Chytrid fungus dynamics and infections associated with movement distances in a red‐listed amphibian

Amphibians are among the most threatened vertebrate taxa due to anthropogenic habitat change and emerging pathogens. The fungus Batrachochytrium dendrobatidis (Bd) may cause trade-offs between the immune responses and other important functions, such as mobility. The pool frog (Pelophylax lessonae) is red-listed in Sweden, and an earlier study conducted in our study area did not detect the fungus. In this study, 67 Swedish pool frogs were tested for Bd. Of these, 28 adults were provided with external radio transmitters and tracked during the summer and autumn to compare movement among infected and non-infected individuals. Additionally, we tested some individuals for Bd twice during the summer to study intraseasonal variation in individual infections. Our results suggest that Bd has recently invaded these pool frog populations. During autumn, movement distances of infected pool frogs were shorter compared to uninfected individuals, and summer movements within ponds were reduced by increasing Bd load. We also found that the frogs can clear (or reduce) Bd infection during their active season. The results from this study increase understanding of the implications of Bd infections in an anuran fringe population, as well as in amphibian metapopulations, and can further guide amphibian conservation planning and management.


Introduction
Pathogens often reduce fitness of individuals and cause declines of animal populations. Apart from direct lethal effects, a decline in fitness may occur because resources required for elevated immune responses come at the expense of other important functions (Sheldon & Verhulst, 2008). A trade-off between resisting infection or tolerating sub-lethal infection and other important functions has previously been found in amphibians (Burrowes, Longo & Rodr ıguez, 2007;Chatfield et al., 2013;Bielby et al., 2015). Locomotion is often affected by these trade-offs, with negative consequences for the ability to escape predators, forage, disperse and reproduce (Chatfield et al., 2013). However, while trade-offs between disease and movement are well studied in birds and mammals, they are poorly studied in other taxa, including amphibians (Dougherty et al., 2018).
The amphibian chytrid fungus Batrachochytrium dendobatidis (Bd) has been detected in ca. 700 species and is regarded as one of the most serious threats to amphibian populations (Lips, 2016). Bd attacks the outer keratinized layers of skin, resulting in disruption of water, oxygen and dissolved solutes transport (Voyles et al., 2010) and has caused mass mortality and population declines of amphibians in many parts of the world (e.g. Wake & Vredenburg, 2009;Scheele et al., 2015;Lips, 2016).
Few studies have investigated the impact of Bd on amphibian movement, particularly in the case of wild animals. A laboratory study reported reduced peak speed of movement in Bdinfected leopard frogs Lithobates pipiens (Chatfield et al., 2013), while Daversa et al. (2018) reported no effect on movement in four infected radio-tracked spiny toads, Bufo spinosus. Bd can also reduce the body condition in amphibians (e.g. Retallick & Meira 2007;Pearl et al., 2009;K€ arvemo, Laurila & H€ oglund, 2019), but several studies have found no such impact (e.g. Deguise & Richardson, 2009;Murray et al., 2009;Woodhams & Alford, 2018).
Some species can, aided by antimicrobial peptides, temperature and immune responses, clear infections (Fisher et al., 2009). Recovery of amphibian individuals both within and between seasons has been reported for several species of treefrogs Litoria spp. (Murray et al., 2009;Kinney et al., 2011;Sapsford et al., 2017) and spiny toads B. spinosus (Daversa et al., 2018). Reduced Bd prevalence between winter/spring and summer/autumn has also been shown for several amphibian species in Australia (Berger et al., 2004;Kriger & Hero, 2007) and North America (Voordouw et al., 2006;Duncan Pullen, Best & Ware, 2010). The level of recovery within a species is suggested to be associated with Bd infection loads (Briggs, Knapp & Vredenburg, 2010) and clearing is therefore likely to reflect the relative risk that Bd will cause to the population (Gervasi et al., 2014). Thus, information about intra-seasonal dynamics of Bd can improve predictions of infection risks.
In this study, we radio-tracked and sampled Bd from pool frogs Pelophylax lessonae. In Sweden, this species is red-listed and occurs as a unique fringe metapopulation (Sj€ ogren-Gulve, 1991;Orizaola, Quintela & Laurila, 2010;Orizaola & Laurila, 2016) distributed over more than 100 ponds in south-central Sweden (Lindgren, Nilsson & S€ oderman, 2014). A study on museum material of European amphibians did not find Bd in 197 Swedish amphibian samples collected between 1994 and 2004, while Bd was detected in other parts of Europe within the same timeperiod (Garner et al., 2005). Additionally, in a previous study on Swedish pool frogs conducted between 2005 and 2008 within the same area as our study, all 137 tested individuals (a sample size large enough to ensure a correct assessment of Bd incidence; Adams et al., 2010) were Bd negative (Sainsbury et al., 2007). Since then, Bd has been found in most Swedish amphibian species (Hallengren, 2013;K€ arvemo et al., 2018;Meurling, 2019). This suggests that Bd has recently appeared in Sweden, but is now widespread. However, the current infection status of the pool frogs in Sweden is unknown.
The aims of this study were to (1) investigate evidence for recent invasion of Bd in pool frog populations in Sweden and if present; (2) evaluate potential costs of Bd infections in terms of movement distance; and (3) study intra-seasonal dynamics of Bd infection in individual frogs. We sampled pool frogs for Bd from six locations within the same region where Bd had not been detected more than 10 years ago ( Fig. 1; Foster, 2006;Sainsbury et al., 2007). Ten infected and 18 non-infected individuals were fitted with radio transmitters, allowing screening of some individuals twice for Bd during the study period and tracking of their movements throughout the active post-breeding period.

