SEASONAL AND INTERSPECIFIC VARIATION IN THE PREVALENCE OF OPHIDIOMYCES OPHIDIICOLA AND OPHIDIOMYCOSIS IN A COMMUNITY OF FREE-RANGING SNAKES Open Access

Anthropogenic habitat loss, climate change, overexploitation, and introduction of invasive species are driving rapid declines in biodiversity (Ceballos et al. 2017; Rosenberg et al. 2019). These pressures can interact with infectious diseases to further threaten declining populations (Daszak et al. 2001; Jones et al. 2008; Fisher et al. 2012). Studying host-pathogen interactions in fungal disease systems may help determine the environmental drivers of disease and inform wildlife conservation planning (Smith et al. 2009; Herrera and Nunn 2019).

The fungus Ophidiomyces ophidiicola (formerly Chrysosporium, Allender et al. 2011; Ophidiomyces ophiodiicola, Sigler et al. 2013) can infect snakes and cause ophidiomycosis, formerly called snake fungal disease (Allender, Baker, Wylie, et al. 2015; Lorch et al. 2015; McKenzie et al. 2020b). Clinical signs of ophidiomycosis include localized crusting, discoloration, or displacement of the scales, as well as superficial pustules and subcutaneous nodules (McBride et al. 2015; Guthrie et al. 2016; McKenzie et al. 2020a). The fungus can invade underlying muscle and bone, as well as the internal organs (Allender et al. 2011; Lorch et al. 2015). Ophidiomycosis is associated with mortality in some individuals

(Allender et al. 2011; Dolinski et al. 2014; McBride et al. 2015), although effects on survival rates and population viability remain unclear (Davy et al. 2021; McKenzie et al. 2021).

Ophidiomycosis has been observed in captive and free-ranging snakes worldwide, for example, North America (Lorch et al. 2016; Allender et al. 2020; Davy et al. 2021), Europe (Allain and Duffus 2019), Australia (Paré 2016), Asia (Grinoi et al. 2021; Sun et al. 2021), and Russia (Ovchinnikov et al. 2021). Prevalence varies among host species and geographic sites, and severity ranges across individuals (Burbrink et al. 2017; Tetzlaff et al. 2017; Haynes et al. 2020), potentially reflecting variation in exposure to the pathogen, susceptibility, or both. Susceptibility may be affected by immune function, skin microbiome, and environmental conditions that favor fungal growth (Lorch et al. 2016; Allender et al. 2018). Variation in environmental conditions and the distribution of the pathogen across the landscape confound comparisons of disease prevalence among sites, but surveillance focused at particular sites also suggests interspecific variation in disease prevalence (Haynes et al. 2020).

Lesions consistent with ophidiomycosis are associated with emergence from overwintering sites (Lorch et al. 2016; Paré and Sigler 2016), potentially reflecting suppressed immune response during brumation or environmental conditions that favor fungal growth (Guthrie et al. 2016; Agugliaro et al. 2020; Haynes et al. 2020; McKenzie et al. 2020b). Understanding seasonal variation in snakes' exposure to O. ophidiicola and susceptibility to ophidiomycosis may inform more effective surveillance of the pathogen and disease (McKenzie et al. 2019). If O. ophidiicola grows faster in particular conditions, then the fungus, and a snake's risk of exposure, might be unevenly distributed among habitat types. For example, aquatic and semiaquatic snakes may be more likely to carry O. ophidiicola, compared with strictly terrestrial species (McKenzie et al. 2019). However, it is unclear if semiaquatic species of snakes are more likely to be infected by O. ophidiicola or if the fungus is more abundant in aquatic habitats.

