Spatiotemporal epizootiology of chronic wasting disease in Wisconsin deer

Abstract

Chronic wasting disease (CWD) is a fatal neurological disease of cervids caused by a misfolded protein with no vaccines or other cures to prevent infection and death. In the past decade, CWD has been recognized as one of the 10 most important challenges facing wildlife management. This paper evaluates the temporal and spatial patterns of CWD prevalence in southern Wisconsin white-tailed deer during the past 20 years. In most CWD areas, prevalence progresses from introduction to enzootic equilibrium in 15–20 years. In some monitoring areas, the disease grows at a much slower rate and remains below the equilibrium levels. Outbreaks are characterized by four distinct epizootic stages. Disease establishment (Stage 1) depends on the distance from a CWD focus. High rates of incidence and prevalence growth (Stage 2) are followed by slower growth as the relative number of susceptible deer declines (Stage 3). The rate of prevalence increase over time is higher in more forested ecoregions and in males (0.48 vs. 0.43) but similar in females (0.43 vs. 0.37) and yearlings (0.44 vs. 0.39). Habitat features, acting as surrogates for deer behavior and contact, may affect the rate of prevalence growth at a landscape (ecoregion) scale. Additionally, prevalence may be affected by deer management practices that favor higher deer abundance and more mature males. Finally, enzootic equilibrium (Stage 4) is higher in males (0.5), followed by females (0.36) and then yearlings (0.26). These equilibrium prevalence levels are high enough to have significant population impacts, reduce the abundance of mature males, and facilitate CWD spread by infected yearlings. Epizootic patterns suggest that CWD transmission has been frequency-dependent and primarily driven by direct contact with infected deer. Evidence for a meaningful change in the epizootic pattern from a 10-year management program to reduce deer abundance is lacking. The trajectory of CWD dynamics in Wisconsin suggests rapid growth in regional prevalence following introduction and increased spread across the landscape.

INTRODUCTION

Existing and emerging pathogens can negatively impact wildlife population dynamics (Anderson & May, 1978; Daszak et al., 2000), affect host geographic distributions (Chafin et al., 2020; LaPointe et al., 2012), and drive host natural selection (Ketz et al., 2022; LaCava et al., 2021; Monello et al., 2017; Van Blerkom, 2003). Spatial and temporal disease transmission or infection patterns can provide clues to factors affecting disease dynamics (Ward & Carpenter, 2000). For example, common epizootic patterns can provide compelling evidence about underlying disease processes that contribute to transmission and geographic spread. Heterogeneity in space or time can also give insights into how local environmental, biological, and other factors affect disease processes. These spatial and temporal patterns offer a powerful means for generating and testing hypotheses about disease dynamics and lead to potential strategies for disease management.

Chronic wasting disease (CWD) is a neurodegenerative disease in deer (Odocoileus sp.), elk (Cervus canadensis), moose (Alces alces), and reindeer (Rangifer tarandus) (cervids) caused by misfolded prion proteins (PrP^CWD). When PrP^CWD is ingested, it can bind to normal prions (PrP^C), converting PrP^C into pathogenic form. Over time, PrP^CWD becomes widely distributed in nervous, lymphatic, blood, and muscle tissues in infected cervids and is shed via urine, feces, and saliva shortly after infection (Haley et al., 2009; Henderson et al., 2015; Mathiason et al., 2006; Plummer et al., 2017; Tamgüney et al., 2009). The relatively slow accumulation of PrP^CWD in lymphoid and nervous systems leads to a prolonged incubation period of 2+ years depending on cervid species, genotype, and sex (Hoover et al., 2017; Ketz et al., 2022; Robinson, Samuel, Johnson, et al., 2012; Samuel & Storm, 2016). The incubation period is followed by clinical signs of infection and ultimately death (Williams & Young, 1980). Due to the prolonged incubation period and generally low rates of infection, deer typically show increased prevalence of infection with age, and infection is higher in males than in females (Grear et al., 2006; Ketz et al., 2022; Miller & Conner, 2005; Samuel & Storm, 2016). Vertical transmission is generally unknown in free-ranging North American cervids, so newborn animals are usually uninfected (but see Bravo-Risi et al., 2021; Nalls et al., 2021). Given the prolonged incubation period, yearling (<2 years old) white-tailed deer (O. virginianus) can become infected but are unlikely to die from CWD before reaching adulthood (Samuel & Storm, 2016; Storm et al., 2013). Adult male white-tailed deer have much higher infection rates than adult females, and yearling infection rates are lower than those in adults (Ketz et al., 2022; Samuel & Storm, 2016). It appears that adult females are primarily infected with CWD from contact with genetically related females in their social group (Cullingham et al., 2010; Grear et al., 2010). In contrast, little is known about male infection, and it has been hypothesized that adult males acquire CWD from females during mating, by more contact with environmental sources due to larger home ranges, or from contact with other males during the non-breeding season when bachelor groups of males travel together and generally follow the same movement patterns (Grear et al., 2006; Samuel & Storm, 2016).

Disease transmission occurs by direct contact (animal-to-animal) with infected animals and indirect (environmental) routes (Miller et al., 2004) through ingestion of infectious prions in soil, water, feces, artificial feed, and potentially on plant surfaces or within plant tissues (Ketz et al., 2019). However, the relative importance of direct and environmental transmission during CWD epizootics is not currently understood (Ketz et al., 2019). Environmental transmission is hypothesized to increase during the later stages of an epizootic as long-lived prions accumulate in the environment (Almberg et al., 2011) and become an important additional source of infection via indirect transmission (Ketz et al., 2019).

Although the origin of CWD is unknown, it was originally documented in Colorado in 1967 and subsequently found in many states and three Canadian provinces (Ketz et al., 2019). Following the 2001 discovery of CWD, the Wisconsin Department of Natural Resources (WDNR) initiated a management program designed to control disease prevalence and spread, and hopefully eradicate CWD. This program involved a decade-long program (2002–2010) of population reduction by intensive harvest of all deer, especially female deer, in areas where disease was present and by conducting disease surveillance throughout the state to determine the distribution of CWD (Wisconsin Department of Natural Resources, 2010). Management activities included mandatory sampling and testing deer for CWD, a ban on baiting and feeding practices used to increase deer hunting or viewing opportunities, incentives to increase deer harvest, disposal of infected carcasses, a venison donation program, and sharp shooting in limited areas. Given the novel discovery of CWD in a high-density white-tailed deer population, the experience of other states with CWD, and computer modeling studies, the WDNR adopted a CWD eradication policy to eliminate the disease in the state's deer population (Wisconsin Legislative Audit Bureau, 2006). This CWD management program established a population goal of 2 deer/km² (which likely required a population reduction >70%) and was primarily applied to five of the established CWD monitoring areas: the Core, Northcentral Iowa County, Southeast Iowa County, Southwest Dane, and the Southeast area (Figure 1). Hunter resistance to the deer reduction program gradually increased during the management program and eventually resulted in its termination. As a result of the management program, hunter harvest appears to have reduced the deer population in this region by approximately 20%, but it has not been sufficient to cause a substantial, widespread population decline (Wisconsin Department of Natural Resources, 2010). This management effort generally assumed that CWD was a density-dependent disease where CWD infection is proportional to deer density and disease can be controlled by deer density reduction (Schauber & Woolf, 2003). However, subsequent research demonstrated that transmission is primarily frequency-dependent (Jennelle et al., 2014; Wasserberg et al., 2009) where CWD infection is proportional to the frequency of infectious deer and transmission can be controlled by reduction in disease prevalence. A 2006 assessment of the management program was conducted by the Wisconsin Legislative Audit Bureau (2006), but a subsequent assessment has not been conducted.