Methods
Sixty-seven pool frogs (25 adult males, 29 adult females and 13 juveniles) from six ponds were tested for Bd in summer and autumn 2017 (48 individuals) and in summer 2018 (19 individuals) in the Forsmark -H alln€ as area in the province of Uppland, central Sweden ( Fig. 1; Table S1). The six locations were surrounded by similar forest structures of pine-dominated mixed forests (Wikstr€ om, 2008). The pool frogs were considered as juveniles when their body mass was less than 10 g. For sampling Bd, we used the standard swabbing protocol of Brem, Mendelson and Lips (2007) with 25 strokes on the ventral skin and feet per individual. The mean number of individuals sampled per pond was 11.2 AE 3.2 (SE). At the sites Sl ataren and Rundsk€ ar, only three and four individuals were sampled, respectively. Bd infection prevalence was compared with a previous disease risk analysis that sampled Bd from adjacent populations in (Sainsbury et al., 2007. Of the 54 sampled adults, 28 were fitted with radio transmitters (PIP3 single cell tag A392; www.biotrack.co.uk) in 2017 with the intention of tracking them daily between 14 June and 25 July (summer) or 21 August and 10 October (autumn). We used external transmitters with waistbands of custom-fitted 1 mm elastic bead cord according to Groff et al. (2015). Terrestrial and pond positions of the pool frogs were recorded using a manual SIKA Radio Tracking Receiver combined with a flexible Yagi antenna (www.biotrack.co.uk). Each tracked position was recorded by a handheld GPS (Garmin GPSMAP 64) and 609 locations were recorded in total (summer: 423, autumn: 186). Even though the pool frog is a semi-aquatic species and spends much of its time in the breeding ponds, 1-12% of the local population can migrate annually to other ponds (Sj€ ogren-Gulve, 1994;Holenweg Peter, 2001). However, in this study, too few of the tracked individuals performed voluntary terrestrial migrations. Thus, to be able to sample movement data, 21 individuals were translocated 500 m from their breeding ponds. Such displacement is a common method to study amphibian movement distances (e.g. Sinsch, 1996;Smith & Green, 1998). However, the translocation can change the amphibian behaviour (Pettit, Greenlees & Shine 2017) and potentially increase the movement distances of amphibians, as the animals often orient back to their breeding ponds (Sinsch, 1996;Smith & Green, 1998). Consequently, translocation was included as a factor in the land-movement models (see Statistical analyses). Fourteen individuals were translocated in the summer and seven in the autumn (Table S1). In addition, seven individuals were recorded after natural (voluntary) terrestrial movements, mainly during the autumn migration to terrestrial hibernation habitats (Table S1).
Air temperature, sex and body size are important factors influencing amphibian movement (e.g. Holenweg Peter, 2001;Mazerolle, 2001 and references therein; Todd & Winnie 2006;Wikstr€ om, 2008) and were consequently evaluated to control for these variations. Data on sex, body mass (g) and snout-vent length (mm) were recorded for all individuals and ambient air temperature for each position was recorded with a Testo 922 thermometer (Testo AG, Lenzkirch, Germany). We estimated a body condition index (BCI) from the residuals of the linear regression between body length and weight. To detect whether the frogs cleared the infection during the summer season, eleven of the radio-tracked individuals were tested for Bd twice, once in early June and once in late July. These individuals were designated as infected in the analyses only if they were Bd-positive in the first sampling occasion in June. All methods were approved by and the research was carried out under permits granted by Uppsala ethical committee for animal experiments (5.8.18-2059/2017) and the county board of Uppsala (522-420-15). We used a Biorad CFX96 system and previously described amplification conditions (Boyle et al., 2004), with standards of 0.1, 1, 10 and 100 genomic equivalents (GE), for the qPCR assays. An individual was scored as Bd positive if one or more of the triplicate samples exhibited a positive signal (i.e. a clear exponential amplification curve). We classified the infection  Table 1. Empty diamonds indicate non-detection of Bd in a previous survey conducted between (Foster, 2006Sainsbury et al., 2007). level for each Bd-positive individual by comparing the sample cq-value to the Bd standards. These values were divided into three classes: <0.1, 0.1 and >0.1 < 1 GE standards, and used in additional models to evaluate the effects of relative differences in Bd infection loads. We used a classification of GE for a relative comparison of infection load among individuals, rather than the conversion into an absolute number of zoospores per individual. This is more suitable in our study as many methodological factors potentially affected fine-scale data (e.g. inhibitors originating from amphibian skin and the pond water, different fieldworkers carrying out the sampling). If the IPC showed signs of inhibition (no curve), negative samples were re-run once before being designated impossible to score. Scoring one of the triplicates as positive may cause some false positives due to potential contamination, which can result in overestimation of Bd-positive individuals. On the other hand, PCR inhibitors for Bd detection are very common in Pelophylax spp. compared to other species, which increases the risk of false negatives (Bal a z et al., 2014). Thus, we used the more inclusive strategy, and regarded all individuals with one (or more) positives as infected. However, to control for possible false positives we also validated our findings by also running analyses when excluding the samples when only one of the triplicates was positive (see Statistical analyses).