We explored interspecific, temporal, and spatial variation in the prevalence of O. ophidiicola and ophidiomycosis at a site with eight co-occurring snake species. We hypothesized that snakes are more likely to be exposed to O. ophidiicola or more likely to develop ophidiomycosis during brumation, predicting that pathogen and disease prevalence is highest at emergence from overwintering and decreases through the active season. We also hypothesized that the pathogen is unevenly distributed among habitats, predicting that the frequency of pathogen detection on snakes would vary among habitats. Finally, we explored factors affecting detection probability of O. ophidiicola. We hypothesized that the fungal load collected during swabbing is proportional to the area swabbed, predicting that the fungus is more likely to be detected on larger snakes because of the greater surface area.

We conducted weekly coverboard transect surveys (?350 boards) from April 2013 to October 2018 at Rondeau Provincial Park, Ontario, Canada. We supplemented coverboard transects with targeted searches for snakes (eight species) in the various available habitats: anthropogenic sites (buildings, roads), dunes, forest, marsh, and oak savannah.

We wore a new pair of nitrile gloves to handle each snake. We also cleaned our processing tools with an alcohol or bleach solution between individuals (Rzadkowska et al. 2016). We recorded sex, snout-vent length (SVL), mass, and the habitat they were found in for each snake. Finally, when we encountered Pantherophis vulpinus (eastern foxsnake) >50 g, we implanted passive integrated transponder tags (HPT8 Pit Tag, Bio-mark, Boise, Idaho, USA).

Following early reports of ophidiomycosis in the US, including Allender et al. (2011), we checked snakes for lesions consistent with the disease, yielding confirmation of ophidiomycosis at Rondeau Provincial Park through quantitative PCR and histopathology. In 2017, we began targeted surveillance for the pathogen (explained soon).

After preliminary detections of O. ophidiicola at our study site (2013–16), we began targeted surveillance for the pathogen. Each snake captured in 2017–18 was swabbed ventrally and dorsally along the length of the snake's body, twice (to increase detection of O. ophidiicola, following Hileman et al. 2018), with two separate swabs (3? Sterile Standard Cotton Swab w/Semi-Flexible Polystyrene Handle, Puritan Medical Products, Guilford, Maine, USA). The two body swabs were placed in a single sample tube containing lysis buffer, providing us with a single body swab sample. When we observed skin lesions consistent with ophidiomycosis (Fig. 1), we collected two additional swabs targeting the lesions (lesion swabs), which were stored together in a single sample tube containing lysis buffer, providing us with a single lesion swab sample. Therefore, if a snake had observable lesions indicative of ophidiomycosis, that individual would have paired samples, a body swab sample plus a lesion swab sample.

For each snake captured 2013–18, we recorded the presence or absence of clinical signs of ophidiomycosis: gross skin lesions, including regional or local edema, crusts, ulcers, dysecdysis, and other noninjury type damage to the skin (McBride et al. 2015; Guthrie et al. 2016). When captures of snakes with lesions coincided with the availability of our team's veterinarian, we biopsied lesions to confirm the presence of fungal hyphae consistent with ophidiomycosis. Biopsies were collected under sterile surgical conditions by veterinarian Sean Egan at Egan Fife Animal Hospital (Chatham, Ontario, Canada). Tissue around the lesion was locally anaesthetized with 0.1 mL of lidocaine (Baxter, Missisauga, Ontario, Canada) infused under the skin. A 2.5-mm biopsy punch was used to sample the affected tissue and a small margin of normal tissue. Biopsy sites were closed with 4-0 polydioxanone (Ethicon, Markham, Ontario, Canada) simple interrupted sutures, and snakes were released within 24 h at the initial capture sites. Individuals with swabs containing the pathogen from which lesion biopsies contained fungal hyphae (confirmed through histopathology; McKenzie et al. 2020a) met the Canadian Wildlife Health Cooperative (CWHC) diagnostic criteria for ophidiomycosis (Davy et al. 2021).

We tracked P. vulpinus >360 g using radio telemetry for a study on habitat use that began in 2013. This tracking also enabled reexamination of snakes during regular relocations, allowing us to assess the progression and potential resolution of lesions indicative of ophidiomycosis in individuals.

