Conceived and designed the experiments: ESA DM DWS JWS RLC. Performed the experiments: DWS JWS RLC. Analyzed the data: ESA. Wrote the paper: ESA DM.
The authors have declared that no competing interests exist.
Gray wolves (
We analyzed sympatric wolf, coyote (
Of the pathogens we examined, none appear to jeopardize the long-term population of canids in YNP. However, CDV appears capable of causing short-term population declines. Additional information on how and where CDV is maintained and the frequency with which future epizootics might be expected might be useful for future management of the Northern Rocky Mountain wolf population.
Several high-mortality disease outbreaks among carnivore populations have demonstrated the potential for pathogen-induced population declines
Yellowstone National Park (YNP) is home to one of the largest, protected, intact suites of carnivores in the contiguous United States. Gray wolves (
Thus we sought to use long-term serological data to identify temporal, spatial, and demographic patterns of pathogen exposure among wolves, coyotes (
We assessed whether each of the pathogens of interest were enzootic or epizootic in the YNP canid community, and whether pathogen exposure varied by region of the park in relation to canid density. We investigated if behavioral differences between resident and transient coyotes, the latter potentially interacting with many more individuals across many different packs, and thus potentially at greater risk for pathogen exposure, contributed to differences in exposure risk. Host age was used primarily to examine temporal patterns of exposure, but it was also evaluated as a risk factor for recent or current infection with CHV and
Survival data were not available for coyotes or foxes. However, we did examine the relationship between pathogen exposure and wolf-pup survival. Furthermore, we used comparisons of exposure patterns among the canids to assess the likelihood of single versus multi-host pathogen transmission within YNP.
All wolves, coyotes, and foxes used in this study were handled in strict accordance with recommendations from the American Society of Mammalogists
YNP encompasses 8,991-km2 of protected land in northwestern Wyoming and adjacent parts of Montana and Idaho in the western United States (44°33′ N, 110°30′ W). YNP is surrounded by the Greater Yellowstone Ecosystem (GYE), a 60,000-km2 area that includes Yellowstone and Grand Teton National Parks, national forests, wildlife refuges, and a mosaic of state and private lands. YNP is mountainous (elevation range: 1,500 to 3,800 m), and its steep gradients in elevation, soil, and climate contribute to varied land cover, including riparian vegetation, shrubland, grassland, alpine meadows, and mixed coniferous forests.
We divided the park into two units, the Northern Range (NR) and the Interior, based on ecological and physiographical differences
YNP has an intact suite of terrestrial carnivores, including gray wolves, grizzly (
Since wolf reintroduction to YNP, the National Park Service has captured and radio-collared an annual average of 26 wolves (range 16–38) spanning all known packs in the park (mean packs sampled per year = 8, range = 4–12). Collaring efforts have generally targeted breeders and ∼50% of each year’s young, with an emphasis on maintaining contact with each pack. We darted wolves from a helicopter during November-March and anesthetized them using a 10 mg/kg dose of Telazol® (tiletamine & zolazepam). We fitted them with radio-collars (Telonics, Inc. Mesa, AZ), drew 6–8 ml of blood from the saphenous vein, and categorized the animals as pups (<12 months) or adults, with precise ages estimated from tooth wear
Staff from the Yellowstone Ecological Research Center (Bozeman, MT, USA) captured coyotes on the NR of YNP during three, multi-year sampling intervals (1991–1992, 1996–1999, and 2003–2005). Foxes were also captured on the NR of YNP, but trapping efforts were less intense and less frequent (1993, 1996, 2003, and 2005). Coyotes and foxes were captured from three regions (Lamar Valley, Blacktail Plateau, and Gardiner River Basin) spanning east to west on the NR inside YNP from September through October. Juvenile and adult coyotes and foxes were captured using padded, offset, center-swivel, foot-hold traps (Victor soft-catch, Woodstream Corp., Lititz, PA, USA) baited with species-specific urine lures. Sex, weight, condition, dentition, and body measurement data were collected for each animal. Individuals were classified as juveniles (0.5–1.5 yrs), young adults (1.6–4.9 yrs), or old adults (≥5 yrs) based on tooth wear
Monitoring of radio-collared coyotes permitted classifying individuals as residents (i.e., member of a territorial pack) or transients (i.e., solitary individuals, typically inhabiting an area overlapping one or more pack territories). However, we did not have detailed information on individual coyotes’ pack membership or territory location. Thus, exposure data from resident and transient individuals captured in the same region were assumed to be non-independent.