FIGURE 1. Eleven chronic wasting disease (CWD) monitoring areas (light blue areas) were established by the Wisconsin Department of Natural Resources in southern Wisconsin. Northcentral Iowa County, A; Northwest Iowa County, B; Southcentral Iowa County, C; Southeast Iowa County, D; Southeast Richland, E; Southwest Sauk, F; Core, G; Southwest Dane, H; Baraboo, I; Northeast Grant, J; Southeast, K. Areas A, D, G, H, and K (outlined in black) were managed to reduce deer abundance and CWD prevalence from 2002 to 2010. Areas A, B, C, D, G, H, and J are considered part of the Southwest Savannah ecoregion. Areas E, F, and I are considered part of the Western Coulee Ridge ecoregion. Area K is considered part of the Eastern Grassland Prairie ecoregion.

Because CWD in free-ranging cervids was not recognized until the 1980s, limited data have been available for studying temporal, spatial, and demographic influences on epizootic dynamics in wild deer populations (Miller & Conner, 2005). The relatively recent expansion of CWD across landscapes, jurisdiction, and cervid populations provides an important opportunity to understand temporal, spatial, and demographic patterns of CWD epizootics that complement our knowledge of CWD transmission dynamics. A more comprehensive understanding of CWD epizootiology is needed to assess the implications of this expanding wildlife disease and craft management strategies. Moreover, insights regarding temporal and spatial trends would be useful in refining strategies for controlling epizootics in free-ranging wildlife (Miller & Conner, 2005). For most wildlife diseases, observed prevalence is the metric typically used for understanding trends and evaluating intervention efforts (Miller & Wolfe, 2021). Previous studies have used CWD prevalence to assess patterns of disease emergence and expansion (Heisey et al., 2010; Joly et al., 2006; Miller et al., 2000; Osnas et al., 2009), demographic prevalence and infection patterns (Grear et al., 2006; Hedman et al., 2021; Miller & Conner, 2005; Nobert et al., 2016; Samuel & Storm, 2016; Storm et al., 2013), genetic selection pressure (Ketz et al., 2022; Robinson, Samuel, Johnson, et al., 2012), geographic spread (Robinson et al., 2013; Smolko et al., 2021), modes of disease transmission (Jennelle et al., 2014; Wasserberg et al., 2009), the influence of environmental and habitat factors (Evans et al., 2016; Farnsworth et al., 2005; Nobert et al., 2016; Robinson et al., 2013; Ruiz et al., 2013; Storm et al., 2013; Walter et al., 2011), and to assess the potential effectiveness of CWD management strategies (Conner et al., 2007, 2021; Manjerovic et al., 2014; Miller et al., 2020; Varga et al., 2021; Wolfe et al., 2018). The goal of this study was to investigate spatial and temporal patterns of CWD prevalence in Wisconsin deer during the first 20 years after discovery. Temporal prevalence was used to identify fundamental characteristics of CWD outbreaks, determine how those characteristics vary spatially, assess the role of ecological factors in epizootics, and consider the role of management actions.

METHODS Study area and data

In 2001, three adult male white-tailed deer harvested in western Dane County, Wisconsin, tested positive for CWD. This discovery launched an intensive surveillance of harvested deer in southern Wisconsin that has continued since 2002. These surveillance efforts showed two focal areas of CWD infection in Wisconsin: one centered in southwestern Wisconsin and a second in northern Illinois extending into southeastern Wisconsin (Robinson et al., 2013). CWD is not evenly distributed within these affected areas. Disease prevalence is typically much higher near these two infection foci and declines with increasing distance from the center (Joly et al., 2006; Osnas et al., 2009). These spatial patterns are consistent with two separate disease introductions at some time in the past (likely >20 years prior to discovery) with growth in prevalence near the points of introduction and geographic spread to the current distribution (Heisey et al., 2010; Jennelle et al., 2014; Osnas et al., 2009; Robinson et al., 2013; Wasserberg et al., 2009).

The WDNR established 11 CWD monitoring areas within these two infected areas of the state. Ten areas were established in the southwestern portion of the state with one in the southeast (Figure 1). These areas were created to monitor disease patterns and trends over time, and to determine whether management efforts were controlling the distribution and intensity of CWD (Wisconsin Department of Natural Resources, 2010). The WDNR established a “Disease Eradication Zone” (Figure 1) where management used liberal hunting regulations including long hunting periods, increased harvest tags, removal of female deer to reduce abundance, and required testing of all harvested deer. The goal of this strategy was to eradicate CWD by reducing density to <2 deer/km². Harvested deer were tested for CWD using brain stem (obex) or retropharyngeal lymphatic tissue with immunohistochemistry and/or enzyme-linked immunosorbent assay (ELISA) (Keane et al., 2008). Each deer was classified by CWD status (CWD+ or CWD not detected), sex, and adult or yearling age group. Records from over 80,000 harvested white-tailed deer from these monitoring areas from 2002 to 2021 were used to assess spatial and temporal trends in CWD prevalence.

