Demographic Risk Factors Vary in the Invasion Front of Chronic Wasting Disease in West Virginia, USA

A wildlife pathogen invasion front (Langwig et al. 2015) can be defined as the initial period when prevalence is absent or increasing while the host population trend is stable or experiencing initial declines. Management goals may include pathogen elimination or slowing the pathogen’s spread. Chronic wasting disease (CWD) is an invariably fatal, transmissible, spongiform encephalopathy of cervids caused by the accumulation of misfolded prion proteins (PRP^CWD; Williams and Miller 2002). Localized elimination of CWD is difficult to achieve and not a probable outcome (Thompson et al. 2023) without a preemptive harvest strategy (Belsare et al. 2021). A proven method to slow CWD’s advance is lacking, compounded by knowledge gaps in the complex epidemiology of the disease including transmission dynamics driven by demographic risk factor variation (Uehlinger et al. 2016). With CWD having been detected in free-ranging cervids in 34 states in the US, four Canadian provinces, Norway, Sweden, and Finland (Ågren et al. 2021; U.S. Geological Survey 2024), a better understanding of the disease ecology of CWD is crucial to slow its spread.

Impeding this better understanding, CWD management is stymied by multiple factors: long incubation time, pervasiveness of PRP^CWD shedding, environmental persistence of PRP^CWD, and often reliance on postmortem testing. Typically, the disease incubates for a minimum of 16 mo before clinical signs appear (Williams 2005), but some prion protein genotypes prolong incubation (Ketz et al. 2022). However, infected animals are shedding PRP^CWD long before clinical signs appear (Plummer et al. 2017; Tennant et al. 2020). The PRP^CWD may be shed in saliva, urine, feces, blood, and antler velvet (Mathiason et al. 2006; Angers et al. 2009; Haley et al. 2011; Tennant et al. 2020) and can persist in the environment for over 2 yr (Miller et al. 2004); thus both direct (deer to deer; Miller and Williams 2003) and indirect (environmental; Miller et al. 2004) transmission may occur. To date, environmental transmission is the most uncertain aspect of CWD ecology, because the process has been documented only in captive experiments (Miller et al. 2004) and probably involves considerable environmental variation (Saunders et al. 2008), and PRP^CWD transfer from a fomite may be difficult to discern from direct contact without controlled conditions (Mathiason et al. 2009). Finally, CWD surveillance often relies on postmortem testing (Evans et al. 2014) that may miss very early infections (Haley and Richt 2017). Antemortem testing is crucial to help answer ecological questions, but the low sensitivity of antemortem rectal biopsies (Thomsen et al. 2012) limits research investigations. New antemortem and environmental testing are currently being investigated (e.g., Pritzkow et al. 2015; Plummer et al. 2018; Ferreira et al. 2021; Burgener et al. 2022).

After several decades on the landscape in the US, CWD has become an epidemic (Langwig et al. 2015) in some locations with population declines in white-tailed (Edmunds et al. 2016) and mule deer (DeVivo et al. 2017) in the western US. Sex is associated with prevalence variation in CWD outbreaks (Miller and Conner 2005; Grear et al. 2006; Osnas et al. 2009). Male white-tailed deer often maintain higher CWD prevalence than females (Jennelle et al. 2014; Samuel and Storm 2016), but this pattern is not always the case (Edmunds et al. 2016). Older males are hypothesized to encounter more CWD exposure risk because of larger home ranges, more dynamic social grouping, and increased conspecific contact during the mating season (Escobar et al. 2020). Additionally, understanding how landscape patterns influence deer social interactions might be used to help identify the riskiest social groups for transmission in an ecological context (e.g., Egan et al. 2023).

Most of the studies mentioned earlier occurred in the western and midwestern US. To that end, we performed a retrospective study of the invasion front of a CWD outbreak in a white-tailed deer population in Hampshire County, West Virginia, US. Here, CWD was first documented in 2005; was detected in surrounding states in subsequent years (Evans et al. 2014); and is now part of a growing regional CWD epidemic involving multiple outbreaks across Pennsylvania, Maryland, and Virginia, US (Evans et al. 2016). The detection of a single, <5% prevalence focus of CWD in Hampshire County afforded an opportunity to survey this invasion front in an eastern white-tailed deer population. We examined demographic risk factors (sex, collection year, age class, and adjusted body weight) to identify risk factor variation to better understand the spread of CWD in this ecosystem.