Transmitter implantation was performed by Sean Egan under sterile conditions. Transmitters (SI-2, Holohil Systems, Carp, Ontario, Canada; 9–11 g) weighed <3% of snakes' body mass and were replaced annually. Field and surgical procedures received approval from the Animal Care Committee of the Ontario Ministry of Natural Resources and Forestry (protocol no. 343) and Trent University (protocol no. 24900). Detailed surgical methods are available by request to the authors. To assess whether snakes were more likely to develop ophidiomycosis if they underwent surgery (as recommended in Hileman et al. 2018), we tracked pathogen detection and lesions on these individuals before and after surgery. All transmitters were removed in 2019 after snakes emerged from brumation.

We held snakes for 24 h at 25–30 C postsurgery to monitor recovery and then released them at the capture location. We tracked them every 24 h for the next 5 d and every 48 h for the following week to monitor healing of the surgical incision. We tracked snakes as they emerged from brumation (April–May) and continued until they returned to the overwintering sites (September–October). We located snakes approximately once weekly from 2013 to 2015, and twice weekly from 2016 to 2018, checking for lesions consistent with ophidiomycosis at each observation.

Swab samples were stored at room temperature and sent to the Animal Health Laboratory (Guelph, Ontario, Canada). Extraction of DNA and testing for O. ophidiicola followed the methods described in McKenzie et al. (2020a) and Davy et al. (2021), using a previously described quantitative PCR (qPCR) assay (Allender, Raudabaugh, Gleason, et al. 2015). Standard curves were generated using the cycle threshold (Ct) values of the positive control plasmid dilutions. We categorized swab samples as O. ophidiicola present (samples with Ct<40) or O. ophidiicola absent (samples with Ct?40, indicating that O. ophidiicola was either absent or was present in quantities below the detection limit of the assay). Snakes exhibiting skin lesions and on whom the pathogen was present (Ct<40) met the CWHC diagnostic criteria for suspect ophidiomycosis (Davy et al. 2021).

Diagnostic criteria for ophidiomycosis vary across the literature (reviewed in Davy et al. 2021). Some published reports of ophidiomycosis are based solely on the observation of gross lesions (e.g., McCoy et al. 2017); others include qPCR confirmation of pathogen presence (e.g., Chandler et al. 2019; McKenzie et al. 2022), but the gold standard for diagnosis includes histopathologic confirmation of fungal hyphae or arthroconidia in lesions or both (Baker et al. 2019). We worked with a free-ranging, endangered population to minimize impact on individuals, and for logistic reasons (as mentioned earlier), we did not surgically biopsy every lesion. Thus, the data we used to assess prevalence of ophidiomycosis represent a combination of CWHC ophidiomycosis and suspect ophidiomycosis, (hereafter ophidiomycosis). The biopsies allowed us to confirm how often biopsied snakes met the full diagnostic criteria, to estimate the accuracy of our field assessments.

We conducted all statistical analyses in R (R Core Team 2018) and considered results statistically significant at ?<0.05. We tested whether the prevalence of O. ophidiicola differed among species of snake using a Fisher exact test (fisher.test, R Core Team 2018) and a pairwise comparison of proportions test (pairwise.prop.test, R Core Team 2018) using Holm P-value adjustment. We also assessed whether prevalence of lesions consistent with ophidiomycosis differed among species, using the same analyses as explained earlier.

We tested whether detection of O. ophidiicola on P. vulpinus captured in 2017–18 varied across the active season using a mixed effects logistic regression of day of year against detection of O. ophidiicola (presence or absence). We included year as a fixed effect in this model because observations are from 2 yr (2017–18), and included individual as a random effect. We summarized the swab result for each snake (if any body or lesion samples were positive, the snake was carrying the pathogen). We also performed a logistic regression of day of year against the presence or absence of lesions for observations between 2013 and 2018, including individual and year as random effects. This model included year as a random effect because observations were from 6 yr (2013–18). We fitted both models with the glmer function (Bates et al. 2015) with a binomial distribution and a logit link function.