Sera from wolves (
Data from wolf and coyote pups were used only for animals ≥5 months old to avoid the influence of maternal antibodies
We identified wolf dens by tracking radio-collared adult females throughout April. Dens were monitored and pups counted weekly in May and June. Pup counts in the remote Interior were primarily conducted from airplanes. Aerial monitoring of NR dens was often supplemented with ground counts using spotting scopes. We estimated pups born per pack based on high counts observed between May-June. We also estimated pup survival per pack between May and December by calculating the proportion of pups in a pack still alive at the end of December based on weekly (at minimum) aerial and ground counts. Survival data were not available for coyote-pups and fox-kits.
To accommodate the available datasets and the biological differences between both the canid hosts and the pathogens, our analyses involved several different approaches outlined below.
The viral pathogens CPV, CDV, and CAV-1 generally produce long-lasting immunity in their hosts
By contrast, CHV, a herpesvirus, produces life-long infections characterized by periods of latency where the virus is present but does not provoke a strong immune response
Canids acquire
Positive and negative test results were analyzed using a logistic, generalized, linear, mixed model with random “pack,” or in the case of coyotes, “region” effects
Factor | Species | Number of categories | Categories | Model notation |
Year | Wolves | 11 | 1997–2007 | Year |
Coyotes | 9 | 1991–1992, 1996–1999, 2003–2005 | Year | |
Location | Wolves | 2 | NR, Interior | Location |
Resident status | Coyotes | 2 | Resident, Transient | Resident |
Age class |
Wolves | 3 | Juvenile (0.5–1.9 yrs), Young Adult (2–4.9 yrs), Old Adult (≥5 yrs) | Age Class |
Coyotes | 3 | Juvenile (0.5–1.5 yrs), Young Adult (1.6–4.9 yrs), Old Adult (≥5 yrs) | Age Class |
Age class was used as a factor in the analysis of canine herpesvirus and
Note the differences in factors and categories considered in the analysis of wolf and coyote data. Factors considered in the analysis of exposure to a particular pathogen are detailed in the text.
Wolf | Coyote |
Intercept | Intercept |
Year | Year |
Location | Resident |
Year + Location | Year + Resident |
Year + Location + Year*Location | |
Age Class |
Age Class |
Age Class + Year |
Age Class + Year |
Age Class + Location |
Age Class + Resident |
Age Class + Year + Location |
Age Class + Year + Resident |
Denotes the additional models considered in the analysis of canine herpesvirus and
Additive effects are expressed with a plus sign, and interactions between factors are connected with an asterisk.
Sets of candidate models for wolves and coyotes were evaluated for each pathogen using model-selection procedures based on Akaike’s Information Criterion, corrected for small samples (AICc)
Year (1995–2007) and location effects on wolf-pup survival were evaluated using a logistic, generalized, linear, mixed model with random pack effects and AICc model-selection procedures. Not all monitored dens were visible from the air or ground, so we did not always have pup counts to match the serological results from a particular pack to directly test the relationship between seroprevalence and survival. Therefore, while the strength of inference was reduced, we used regression analyses to examine the relationship between annual wolf-pup survival and annual wolf-pup seroprevalence (
Wolf CPV seroprevalence was 100% across all years, locations, pups, and adults (
Pathogen | Category |
Wolf Seroprevalence | Coyote Seroprevalence |
Pup/Juvenile | 100% (117/117) | R: 92% (24/26) | |
T: 100% (9/9) | |||
Adult | 100% (92/92) | R: 98% (45/46) | |
T: 86% (18/21) | |||
Pup/Juvenile | 91% (106/116) | R: 23% (6/26) | |
T: 11% (1/9) | |||
Adult | 96% (89/93) | R: 89% (41/46) | |
T: 71% (15/21) | |||
Total Population | 87% (181/209) | R: 51% (39/77) | |
T: 40% (12/30) | |||
Juvenile | 84% (137/164) | 23% (8/35) | |
Young Adult | 100% (39/39) | 52% (28/54) | |
Old Adult | 83% (5/6) | 87% (13/15) |
See
Seroprevalence reported for canine parvovirus (CPV), canine adenovirus type-1 (CAV-1), and canine herpesvirus (CHV). Coyote seroprevalence is divided into residents (R) and transients (T). The fraction of (positives/total samples) are noted parenthetically. CHV analysis included age class as a risk factor, so analyses were not divided by pups/juveniles and adults.
