Do Frogs Get Their Kicks on Route 66? Continental U.S. Transect Reveals Spatial and Temporal Patterns of Batrachochytrium dendrobatidis Infection

The chytrid fungus Batrachochytrium dendrobatidis (Bd) has been devastating amphibians globally. Two general scenarios have been proposed for the nature and spread of this pathogen: Bd is an epidemic, spreading as a wave and wiping out individuals, populations, and species in its path; and Bd is endemic, widespread throughout many geographic regions on every continent except Antarctica. To explore these hypotheses, we conducted a transcontinental transect of United States Department of Defense (DoD) installations along U.S. Highway 66 from California to central Illinois, and continuing eastward to the Atlantic Seaboard along U.S. Interstate 64 (in sum from Marine Corps Base Camp Pendleton in California to Naval Air Station Oceana in Virginia). We addressed the following questions: 1) Does Bd occur in amphibian populations on protected DoD environments? 2) Is there a temporal pattern to the presence of Bd? 3) Is there a spatial pattern to the presence of Bd? and 4) In these limited human-traffic areas, is Bd acting as an epidemic (i.e., with evidence of recent introduction and/or die-offs due to chytridiomycosis), or as an endemic (present without clinical signs of disease)? Bd was detected on 13 of the 15 bases sampled. Samples from 30 amphibian species were collected (10% of known United States' species); half (15) tested Bd positive. There was a strong temporal (seasonal) component; in total, 78.5% of all positive samples came in the first (spring/early-summer) sampling period. There was also a strong spatial component—the eleven temperate DoD installations had higher prevalences of Bd infection (20.8%) than the four arid (<60 mm annual precipitation) bases (8.5%). These data support the conclusion that Bd is now widespread, and promote the idea that Bd can today be considered endemic across much of North America, extending from coast-to-coast, with the exception of remote pockets of naïve populations.

Because of the variable nature of the available datasets (a situation we do not criticize), there have been questions about the occurrence and spread of Bd. In response, two general scenarios have been proposed that have strong empirical support [36,37,38]. In the first scenario, Bd is an epidemic (arising outside the population), spreading as a wave and wiping out individuals, populations, and species in its path. This has been well documented and is occurring in Central America, in eastern Australia, and in parts of California [35,39,40,41,42,43,44,45]. The second scenario suggests that in certain regions of the world, such as North America, much of the spread of Bd occurred decades ago (when it was an epidemic) and that in these places it is now endemic (arising within the population) [35,45,46]. Indeed, Bd is now widespread throughout many geographic regions and is known to occur on every continent inhabited by amphibians (though some land masses and regions appear to remain naïve); therefore, this infection may be considered global [15,16,18,38,47,48,49,50,51,52,53,54,55,56,57,58]. A third scenario, the Bd thermal optimum hypothesis, combines the first two hypotheses and has been more controversial. It suggests widespread benign Bd distribution has been triggered to lethality in regions by increased temperatures due to global warming [59], but this interpretation has been contentious [60].
Exploring the epidemic and endemic hypotheses requires, in part, broad-scale studies using standardized techniques. Further, due to the confounding factor of human disturbance on broadscale patterns, it is best to examine low-impact (i.e. ''natural''), or well-protected areas. Perhaps the most widely available habitats that remain ''undisturbed'' (a relative term, with perhaps the exception of the deep sea floor it is likely there are no longer any truly undisturbed environments left on earth [61]) in the United States today are United States Department of Defense (DoD) installations, which are secured as a matter of national interest. Military installations are protected against the indiscriminate human traffic experienced by parks, wildlife refuges, and other public areas. Moreover, following the tragic events of September 11 th , 2001, access to these installations has been further limited, and in some cases severely restricted. DoD installations encompass over 12 million ha and occur throughout the United States, making continent-wide surveys possible. DoD lands are managed differently than typical surrounding landscapes, using ecosystem management techniques. Indeed, American military lands are thought to harbor the greatest concentrations of endangered and threatened species in the United States [62].
We conducted a transcontinental transect designed to assess the presence of Bd on military lands across the North American continent. DoD installations were sampled from west to east along U.S. Highway 66 (the ''Mother Road'') from California into central Illinois, and continuing eastward from there to the Atlantic Seaboard along U.S. Interstate 64 (in sum from Marine Corps Base Camp Pendleton in California to Naval Air Station Oceana in Virginia, between 33u and 39u N latitude). We sampled across warm seasons, and used standardized collection and analytical techniques to address the following questions: 1) Does Bd occur in amphibian populations in these relatively undisturbed environments? 2) Is there a temporal (seasonal) pattern to the presence of Bd? 3) Is there a spatial pattern to the presence of Bd? and 4) In secured, limited-traffic areas of the country, is Bd acting as an epidemic (i.e., is there evidence of recent introduction and/or dieoffs due to chytridiomycosis), or as an endemic (is it present without clinical signs of disease)?