Statistical analyses
Movements (distance travelled) of pool frogs were analysed in linear mixed models (LMM) with Bd infection as a fixed factor, air temperature and body mass as covariates and individual as a random effect. To control for movement effects caused by translocations, this covariate was used in models including terrestrial movements, except for the Bd-load autumn-land models due to rank deficiency. Movement data were determined from the linear distances between two subsequent logger positions of each individual and were transformed to 1 + log in the analyses to attain normal distribution of the residuals. Body mass was nested within the random effect (individual), which effectively controls for pseudoreplication for all predictors (Schielzeth & Forstmeier, 2008). Explained model variance (R 2 ) was determined from the MuMIn package function r.squaredGLMM in program R (Barton, 2018). The LMM analyses included data with (1) pooled land and pond movements, as well as separate analyses of (2) land and (3) pond movements. These analyses were in turn divided into separate models for summer and autumn movements, resulting in six different main models of Bd occurrence effects on movements. This was done because we were interested in the separate movements across land and ponds, and because no time gap should occur in movement analyses (Calenge, 2006). For occasional days without field work, or if no signal was recorded, missing positions (NAs) were included in the data, resulting in regular daily trajectories (Calenge, 2006). For comparisons, data for movements from translocated and natural individuals were additionally analysed separately. We also validated the Bd-occurrence models where Bd was significant with additional models where we excluded samples with only one positive amplification (and possible risk of false positives; see Molecular analyses).
Four Bd-load models (including pooled land and pond movements across both seasons, pond movements in summer and land movements in autumn) were constructed to evaluate if movements of the infected individuals were associated with the amount of zoospores. Summer-land and autumn-pond movements were excluded, as the variation in Bd loads was too low, or there were too few pond observations of infected individuals. Potential false positives in Bd-load models (exclusion of single amplifications) could not be validated, as there was too little variation in Bd loads. Some of the models did not converged and were fitted by an optimization of the model algorithm via the Nelder-Mead function of the lmerControl parameter (Bates, 2010). This procedure minimizes the model function in a multidimensional scale (Nelder & Mead, 1965). This was done for the: pooled summer-land and pond models and summer-land models for Bd occurrence, and pooled autumn-land and pond models for Bd loads.
Chi-square tests were used to compare infection prevalence between males, females and juveniles. Differences in BCI and body mass between infected and non-infected males, females and juveniles were analysed by t-tests and by a Wilcoxon test for juvenile BCI. Daily summer and autumn movements (distances divided by days (log)) of translocated individuals were compared with t-tests. The program R (ver. 3.5.0; Foundation for Statistical Computing, Vienna, Austria, https://www.R-pro ject.org/) was used for all statistical analyses with the package lme4 (Bates, 2010) for mixed model analyses.