Statistical analysis

A nonlinear logistic regression model was used to evaluate temporal patterns of CWD prevalence within the 11 monitoring areas. A binomial response of the number of CWD+ deer and number of deer tested each year was regressed versus time (in years) since 2002, when surveillance was initiated. Separate preliminary analyses (Appendix S1: Tables S1 and S2) were done for each monitoring area and for adult males aged 2+ (hereafter males), adult females aged 2+ (hereafter females), and yearlings (sexes combined) within each monitoring area using SAS Proc NLMIXED (SAS Institute Inc., 2015). The typical logistic regression equation assumes that the binomial response asymptotes at 0 and 1. It was evident from the prevalence data and preliminary analyses that CWD prevalence in wild deer reaches an asymptote considerably below 1; therefore, a modified logistic equation was used to estimate the upper asymptote (enzootic equilibrium) of prevalence. The modified logistic equation is:[Image Omitted. See PDF]where P is the observed binomial CWD prevalence, βU estimates the upper asymptote (enzootic equilibrium) of prevalence, β0 is the estimated intercept that predicts P at 2002 (time = 0), and βt (thereafter called transmission rate) can be considered correlated with the constant disease transmission coefficient for a frequency-dependent disease (e.g., McCullum et al., 2001, box 1, eqs. 3 and 4) among deer. In this model, CWD transmission rate (βt) is constant, but the rate of change in disease prevalence shifts over time because the relative abundance of susceptible deer declines as the epizootic stage progresses. Management actions that reduce the rate of transmission between infected and susceptible deer will reduce βt and delay epizootic progression. When prevalence is low, CWD transmission will primarily be to susceptible deer, and disease incidence (number of new cases) is high, causing prevalence to increase rapidly. As prevalence increases, potential CWD transmission will be distributed to both susceptible and infected deer, reducing both incidence and the rate of prevalence increase. Prevalence reaches the enzootic equilibrium as incidence is balanced by disease mortality of infected deer and recruitment of susceptible fawns, as a result prevalence stabilizes. In several monitoring areas, the estimated equilibrium prevalence values were unreliable, or the model did not successfully converge, indicating prevalence remains below an equilibrium level. To get reliable estimates for β0 and βt for these areas, the estimated βU for monitoring areas that reached equilibrium was used as a constant in Equation (1) (Appendix S1: Table S2). The predicted annual prevalence from Equation (1) was used to estimate the year when area-specific prevalence likely exceeded 0.01 for males and females. This model-based prevalence estimate provides a more consistent approximation of CWD establishment than observed prevalence because the number of deer tested varies considerably across areas and years. A rearranged version of Equation (1) was used to estimate the year when the predicted prevalence was within 0.01 of the equilibrium prevalence using the estimated β's for males and females in monitoring areas with a predicted equilibrium prevalence. For many of the monitoring areas, testing model fit to the observed prevalence data was difficult because expected cell frequencies were <5 due to low prevalence and/or limited testing. Therefore, visual inspection of the observed versus expected prevalence curves was also used to assess model fit.

Following this preliminary evaluation of separate areas (Appendix S1), a modeling approach was used to evaluate how several biological factors influence epizootic patterns of CWD prevalence. Hypotheses on CWD spatial and temporal patterns were evaluated for the 10 monitoring areas in southwestern Wisconsin (Figure 1: excluding the separate outbreak area K in southeastern Wisconsin) using SAS Proc NLMIXED. Candidate models included the effect of distance from the Core monitoring area (Figure 1: area G) to each monitoring area on the model intercept (β0); the effects of deer demographic categories (male, female, and yearlings), ecoregion classification, which included the Southwest Savannah (SWS) and Western Coulee Ridge (WCR) considered to be regions south (Figure 1: areas A–D, G, H, J) and north (Figure 1: areas E, F, I), respectively, of the Wisconsin River (Robinson et al., 2013, fig. 2D); and the demographics–ecoregion interaction on the rate of infection (βt) and the prevalence asymptote (βU). Akaike information criterion (AIC) was used to determine the relative model fit to the prevalence data with ΔAIC > 2 used for model selection. Biological factors were modeled as fixed effects and final model parameters were estimated using SAS Proc NLMIXED.

To evaluate whether the WDNR management program impacted CWD prevalence, predicted enzootic equilibrium levels (βU) and transmission rates (βt) for five managed areas (Figure 1: areas A, D, G, H, K) were compared with unmanaged areas. In addition, the observed prevalence data for managed areas were evaluated using a change point model:[Image Omitted. See PDF]where t1 = 1 and t2 = 0 for years 2002–2010, t1 = 0 and t2 = 1 for years 2011–2021, βt1 represents the transmission rate for the 2002–2010 management period, and βt2 represents the 2011–2021 post-management period. Yearlings were excluded because the limited number of samples/prevalence made estimation of βt inconsistent. AIC values for the single parameter model (βt) were compared with the change point model (βt1 and βt2) with ΔAIC values ≥2.0 used to select the best model. Changes in transmission rate (βt2 − βt1) were calculated separately for males and females. If the difference in rate of transmission had similar trends in both males and females (e.g., both higher in the late period), then the average difference between both sexes was estimated using a simple Bayesian model that included the estimated prevalence rate changes and variances for each sex as model priors.

The geographic rate of CWD spread in southwestern Wisconsin was estimated using linear regression of the distance (in kilometers) from the center of the Core monitoring area (where the disease was first discovered) to the center of each monitoring area (n = 9) and the center of each County with ≥0.01 CWD prevalence (n = 10) as the response variable. Time since 2002 when predicted CWD prevalence in male deer in monitoring areas and observed prevalence in Counties >0.01 was the predictor variable. The slope of the linear regression estimated the rate of CWD spread.

RESULTS CWD prevalence

Between 2002 and 2021, more than 80,000 deer were tested for CWD within the 11 monitoring areas (Table 1). Overall, apparent prevalence was 0.080 but prevalence was highest in adult males (0.149 = 3948/26,524), followed by adult females (0.058 = 1597/27,424) and then yearlings (0.036 = 980/27,330). Prevalence increased over time and varied considerably among monitoring areas (Appendix S1). Because previous studies found similar infection rates for male and female yearlings (Ketz et al., 2022; Samuel & Storm, 2016), yearlings were not separated by sex. Although CWD testing was initiated in all monitoring areas in 2002, the highest amount of testing occurred in the Northcentral Iowa County and Core monitoring areas, where CWD was discovered in 2001. Because the probability of CWD detection in deer is strongly dependent on sample size (Samuel et al., 2003), model-predicted prevalence was considered a more consistent estimate of CWD establishment (Stage 1). Predicted prevalence exceeding 0.01 in males occurred in 2002 in the Core (0.08), Northcentral Iowa County (0.025), and the Southeast monitoring (0.02) areas. For females, predicted prevalence exceeded 0.01 in the Core (0.03) and the Southeast monitoring (0.01) areas in 2002. Predicted prevalence exceeding 0.01 in the remaining monitoring areas varied between 2005–2012 and 2008–2012 for males and females, respectively (Table 2).

TABLE 1 Number of adult males, adult females, and yearling white-tailed deer tested and chronic wasting disease (CWD) positive for 11 CWD monitoring areas in southern Wisconsin from 2002 to 2021.

CWD monitoring area	No. positive of no. tested
	Males	Females	Yearlings	Total
Northcentral Iowa Countya	511/3506	240/3469	156/2967	907/9942
Northwest Iowa County	328/2482	145/2687	81/2683	554/7852
Southcentral Iowa County	244/1597	117/1645	53/1702	414/4944
Southeast Iowa Countya	265/2104	113/2305	48/2227	426/6636
Southeast Richland	440/1953	157/1949	78/1652	675/5554
Southeast Sauk	473/2442	219/2739	116/2139	708/7320
Corea,b	1076/5850	390/5356	280/5793	1746/16,999
Southwest Danea,b	183/2416	80/2602	52/2657	315/7675
Baraboob	198/1387	61/1628	51/1791	310/4806
Northeast Grantb	120/838	26/821	14/797	160/2456
Southeasta,b,c	110/1949	49/2223	51/2922	210/7094

Note: See map (Figure 1) for location of monitoring areas. a Monitoring areas where population management was used from 2002 to 2010 to reduce CWD prevalence. b Monitoring areas that are below equilibrium prevalence. c Southeast monitoring area is considered part of the CWD outbreak originating in Illinois.

