Chytrid Infection Dynamics in Cricket Frogs on Military and Public Lands in the Midwestern United States

A number of factors have been linked to the amphibian decline crisis over the past two decades. Although habitat alteration is widely accepted as the major cause (Cushman 2006; Gallant et al. 2007), there is growing concern over widespread disease outbreaks in amphibian populations (e.g., Lips et al. 2006; Une et al. 2008; Martel et al. 2013). One such disease is caused by the highly contagious pathogenic fungus Batrachochytrium dendrobatidis (hereafter, Bd), the causative agent of amphibian chytridiomycosis first identified in 1998 (Berger et al. 1998). Until recently (Martel et al. 2013), this organism was the only member of the phylum Chytridiomycota known to parasitize vertebrates (Berger et al. 1998; Longcore et al. 1999). The origin of this pathogen is unknown, but the earliest Bd-positive specimen is from 1888 (found in Adams County, Illinois, USA; Talley et al. 2015). Recent evidence suggests a sudden range expansion facilitated by human transport or introduction of a carrier organism (Daszak et al. 2003; Weldon et al. 2004; Huss et al. 2013). Mass mortality events caused by chytridiomycosis have been documented as a contributing or causative factor in the declines of amphibian populations worldwide (Lips et al. 2006; Skerratt et al. 2007; Wake and Vredenburg 2008). However, not all amphibian species are affected and a clear mechanism of resistance that affords some species protection from the fungus has not been established (Longcore et al. 2007; Rothermel et al. 2008). Additionally, there is some evidence that selection pressure may favor individuals with stronger immune systems, thereby altering amphibian populations and providing long-term resistance (Richmond et al. 2009). Further, the thermoregulatory behavior of certain species has been shown to affect Bd infection rates in wild populations; Golden Frogs Atelopus zeteki that raised their body temperatures above normal set points had decreased odds of infection (Richards-Zawacki 2010).

Accurate identification of amphibian populations exposed to Bd is required for effective disease management (Kriger and Hero 2007a). In Australia, climatic variables, such as rainfall and temperature, have been implicated in the prevalence and intensity of Bd infections between infected and uninfected sites (Kriger and Hero 2007a). However, snout–vent length was found to consistently be the best predictor of infection levels across infected sites; small frogs were more likely to be infected and carried more intense infections than larger frogs (Kriger and Hero 2007b). Further, wild frogs have been shown to be capable of acquiring Bd as adults but can subsequently clear an infection (Kriger and Hero 2007b). Overall, infections are not evenly distributed across the landscape and are more prevalent in lentic versus lotic systems (Kriger and Hero 2007a; Hossack et al. 2010). Additionally, species hibernating in terrestrial habitats seem to have lower infection rates than species hibernating in aquatic habitats (Longcore et al. 2007).

In the United States, Bd has been detected in numerous amphibian species in nearly all states (Olson et al. 2013; also see http://www.bd-maps.net/), but mortality and population declines positively linked to Bd have only been documented in Arizona (Bradley et al. 2002), the Sierra Nevada mountains of California (Vredenburg et al. 2010), and Indiana (Kinney et al. 2011). A number of studies have been conducted to identify the distribution of Bd in the Pacific Northwest (Pearl et al. 2007), Rocky Mountains (Muths et al. 2008), Northeastern United States (Longcore et al. 2007), and Southeastern United States (Rothermel et al. 2008). However, in the Midwestern United States, less is known about the status and distribution of Bd in amphibian populations (Sadinski et al. 2010, Phillips et al. 2014). Further, most studies rely on sampling a number of different species to determine occurrence and prevalence, but this may lead to faulty conclusions because susceptibility to Bd can differ between (Woodhams et al. 2007) and within species (Gervasi et al. 2013).

Research focused on understanding the parameters leading to mass mortality in amphibian populations in some regions during certain time frames is important for the development of conservation and management strategies. Sampling for Bd up to this point has largely been opportunistic and haphazard, so little information exists on the true occurrence and prevalence of the disease in the Midwestern United States, and therefore, how to most appropriately sample for it. Further, there is a lack of information on the occurrence of the diseases as it relates to different types and levels of human land use. The ability to detect Bd with a standardized protocol is necessary for the management of amphibian populations and determining how environmental and temporal factors affect occurrence and prevalence rates is crucial. The objectives of our study were to 1) estimate the occurrence and prevalence of Bd in Midwestern U.S. amphibian populations, 2) determine whether military sites have lower occurrence and prevalence rates when compared with public-use sites, and 3) determine how environmental and temporal factors affect estimated prevalence rates.