Results
In contrast to the previous study from 2005 to 2008, which found zero incidence of Bd-positive individuals (Sainsbury et al., 2007), all six ponds tested positively for Bd in 2017-2018 (Fig. 1), that is at least one frog was tested positive on at least one sampling occasion. However, this includes Sl ataren and Rundsk€ ar with only one positive individual each, which both were based on one single amplification. The total individual prevalence was 31% (21 of 67; range 11-50% among sites/ponds; Table 1) with a mean of 29.7% (95% CI: 14.7%, 44.6%) across the populations. Infection load class in infected individuals varied between < 0.1 and 1 GE (Table S1). The percentage of infected individuals was 40% for males (10 of 25), 17% for females (five of 29) and 46% for juveniles (six of 13). We found no significant differences in mean Bd prevalence among males and females or juveniles (males-females: v 2 = 2.42, P = 0.119; males-juveniles: v 2 < 0.01, P = 0.798; juveniles-females: v 2 = 2.53, P = 0.112). We did not find any differences in mean body condition or body mass between infected and non-infected individuals, either for males, females or juveniles (Table S2).

Movements
Analyses of pooled data for land and pond movement distances of the radio-tracked frogs indicated that movements were negatively associated with occurrence of Bd infection, but only during the autumn (Fig. 2a-c; Table S3), and were positively associated with body mass during the summer ( Fig. 2a; Table S4). When analysing land and pond data separately however, neither the association between movement distance and body mass, nor movement distance and Bd occurrence, were significant. The main Bd-occurrence models without the single-amplified Bd positives (to examine possible effect of false positives), also showed negative effects of Bd on pooled land and pond movements, and additionally on land movements in autumn (Table S5). Only including translocated individuals in analyses of autumn movements resulted in negative effects of Bd and positive effects of temperature on frog movement distances (Table S6). Daily mean distances (log) of translocated individuals were shorter in the summer than in the autumn (t-test: À3.56; P < 0.001), in line with their more dispersal-prone behaviour during autumn. Thus, the strongest negative effect of Bd occurrence on movements was observed when individuals moved during (relatively) long-distance migration to ponds. The voluntary movements that occurred, mainly towards hibernation sites, were in contrast, not affected by any of the explanatory variables (Table S7).
Bd loads were negatively associated with movement distances, but only within ponds during the summer (Fig. 3a-c; Table S8-S9). There was no association between movement distances and ambient air temperature, either in pooled or separated land and pond models. The number of radio-tracked Bdinfected individuals, Bd-load classes of tracked individuals and number of recorded positions on land (voluntary and translocated) and in ponds are presented in Table S1.

Intra-seasonal dynamics of Bd
Eleven radio-tracked individuals were sampled for Bd twice, once in early June and once in late July (Table 2). Over this period, individual prevalence of Bd was reduced from 55% to 27%. Four of eleven individuals remained Bd negative, two remained positive, four of the positive individuals lost or reduced their infection and one of the previous negative individuals became infected. These positives also include three single amplifications (Table 2).

Discussion
Our six study sites were located within the same area as the six sites in the previous study on Bd infection in Swedish pool frogs, which took place from 2005 to 2008. That study found that all 137 tested individuals (72 larvae, 29 juveniles and 36 adults) were Bd negative (Sainsbury et al., 2007). In sharp contrast, we found Bd-infected pool frogs in all of our six studied ponds, with a total infection prevalence of 31%, strongly suggesting that Bd has either invaded the area within the last decade or become much more common. A previous study showed no difference in Bd detection between the two sampling methods that were used in the two compared studies, that is toe clipping and skin swabs, respectively (Livo et al., 2004). The indication of a recent invasion of Bd in Sweden are consistent with a study on museum material that did not find Bd in 197 Swedish amphibian samples collected between 1994 and 2004 (Garner et al., 2005), and more recent findings (2015-2019) of Bd in all nine Swedish amphibian species studied, with prevalence varying from two to 61% among