TABLE 2 Year adult male and female prevalence (prev.) was predicted (Equation 1) to exceed 0.01 chronic wasting disease (CWD) prevalence and when CWD was detected (+) for 11 CWD monitoring areas in southern Wisconsin from 2002 to 2021.

CWD monitoring area	Year male+ predicted >0.01	Year male+ detected	Year female+ predicted >0.01	Year female+ detected
Northcentral Iowa Countya	<2002; 0.025 prev.	2002	2003	2002
Northwest Iowa County	2009	2006	2009	2006
Southcentral Iowa County	2006	2003	2008	2006
Southeast Iowa Countya	2005	2002	2010	2006
Southeast Richland	2008	2009	2009	2009
Southeast Sauk	2008	2004	2009	2005
Corea,b	<2002; 0.08 prev.	2002	<2002; 0.03 prev.	2002
Southwest Danea,b	2006	2003	2009	2006
Baraboob	2010	2003	2010	2005
Northeast Grantb	2012	2012	2012	2010
Southeasta,b,c	<2002; 0.02 prev.	2003	<2002; 0.01 prev.	2003

Note: See map (Figure 1) for location of monitoring areas. a Monitoring areas where population management was used from 2002 to 2010 to reduce CWD prevalence. b Monitoring areas that are below equilibrium prevalence. c Southeast monitoring area is considered part of the CWD outbreak originating in Illinois. Epizootic patterns

The preliminary analyses showed considerable variation in epizootic patterns among monitoring areas (Appendix S1). Enzootic equilibrium prevalence was predicted in 6 of the 11 CWD monitoring areas, indicating these areas are at or near Stage 4 (Appendix S1: Table S1 and Figure S1). Some of these areas reached equilibrium prevalence in less than 15 years following disease introduction. The remaining five CWD monitoring areas had prevalence patterns that prevented the estimation of an equilibrium level, even after 20+ years of disease presence in the Core and Southwest Dane monitoring areas. Two (Baraboo and Northeast Grant) of these monitoring areas are rapidly increasing toward equilibrium for males and females (Appendix S1: Figure S2) with transmission rates similar to those found in the six monitoring areas with a predicted equilibrium (Appendix S1: Tables S1 and S2). The Southeast monitoring area is part of a CWD outbreak centered in Illinois (Robinson et al., 2013) and should be considered a separate epizootic from the southwest monitoring areas. Unfortunately, comparable data for the CWD epizootic are not available from Illinois for spatial or temporal comparison.

Seven alternative models were used to evaluate potential ecological factors shaping CWD epizootics for eight monitoring areas within southwestern Wisconsin (Table 3). The Core and Southwest Dane (Figure 1: areas G and H) were removed from this analysis because they are considerably below an enzootic prevalence threshold and had lower transmission rates compared with other SWS monitoring areas (Appendix S1: Tables S1 and S2). The best model (Model 7) showed that distance from the original CWD outbreak location (Core monitoring area) (Joly et al., 2006; Osnas et al., 2009) was an important predictor of when CWD occurred at each monitoring area. Farther distance decreased (negative slope for distance) the predicted regression model intercept (β0) and predicted a later year of disease initiation (Table 4). Model 7 also showed that the CWD prevalence asymptote (βU) was best predicted by differences in males, females, and yearlings. Enzootic prevalence was highest for males (0.50, 95% CI = 0.47–0.53), followed by females (0.36, 95% CI = 0.32–0.40) and then yearlings (0.26, 95% CI = 0.22–0.29). Transmission rates (βt) were higher in males, than in females or yearlings, and differed between ecoregions (Table 4 and Figure 2). The transmission rate was consistently higher in the WCR compared with the SWS ecoregion for males (0.48 vs. 0.43), females (0.42 vs. 0.37), and yearlings (0.44 vs. 0.39) (Table 4). While transmission was significantly higher in the WCR (p < 0.02) for males, there was overlap between ecoregions for females and yearlings, perhaps due to lower precision of their estimated rates. Predicted prevalence for males, females, and yearlings shows that prevalence increases more rapidly (by 2–3 years) during the middle part of a CWD epizootic (Stages 2 and 3) in areas (Figure 1: areas E, F, I) north of the Wisconsin River (WCR) (Figure 2).

TABLE 3 Alternative nonlinear logistic regression models for chronic wasting disease epizootics in southwestern Wisconsin (Equation 1).

Model no.	Description	Parameters	AIC	ΔAIC
1	β0 = Dist, βt = Eco, βU = Dem × Eco	6	2641
2	β0 = Dem, βt = Dem, βU = Dem	9	2638	3
3	β0 = Dist, βt = Dem × Eco, βU = Eco	10	2034	604
4	β0 = Dist, βt = Dem, βU = Dem	8	2018	16
5	β0 = Dist, βt = Dem, βU = Dem × Eco	11	1980	38
6	β0 = Dist, βt = Dem × Eco, βU = Dem × Eco	14	1976	4
7	β0 = Dist, βt = Dem × Eco, βU = Dem	11	1973	3

Note: Model descriptions indicate biological factors included for each model parameter: Dem, (male, female, and yearling); Dist, linear model of distance from Core monitoring area to each monitoring area; Eco, Southwest Savannah (SWS) and Western Coulee Ridge. Two monitoring areas (Core and Southwest Dane) in the SWS ecoregion were removed because they remain considerably below equilibrium prevalence. ΔAIC indicates model reduction in fit from previous model number. Abbreviation: AIC, Akaike information criterion.

TABLE 4 Estimated chronic wasting disease epizootic model parameters, coefficients with SEs, and significance levels from Model 7 (Table 3) with Core and Southwest Dane monitoring areas removed.

Model parameter	Coefficients	Estimate	SE	t	p
β0 (initial outbreak year)	Intercept for Dist	−2.34	0.12	−19.5	<0.001
	Slope for Dist	−0.09	0.004	−21.3	<0.001
βU (prevalence threshold)	Male	0.50	0.012	40.0	<0.001
	Female	0.36	0.021	17.4	<0.001
	Yearling	0.26	0.019	13.2	<0.001
βt (transmission rate)	Male in SWS	0.43	0.012	35.1	<0.001
	Male in WCR	0.48	0.013	36.2	<0.001
	Female in SWS	0.37	0.014	27.1	<0.001
	Female in WCR	0.42	0.015	27.3	<0.001
	Yearling in SWS	0.39	0.018	21.2	<0.001
	Yearling in WCR	0.44	0.02	22.8	<0.001

Note: Dist, linear model of distance from Core monitoring area to each monitoring area; Eco, Southwest Savannah (SWS) and Western Coulee Ridge (WCR).