Ethics Statement
Lee, Fort A.P. Hill, Fort Belvoir, and Naval Air Station Oceana in Virginia. Each base was sampled three times: once in the spring/ early summer (April, May, or the first week in June), once in midsummer (July, August), and once in the late summer/fall (September, October), conforming to environmental conditions when Bd is most likely detectable [63]. Generally, three wetlands were sampled at each installation. Most sampling occurred at night, when amphibians are active, using dip nets. Captured amphibians were placed in new, individual plastic bags for processing and handling. Bags were discarded after one use; boots and nets were rinsed to clean off mud and debris, and sterilized with a dilute bleach solution between wetland sites.
Amphibians are the only known animate host for Bd [4,45,64], although zoospores can survive for up to 12 weeks under favorable (cool, moist) soil conditions [65,66,67,68,69]. Therefore our survey effort focused exclusively on amphibian populations. At each installation, postmetamorphic animals were sampled as they were encountered-because we were broadly interested in the presence of Bd we did not discriminate between salamanders and frogs. All animals were handled using sterile techniques and sampled using cotton, wooden-handled swabs. Swabs were rubbed while rolling the cotton over the body surface; five rubs each on the dorsum, flanks, ventrum, cranium, inguinal region, and the palmar/plantar surface of each foot for a total of 50 rubs [64,70]. The head of the swab was then broken off in an individually labeled 0.5 ml free standing polypropylene screw cap microcentrifuge tube (Fisherbrand 02-681-333) and stored and shipped cold prior to analysis.

Laboratory Analyses
Swabs were analyzed for Bd using conventional PCR (polymerase chain reaction) techniques [71,72,73]. Briefly, to extract Bd DNA from field samples, 1 ml of 70% ethanol was added to microcentrifuge tubes containing sample swabs and stored overnight at 220uC. Swabs were removed and the supernatant was centrifuged (16,0006 g for 10 min). ATL-PK (Qiagen) tissue lysis buffer (200 ml) was added to the pelleted fraction and incubated overnight (55uC). To detect Bd zoospores, we used a nested PCR approach (18). Amplification products were visualized on a 3% agarose gel (Ameresco agarose 3:1 HRB). Presence or absence of a 300-bp band was compared against the EZ Load 100bp molecular ruler (Bio-Rad) and a positive control. Negative controls were run with each sample. Samples were analyzed twice; if either sample tested Bd positive the animal was considered infected. Results were recorded on digital spreadsheets (Microsoft Excel).

Temperature and Precipitation Data, and Statistics
Mean monthly and annual temperature and precipitation data for a 30-yr period  were obtained from stations near or at each base by searching National Oceanic and Atmospheric Administration (NOAA) databases (http://cdo.ncdc.noaa.gov/cgibin/climatenormals/climatenormals.pl).
We used Akaike's information criterion (AIC) [74,75] to compare models (general linear mixed models; SPSS v. 17) fitting the pattern of Bd presence to four variables (season [S]-spring/ early summer, mid-summer, and late summer/fall; geographic location [G]-east, central, west; mean seasonal rainfall [R]; and mean seasonal temperature) and each possible combination of each of the four variables (23 candidate models in all). Under this model selection framework, one model (see below) was clearly the best fit. However, because of the danger in underfitting models (leaving out models with important biological inferences) [75], we explored the next two models using a hypothesis testing framework. In particular, because only temperature data from among our datasets were normally distributed (Shapiro-Wilk normality test, Program R). We used nonparametric Kruskal Wallis tests (SPSS v. 17) to refine our examination of geographic location by comparing Bd infection prevalences, temperatures, and precipitation values between arid and temperate installations. Arid installations were located west of the 100 th meridian, where mean annual precipitation is ,60 cm, and included Marine Corps Base Camp Pendleton, Camp Navajo, Kirtland Air Force Base, and Cannon Air Force Base. Temperate bases were Fort Sill, Camp Gruber, Fort Leonard Wood, Sparta Training Center, Naval Support Activity Crane, Fort Knox, Radford Army Ammunition Plant, Fort A.P. Hill, Fort Belvoir, Fort Lee, and Naval Air Station Oceana. Significance levels were set at p#0.05.