We recorded the progression of clinical signs in tracked P. vulpinus in 2017–18, counting the number of lesions at least once a month. We used a generalized linear mixed effects model (glmer, Bates et al. 2015) to assess temporal variation in the number of lesions on tracked snakes over two active seasons, including individual as a random effect and year (2017 or 2018) as a fixed effect. We fitted this model with a log link (Poisson family).

Older snakes have had more time to be exposed to O. ophidiicola, and we hypothesized that snakes with longer SVL (an imperfect but general proxy for relative age) might be more likely to carry the pathogen. However, the surface area of snakes also increases with snake length, and if O. ophidiicola was relatively evenly distributed on a snake's skin, the probability of detecting the pathogen might increase with snake size, as more fungal DNA would be picked up on the swabs. To test this hypothesis, we tested for an association between SVL and Ct value.

We used generalized linear mixed effects models (glmer,Bates et al. 2015) to test whether pathogen detection or lesions on P. vulpinus were associated with snake length and habitat type. We included a fixed effect of year (2017–18) and a random effect for individual. This analysis was performed only for P. vulpinus, because in other species, sample sizes were too low or the detected prevalence of the pathogen was too low for this analysis.

We also used a linear mixed effects model (lmer, Bates et al. 2015) to explore the association between the SVL, habitat type, and fungal load (Ct values) for snakes whose swab samples contained O. ophidiicola. We included a fixed effect for year (2017–18) and a random effect of individual because we sampled the same snakes multiple times.

We calculated Cohen kappa to quantify the agreement in O. ophidiicola detection between body and lesion swab samples collected from P. vulpinus (Cohen kappa ranges from –1, no agreement between body and lesion samples collected at each sampling event to 1, complete agreement; Cohen 1960). Data and code to reproduce all analyses are available (Dillon et al. 2021).

We examined 336 individual snakes (seven species) for lesions and collected a total of 464 swab samples, which were included in a previously published Ontario-wide data set (Davy et al. 2021). We collected 14 surgical biopsies from different individuals; nine tested positive for O. ophidiicola, of which seven had detectable fungal hyphae.

Ophidiomyces ophidiicola was not detected on Heterodon platirhinos (n=1) or Diadophis punctatus (n=1); therefore, we excluded these species from further analyses. Prevalence of O. ophidiicola detection varied among the five species analyzed (P<0.001; Fig. 2): P. vulpinus (24/84); Nerodia sipedon (1/6); Storeria dekayi (1/60); Thamnophis sirtalis (2/160); and Thamnophis sauritus (0/23). Pairwise comparisons by proportions revealed that O. ophidiicola prevalence on P. vulpinus was higher than the prevalence on S. dekayi (P<0.001) and T. sirtalis (P<0.001) but only marginally higher than on T. sauritus (P=0.07). The sample size for N. sipedon (n=6) was insufficient for statistical testing. We observed clinical signs consistent with ophidiomycosis on 34/84 P. vulpinus, 6/60 S. dekayi, 15/160 T. sirtalis, 2/23 T. sauritus, and 1/6 N. sipedon in 2017–18. Lesion prevalence varied among the five species analyzed (P<0.001). Pairwise comparisons by proportions revealed that lesions indicative of ophidiomycosis were observed more often on P. vulpinus than on S. dekayi (P=0.001) and T. sirtalis (P<0.001) but only marginally higher than on T. sauritus (P=0.07). Again, the sample size for N. sipedon (n=6) was insufficient for statistical testing. Given the low detected prevalence of the pathogen on most species, further analyses were restricted to P. vulpinus.

Of 84 P. vulpinus, 19 had ophidiomycosis (positive swab sample for O. ophidiicola in addition to displaying lesions indicative of ophidiomycosis) at some point during 2017–18. Ophidiomycosis could not be confirmed in sampled individuals from other species based on the CWHC criteria.