FIGURE 2. Predicted male, female, and yearling prevalence for 30 years following the introduction of chronic wasting disease (CWD). Prevalence predictions for ecoregions north (Western Coulee Ridge [WCR]) versus south (Southwest Savannah [SWS]) of the Wisconsin River for males, females, and yearlings based on transmission rates and prevalence asymptotes from CWD monitoring areas (Table 4).

Rate of spread

A linear regression of distance from the Core monitoring area to other monitoring areas and surrounding counties (response variable) versus the number of years since 2002 when male predicted prevalence exceeded 0.01 in monitoring areas or observed prevalence exceed 0.01 in counties (Figure 3) provided a good fit to the data (R² = 0.85). The slope of the linear model indicated that CWD was spreading from the Core monitoring area at about 5.0 km/year (95% CI = 4.0–6.0). The linear model intercept was 14.0 km (95% CI = 3.6–24.4), which generally corresponds to CWD being present at 0.025 prevalence in the Northcentral Iowa County monitoring area (20 km from the Core monitoring area) in 2002 (Table 2). These results are larger than the estimated rate of CWD spread (1.13 km/year) early in the Wisconsin epizootic (Jennelle et al., 2014) and suggest that the rate of CWD spread has increased over time, probably due to increasing CWD prevalence in dispersing yearlings.

FIGURE 3. Linear regression of distance (in kilometers) from the Core monitoring area (Figure 1: area G) versus year when predicted chronic wasting disease prevalence in male deer >0.01 for nine monitoring areas (Figure 1: areas A–F and H–J) or observed chronic wasting disease prevalence >0.01 for Green, Lafayette, Grant, Crawford, Vernon, Adams, Juneau, Sauk, Marquette, and Columbia counties. See Figure 1 for the map of locations.

CWD management

Only two of the four managed areas (Figure 1), Northcentral and Southeast Iowa County, had predicted prevalence equilibria. In both cases, the predicted equilibrium values for males and females were similar to those in the unmanaged areas (Appendix S1: Table S1). Both managed areas had overall transmission rates that were lower than unmanaged areas for males but were not statistically different for females (Appendix S1: Table S1).

To evaluate management impacts on CWD progression, separate transmission rates (βt) for males and females were compared for the management period (βt1, 2002–2010) and the following decade (βt2, 2011–2021) using a change point model (Equation 2). In Northcentral Iowa County, the transmission rate was higher after management for males (ΔAIC = 5.8) and lower but not significant for females (ΔAIC = 1.2). Males showed a higher transmission rate after management (Table 5) compared with the management period (+0.1, 95% Bayesian credible interval [BCI] = 0.05–0.15). The lower management period transmission rate could delay the time required to reach equilibrium by approximately 2–4 years (Figure 2). However, in Southeast Iowa County, transmission rates for both males (ΔAIC = 1.0) and females (ΔAIC = 1.6) were similar during and after management. The remaining three management areas (including the Southeast) have stayed below the predicted enzootic equilibrium levels for males and females. Whether these areas will eventually reach equilibrium levels similar to the other monitoring areas (both managed and unmanaged) or different equilibria remains to be determined. Estimating transmission rates for these three areas using Equation (2) required assuming a fixed equilibrium level (Table 5), which could make comparison with transmission rates to unmanaged areas biased. In the Core monitoring area, the post-management transmission rate was significantly higher for males (ΔAIC = 3.5) and higher but not significant for females (ΔAIC = 1.8). The combined differences for males and females indicated the average transmission rate was higher in the post-management period (+0.03, 95% BCI = 0.002–0.06). For comparison to other managed areas, the apparent effect of management in the Core is unlikely to have a meaningful impact on epizootic progression (Figure 2). For Southwest Dane, the transmission rates were lower post-management, but the change point model was not better than the single slope model; for males (ΔAIC = 0.1) and females (ΔAIC = 0.2). The average difference in male and female transmission rates was lower in the late period (−0.01; 95% BCI = −0.06 to 0.04) but statistically similar to the management period. For the Southeast monitoring area, which is part of the Illinois CWD epizootic, the post-management transmission rate was lower but not statistically different from the management period for males (ΔAIC = 0.0) and females (ΔAIC = 1.8). In addition, the average transmission rate in the post-management period was not significantly lower (−0.04; 95% BCI = −0.10 to 0.02).

TABLE 5 Estimated model (Equation 2) slope parameters βt1 (2002–2010) and βt2 (2011–2021), ΔAIC for single slope (βt, Equation 1) and two-slope (βt1 and βt2, Equation 2) models, and estimated slope differences (βt2 − βt1) for chronic wasting disease monitoring areas during management actions (2002–2010) and post-management actions (2011–2021).

Cohort	Parameter	Northcentral Iowa County	Southeast Iowa County	Core	Southwest Dane	Southeast
Male	βt1	0.271 (0.033)	0.294 (0.082)	0.107 (0.020)	0.291 (0.075)	0.119 (0.049)
	βt2	0.373 (0.048)	0.350 (0.044)	0.135 (0.010)	0.283 (0.029)	0.115 (0.024)
	ΔAIC	5.8	1.0	3.5	0.1	0.0
	βt2 − βt1	0.102 (0.047)	0.055 (0.056)	0.028 (0.016)	−0.009 (0.054)	−0.004 (0.035)
Female	βt1	0.376 (0.049)	0.554 (0.174)	0.099 (0.027)	0.332 (0.110)	0.169 (0.069)
	βt2	0.343 (0.043)	0.488 (0.092)	0.129 (0.013)	0.322 (0.043)	0.100 (0.039)
	ΔAIC	1.2	1.6	1.8	0.2	1.8
	βt2 − βt1	−0.033 (0.035)	0.066 (0.100)	0.030 (0.022)	−0.010 (0.078)	−0.069 (0.057)
Male + female averagea				0.029 (0.014)b	−0.01 (0.048)	−0.037 (0.034)

Note: For Core, Southwest Dane, and Southeast monitrng areas model slope parameters (βt) were estimated using a fixed value of βU = 0.49 and 0.34 for males and females, respectively. Abbreviation: AIC, Akaike information criterion. a (Male (βt2 − βt1) + female (βt2 − βt1))/2. Calculated using a Bayesian model when both males and females have the same trend. b Results that do not overlap 0.0. DISCUSSION

The CWD epizootics in Wisconsin vary temporarily and spatially (Figure 2 and Appendix S1) but are typically characterized by four epizootic stages (Figure 4), following those identified by Kaplan and Windsor (2021): introduction/establishment (Stage 1), acceleration of incidence and prevalence (Stage 2), compounding prevalence (Stage 3), and enzootic equilibrium (Stage 4). Stage 1 is characterized by low disease prevalence (≤0.05) and incidence in males and females with minimal impact on deer population structure and abundance. Routes of CWD infection are primarily direct contact among deer (Grear et al., 2006; Ketz et al., 2019; Storm et al., 2013). Disease spread is primarily by dispersal of a limited number of infected yearlings and local movement of infected males or females (Robinson, Samuel, Lopez, et al., 2012). Discovery of Stage 1 requires early detection of disease when prevalence is low (ideally ≤0.01) and has a limited geographic distribution. Management actions during Stage 1 should consider CWD eradication using intensive removal of high-prevalence age–sex cohorts, local depopulation of animals (e.g., AFWA, 2018; Ketz et al., 2019; Ytrehus et al., 2018) when possible, and comprehensive testing of harvested deer to assess disease prevalence and spatial distribution.