Pathogen or Survival | Species & Age | Best-Supported Models | K | -Log Likeli-hood | AICc | Δ | ||
CPV∼1 | 1 | 35 | 7.67 | 19.46 | 0.00 | 0.62 | ||
CPV∼1+Resident | 2 | 35 | 7.05 | 20.48 | 1.01 | 0.38 | ||
CPV∼1+Resident | 2 | 67 | 13.43 | 32.86 | 0.00 | 0.69 | ||
CPV∼1 | 1 | 68 | 15.21 | 34.49 | 1.63 | 0.31 | ||
CAV∼1 | 1 | 116 | 14.34 | 32.73 | 0.00 | 0.74 | ||
CAV∼1+Location | 2 | 116 | 14.34 | 34.79 | 2.06 | 0.26 | ||
CAV∼1 | 1 | 93 | 4.82 | 13.67 | 0.00 | 0.74 | ||
CAV∼1+Location | 2 | 93 | 4.82 | 15.76 | 2.09 | 0.26 | ||
CAV∼1 | 1 | 35 | 17.47 | 39.06 | 0.00 | 0.67 | ||
CAV∼1+Resident | 2 | 35 | 17.05 | 40.47 | 1.40 | 0.33 | ||
CAV∼1+Resident | 2 | 67 | 28.37 | 62.93 | 0.00 | 0.88 | ||
CHV∼1+AgeClass | 3 | 209 | 56.97 | 122.02 | 0.00 | 0.98 | ||
CHV∼1+AgeClass | 3 | 104 | 61.52 | 131.24 | 0.00 | 0.46 | ||
CHV∼1+Resident+AgeClass | 4 | 103 | 60.43 | 131.31 | 0.07 | 0.45 | ||
Neo∼1+AgeClass+Year | 13 | 202 | 53.10 | 136.14 | 0.00 | 0.28 | ||
Neo∼1+AgeClass | 3 | 202 | 64.17 | 136.42 | 0.29 | 0.24 | ||
Neo∼1+Location+AgeClass | 4 | 202 | 63.58 | 137.40 | 1.27 | 0.15 | ||
Neo∼1+Year+Location+ AgeClass | 14 | 202 | 52.73 | 137.75 | 1.61 | 0.12 | ||
CDV ∼1 | 1 | 114 | 42.46 | 88.97 | 0.00 | 0.65 | ||
CDV ∼1+Year+Location | 12 | 114 | 30.96 | 91.01 | 2.04 | 0.24 | ||
CDV ∼1+Year | 11 | 97 | 42.68 | 112.51 | 0.00 | 0.74 | ||
CDV ∼1+Year+Location | 12 | 97 | 42.45 | 114.61 | 2.11 | 0.26 | ||
CDV ∼1 | 1 | 35 | 4.743 | 13.61 | 0.00 | 1.00 | ||
CDV ∼1+Year | 9 | 69 | 27.36 | 77.77 | 0.00 | 0.67 | ||
CDV ∼1+Year+Resident | 10 | 68 | 26.88 | 79.63 | 1.86 | 0.27 | ||
Survival∼1+Year+Location | 14 | 723 | 363.10 | 756.79 | 0.00 | 0.82 | ||
Survival∼1+Year+Location+ Location*Year | 27 | 723 | 351.83 | 759.78 | 2.98 | 0.18 |
Models presented are those best-supported (Δ AICc <3) under the Information-theoretic approach
The best-supported models for wolf-pup and adult CAV-1 seroprevalence suggested a constant, very high probability of exposure (for both pups and adults: Pr[E] = 1, 95% CI: 0, 1), irrespective of year or location. Similar to CPV, the best-supported models for both juvenile and adult coyote CAV-1 exposure included a covariate for resident status. Although not significant, both juvenile and adult resident coyotes had greater probabilities of CAV-1 exposure (juveniles: Pr[E] = 0.19; 95% CI: 0.02, 0.70; adults: Pr[E] = 0.89, 95% CI: 0.76, 0.96) than their transient counterparts (juveniles: Pr[E] = 0.07, 95% CI: 0.03, 0.18; adults: Pr[E] = 0.72, 95% CI: 0.49, 0.87).