Emergence of P. vulpinus from brumation for the 38 snakes tracked from 2013–18 ranged from April 25 to May 23. Detection of O. ophidiicola on P. vulpinus was negatively associated with day of year (Fig. 3; ?²=4.90; df=1; P=0.03; n=134 observations from 84 snakes). Detection was most frequent after snakes emerged from brumation (May; 16/37 snakes swabbed) and decreased as the active season progressed. As snakes returned to overwintering sites (September–October), we detected O. ophidiicola on only 2/18 P. vulpinus swabbed. The predicted probability of detecting O. ophidiicola on P. vulpinus decreased from 29.8% (95% confidence interval [CI], 10.5–60.8) on 7 May (first day in our sample) to 3.9% (95% CI, 0.5–22.1) on 10 October (last day in our sample).

In P. vulpinus examined from 2013–18 (n=560 observations from 324 snakes), lesions associated with ophidiomycosis in P. vulpinus typically resolved during the active season. The frequency of lesions consistent with ophidiomycosis decreased with day of year (Fig. 3; ?²=17.0; df=1; P=3.71×10^–5). The probability of observing lesions in P. vulpinus decreased from 36.1% (95% CI, 21.2–54.2) on 26 April (first day in our sample) to 5.5% (95% CI, 2.2–13.1) on 24 October (last day in our sample).

We radiotracked 38 P. vulpinus from 2013–18; 17 of these were tracked in 2017–18, concurrent with standardized observations of lesions and collection of swab samples. Of the 17 P. vulpinus that underwent surgery during 2017–18, nine had no detectable fungal load before surgery, and O. ophidiicola was detected on only one of these in the week following surgery. Three individuals with detectable fungal loads presurgery had no detectable fungal load a week postsurgery. Nine of the 17 snakes entered surgery without lesions, and none developed lesions in the months between transmitter implantation and onset of brumation. Five of these developed lesions over the following winter, but we have no reason to attribute these lesions to surgery rather than the naturally occurring development of ophidiomycosis in some brumating snakes.

For the 17 snakes tracked in 2017–18, the number of lesions decreased with day of year (?²=5.08; df=1; P=0.02; n=63 observations). Of 13 individuals tracked in 2017, those that exhibited many lesions (3+) on emergence in spring 2017 exhibited a general decrease in number of lesions, with most lesions resolved by the end of the active season (Fig. 4). Similarly, lesions generally resolved over the active season in the 10 individuals tracked in 2018 (Fig. 4).

Detection of O. ophidiicola and lesions consistent with ophidiomycosis were positively associated with observed SVL of P. vulpinus. The SVL ranged from 22.6 cm to 130.5 cm (mean±SD=83.9±25.0 cm; n=127 observations from 77 individual P. vulpinus; seven individuals were excluded for missing SVL data). Detection of O. ophidiicola was more likely on longer P. vulpinus (Fig. 5; ?²=5.57; df=1; P=0.02); a 1-cm increase in SVL increased the odds of detecting the pathogen by 3.4% (95% CI, 0.6–6.2). Detection of O. ophidiicola was also associated with habitat type (?²=9.50; df=4; P=0.05). The model-predicted probability of O. ophidiicola detection was highest in snakes swabbed in marsh habitats (Fig. 6; 0.37; 95% CI, 0.19–0.59) and lowest in snakes swabbed in anthropogenic habitats (0.05; 95% CI, 0.01–0.25), but the CIs of all habitat types overlapped.