FIGURE 4. Four stages of chronic wasting disease epizootics in Wisconsin deer. Stage 1, introduction/establishment of disease. Stage 2, acceleration of prevalence in adults and yearlings. Stage 3, decline in incidence of disease. Stage 4, equilibrium prevalence.

Stage 2 is characterized by rapidly accelerating prevalence and incidence in adult deer and yearlings (Figure 2 and Appendix S1). Routes of infection are still primarily by direct contact among deer, with the potential for increasing deposition of environmental prions. There is rising potential for disease spread due to dispersal of more infected yearlings, especially males. Population impacts are still limited, but the magnitude of CWD-associated mortality in adult deer is beginning to increase. The initial phases of genetic selection are beginning (depending on the level of genetically neutral hunting pressure) with some age reduction in males and females (Ketz et al., 2022). Deer abundance is somewhat affected by lower survival of older females with CWD infection. Management efforts should focus on aggressive removal of high-prevalence age–sex cohorts to control prevalence and incidence (AFWA, 2018; Jennelle et al., 2014; Ketz et al., 2019) and prevent progression to Stage 3. Some population reduction may be warranted to reduce the number of infected yearling males that disperse.

During Stage 3, the rate of prevalence growth is beginning to decline (Figure 2 and Appendix S1) as incidence declines due to the relative decrease in susceptible deer. Routes of infection may include both direct and environmental prion transmission. There is an accelerating risk of CWD spread as infection in dispersing yearlings continues to increase toward an equilibrium level. In Wisconsin, CWD is currently spreading at approximately 5 km annually, probably because epizootic progression has increased the number of infected yearling dispersers. The estimated rate of spread also provides a potential guideline for conducting monitoring in surrounding areas to detect disease spread. Increased genetic selection pressure on males and females favors a higher frequency of less susceptible deer, depending on genetically neutral hunting pressure (Ketz et al., 2022). The increasing potential for environmental prion contamination could provide a long-term source of infection exceeding that from direct contact with CWD+ deer. Population impacts increase (likely approaching a measurable level) from lower survival of more CWD+. Deer abundance and trophy males decline due to lower survival of older CWD-infected males and females. Management needs to be aggressive and long term (likely >10 years depending on effectiveness), to reduce disease prevalence by removal of CWD+ deer with a goal of reaching Stage 1 levels (Jennelle et al., 2014).

During Stage 4, CWD prevalence reaches a stable enzootic equilibrium (Figure 2 and Appendix S1) created by the balance between incidence, lifetime of infected deer (Freeman & Hutchison, 1980), and birth of susceptible fawns. To mitigate population reduction from CWD requires reduced hunting rates on females to slow population declines. The number of trophy males will also be reduced due to disease mortality. Routes of infection continue to be via direct contact, but potential environmental routes could become increasingly important as environmental contamination is likely to accumulate due to prion shedding by CWD+ deer and carcasses. Increasing environmental prions may cause prevalence to increase beyond the current enzootic equilibrium as long-term environmental routes of infection become more important than direct contact. Disease spread has reached a maximum level with approximately 25% of yearling dispersers infected with CWD. Genetic selection pressure favoring less susceptible genotypes is high, depending on genetically neutral hunting pressure. A shift in genetic frequency to less susceptible genotypes may also cause prevalence to increase (Ketz et al., 2022) beyond the current enzootic equilibrium. Management should be similar to Stage 3, but population reduction may also be helpful when the goal is to reduce long-term environmental contamination and dispersal of infected yearlings to slow CWD spread.

In addition to the impacts of CWD on cervid populations (DeVivo et al., 2017; Edmunds et al., 2016; Galloway et al., 2017; Jennelle et al., 2014), the progression of CWD prevalence from Stage 1 to Stage 4 has many other implications. The cost of managing CWD epizootics, especially the cost of testing harvested animals, can quickly outpace the resources available to state management agencies and divert resources from other conservation programs (Wisconsin Legislative Audit Bureau, 2006). These costs can become especially burdensome as CWD continues to spread across an entire state or province. Other potential costs and impacts of CWD include detection and management activities, reduced hunter participation, loss of public support for agency missions, and loss of license fees and excise tax revenues that fund general wildlife conservation (AFWA, 2018). Hunters, for example, are concerned about the potential exposure to CWD-infected venison, which may result in a decline in hunter numbers, resulting in lost conservation and hunting-related economic revenue (Bishop, 2004; Vaske et al., 2004). The possibility of cross-species transmission of CWD to humans remains an important argument for management efforts to reduce CWD prevalence and spread across North America as a long-term public health measure (Osterholm et al., 2019; Samuel & Miller, 2017).

CWD in Wisconsin deer has reached or is nearing predicted enzootic equilibrium (Stage 4) within 15–20 years in 6 of the 11 monitoring areas (Appendix S1: Figure S1). The rapid growth of prevalence (≤20 years from establishment to enzootic equilibrium) is much faster than previously estimated for southcentral Wisconsin (Jennelle et al., 2014; Wasserberg et al., 2009). The disease also arrived later in two additional monitoring areas (Baraboo and Northeast Grant) where it will likely reach equilibrium prevalence within the next decade (Appendix S1: Figure S2). For monitoring areas that have reached (or are approaching) Stage 4, predicted equilibrium prevalence differs among sex and age groups (Table 4). Males reach the highest enzootic prevalence (0.50), compared with females (0.36), followed by yearlings (0.26) (Table 4). These relative differences in prevalence are similar to the sex and age patterns (Grear et al., 2006; Miller & Conner, 2005; Nobert et al., 2016) and rates of infection (Ketz et al., 2022; Samuel & Storm, 2016) previously reported in Wisconsin and elsewhere. At these equilibrium levels, disease-associated mortality is expected to result in a younger age distribution of males and females, and to lower population abundance due to disease-associated mortality in males and especially in reproductive females (Edmunds et al., 2016; Jennelle et al., 2014; Ketz et al., 2019, 2022; Samuel & Storm, 2016; Wasserberg et al., 2009).