Installations
Each base was visited three times (Table 1). In April, Camp Navajo and Fort Sill were too cold for amphibian activity, and bases were visited but no amphibians were found and therefore no samples were collected. During the mid-summer and latesummer/fall trips the ponds at Fort Belvoir were dry and no amphibians were detected. At all other sites, during all other sampling times, animals were collected and sampled. In total, from all bases, during all visits, 1,306 amphibians were sampled for this project, 217 (16.6%) swabs tested positive for Bd. In general, the more arid the site the more difficult it was to detect amphibians, especially later in the year, and fewer animals were collected from these bases.
We did not detect Bd at two bases, Camp Navajo, AZ and Fort Sill, OK. This was not due to sample size, per se. Thirty five samples were taken at Camp Navajo; 34 in July (mid-summer), 1 in September (late-summer/fall). At Fort Sill, a total of 43 samples were taken; 12 during June (mid-summer), 31 during September (late-summer/fall). This result could have been caused, in part, to a lack of samples during the spring/early summer sampling period (due to cold and snow), when the majority of positive samples at other bases were collected (see below).
Bd was detected at the remaining 13 bases (Table 1

Temporal Patterns (Seasonality)
Among 23 candidate models based on season, geographic location, mean seasonal rainfall, and mean seasonal air temperature examined individually and in combination, seasonality produce the lowest AICc score (Table 3). We agree there was a strong seasonal component to our results (Table 3, Fig. 3). During the spring/earlysummer sampling period, 39.3% of all samples were positive. This number dropped to 6.1% for mid-summer samples, and to 4.5% for late-summer/fall samples. Most bases followed this pattern (Table 1) including Marine Corps Base Camp Pendleton, Cannon Air Force Base, Camp Gruber, Fort Leonard Wood, Sparta Training Center, Naval Support Activity Crane, Fort Knox, and Fort Lee. Camp Navajo and Fort Sill had no positive samples and Fort Belvoir had animals collected only in the spring/early summer period, and therefore these bases had no temporal pattern of infection. Radford  Army Arsenal had 15.8% positive in spring/early summer, 33.3% positive samples in mid-summer, and 23.5% positive in latesummer/fall. Three East-Coast installations-Fort A.P. Hill, Fort Lee, and Naval Air Station Oceana-had higher percentages of animals infected during the late-summer/fall than in the midsummer period.

Spatial Patterns
Among candidate models, seasonality plus mean annual rainfall produced the second lowest AICc score (Table 3). While this model was much less predictive, aridity did appear to have a negative effect on Bd prevalence. Five of the six sites with the lowest prevalences (Camp Navajo, 0%; Fort Sill, 0%; Kirtland Air Force Base, 2%; Cannon Air Force Base, 6%, and Camp Gruber, 8%) occur in the arid southwest (Arizona, New Mexico) or on or near the Great Plains (Oklahoma); the exception was Fort Lee (7%) in Virginia. Remaining sites occur in coastal areas, or inland areas that receive higher levels of precipitation (Table 4). A second way we explored this trend was to compare the data for the western arid bases (Marine Corps Base Camp Pendleton, Camp Navajo, Kirtland Air Force Base, and Cannon Air Force Base) with data for the eastern temperate sites. The prevalence of positive samples for the arid installations was 8.5611.7% (x695% C.I.)); the prevalence of positive samples for the eastern temperate sites was 20.868.4%. This difference was statistically significant (Kruskal-Wallis, p = 0.027). Precipitation levels were also different between the western arid and eastern temperate bases (x = 39.5613.2 vs. 107.8611.8 cm annually; Kruskal-Wallis, p = 0.002), although temperatures were not (x = 13.762.9 vs. 13.961.1uC; Kruskal-Wallis, p = 0.3).