During the 2009 field season (April–October), we sampled six U.S. military installations and three public-use areas to investigate trends in seasonal occurrence and prevalence of Bd in the Midwestern United States. Additionally our design allowed us to examine the impact of site type (i.e., military vs. public-use) on Bd occurrence and prevalence. The six military installations sampled were 1) Sparta Training Center, Illinois (38°09′07″N, 89°43′00″W); 2) Camp Atterbury, Indiana (39°21′36″N, 86°01′47″W); 3) Naval Surface Warfare Center Crane Division, Indiana (38°52′12″N, 86°50′02″W); 4) Fort Campbell, Kentucky (36°39′48″N, 87°28′38″W); 5) Fort Knox, Kentucky (37°54′57″N, 85°57′22″W); and 6) Fort Leonard Wood, Missouri (37°47′37″N, 92°08′04″W; Figure 1). The public-use areas sampled were 1) Hillenbrand Fish and Wildlife Area, Indiana (39°07′21″N, 87°10′34″W); 2) Peabody Wildlife Management Area, Kentucky (37°14′39″N, 87°01′21″W); and 3) Woodson K. Woods Memorial Conservation Area, Missouri (37°55′55″N, 91°29′13″W; Figure 1). At each of these sites we sampled 6 wetlands during 3 different visits (spring = 4 May–21 May; summer = 29 June–16 July; autumn = 31 August–30 September) for 54 total wetlands; however, we excluded 3 wetlands from analyses because of low capture numbers (1 wetland at Camp Atterbury and 2 wetlands at Fort Knox). Within a site, wetlands were separated by a minimum distance of 250 m. To determine if any seasonal trends were consistent across years, we sampled a reduced number of wetlands in 2011. We sampled 5 wetlands at each of 2 military installations (Sparta Training Center and Fort Leonard Wood) and 5 wetlands located on nonmilitary sites in Illinois and Missouri for 15 total wetlands. We sampled each wetland during the spring (9 May–3 June) and summer (9 August–16 August) periods only.

During each visit to a wetland, we captured and swabbed the first 20 postmetamorphic Cricket Frogs Acris blanchardi or A. crepitans (Gamble et al. 2008) encountered. The range of Cricket Frogs swabbed per wetland per sampling period was 20–21 in 2009 and 17–26 in 2011. Cricket Frogs are small anurans that use permanent bodies of water (wetlands and streams) for breeding, and overwinter near these aquatic habitats. Additionally, Cricket Frogs have experienced recent enigmatic declines throughout many northern Midwestern states (Gray and Brown 2005; Lehtinen and Skinner 2006), making them a species of special conservation concern throughout the region. We chose Cricket Frogs as the focal species for this study because of their diurnal habits, which make them available for collection during the day, and their recent unexpected declines in the Midwest, which may be disease related. All sampling occurred during the daytime when Cricket Frogs were most active at the periphery of the wetlands. We captured Cricket Frogs by placing a clean plastic zipper bag in front of the individual and coaxing it to jump into the bag. We swabbed the Cricket Frogs by allowing the animal to the top of the zipper bag, gently grasping the hind leg through the plastic bag, then opening the bag. We followed the protocols of Brem et al. (2007), swabbing the dorsum, flanks, ventrum, cranium, inguinal region, and both the palmar and plantar surfaces of each foot with a sterile swab; individuals who swabbed the frogs wore latex gloves. Researchers rubbed each region 5 times for 50 total strokes/frog. We stored swabs in individually labeled vials of 70% ethanol and released frogs at the point of capture after swabbing. We discarded all plastic bags after one use, and all researchers cleaned their rubber sampling boots with a brush and dilute bleach solution at the conclusion of sampling at each wetland.

In addition to date of swabbing, we obtained the following temperature and precipitation variables for relationship to Bd prevalence: average maximum temperature 30 d prior to swabbing (30d Tmax); average maximum temperature 60 d prior to swabbing (60d Tmax); and total precipitation 30 d prior to swabbing (30d Precip). These weather variables have been shown to explain a significant amount of variance in Bd prevalence in amphibians across sites and years (Lannoo et al. 2011; Murray et al. 2013). We took weather data from weather stations nearest to the study sites using the National Climatic data center's website: http://www.ncdc.noaa.gov/cdo-web/.