Movements
Amphibian activity may be affected by environmental (context dependent) and behavioural (phenotype dependent) factors (Joly, 2019). For example, landscape structure and fragmentation, as well as sex, age and condition, may directly or indirectly influence migration and are consequently important features in amphibian conservation. Based on our results, we cannot exclude that pool frog movements are negatively affected by Bd infection, particularly during relatively long-distance movement between ponds. In addition, pond movements in the summer were negatively associated with Bd loads of infected individuals. Reduced movement distances of infected individuals have been reported also in other systems such as house sparrows (Bonneaud et al., 2003) and wolves (Cross et al., 2016). The impact of infections on amphibian movement distances is poorly investigated, but bacterial infection load was associated with reduced movement distance in cane toads Rhinella marina (Brown & Shine, 2014), and peak velocity and acceleration were reduced in Bd-infected leopard frogs Lithobates pipiens (Chatfield et al., 2013). The reduced movement distances associated with Bd can be caused by a trade-off between immune system and mobility, suggesting that the energetic cost caused by the fungal infection affects movement (Chatfield et al., 2013;Brown & Shine, 2014). Our results may alternatively be due to sedentary individuals that spend more time in the water where they are more exposed to infections, compared to individuals with a more mobile behaviourthat spend more time on dry landand thus will be less exposed to Bd (Johnson & Speare, 2005;Terrell et al., 2014). Assuming a causal link between infection and reduced mobility, the Bd-mediated reduced mobility may result in higher predation risk, lower reproduction (Johnson et al., 2006) and other fitness costs (Burrowes et al., 2007;Bielby et al., 2015). In our study populationas well as in many other amphibians (Smith & Green, 1998)reduced movement caused by Bd may also have a negative impact on the metapopulation structure and survival, as proximity and migration rate to neighbouring sites is a key factor preventing local extinctions (Sj€ ogren-Gulve, 1991). Extinctions of local populations may potentially increase the number of isolated patches in a metapopulation, which in turn can be critical for persistence of many amphibian populations (Marsh & Trenham,Figure 3 Estimates and standard errors of translocation, ambient temperature, body mass and Bd loads (three classes) in separate models of summer and autumn movement distances, including (a) pooled movements on land and in ponds, and (b) on land, and (c) in ponds. Summerland movements and autumn-pond movements are excluded due to a too low variation in Bd-load classes and too few pond observations of infected individuals in the autumn. Significant effects are indicated ** = 0.01> P > 0.001. Table 2 Results from repetitive Bd sampling of radio-tracked adult pool frogs in June and end of July 2017, from Vedl€ osa (V), G€ addalen (G) and Rundsk€ ar (R)

Ind
Sex 170608-170615 170725-170728 Male Bd+ a BdÀ % Bd positives 55 27 BdÀ indicates no detection, and Bd + indicates detection of Bd in each sampling period.  (Sj€ ogren, 1991). However, to draw general conclusions about movements associated with the metapopulation viability, further studies across different spatial and temporal scales and life-stages would be needed (Bailey & Muths, 2019). While we note that three of four infected individuals tracked during autumn had the lowest movement distances of all late-season individuals (Fig. 4), the limited number of radio-tracked Bd-infected individualsand on some occasions a limited number of observed movementsmeans that broader conclusions from the present study should be made with caution. The result of reduced pond movements associated with infection loads in the summer was however, based on seven infected individuals. We found that Bd occurrence and translocations had a stronger effect on autumn than on summer movements. In addition, the mean of daily distances of translocated individuals in the autumn were more than twice as long, compared to translocated individuals in the summer and the weak effect on movements during the summer may thus be caused by an unnatural migration situation for the season (Sj€ ogren Gulve, 1998;Holenweg Peter, 2001). Translocated individuals were also moving longer distances than voluntary individuals, which may be caused by a homing behaviour that can increase movement distances considerable compared to the normal range of movements (Phillips, Adler & Borland 1995). In contrast to Bd occurrence, the amount of detected Bd zoospores in infected individuals was negatively associated with the summer-pond movements (Fig. 5). While this result may reflect a true negative effect of Bd load on frog movements, an alternative explanation predicts higher loads on individuals remaining in the ponds, where conditions for Bd growth are better than in the terrestrial habitat (e.g. Berger et al., 2004;Terrell et al., 2005). Summer movement distances in the Bd-occurrence models increased with body mass, which may be associated with higher investment in reproduction and territorial behaviour in larger individuals (Camargo et al., 2005). Unlike some earlier studies on other amphibian species (Murray et al., 2009;Chatfield et al., 2013;Woodhams & Alford, 2005;K€ arvemo et al., 2019), Bd infection in pool frogs was not linked with body size or condition.