The spatially consistent pattern of enzootic equilibrium in Wisconsin is characteristic of frequency-dependent CWD transmission, compared with an expected pattern of damped oscillations for a density-dependent transmission process (Almberg et al., 2011; Wasserberg et al., 2009). Almberg et al. (2011) evaluated alternative CWD prevalence patterns using a theoretical model to compare the potential importance of direct and environmental transmission. They concluded that the “only plausible simulations that captured high prevalence of 35%–50%” were those assuming frequency-dependent transmission or intermediate frequency dependence with indirect transmission of environmental prions with a half-life of approximately 8 years. If environmental prions are long-lived, then indirect transmission was also hypothesized to increase over time as this prion reservoir grows (Almberg et al., 2011). Potential accumulation of environmental prions is predicted to increase in areas with high and sustained disease prevalence, suggesting a future increase in indirect transmission and higher equilibrium prevalence levels if this prion reservoir supplements direct transmission from shorter lived infected deer (Almberg et al., 2011; Ketz et al., 2019). Although Miller et al. (2004) demonstrated that indirect transmission of CWD from a contaminated environment is feasible in captive deer, the relative importance of direct and indirect CWD transmission in wild deer is currently unknown. Captive deer have higher rates of direct and indirect transmission than wild deer (Keane et al., 2008), can have explosive prevalence growth from <0.05 to >0.60 in 3 years (Haley et al., 2021), and reach prevalence rates ≥0.80 (Keane et al., 2008). Similar patterns of rapid increase and high prevalence levels are not typical, and would likely be devastating, in wild deer populations. The current patterns of stable enzootic equilibrium in Wisconsin suggest that indirect transmission in wild deer may be less important than previously hypothesized or that more time is needed before the impact of environmental prions becomes apparent.

Modeling results in this paper show that rates of prevalence growth are consistently higher north of the Wisconsin River (Figure 1: areas E, F, I) (WCR ecoregion composed of rolling hills, about 40% forest cover, and small agricultural fields) and slower in the less forested areas south of the river (Figure 1: areas A–D, G, H, J) (SWS ecoregion composed mostly of agriculture or grassland with 10% forest cover) (Table 4) for males (0.48 vs. 0.43), females (0.42 vs. 0.37), and yearlings (0.44 vs. 0.39). CWD progression was also slower in the grassland-dominated Southeast Glacial Plains (EGP ecoregion composed of agricultural and urban areas with about 10% forest cover) (Figure 1: area K, Appendix S1: Table S2). Other studies also reported that CWD prevalence was associated with a variety of environmental conditions. In Colorado, mule deer prevalence was related to clay soils (Walter et al., 2011); however, this was not the case with Wisconsin (Robinson et al., 2013; Storm et al., 2013) or Illinois (Ruiz et al., 2013) white-tailed deer. In West Virginia, CWD prevalence in white-tailed deer using grassland, pasture, and crop fields was lower than those in the areas near urban development (Evans et al., 2016). Higher prevalence in mule deer in urban areas was also found in Colorado (Farnsworth et al., 2005). In the Canadian prairie parklands, CWD prevalence was highest in areas of high stream density associated with agriculture (Nobert et al., 2016). In previous Wisconsin studies, CWD prevalence in yearlings was higher in areas with more forest cover and forest edge density (Storm et al., 2013). Robinson et al. (2013) reported that CWD prevalence in southern Wisconsin and northern Illinois was associated with distance from each disease focus and higher in ecoregions with more forest cover. And CWD prevalence was higher in the WCR ecoregion than that in the EGP and SWS ecoregions, which had intermediate levels of prevalence (Robinson et al., 2013). In addition, genetic relatedness extended farther across the landscape in the highly forested WCR, was lower in the agricultural/grassland EGP, and least in the SWS (Robinson, Samuel, Lopez, et al., 2012). Genetic relatedness expanded spatially with both the amount and fragmentation of forested habitat. These results suggest higher density, increased philopatry, and likely higher contact of related social groups in more forested areas (Robinson, Samuel, Lopez, et al., 2012). Many researchers believe that landscape features affecting CWD transmission are proxies for deer concentration, grouping behavior, spatial overlap, and social contact (Ruiz et al., 2013; Storm et al., 2013), results supported by the differing extent of genetic relatedness across the landscape (Robinson, Samuel, Lopez, et al., 2012), and the propensity for infectious contact among related deer (Grear et al., 2010; Magle et al., 2013; Schauber et al., 2015).

In southwestern Wisconsin, more than 90% of the land is private with landowners/hunters typically managing local deer populations and trophy males. Higher deer density, higher density of harvested bucks, and low percentage of yearlings in the buck harvest occur in areas north of the Wisconsin River where CWD shows the highest growth in prevalence (Figure 2 and Appendix S2: Figures S1–S3). These patterns can occur where deer are managed for both high abundance and trophy males. In these areas, increasing the male harvest rate to lower the abundance of mature males may be helpful in reducing CWD prevalence (Conner et al., 2021; Jennelle et al., 2014) and epizootic progression. Further research is needed to determine how both habitat characteristics and deer management practices influence deer contact and CWD transmission. A better understanding of how these factors influence apparent landscape differences in the progression of CWD epizootics (Stages 1–4) could improve management strategies and our ability to predict the impact of CWD on deer populations.

Within both ecoregions, females and yearlings show similar patterns in transmission rate (Table 4). This pattern suggests a similar route of CWD transmission for females and yearlings; perhaps by direct contact between related females and yearlings (Grear et al., 2010; Magle et al., 2013; Schauber et al., 2015). However, reasons for higher transmission in males are currently unknown. Overall, the transmission rate in males is also predictive of the time required for a Wisconsin CWD epizootic to progress from Stages 1 to 4 (Appendix S1). These results are consistent with previous findings of frequency-dependent CWD population-level transmission driven by higher infection and disease prevalence in males (Jennelle et al., 2014; Ketz et al., 2019, 2022; Samuel & Storm, 2016). Further, they suggest that controlling male prevalence is likely a very important key to affecting CWD epizootics (Jennelle et al., 2014; Ketz et al., 2019). Recent studies found that increased harvest of male mule deer can lead to reduction in CWD prevalence and management actions promoting an abundance of mature male deer likely contribute to rapid growth of CWD prevalence (Conner et al., 2021; Miller et al., 2020) and higher transmission rates in males. A better understanding of male transmission would be beneficial to the development of alternatives to current CWD management recommendations focused on reducing male abundance to reduce CWD prevalence and transmission (Jennelle et al., 2014; Ketz et al., 2019, 2022; Miller et al., 2020; Samuel & Storm, 2016).

