion, since that would require multiple infected animals crossing apparent barriers to gene flow, warranting investigation of human-mediated spread.

The team then tested whether genetics could pinpoint individual origins. Categorical assignment tests, which sort individuals into predefined clusters, achieved perfect accuracy in cross-validation using only the strongest genetic representatives of each of the ten clusters, but accuracy dropped to an average of 71 percent, ranging from 41 to 100 percent, when admixed individuals were included as test subjects. The authors concluded the categorical approach is unsuitable for tracing wild deer origins, though it may retain value for distinguishing captive from wild animals. A continuous assignment method called SPA, which models allele frequencies as smooth two-dimensional functions of geography, failed outright, producing inferred locations wildly distant from sampling sites, likely because the method requires a monotonic allele frequency surface and cannot accommodate the multiple peaks created by genetic structure.

The clear winner was LOCATOR, a deep neural network method that learns genotype-location relationships without assuming any particular spatial model. Using an iterative leave-one-out design in which each deer was predicted against all others, LOCATOR placed the average animal within 29.9 kilometers of its sampling location, and 82 percent of samples within 50 kilometers. Among the less accurate assignments, nearly half involved deer whose sampling coordinates were only known to a county centroid, meaning the true accuracy may be better than measured. The method also flagged eight individuals whose inferred origins lay more than 150 kilometers from their sampling sites, including two collected under the same project with the same reported location, distances of 247 and 495 kilometers, a discrepancy suggesting coordinate errors or possibly escaped captive animals. The authors note that natal dispersal, overlapping generations, and polygynous mating all blur the genotype-location association, so some error is inevitable in a wide-ranging species, yet they consider the performance highly accurate given the spatial scale involved.

Beyond disease tracing, the study argues that genetic structure should inform how management units are drawn in the first place. Density estimates and harvest regulations assume that spatial units match the distribution and movement of the populations being monitored; when they do not, management actions can appear ineffective simply because deer redistribute across artificial boundaries. Because both gene flow and disease transmission depend on the same animal movements, genetically defined clusters offer a biologically grounded template for surveillance zones and for identifying areas within which the disease could spread naturally. The authors suggest that agencies sharing sample locations and genotypes across large regions could build a lasting reference resource, one that turns every disease-positive deer into a clue about how the fatal prion pathogen is really moving across the landscape.

Subject of Research: Using SNP-based genetic assignment to infer the geographic origin of white-tailed deer for chronic wasting disease management

Article Title: Inferring Geographic Origin of White‐Tailed Deer (Odocoileus virginianus) Using Single Nucleotide Polymorphisms in a Region Experiencing Long‐Term Effects of Chronic Wasting Disease

Article References: Fameli, A. F., Edson, J. E., Schuler, K. L., & Walter, W. D. (2026). Inferring Geographic Origin of White‐Tailed Deer ( Odocoileus virginianus ) Using Single Nucleotide Polymorphisms in a Region Experiencing Long‐Term Effects of Chronic Wasting Disease. Ecology and Evolution, 16(10), Article e74454. https://doi.org/10.1002/ece3.74454

Image Credits: AI Generated

DOI: 10.1002/ece3.74454

Keywords: white-tailed deer, chronic wasting disease, SNP genotyping, genetic assignment, population structure, prion disease, wildlife management, landscape genetics, LOCATOR, mid-Atlantic United States, gene flow barriers, disease surveillance