Observable lesions consistent with ophidiomycosis were positively associated with longer SVL in P. vulpinus (Fig. 5; ?²=14.14; df=1; P<0.001); a 1-cm increase in SVL increased the odds of detecting clinical signs by 4.0% (95% CI, 1.9–6.1). Lesions consistent with ophidiomycosis were also associated with habitat type (?²=16.08; df=1; P=0.003). The model-predicted probability that an individual P. vulpinus had lesions was highest in snakes from marsh habitats (Fig. 6; 0.64; 95% CI, 0.47–0.79) and lowest in snakes from anthropogenic habitats (0.15; 95% CI, 0.05–0.35).

Our hypothesis that greater prevalence of O. ophidiomycosis on larger snakes might reflect the amount of surface area swabbed was not supported. Fungal load (Ct value) on P. vulpinus was not associated with snake size (SVL; F=0.06; df=1; 25; P=0.81) or with habitat type (F=0.42; df=4, 23; P=0.79).

Of the 54 pairs of body and lesion swab samples collected from 34 P. vulpinus that had skin lesions consistent with ophidiomycosis, 28 pairs produced only negative qPCR results (no fungus detected), while in 12 pairs, the fungus was detected on both swab samples. In 12 further pairs, the fungus was detected on the lesion samples but not the body sample. In the last two pairs, the fungus was detected on the body samples but not the lesion sample. Overall, there was 74% agreement between body and lesion swab samples collected from the same snake at the same time. Cohen kappa was 0.452 (z=3.61; P=0.003).

Our data show interspecific variation in prevalence of O. ophidiicola and of lesions consistent with ophidiomycosis at our site. This suggests interspecific variation in exposure to the pathogen or susceptibility to the disease or both, although we cannot distinguish between these hypotheses with our current data. We also detected the pathogen and lesions most frequently as snakes emerged from overwintering, and prevalence of both decreased over the active season. This supports the hypothesis that ophidiomycosis develops during brumation (Lind, Moore, Akçay, et al. 2018; McKenzie et al. 2019). Detection of O. ophidiicola more frequently on P. vulpinus sampled in marsh habitats than in other used habitats, such as dunes and anthropogenically altered areas, suggests that risk of exposure varies among habitats. Longer P. vulpinus were more likely to carry the pathogen and more likely to exhibit lesions consistent with ophidiomycosis, suggesting a potential association with age (time available for exposure). Conversely, fungal load was not associated with SVL, contrary to our hypothesis that the surface area of a swabbed snake determines the amount of pathogen sampled. Finally, our results confirm that repeated sampling of individuals can increase detection likelihood for O. ophidiicola (Hileman et al. 2018).

Pantherophis vulpinus at our study site share hibernacula, coverboards, and other habitat features with co-occurring species but are far more likely to exhibit ophidiomycosis. We swabbed almost twice as many T. sirtalis than P. vulpinus, but O. ophidiicola was only detected on two T. sirtalis, compared with about 30% of P. vulpinus. It remains unclear how O. ophidiicola is transmitted to and among individuals or how often it occurs as an environmental pathogen. Ophidiomyces ophidiicola can use multiple sources of carbon and nitrogen, tolerates pH ranges (5–11), temperatures (10–35 C), and high sulfur levels common in soil and grows well on dead organisms in vitro (Allender, Raudabaugh, Gleason, et al. 2015). However, certain soil microbe and fungal communities suppress growth of O. ophidiicola, and it is more likely to be detected in soils in underground hibernacula than in topsoils (Campbell et al. 2021), suggesting that it may have limited ability to thrive when not on a snake host.

Increasing disease severity during overwintering may reflect reduced snake immune responses during brumation (Muñoz and De la Fuente 2004; Ferguson et al. 2018). Reduced shedding frequency during brumation may also increase disease severity by allowing the pathogen to invade deeper tissues; additionally, snakes with advanced ophidiomycosis may become susceptible to secondary bacterial infections or pneumonia (Rajeev et al. 2009; Lorch et al. 2016; Allender et al. 2018). Snakes infected experimentally with O. ophidiicola increased shedding frequency, and shedding was associated with resolution of lesions (Lorch et al. 2015). However, snakes may be unable to shed their skin (and associated epidermal fungi) while overwintering: shed skins are often found just outside hibernacula but not within (Brown and Parker 1976), and other reptiles are known to pause shedding while overwintering (Zena et al. 2021).