Discussion
We used standardized collection and analytical techniques to address the following questions: 1) Does Bd occur in amphibian populations in these relatively undisturbed environments? 2) Is there a spatial pattern to the presence of Bd? 3) Is there a temporal pattern to the presence of Bd? and 4) In secured, limited-traffic areas of the country, is Bd acting as an epidemic (i.e., is there evidence of recent introduction and/or die-offs due to chytridiomycosis), or as an endemic (is it present without clinical signs of disease)?
Before we formally consider these questions, we offer some background on the life history and physiological ecology of Bd. The life history of Bd is composed of two stages: a thallus (body), which is present in amphibian skin and free-living zoospore, which is flagellated and motile in aquatic environments [76]. Zoospores can swim about 2 cm [77] and infect keratinizing squamous epithelial cells [78]. Favorable environments, where the infection can spread, are cool and wet. Hot and dry environments are considered hostile, and temperatures .25uC may assist infected amphibians in clearing the infection [76,79,80]. Given this, across amphibian species, behavioral and life history features of the animal, and ecological  features of the geographic region (ecoregion), will affect the course and extent of Bd infection [81,82,83,84].
In the present study, Bd was detected in 13 of 15 DoD installations, spanning the width of the North American continent.
Fifteen of 30 amphibian species sampled tested positive for Bd. There were both spatial and temporal patterns to Bd prevalence, as follows.
Bd is found in the highly secure environments of U.S.