West Virginia Division of Natural Resources (WVDNR) personnel culled white-tailed deer on cooperating private lands for CWD monitoring within Hampshire County, West Virginia, US, during March and April of every year from 2006 to 2012, and again in 2017. Hampshire County comprises 1670 km² in the Ridge-and-Valley ecological region along the eastern slope of the Appalachian Mountains and is primarily farmland and deciduous forest. West Virginia’s first positive case of CWD (based on immunohistochemistry [IHC] at the Southeastern Cooperative Wildlife Disease Study [SCWDS], Athens, Georgia, USA, and National Veterinary Services Laboratories, Ames, Iowa, USA), a 2.5-yr-old road-killed male, was collected in Hampshire County in October 2004 as part of the state’s CWD surveillance efforts. The WVDNR initiated long-term sharpshooting monitoring in a 101-km² portion of Hampshire County (Fig. 1) in 2006 during March and April to reduce potential conflicts with the fall deer hunting season. To achieve as random a sample collection as possible, agency staff secured permission from approximately 50 landowners within the long-term monitoring area and opportunistically collected any deer regardless of age or sex if that deer occurred on those properties. As such, samples collected should be representative of deer age and sex distribution on the landscape at the time sharpshooting was conducted. Sharpshooting teams operated after dusk with suppressed rifles and spotlights. Deer were sampled annually until 2012, when sharpshooting shifted to sampling every 5 yr because of agency and landowner fatigue. We weighed each fresh carcass with a 453.5±0.2-kg beam balance scale (Model 41-31-32, Fairbanks Scales, Overland Park, Kansas, USA) and estimated the animal’s age by assessing tooth wear and replacement (Severinghaus 1949). Fetuses, kidney, and kidney-associated fat were measured during necropsy. Carcasses were transported in pickup truck beds (lined with heavy duty disposable plastic to minimize environmental contamination) to a nearby WVDNR district headquarters. We collected the retropharyngeal lymph nodes (RPLNs) and the obex from every animal, regardless of age, for CWD testing.

From 2006 to 2012 and during 2017, we tested 853 white-tailed deer from the CWD sharpshooting monitoring effort. Both ELISA and IHC staining of the RPLNs and/or obex were performed according to approved standard protocols at either SCWDS or the Wisconsin Veterinary Diagnostic Laboratory (Madison, Wisconsin, USA). Any ELISA initial reactor was confirmed with IHC, and we considered CWD status positive with PRP^CWD detection by IHC of either tissue. Excluded samples included an ELISA initial reactor with negative IHC tests and a sample with a negative obex IHC test and missing RPLN IHC test.

For these models, we generated posterior distributions using weakly informative priors and four Markov chains with 10,000 discarded warmup iterations and 10,000 test iterations. To identify the interaction combinations that best generalized to unseen data, we employed leave-one-out cross-validation (LOOCV) for model selection (Goodrich et al. 2020). In LOOCV, each data point is held out as a test set once, and the remaining data are used to train the model. This process is repeated for all data points, providing a robust estimate of the model’s performance. The LOOCV analysis was used to rank the interaction combinations based on their expected log pointwise predictive density, a measure of overall model fit. We evaluated the models with a diagnostic procedure previously described (Muth et al. 2018): no divergent transitions occurred, effective sample sizes were greater than 16,000, and was less than 1.1 for all explanatory parameters.

We used 89% and 50% equal-tailed Bayesian credible intervals (BCIs) and probability of direction (P_D; Makowski et al. 2019) to describe our results. Model estimates were exponentiated and reported as odds ratios. Covariate significance was assessed by whether the 89% BCI for the odds ratio excluded 1. We suspected that the combined model might mask sex-specific effects because of opposing temporal trends between males and females. To investigate this possibility and gain a clearer understanding of the best fit interaction effect, we split the best fit model by sex. To check for bias from the unbalanced age structure between the sexes, we also ran a <4-yr-old model from the best fit model.