Increased prevalence of O. ophidiicola and lesions on longer P. vulpinus (Fig. 5) suggests that older snakes have had more opportunities to encounter fungal pathogens or that the probability of contact with environmental reservoirs of O. ophidiicola increases with the snake's surface area. Alternately, the association between SVL and pathogen detection may reflect decreased moulting frequency in larger snakes (Brown 1956; Paterson 2006); more frequent shedding may allow smaller snakes to avoid or resolve infections faster than larger individuals.

The mechanism driving the association between marsh habitats and prevalence of O. ophidiicola is unclear. Previous research suggested that O. ophidiicola may be more prevalent on semiaquatic species than on terrestrial species (McKenzie et al. 2019), but we were unable to test this with our current data. Soil and moisture conditions in marsh habitats may provide optimal conditions for the fungus to grow or for infection to develop (e.g., Zuber and Badam 2001), but further research is needed to test this, given the apparent reliance of O. ophidiicola on snake hosts (Campbell et al. 2021).

Our results should be interpreted with the following considerations. First, our detections of O. ophidiicola represent minimum prevalence, because some samples may have contained fungal loads below the qPCR detection limit. Discrepancies between pathogen detection from body and lesion sample pairs also highlight the importance of collecting multiple samples per individual to increase detection of O. ophidiicola (Hileman et al. 2018). Second, the gold standard for diagnosing ophidiomycosis includes histopathology of lesions to confirm presence of fungal hyphae and erumpent arthroconidia (Baker et al. 2019; Davy et al. 2021). We could not biopsy each lesion we observed. Thus, as with other studies that used presence of gross lesions to identify likely cases of ophidiomycosis (Guthrie et al. 2016; Lind et al. 2018), some of the observed lesions may have had causes other than ophidiomycosis. Thus, although qPCR results may have underestimated the prevalence of O. ophidiicola, we may have overestimated the prevalence of the disease itself.

Future research should continue to unravel the transmission dynamics of ophidiomycosis by identifying factors that predict exposure to the pathogen and interspecific variation in susceptibility. Further characterizing the environmental distribution of O. ophidiicola will clarify risk factors for infection and reveal which overwintering conditions are most conducive to infection. The seasonality of ophidiomycosis should be explicitly considered when designing surveillance programs. For temperate, Nearctic snakes, surveillance should prioritize sampling in spring after snakes emerge from brumation, when the pathogen and disease are most prevalent.

Funding for this study was provided by the Government of Ontario, the Government of Canada (Mitacs Accelerate Program), and Wildlife Preservation Canada and was supported by Ontario Parks. Sean Egan surgically implanted radio transmitters in P. vulpinus. For fieldwork, logistic, and other assistance, we thank Jennifer Angoh, Hower Blair, Jeff Bowman, Tony Braithwaite, Mhat Briehl, Courtney Butler, Pauline Catling, Patricia Charlebois-Page, Maria Ciancio, Brad Connor, Mike Donaldson, Jeffrey Ethier, Thomas Hossie, Rebecca Lewis, Christina McKenzie, Steve Marks, Jess Matthews, Ric and Anne McArthur, Sheeva Nakhaie, Mike Nettle-ton, Gavin Prince-Badke, Juliana Skuza, Caitlin Sparks, Brittany Talarico, Kristin Thiessen, Xiaotian Wang, Amelia Whitear, Jonathan Wild, Kyle Yurkiw, and others. Three anonymous reviewers provided helpful feedback on an earlier version of this manuscript. Finally, the fieldwork for this study was conducted on the traditional territories of the Attiwonderonk, and we gratefully acknowledge the historic and ongoing contributions of First Nations in stewardship of wildlife around the Laurentian Great Lakes.