By contrast, wolf and coyote exposure to CDV varied annually. The best-supported models for CDV exposure suggested constant, low pup exposure and a year effect among adults. There was also marginal support for a model with year and location effects among both wolf pups (Δ AICc = 2.04, weight = 0.24) and adults (Δ AICc = 2.11, weight = 0.26) which exhibited a much better fit, particularly to the pup data. Among adult coyotes, the best-supported models included year and resident effects. While the best-supported model for juvenile coyote seroprevalence suggested constant, near-zero exposure, the model exhibited poor fit to 2 years of the data (i.e., 1999 and 2005;
Among wolves, data are divided by location (Northern Range [NR] and Interior), whereas coyotes were sampled only on the NR. Sample sizes are displayed above seroprevalences (see
The probability of CDV exposure among wolf pups was highest in 1999, 2002, and 2005, a pattern less clearly mirrored in the adult data (no year effect was significant) (
Both juvenile and adult coyote seroprevalence mirrored the temporal patterns among NR wolf pups; CDV seroprevalence was 100% in 1999 and 2005 among both age groups and 0% otherwise among juveniles (year effects were not significant;
The best-supported model for wolf exposure to CHV included a covariate for age class; however, wolf CHV seroprevalence was uniformly high (87%) and estimated probabilities of exposure were 1.0 for all three age classes (95% CIs, juveniles: 0.97–1.0; young adults: 0–1.0; old adults: 0–1.0). Among coyotes, the two competing models with nearly equal AICc weights suggested support for age class and resident status covariates in the risk of CHV exposure. The probability of CHV exposure among coyotes significantly increased with age class; juveniles had the lowest probability of exposure (Pr[E] = 0.20, 95% CI: 0.09, 0.38; seroprevalence = 23%), followed by young adults (Pr[E] = 0.81, 95% CI: 0.69, 0.89; seroprevalence = 51%), and old adults (Pr[E] = 0.96, 95% CI: 0.83, 0.99; seroprevalence = 87%). Although not statistically significant, resident coyotes had as much as an 11% positive difference in their probability of CHV exposure compared to transients (OR = 1.58, CI: 0.59, 4.18).
The four best-supported models for
Data are divided by age class: juvenile (0.5–1.9 yrs), young adult (2–4.9 yrs), and old adult (≥5 yrs). Sample sizes are displayed above seroprevalences. Where points overlap, numbers refer to juveniles, young adults and old adults, respectively.
There were too few fox samples to look for patterns of exposure, but we did find evidence for fox exposure to CPV, CAV-1, CDV, and CHV (
Year | CPV |
CAV-1 |
CDV |
CHV |
|
1993 | 3 | 2 | 0 | 0 | 0 |
1996 | 1 | 0 | 1 | 1 | 0 |
2003 | 3 | 2 | 3 | 0 | 1 |
2005 | 2 | 0 | 0 | 2 | 0 |
Canine parvovirus.
Canine adenovirus type-1.
Canine distemper virus.
Canine herpesvirus.
Small samples (
Between 1995 and 2007, we annually monitored an average of 9 (SD = 4, range = 2–15) wolf dens, or an average of 84% (SD = 14%) of reproducing packs. Although the best-supported model for annual wolf-pup survival included only year and location covariates, there was also model support for a year*location interaction (
Error bars are 95% confidence intervals and the numbers at the bottom of the graph represent the number of pups monitored/the number of packs observed (NR listed on top).
Annual wolf-pup CDV seroprevalence coincided with significant variation in annual pup survival on the NR (
The discussion that follows must be qualified by the fact that overall, our sample sizes were quite small. Small samples reduced our accuracy and precision in estimating exposure rates as well as our power to detect significant differences in exposure between groups, hence limiting the strength of our inferences based on our data. This is particularly apparent in our analysis of CDV exposure, where our supported models included many estimable parameters. However, in some cases, small samples were unavoidable. For example, in 1999 and 2005, pup survival was so poor that only 13 and 8 pups, respectively, were known to be alive on the NR, making it very difficult to capture pups in those years.