Between 2006 and 2017, WVDNR sharpshooters collected CWD samples from 534 female and 319 male white-tailed deer. Mean (±SD) male age was 1.6±0.9 yr (range 0.8–4.8), and mean age in females was 2.9±1.7 yr (range 0.8–9.8). Mean male weight was 37.9±10.4 kg (range 15.0–62.6), and mean female weight was 41.4±9.5 kg (range 15.9–60.3). Mean male KFI was 8.8±6.5% (range 2.1–42.4), and mean KFI in females was 20.8±14.5% (range 2.2–70.9). Overall, 77% of female deer were pregnant, and the mean number of fetuses was 1.2±0.8 (range 0–3). The mean age of pregnant deer was 3.4±1.4 yr (range 0.8–8.8), whereas for nonpregnant deer mean age was 1.0±1.0 yr (range 0.8–9.8).

Over the study period, 9.9%±1.0% (SE) of the sampled deer were CWD positive (females, 9.9%±1.3%; males, 9.7%±1.7%; Table 1). The apparent prevalence in <1-yr-old deer was 4.3%±1.3% whereas it was 12.4%±1.4% for older deer. Apparent prevalence increased over time from a minimum of 2.4%±1.4% in 2007 to a maximum of 28.0%±4.7% in 2017 (Fig. 2). Sharpshooting collection was slightly more sensitive for finding positive animals compared with hunting (β=1.12, SE=0.35, P=0.02, R²=0.63). In a comparison of models with and without interaction terms, the best-fitting model (Table 2) included an age-weight interaction (Fig. 3a). The best model parameters fit the posterior predictive distributions well (Supplementary Material Fig. S1). In this model, year (mean odds ratio, 4.60; 89% BCI, 3.20–6.57; P_D, 100.0%), age (2.27; BCI, 1.08–4.69; P_D, 96.1%), and adjusted body weight (2.28; BCI, 1.07–4.91; P_D, 95.9%) were associated with increased odds of CWD infection (Fig. 3b, Supplementary Material Table S1), whereas the age-weight interaction (0.08; BCI, 0.02–0.35; P_D, 99.8%) had a negative relationship.

The best whole-herd model did not reveal a significant male sex effect (0.77; BCI, 0.35–1.71; P_D, 70.4%). This aligns with our observation of similar overall CWD apparent prevalence between males and females. However, this finding might have been masked by confounding trends in CWD risk for each sex. Separate models for males and females revealed distinct risk factors, supporting this possibility. Females were found to be driving the decreased risk from the interaction effect (0.12; BCI, 0.02–0.65; P_D, 97.9%; Fig. 3c), but males alone had increased age (4.21; BCI, 1.18–15.9; P_D, 96.5%) and adjusted body weight associated risk (5.88; BCI, 1.50–24.3; P_D, 98.3%; Fig. 3d). Collection-year risk in males (6.25; BCI, 3.22–12.4; P_D, 100.0%) was highest when compared with the whole-herd model (mentioned earlier) and females (4.18; BCI, 2.67–6.57; P_D, 100.0%). To confirm that the unbalanced age structure between males and females was not biasing the results, a <4-yr-old model had the same patterns as the whole-herd model (Supplementary Material Fig. S2).

In our study of West Virginia white-tailed deer sampled during the CWD invasion front in a mix of farmland and deciduous forest, we observed CWD apparent prevalence growth as a rapid, 10-fold increase to 28% over a period of 11 yr. Prevalence is probably still increasing in this population. With endemic equilibrium in Wisconsin, prevalence reached 50% in males and 36% in females (Samuel 2023). Increasing apparent prevalence over time was the strongest risk factor in this West Virginia outbreak (Fig. 3), and animals collected 6.2 yr later were 4.6 times more likely to be infected with CWD. Age and adjusted body weight increased odds of infection in males despite apparent prevalence being similar between males and females. An additional 20 kg of adjusted body weight increased CWD odds 2.3 times, and in the male-only model these odds more than doubled. Conversely, adult females appeared to drive the age-weight interaction where increasing weight in older animals decreased odds of CWD infection. Interestingly, these findings occurred against a backdrop of a skewed age structure between the sexes with fewer older males on the landscape (Table 1, Fig. 2b). The single oldest male was 4.8 yr old, whereas 96 females were older than 4 yr (9.8 yr maximum).