DoD installations
In aggregate, the data for all bases over all three sampling periods (spring/early-summer, mid-summer, late-summer/fall) show a 16.6% prevalence of Bd infection. Bd was found in all but two installations, Camp Navajo in Arizona and Fort Sill in western Oklahoma. Lack of Bd detection on these bases may be the result of insufficient sampling during the first sampling period due to inclement weather (cold/snow). Amphibians were not active during the first sampling period at either of these bases and spring and early-summer was the time when Bd was most likely to be detected (78.5% of our positive samples came from this first sampling period; see below).
During this study we sampled about 10% (30) of all known United States amphibian species and found Bd in half of them, including four plethodontids, three bufonids, five hylids, and three ranids ( Table 2). While Bd absence in the remaining species may be due to inherent resistance [85,86] or ecological avoidance [81], it is most probable that in cases of no detection, individuals sampled happened to be negative, or to test negative at the time of sampling. It is likely that all amphibian species are susceptible to Bd infection, although species-specific variation in susceptibility has been shown [87], as has intraspecific variation in susceptibility [25]. Several of the species that tested Bd positive have tested positive in other studies; salamanders and ranids, including American Bullfrogs (Lithobates catesbeianus), may be carriers of this infection [88,89,90,91].  Table 4. Mean annual precipitation (cm) and temperature (uC) at each of the fifteen DoD installations sampled. There is a temporal (seasonal) pattern to the presence of Bd There was a strong temporal component to our dataset (Table 1; Fig. 3). In total, 78.5% of all positive samples came in the first (spring/early-summer) sampling period, and broken out by sampling period, the percent positive samples were 39.3% (168 of 427), 6.1% (29 of 477), and 4.5% (17 of 374). The data for the majority of bases (Marine Corps Base Camp Pendleton, Cannon Air Force Base, Camp Gruber, Fort Leonard Wood, Sparta Training Center, Naval Support Activity Crane, Fort Knox, and Fort Lee) followed a temporal pattern, but the data from some bases did not (Table 1). For example, three bases had no pattern of infection: Camp Navajo and Fort Sill had no Bd positive samples and Fort Belvoir had positive samples only in the spring/early summer period. Other bases had infection patterns that differed seasonally: Radford Army Arsenal had 15.8% positive in spring/ early summer, 33.3% positive samples in mid-summer, and 23.5% positive in late-summer/fall. Interestingly, and perhaps reflecting more favorable conditions for Bd (high moisture), three East-Coast installations-Fort A.P. Hill, Fort Lee, and Naval Air Station Oceana-had higher percentages of animals infected during the late-summer/fall than in the mid-summer period (Table 1). Overall, our data suggest a strong seasonal component to Bd infection, with the earliest sampling period showing the greatest incidence (Fig. 3).
Seasonality in Bd incidence has been previously demonstrated [27,92,93,94]. As summer proceeds, Bd-positive frogs appear to lose their infection [66,77,79,84,95]. It is also true that infected animals can develop chytridiomycosis and die, and thus be lost to later surveys. Just as we suggest the spatial pattern of Bd presence is due to variations in moisture levels (with moisture promoting infection incidence) we suggest the temporal (seasonal) pattern is due to moisture availability, with Bd present at the highest incidences during the wettest times of the year. Corroborating this hypothesis, our data suggest that Bd prevalence is lower in arid areas (xeric deserts and the arid Great Plains) and during drier times of the year (mid-to late-summer and fall). Temperature may be a covariate, with cooler temperatures promoting the infection, although in our study, by minimizing variation in latitude among sites in the transect [96], we also minimized, as much as possible in a continental transect, temperature differences.
There is a spatial pattern to the presence of Bd The eleven eastern temperate DoD installations had significantly higher prevalences of Bd infection (20.8%) than the four bases situated in the arid western ecosystems (8.5%). Bd went undetected at one of these bases (Camp Navajo); two arid bases each had single-digit levels of detection (Kirtland Air Force Base, 2%; Cannon Air Force Base, 6%). Marine Corps Base Camp Pendleton was the one exception. It had a 26% prevalence, but this installation is characterized by maritime Mediterranean climate, with more moisture than inland installations. Fort Sill was the second exception; Bd was not found on this eastern Oklahoma base. Bd is known to favor cool, moist conditions [38,97]. It therefore follows that warm and dry (i.e., arid) conditions may inhibit this pathogen. Our data are consistent with this interpretation (Fig. 2). Further, animals that bask have been shown to have an increased resistance to chytridiomycosis [79,80]. We find no evidence that particular species are driving geographic trends.
Our data both support and augment data on known distributions of Bd in the United States [14] (Fig. 1). Our data support the observations that Bd is widespread in California and common across portions of the southwest and throughout the eastern half of the country. The relatively few positive samples previously reported from the Great Plains may in part be due to the relatively few samples recorded from this region [14] (Fig. 1). In our study, positive samples from New Mexico, Oklahoma, and Illinois expand upon the sparse reporting from these states. Positive samples from Ft. Knox confirm the presence of Bd in Kentucky.
Our results suggest Bd may be endemic across much of the United States middle latitudes These data lend support to the conclusion [46] that Bd is now widespread across much of North America. Samples from DoD installations located in mesic habitats show a range of Bd infection from 7% (5/67) positive samples at Fort Lee, in Virginia, to 39% positive samples at Fort Belvoir. This spatial pattern-from coastto-coast-supports the assertion [46] that Bd, while once likely epidemic, today is endemic across much of the United States. Further, the phylogenetic range of species that tested positive-plethodontid salamanders in three genera, bufonid toads, treefrogs in three genera, and ranid frogs-suggests that this infection has been associated with aquatic ecosystems long enough to infect numerous taxonomic groups.
The exceptions to this generalization of endemism are two arid bases, Camp Navajo in Arizona and Fort Sill in Oklahoma. These installations warrant further study; there would be greater confidence with the conclusion that Bd is absent at these bases had we sampled large numbers of animals during the spring/earlysummer period. Our data do not distinguish between ''Bd present but not detected'' and ''Bd absent.'' There are known to be pockets of wilderness, for example in regions of the Sierra Nevada that Bd has yet to reach [45]. At these places, when Bd arrives it is predicted to be an epidemic infection, and we suspect amphibian extirpations will follow. Camp Navajo and Fort Sill may be examples of such Bd-negative outposts, but further study is needed.
Our results contribute to the recent and widespread literature being generated on the spatial patterns of Bd distribution tied to assessing risks and identifying refugia [98,99,100,101,102]. In particular, these authors suggest that amphibian species with populations inhabiting the drier interior of the United States may be less at risk than populations inhabiting more mesic regions. Attempts to ameliorate the effects of Bd in the field are being proposed [103]. Our data also suggest that broad geographic surveys, whether they be current or historical, be interpreted in light of seasonality in collection times. For example a directional transect beginning in one region and ending sometime later at a distant region, may interpret a geographic trend, when in fact the cause might be a seasonal trend. and support in the field. Special thanks to Mike Dreslik, Jennifer Heemeyer, Vanessa Kinney, and Nate Engbrecht for the statistical consultation. Danette Pratt created Figure 1. Susie Lannoo proofread early drafts. Thanks also to Allan Pessier and Joe Mendelson for numerous conversations on the nuiances of Bd sampling and interpretation.