Our conclusions must further be qualified by the fact that our serological assays were not specifically validated or optimized for wolves, coyotes, or foxes. Without knowing the sensitivity and specificity of our tests for these species, we do not know the degree to which our positive and negative test results reflect true exposure status. We cannot rule out, for example, false positive results caused by non-specific antibody binding or exposure to closely related or cross-reacting viruses. However, there is good biological reason to believe that wolf and coyote immune systems would behave very similarly to those of closely related domestic dogs, for which the tests have been optimized. Previous serological work with foxes (including CDV assay validation via vaccination trials) suggests our titer cutoffs were appropriate for this species as well
Our findings suggest CPV, CAV-1, and CHV are enzootic, and that CDV is epizootic, within Yellowstone’s canid community. Among wolves,
Resident status of coyotes was the one variable that consistently emerged as a possible risk factor, regardless of age group or pathogen. Contrary to our original hypothesis, resident coyotes tended to have a greater probability of pathogen exposure than their transient counterparts. We had hypothesized that transient coyotes, whose home-ranges overlap multiple resident packs’ territories, might contact a greater variety of individuals and be at greater risk for pathogen exposure. However, it is possible that transients actually make fewer contacts with other coyotes compared to residents, whose frequent interactions among pack-mates may provide the best opportunity for pathogen transmission. An alternative explanation, at least among adult coyotes (≥1.6 yrs), is that the sampled residents tended to be slightly older (10% of transient adult coyotes were old adults [i.e., the other 90% were young adults] compared to the 28% of resident adult coyotes that were old adults), and that perhaps age, which should be positively correlated with exposure risk, was a confounder. From our study alone, it is not clear whether social status among coyotes has a true effect on the pathogen-exposure risk as hypothesized for other social-mammal systems
Following the emergence of CPV in the late 1970s, studies throughout North America have reported high seroprevalences for CPV among wild canids
Nearly all wolves also exhibited exposure to CAV-1, but there was no evidence for or against disease-induced mortality. CAV-1 seroprevalence has generally been high in other canid surveys
None of the studies that screened for CHV antibodies among wild canids found evidence for exposure (
The dynamics of highly immunizing, fast acting, epidemic-type pathogens such as CDV are challenging to decipher within the usual 3–5-year time frame of most wildlife studies. In these situations, reports of average seroprevalence, without regard to year or age of the sampled animal, can be misleading and of limited value for comparisons across different study sites and populations. In many of the serosurveys among coyotes
The supported CDV seroprevalence models suggested that (1) coyotes experienced CDV outbreaks in 1999 and 2005, (2) all wolves experienced CDV outbreaks in 1999, 2002, and 2005 (although 2000 and 2006 adult wolf seroprevalence was also high, these were likely individuals that were exposed in 1999 and 2005 and were thus positive upon capture the following year), and (3) NR wolves experienced a greater probability of CDV exposure than Interior wolves. This last finding was consistent our hypothesis that high wolf densities on the NR may result in higher inter-pack contact rates and thus higher levels of pathogen exposure compared to the less-dense Interior. Although we do not have Interior density estimates for the other canids, it is quite possible that coyote and fox densities are also higher on the NR than in the Interior, and thus higher canid densities in general may contribute to higher rates of wolf exposure observed on the NR.
However, as the seroprevalence data suggested, these aforementioned generalities obscured some potentially important differences in spatial and temporal CDV dynamics. For example, none of the Interior wolf pups handled in 1999 and 2005 had been exposed to CDV in contrast to the high levels of exposure found among the limited samples of Interior adults and NR adults and pups. These inconsistencies may be the result of small samples or differences in case-fatality rates across sampling locations. If all infected pups in the Interior died due to disease, those available for sampling would all be negative. It is also possible that the timing and point of disease introduction into YNP could account for these differences. CDV is generally thought to move quickly through populations as it is highly contagious, infected individuals shed virus for a relatively short time (mean duration of infectiousness = 14 days, maximum 90 days), and the virus rapidly degrades in the environment (within hours at ≥20°C, and within several weeks at 0–4°C)
Furthermore, if there was in fact a 2002 outbreak, it seems to have been confined to the Interior wolves; none of the NR pups and only a few of the NR adults were exposed that year. However, there may be reason to suspect false positives in this particular case. In 2002, the two positive Interior pups had antibody titers just over the positive titer cutoff value (Positive antibody titer: ≥16), in contrast to marked increases in the titers observed among NR pups in 1999 and 2005 (