A combination of lower natural mortality of females and higher hunter harvest of males (Campbell et al. 2005) probably drives this dramatic difference in longevity between the sexes in white-tailed deer in Hampshire County. Sharpshooting collections in the spring followed deer hunting seasons occurring from September to December of the previous year, and Hampshire County deer harvest data from 2005–2017 indicated that harvest favored antlered males (mean approximately 55% of the total annual harvest) over females (mean approximately 40% of the total annual harvest; E. Barton pers. comm.). After this male harvest, sharpshooting collection had a bias towards female deer, but this resulted from the population’s current sex ratio rather than sharpshooter preference. The observed age risk in males might potentially contribute to their shorter lifespan. However, without survival data, we cannot definitively determine the extent to which CWD infection itself shortens lifespan in males. A research model predicted that CWD infection in white-tailed deer males skewed the age class to younger animals (Foley et al. 2016), and CWD mortality was higher in males in Wisconsin (Samuel and Storm 2016) and influenced by more susceptible genotypes (Ketz et al. 2022). Despite their shorter lifespans, male deer showed a strong correlation between age and CWD risk, unlike females. This highlights the contrasting disease transmission dynamics experienced by the sexes on the landscape.

Male behavior probably explains their increased exposure risk (Escobar et al. 2020) with age and adjusted body weight acting as mediating factors. After reaching 1 yr old, male deer typically disperse from their natal range (Rosenberry et al. 1999) and form bachelor groups (Hirth 1977). These groups expose yearling males to individuals from diverse locations and riskier older age groups (Grear et al. 2006). During the breeding season, adult males further increase their risk by visiting numerous female groups, leading to close contact with potentially infected females (Grear et al. 2006). In contrast, female fawns generally remain within matrilineal groups composed of close relatives comprising up to four generations (Hawkins and Klimstra 1970), potentially offering some protection. These stable social groups limit contact with new deer outside the breeding season. However, this close association also presents a vulnerability: If a single member contracts CWD, the constant close contact within the group puts all members at a higher risk of transmission (Grear et al. 2010; Edmunds et al. 2016).

The limited mixing between female groups compared with males suggests a potentially patchier distribution of CWD infection across matrilineal groups. This female social behavior might contribute to the unexpected finding in our study: the lack of a significant age effect on odds of infection in females. An infected matrilineal group would probably encompass a range of ages, potentially obscuring the impact of age in our analysis. Additionally, the presence of multiple matrilineal groups remaining infection free by way of their smaller social networks within the sampled population might further dilute the age effect. This lack of an age risk in females is contrary to other studies (Grear et al. 2006; Samuel and Storm 2016), but prevalence has been found to decrease in older females (Osnas et al. 2009). Our findings may reflect disease dynamics during the invasion front, possibly related to stochasticity and lower environmental contamination. Future studies measuring variation in these sex-segregated group behaviors (e.g., cohesiveness, philopatry, and out-of-group contacts) in the context of CWD risk (e.g., Dobbin et al. 2023) would be critical to understanding this heterogeneity of CWD transmission on the landscape.

Body size is an often-studied demographic factor in wildlife research that forms second-order behavioral and physiological effects (Peters 1986). Mechanistically, adjusted body weight might manifest a synergistic effect on risky male behavior. Heavier male white-tailed deer generally obtain a higher social rank and are more likely to win fights with other males during the rut (Townsend and Bailey 1981), which would lead to successive fights with new challengers. Females are more likely to select these males for mating (Newbolt et al. 2017); thus, these males initiate more contacts with other deer stemming from mating and related behaviors. Larger body size in young male deer can increase their propensity for dispersal behavior (Haus et al. 2019). This potential for increased dispersal, even within our study’s age range of male deer, might translate into a greater chance of encountering infected individuals or contaminated environments, thereby compounding their risk of CWD exposure. A potential physiological explanation for this adjusted body weight risk is that larger males must consume more potentially contaminated food resources to maintain weight (Potapov et al. 2013). To make this mechanistic leap, advances in PRP^CWD detection in soil and plant samples (Pritzkow et al. 2015; Plummer et al. 2018) that can demonstrate infectivity to deer (in mice: Carlson et al. 2023) are a crucial next step in finding environmental sources and temporal changes of PRP^CWD contamination in the environment.