To extract DNA, we briefly vortexed sample tubes and removed swabs before centrifugation at 16,000 × g for 10 min. We added 200 μL of ATL-PK (Qiagen Inc., Germantown, Maryland) tissue lysis buffer to the pelleted fraction and incubated it overnight at 55°C. For the remainder of the DNA extraction procedure, we used Qiagen's (Germantown, Maryland) DNeasy Blood and Tissue Kit according to the manufacturer's instructions. We used nested polymerase chain reaction (PCR) to detect the presence of Bd zoospores (Gaertner et al. 2009). We performed initial reactions with the universal fungal primers ITS1f (5′-CCT GGT CAT TTA GAG GAA GTAA-3′) and ITS4 (5′TCC TCC GCT TAT TGATAT GC-3′; White et al. 1990) in 50-μL volumes using the following: Platinum Taq PCR buffer (1×), MgCl₂ (2 mM), dNTPs (1:1:1:1, 0.1 mM), ITS1f primer (0.5 mM), ITS4 primer (0.5 mM), BSA (0.2 μg/μL), Platinum Taq DNA Polymerase (1 U), and Extracted DNA Sample, 2 μL. Thermocycling conditions consisted of an initial 10-min denaturation at 94°C followed by 30 cycles of the following: 45 s at 94°C, 45 s at 57°C, and 60 s at 72°C. A final extension of 7 min at 72°C completed the reaction. We produced internal fragments of the nested PCR approach using primers Bd1a (5′-CAG TGT GCC ATA TGT GAC G-3′) and Bd2a (5′-CAT GGT TCA TAT CTG TCC AG-3′; Annis et al. 2004). Reaction conditions were identical to those described above except the denaturation temperature was 93°C and the annealing temperature was 60°C. We visualized amplification products on a 2% agarose gel (Ameresco agarose 3:1 HRB) containing 1× GelStar nucleic acid gel stain (Lonza, Allendale, New Jersey). We compared presence or absence of a 300-bp band against the EZ Load 100-bp molecular ruler (Bio-Rad Laboratories, Hercules, California) and a positive control. We analyzed all samples twice, and if ≥1 of the 2 samples tested positive we considered the frog to be infected.

In all analyses, we calculated prevalence as the number of individuals that tested positive for Bd out of the total number of individuals sampled at a wetland within a sampling period. We then arcsine-transformed prevalence values (percentages) for univariate analysis of variance (ANOVA). With the combined 2009 and 2011 data sets, we used a univariate ANOVA (weighted for sample size) to test the fixed effects of site type (military vs. public-use), sampling season, and sampling year on the response variable (Bd prevalence). We conducted post hoc mean comparisons using Tukey's Honest Significant Difference test if the response variable was significantly different in the univariate ANOVA to assess differences between the treatment groups (i.e., sampling seasons). We considered all results significant at P ≤ 0.05.

To investigate the effects of environmental parameters on Bd prevalence, we analyzed the combined 2009 and 2011 data sets using a binomial generalized linear mixed-effects model and evaluated all candidate models simultaneously. Covariates in the mixed-effects model included site type, average maximum daily temperature 30 d prior to a survey (°C), 30-d cumulative precipitation prior to a survey (mm), and sampling year. Previous research has suggested that Bd prevalence reaches peaks at values of precipitation and temperature optimal for Bd growth (Murray et al. 2013), so we also fit quadratic temperature and precipitation terms to our fixed-effects model. To control for the hierarchical and repeated sampling structure of our data, we fit nested random effects to our model: survey within wetland, wetland within site, site within year. We also considered average maximum daily temperature 60 d prior to surveys, but this covariate was highly correlated with 30-d maximum temperature (r = 0.92) and resulted in substantially poorer model fit (Akaike's Information Criterion [AIC] = 692.1 vs. 667.0). We performed all analyses in Program R (version 3.1.1; R Development Core Team 2013).

We detected Bd infections in 100% (n = 51; 33 military and 18 public-use) of the wetlands we sampled in 2009 (Data S1, Supplemental Material). Within the spring 2009 sampling period Bd occurrence was 100%, while it dropped to 58.8% during the summer period and 37.3% during the autumn period. Prevalence per wetland site ranged 0.03–0.38 across sampling seasons (i.e., average of the three sampling periods). Within the spring 2009 sampling period, prevalence ranged 0.10–1.00; while during the summer period prevalence ranged 0.00–0.45 and prevalence ranged 0.00–0.20 during the autumn period. During the spring 2009 sampling period, 98.0% of wetlands (n = 50) had prevalence rates ≥0.25, 70.6% of wetlands (n = 36) had prevalence rates ≥0.50, and 31.4% of wetlands (n = 16) had prevalence rates ≥0.75. In 2011 we detected Bd infections in 100% (n = 15; 10 military and 5 nonmilitary) of the wetlands we sampled (Data S1, Supplemental Material). Within the spring 2011 sampling period, Bd occurrence was 93.3% and it dropped to 66.7% during the summer period. Prevalence per wetland site ranged 0.08–0.74 across sampling seasons (i.e., average of the two sampling periods). Within the spring 2011 sampling period, prevalence ranged 0.00–1.00; while during the summer period, prevalence ranged 0.00–0.47. During the spring 2011 sampling period, 66.7% of wetlands (n = 10) had prevalence rates ≥0.25, 66.7% of wetlands (n = 10) had prevalence rates ≥0.50, and 66.7% of wetlands (n = 10) had prevalence rates ≥0.75.