The wolf-pup data suggested low rates of seroconversion between the discrete outbreak years of 1999, 2002, and 2005. Once a wild or domestic canid is infected with CDV, the animal either recovers rapidly (mean time from infection to recovery [including latency and infectiousness] = 21 days, maximum 120 days) with life-long immunity or dies
Although a thorough analysis of factors influencing wolf-pup survival would evaluate multiple hypotheses such as population density and food availability, the strong negative correlation between NR CDV seroprevalence and NR wolf-pup survival supports the hypothesis that CDV may have contributed to high NR pup mortality in 1999 and 2005. Although ≥8 young wolf-pup carcasses were located in 2005, all were too degraded for CDV isolation. We found several pup mandibles (
We found no relationship between Interior CDV seroprevalence and Interior wolf-pup survival. Aside from the hypothesis that the timing of CDV introduction into the Interior either happened to be too early (e.g. 1999 and 2005) or too late (e.g. in 2002) to cause significant pup mortality, other plausible explanations for this lack of relationship include 1) that there was no CDV outbreak in 2002, and thus insufficient variation in exposure to detect a relationship with survival, and 2) that we failed to detect pup mortality due to bias in our sampling methods. The Interior packs’ dens were remote and only visible from the airplane, and thus, on average, we made our first pup observations and obtained our first high counts of pups over a month later than those made on the NR (First pup observations, NR: μdate = 5/24, sd = 21 days, Interior: μdate = 6/26, sd = 27 days; First high pup count, NR: μdate = 6/19, sd = 34 days, Interior: μdate = 7/22, sd = 37 days). Because much microparasite-induced (e.g., viruses and bacteria) pup mortality takes place following weaning (i.e., at 10–12 weeks of age) in late June through early July, it is quite possible that we failed to detect most Interior pup mortality, yielding artificially high survival estimates.
The results of two previous studies on pathogen exposure in YNP carnivores further support the patterns of CDV exposure that we observed in wolves and coyotes. Gese
These correlations among multiple hosts suggest regular CDV spillover but might also suggest multi-host transmission contributing to CDV persistence in the larger region. Domestic animals cannot be ruled out as a reservoir for CDV. However, reported CDV cases in Montana’s domestic animals are uncommon, with 18 possible cases recorded between 1994 and 2008 (Montana Veterinary Diagnostic Lab, Bozeman, MT, USA, unpublished data). Furthermore, while the percentage of local domestic animals vaccinated for CDV is unknown, it is probably safe to assume that the unvaccinated population of dogs and cats is too small to serve as a CDV reservoir
YNP and the GYE are not closed biological systems. On an annual basis, an unknown number of visitors from around the U.S. bring their pets to YNP and the GYE. There is currently no proof of dog health or immunization required for entry into the national parks. Visiting domestic animals certainly constitute a plausible route for new or emerging pathogens (particularly those that are vector-borne or indirectly transmitted) to enter into local, wild canid populations. Furthermore, YNP is a small fraction of the overall GYE, and pathogen dynamics within YNP may be in part a product of much larger-scale dynamics driven by inter-connected canid and carnivore populations in the Rocky Mountains.
In summary, the constant high canid exposure to CPV, CAV-1, and CHV in YNP suggest that these pathogens are established in the wolf and coyote populations and that they are unlikely to be causing acute mortality in their hosts
Epidemiological characteristics of selected canid pathogens. Data are largely based on the study of domestic dogs.
(0.04 MB DOC)
Models of disease seroprevalence and survival considered and evaluated for Yellowstone National Park's canids. Response variables include seroprevalence of canine parvovirus (CPV), canine adenovirus (CAV-1), canine herpesvirus (CHV), Neospora caninum (Neo), and canine distemper virus (CDV), as well as wolf-pup survival (Survival). Covariates are detailed in
(0.14 MB DOC)
Wolf and coyote canine distemper seroprevalence and associated 95% score confidence intervals. Sample sizes and the number of packs (for wolves) or regions (for coyotes) sampled are noted.
(0.14 MB DOC)
Mean wolf antibody titers to canine distemper virus in Yellowstone National Park, 1997–2007. Mean log2(antibody titers) are displayed with corresponding 95% confidence intervals for Northern Range (NR) and Interior pups (A) and adults (B).
(7.71 MB TIF)
The authors would like to thank staff from the National Park Service and the Yellowstone Ecological Research Center for their assistance with data collection, analysis, and interpretation. We thank D. Andersen, R. Singer, S. Weisberg, M. Atkinson, E. Dubovi, E. Bangs, P.J. White, and P. Cross for their insightful discussions and comments on early versions of this manuscript and B. Layton and the Montana State Livestock Diagnostic Laboratory for access to their canine distemper records.