The decreased odds of CWD with the female-driven age-weight interaction was unexpected. This finding might be partially explained by the unique exposure history of older females in our study. These larger females might have belonged to early cohorts that were on the landscape when CWD prevalence was absent or very low (Fig. 2c). However, this observed difference in exposure history is unlikely to fully explain the interaction. There might be additional unobserved factors, potentially related to sex and life stage that contribute to the differential exposure between younger and older females mediated by adjusted body weight. To further strengthen the understanding of body size as a potential factor, a follow-up study could perform morphometric measurements on sampled animals to confirm our results that used adjusted body weight as a proxy for body size.

Understanding the factors influencing CWD transmission is essential for developing effective management strategies. Although retrospective studies have limitations, they can still provide valuable clues. Here, we found age and adjusted body weight risks in males that show potential for interventions. Our results echo past work in the Midwest where infection probability increased with age in white-tailed deer, with a stronger effect in males (Grear et al. 2006; Osnas et al. 2009; Jennelle et al. 2014; Samuel and Storm 2016). Intensive male harvest might help stabilize deer densities, decrease CWD prevalence, and limit spread on the boundaries of the epidemic (Jennelle et al. 2014). In polygynous mating systems such as that observed in white-tailed deer, males are not limiting for population growth unless the sex ratio becomes extremely skewed (Mysterud et al. 2002). A targeted male cull experiment could show if this approach holds merit, but sharpshooting can be controversial (Harper et al. 2015). With the advent of new antemortem tests being developed, testing and culling deer social groups under exceptional circumstances (e.g., in habituated urban deer populations) might enable a finer-scale intervention (Wolfe et al. 2018) and determine if removing larger, dominant males reduces transmission. Finally, for underpopulated deer herds experiencing a CWD outbreak, understanding the potential impact of CWD on female lifespan may be crucial for informing management decisions. It is known that CWD can reduce female survival (Samuel and Storm 2016), and antlerless harvest to facilitate genetic selection of reduced susceptibility genotypes in males decreased overall population growth (Ketz et al. 2022).

Our unique sampling scheme, employed in the context of this CWD invasion front in West Virginia, revealed an increased odds of CWD associated with both male age and adjusted body weight. The age effect in males remained significant even with a right-skewed (i.e., mostly low) age distribution, suggesting a prompt, elevated risk when males disperse from their maternal groups. Our finding that female deer exhibited a different pattern, with age not appearing as a significant risk factor but as an age-weight interaction being protective, runs contrary to other studies and requires further investigation to possibly identify unobserved mediating factors. Furthermore, our novel finding of an adjusted body weight risk factor accentuates the distinct social behaviors between sexes and potentially reflects a combination of behavioral and physiological mechanisms. We hope our findings will help to inform future experiments of social behavior CWD risk and management decisions.

We are extremely grateful to member state wildlife management agencies of Southeastern Cooperative Wildlife Disease Study in Alabama, Arkansas, Florida, Georgia, Kentucky, Kansas, Louisiana, Maryland, Mississippi, Missouri, Nebraska, North Carolina, Oklahoma, South Carolina, Tennessee, Virginia, and West Virginia, USA. Further, we thank SCWDS partnering federal agencies, including the U.S. Fish and Wildlife Service and the U.S. Geological Survey Ecosystems Mission Area. We thank SCWDS faculty and staff involved with CWD diagnostic testing, especially John Wlodkowski, John Bryan, and Richard Gerhold, as well as Dan Barr and other staff at the Wisconsin Veterinary Diagnostic Laboratory. We would also like to thank WVDNR staff who have worked tirelessly and diligently on the agency’s CWD surveillance efforts since 2002 and disease response and monitoring efforts since 2005. We thank our external reviewer, Sonja Christensen. Finally, we thank Paul Cross at the U.S. Geological Survey Northern Rocky Mountain Science Center for manuscript assistance. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. government.

Supplementary material for this article is online at http://dx.doi.org/10.7589/JWD-D-22-00160.