There were no significant interactive effects of site type, sampling season, and sampling year on Bd prevalence (Table 1). There was also no direct effect of site type on Bd prevalence; however, both sampling year and sampling season significantly affected Bd prevalence (Table 1). Batrachochytrium dendrobatidis prevalence in 2011 was significantly higher than in 2009 (Figure 2A) and Bd prevalence was significantly higher during the spring season than in the summer or autumn (Figure 2B). Across years we found Bd prevalence was best predicted by site type, sampling year, a quadratic relationship with 30-d maximum temperature, and 30-d precipitation (Table 2). Of the variables in this top model, Bd prevalence was most affected by the average 30-d maximum temperature prior to sampling (Figure 3A; Table 3). Prevalence exceeded 70% when maximum daily temperatures were <19.5°C, but decreased substantially as maximum daily temperatures increased (Figure 3A). Cumulative 30-d precipitation was also a significant predictor, with Bd prevalence increasing as total precipitation increased (Figure 3B; Table 3). Precipitation had a smaller overall effect on prevalence, with predicted prevalence varying from 6.8 to 46.9% across the range of observed cumulative precipitation values. Consistent with the univariate ANOVA results, site type was not a significant predictor of Bd prevalence, while sampling year was a significant predictor (Table 3).

Emerging infectious diseases have rapidly increased in the past two decades and now pose a major threat to maintaining biodiversity on a global scale (Jones et al. 2008; McCallum 2012). Obtaining information on the occurrence and prevalence of Bd in the Midwestern United States is a critical first step in assessing the threat this pathogen may pose to amphibian populations and directing conservation actions toward affected species. In this study, we found that Bd occurred in 100% of wetlands we sampled in 2009 (n = 51) and 2011 (n = 15). Overall prevalence was 22.7% in 2009 and 40.5% in 2011. Although our prevalence rates are consistent with those found in other regions of the United States, our occurrence rates are somewhat higher than those found in prior regional studies. In the Pacific Northwest, Pearl et al. (2007) found Bd occurred at 43% of sites sampled with an overall prevalence rate of 28%; whereas, Muths et al. (2008) found Bd occurred at 64% of clusters (sites separated by <3 km were considered a functional cluster) sampled in the Rocky Mountains region with an overall prevalence rate of 24%. In the Southeastern United States, Rothermel et al. (2008) found Bd occurred over a broad geographic region (6 different states) in multiple populations of 10 different species with an overall prevalence rate of 17.8%; whereas, in the Midwestern United States, Sadinski et al. (2010) found Bd occurred in 45% of sites sampled (no prevalence data were available).

Currently, there are >12 million ha of land managed by the U.S. Department of Defense, and a large number of threatened and endangered species (at both the state and federal levels) are found on military installations (Stein et al. 2008). We suspected occurrence and prevalence rates may be lower on sites such as military installations because of the potential of increased transmission rates at high-traffic sites. In a country-wide survey of U.S. military installations, Lannoo et al. (2011) found that Bd was present at 86.7% (13/15) of bases sampled, although the authors noted the lack of detection at the two remaining sites could have been due to insufficient sampling. Across their three sampling periods (spring, summer, and autumn of 2009), they found an overall prevalence rate of 16.6%. These occurrence results are in line with our 2009 results, where we found Bd on all six military installations we sampled, whereas our overall prevalence rate of 22.7% was slightly higher. In this study we directly tested whether military sites have significantly lower prevalence of Bd when compared with public-use sites. We found that Bd prevalence was not significantly lower (Tables 1 and 3) in military wetlands and our results from the military installations are similar to other reported values for prevalence from nonmilitary sites from other studies (e.g., Pearl et al. 2007; Muths et al. 2008; Rothermel et al. 2008).