Although eradication of CWD is the preferred goal, this appears unlikely once the disease has become established (Stage 2). Instead, the secondary goal of CWD management should be a substantial reduction in disease prevalence to reduce transmission, disease spread, negative impacts on deer populations, and the potential risk of human exposure (Jennelle et al., 2014; Ketz et al., 2019; Miller & Fischer, 2016; Samuel & Miller, 2017). A reduction in CWD transmission would also delay epizootic progression, providing benefits for deer demographics and reduction in disease spread by infected yearlings. Management efforts by WDNR to reduce deer populations are a potential reason for lower transmission and spatial variation in CWD epizootics. The Core, Southwest Dane, Northcentral and Southeast Iowa County, and Southeast monitoring areas were part of the 10-year management program to reduce CWD prevalence (Figure 1). The largest impact of population management was found in Northcentral and Southeast Iowa County (Table 5) where management was expected to delay the progression of CWD by a few years; however, CWD prevalence in these areas was predicted to reach equilibrium after approximately 20 years, which is similar to areas outside the management zone (Appendix S1: Figure S1), results that are inconsistent with a substantial management impact. In contrast, the Core, Southwest Dane, and the Southeast management areas appear to be stuck in early stages (Stages 2 or 3) of the epizootic and generally have lower prevalence and transmission for males and females than other monitoring areas (Appendix S1). However, a comparison of the rate of male and female transmission between the management period and the following decade using a change point model showed only minor changes due to management actions. Population reduction in the Core monitoring area appears to have reduced the combined transmission rate for males and females, but this decrease was not sufficient to substantially reduce epizootic progression. Two managed areas in Iowa County showed small reductions in male transmission that may have delayed the enzootic equilibrium by 1–3 years. The Core monitoring area had evidence for an even smaller reduction in transmission, and two other management areas had no consistent reduction in transmission during a decade of management. In summary, there was limited evidence of a reduction in CWD prevalence and transmission from nearly a decade of deer population management. The WDNR management program, which assumed density-dependent transmission, was unlikely to substantially reduce prevalence growth within the CWD management zone; perhaps in part because population reduction was insufficient due to hunter resistance. Subsequent to the management program, it was shown that CWD transmission in Wisconsin was frequency-dependent, resulting in management recommendations that focus on reduction in CWD prevalence by removal of the deer with higher prevalence (e.g., adult males) rather than reduction in deer density (Jennelle et al., 2014). Future monitoring is needed to determine whether three of the managed areas will reach a CWD enzootic equilibrium level like the other managed and unmanaged areas.

There is a recognized need for increased surveillance and research to understand the epizootiology and management of emerging wildlife and zoonotic diseases (Stallknecht, 2007). Whether wildlife diseases have a population threshold for invasion and/or characteristics that enable enzootic persistence are critical topics in wildlife disease ecology and control (Jolles et al., 2021; Lloyd-Smith et al., 2005). However, lesser attention has been given to the temporal and spatial patterns of wildlife epizootics and enzootic equilibrium thresholds. This paper illustrates that enzootic equilibria are a key characteristic of CWD epizootics in Wisconsin white-tailed deer and these equilibria differ based on age–sex groups. Similar patterns seem likely for CWD epizootics in other cervid species and geographic locations. The four stages of epizootic progression found in CWD may also be characteristic of other frequency-dependent or chronic wildlife diseases such as brucellosis (Brucella abortus) in bison (Bison bison) and elk (Cotterill et al., 2018; Joly & Messier, 2004), and bovine tuberculosis (Mycobacterium bovis) for African buffalo (Syncerus caffer) in Kruger National Park (Cross et al., 2009) and white-tailed deer in Michigan (Schmitt et al., 2002). In other wildlife epizootics, disease characteristics such as reduced virulence or low transmission rates may also be important factors that facilitate pathogen persistence in carrier animals, which can become the source of periodic outbreaks (Almberg et al., 2022; Jolles et al., 2021). A better understanding of the spatial, temporal, and demographic patterns of wildlife epizootics is needed to evaluate transmission risks, potential population effects, spillover to other species, and the efficacy of disease management strategies. A significant limiting factor in understanding these patterns is the lack of long-term surveillance and monitoring of most wildlife epizootics.

CONCLUSIONS

At the landscape level, CWD is spreading across southern Wisconsin and northern Illinois from two disease foci, one in southwestern Wisconsin and another in northern Illinois, where disease outbreaks were initially discovered. The current rate of spread is approximately 5 km/year in southwestern Wisconsin, faster than estimated during early epizootic stages. Once CWD is established in a new area, there is a steady increase in prevalence over time, slowly at first then rapidly until an enzootic equilibrium is reached. However, the rate of change varies spatially, which is higher in forested areas north of the Wisconsin River than in less forested areas south of the river. In many areas of southern Wisconsin, CWD has rapidly progressed to enzootic equilibrium in ≤15–20 years. Within the Wisconsin CWD epizootics, equilibrium prevalence is characteristically higher in males, intermediate in females, and lowest in yearlings. The resulting equilibrium prevalence levels are expected to lead to lower survival, producing reduced abundance and a younger age structure (especially in males), lower abundance of reproductive females causing population decline, and faster disease spread from dispersal of infected yearlings. Transmission rates and epizootic progression are highest in areas with more forest cover and lower in areas with a more open landscape (more grassland and agriculture). Some areas in less forested (grassland-dominated) habitats have lower rates of infection and remain in earlier epizootic stages >20 years after establishment. These habitat characteristics are possible surrogates for the spatial extent of social contact among deer, which are stronger in forests and weaker in open habitats. In addition, these forested habitats in southern Wisconsin are characterized by higher abundance of deer and higher density of mature males, both factors likely represent deer management practices by private landowners and hunters to maintain more deer and encourage higher abundance of older/trophy males. Spatially consistent levels of equilibrium prevalence for males, females, and yearlings support the inference that CWD is a frequency-dependent disease that is primarily transmitted by direct contact. If indirect transmission from environmental reservoirs becomes a significant future source of infection, a secondary increase in the enzootic equilibrium prevalence seems likely. If environmental sources of infection are widely distributed on the landscape, rather than concentrated at specific sites, increases may be especially notable in males, which have larger home ranges and higher food consumption than females. Over time, CWD is expected to select for less susceptible genotypes, also resulting in a higher future equilibrium prevalence. The potential effects of that genetic shift on future deer demographics remain an important concern.

ACKNOWLEDGMENTS

This paper is dedicated to the memory of my colleague Dr. Joel A. Pedersen for his many contributions to the environmental science of prions. Many hunters in southern Wisconsin facilitated sampling of their deer. Staff from the Wisconsin Department of Natural Resources and the Wisconsin Veterinary Diagnostic Laboratory spent numerous hours collecting, processing, and analyzing deer samples. M. Foy and T. Hauge provided initial encouragement for this study. Discussions with C. S. Jennelle substantially improved the paper. T. Hauge, W. C. Turner, and two referees provided many useful suggestions that improved the paper. D. J. Storm provided information on the percentage of yearling deer in the buck harvest.

CONFLICT OF INTEREST STATEMENT

The author declares no conflicts of interest.

DATA AVAILABILITY STATEMENT

Data supporting this research are sensitive and are not available publicly. Graphical representation of the CWD prevalence data used in this paper is available on the Wisconsin Department of Natural Resources website: https://dnr.wisconsin.gov/topic/WildlifeHabitat/prevalence.html. Numerical representation of the prevalence data must be requested from the Wisconsin Department of Natural Resources by contacting [email protected].