Numerous studies have shown higher Bd prevalence during cooler months (Woodhams and Alford 2005; Kriger and Hero 2007a; Lannoo et al. 2011). In Virginia, Bd prevalence was significantly higher during the spring (45%) than during late summer (2%; Pullen et al. 2010) and Rothermel et al. (2008) identified a similar trend. Further, Lannoo et al. (2011) found the majority of positive tests resulted from the first sampling period and reported prevalence rates of 39.3% in the spring, 6.1% in the summer, and 4.5% in the autumn. We found similar results for prevalence rates in 2009 for all three periods, with 59.4% in the spring, 5.8% in the summer, and 2.8% in the autumn; Bd prevalence in Cricket Frogs was significantly higher during the spring sampling period when compared with summer and autumn samples in 2009. This pattern remained consistent during 2011, when we found significantly higher prevalences during the spring versus summer sampling periods (autumn samples were not collected in 2011).

Batrachochytrium dendrobatidis can tolerate a wide range of temperatures and is capable of growth between 4 and 28°C (Piotrowski et al. 2004; Woodhams et al. 2008). Optimal growth occurs between 17 and 25°C (Piotrowski et al. 2004); however, at 10°C Bd produces more zoospores, the zoospores live longer (up to 48 h), and are capable of traveling longer distances when compared with zoospores at 23°C (Piotrowski et al. 2004; Woodhams et al. 2008). Additionally, there is variation in the thermal performance of different Bd isolates (i.e., different thermal maxima dependent upon the specific isolate; Stevenson et al. 2013). Climatic variables such as rainfall, relative humidity, and temperature have explained a significant amount of variation in the occurrence and prevalence of Bd among sites and years (Kriger and Hero 2007a; Lannoo et al. 2011; Murray et al. 2013) and underlie the seasonal trend observed above. Interestingly, Raffel et al. (2010) found that Bd infection levels in Red-spotted Newts Notophthalmus viridescens were best predicted by the proportion of the pond substrate consisting of leaf litter or vegetation (which would serve to moderate water temperatures), along with a significant effect of water temperature. When evaluating environmental factors influencing prevalence, prevalence was highest when maximum daily temperatures were <19.5°C, but decreased substantially as maximum daily temperatures increased (Figure 3A). However, precipitation also played a significant role (Figure 3B).

This study provides data on the occurrence and prevalence of Bd in the United States and fills an important gap in the Midwest, while also corroborating the seasonal trends of increased prevalence in the cooler spring season. Batrachochytrium dendrobatidis appears to be endemic in the Midwestern United States, and military sites do not appear to provide any significant refugia for amphibian populations. Although we did not conduct a capture–mark–recapture study to sample specific individuals from period to period to determine if prevalence declines were due to mortality or animals clearing themselves of the infection, Gaertner et al. (2009) found that individual Cricket Frogs Acris blanchardi that tested positive for Bd in the spring subsequently tested negative during summer sampling periods. Thus, additional studies are needed to determine if the seasonal trends are driven by animals clearing themselves of the infection as environmental conditions change or if the reduced prevalence rates are due to die-offs in the population (although these two hypotheses are not mutually exclusive). Further, information is needed on how Bd infection of a host influences infection rates of other pathogens found in the system and the resulting impacts on population dynamics (Jones et al. 2008; Hoverman et al. 2012). Greater insight into the disease ecology of amphibian populations is needed to direct future conservation and management efforts.

Data S1. Chytrid fungus Batrachochytrium dendrobatidis (Bd) occurrence and abundance in Cricket Frog Acris blanchardi or A. crepitans populations was studied in the Midwestern United States in 2009 and 2011. Data file (.xls), including description of data, used to 1) estimate the occurrence and prevalence of Bd in Midwestern U.S. amphibian populations, 2) determine whether military sites have lower occurrence and prevalence rates when compared with public-use sites, and 3) determine how environmental and temporal factors affect estimated prevalence rates.

Found at DOI: http://dx.doi.org/10.3996/012017-JFWM-003.S1 (22 KB XLSX)

We thank S. Andrews (Naval Surface Warfare Center Crane Division), M. Bradenburg (Fort Knox), C. Carmichael (Sparta Training Center), K. Lohraff (Fort Leonard Wood), M. Peterkin (Camp Atterbury), and G. Zirkle (Fort Campbell) for access to their military installations and help with wetland locations. We also thank N. Engbrecht, E. Kessler, A. Robinson, J. Tiemann, D. Wylie, and S. Baker-Wylie for field assistance and J. Kelsey for help with compiling the weather data. We thank J. Longcore for providing Bd strains and J. Woods for expert advice on Bd PCR analysis and interpretation. Finally, we thank the Associate Editor and three Journal Reviewers for helpful comments and suggestions that greatly improved this manuscript. Funding for the project was provided by the U.S. Army Corps of Engineers (Construction and Engineering Research Laboratory). All research was conducted in accordance with animal care protocols of the University of Illinois animal care and use committee (permit # 09029) and state collecting permits issued to CAP.

Any use of trade, product, website, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government.