2001), including the pool frog in Sweden
Models without the individuals with only one positive amplification showed similar negative effects of Bd on movements as the models including all individuals, indicating that the results are not driven by the single amplifications. Bd detection from Pelophylax spp. is strongly affected by PCR inhibitors (Bal a z et al., 2014) which increase the risk of single positive amplifications. Thus, for pool frogs, false negatives may be a larger problem than false positives. The relatively Figure 4 Boxplots of autumn movements distances per day (pooled pond and land movements) of individual Bd-negative and Bd-positive adult pool frogs (x-axis). Bar: median; box: interquartile range (IQR); whiskers: min/max. values < 1.5 9 IQR below/above box; dots: outliers.
high prevalence in Swedish pool frogs is comparable with other studies on this species (Tobler, Borgula & Schmidt, 2012;Bal a z et al., 2014).

Intra-seasonal dynamics of Bd
Four of the eleven individuals that were sampled for Bd twice lost their infection between June and July (prevalence decreasing from 55% to 27%), indicating that pool frogs can clear Bd infections, or that Bd is reduced to such low doses that it is no longer detectable (Bielby et al., 2015). This is in accordance with previous studies reporting recovery of amphibian individuals within and between seasons (Murray et al., 2009;Kinney et al., 2011). Earlier studies have shown reduced Bd prevalence in temperate regions between spring and summer (Gaertner et al., 2009;Duncan Pullen et al., 2010;Voordouw et al., 2006, but see Kinney et al., 2011. In our case, this can potentially be explained by differences in temperature between June and July, with more suitable temperatures for Bd occurring in early summer (Kriger & Hero, 2007). Another possible explanation for this pattern may be that three of the four individuals that cleared the infections were translocated to dry and warmer terrestrial habitats, which may help to clear infections (Johnson & Speare, 2005;Terrell et al., 2005;Chatfield & Richards-Zawacki, 2011). However, two individuals remained infected despite being translocated.

Conclusions
The pool frog population in central Sweden has probably become exposed to Bd within the last decade. They show relatively high pond and population prevalence of Bd infections, possibly due to their aquatic lifestyle. This isolated metapopulation is one of the very few populations of this species in the Nordic countries (Zeisset & Hoogesteger, 2018) and is included in the Swedish Red List (VU). Our results suggest that while pool frogs can clear or reduce Bd infection, the infection may cause a reduction in movement. This in turn may affect their foraging performance, reproduction and dispersal, and increased susceptibility to predation. It is therefore critical to temporally monitor the present population in order to detect potential declines, and should they occur, mitigate adverse effects of Bd via conservation measures.

Supporting Information
Additional Supporting Information may be found in the online version of this article: Table S1. Pool frogs (Pelophylax lessonae) included in the study. Site denote names of the six studied locations, in the Sex column is M = male, F = female and J = Juvenile, Date is when the Bd sample was taken, Body mass and Body lenght is weight in gram and Snout-Vent-Length in mm.  Table S2. Statistical evaluation of mean body condition (BCI) and body mass between infected (Bd+: AEse) and non-infected (Bd-:AEse) pool frog individuals. Table S3. Outcomes from the linear mixed models of autumn movement distances of pool frogs with translocation (a-b; models including land data), ambient temperatures body mass and Bd occurrences as explanatory variables (a) on land and in ponds, (b) on land and (c) in ponds. Individual (random effect) were nested with body mass in the models. Deviations are based on the ML of the fixed effect variables. Table S4. Outcomes from the linear mixed models of summer movement distances of pool frogs with translocation (a-b; models including land data), ambient temperatures body mass and Bd occurrences as explanatory variables (a) on land and in ponds, (b) on land and (c) in ponds. Individual (random effect) were nested with body mass in the models. Deviations are based on the ML of the fixed effect variables. Table S5. Outcomes from the linear mixed models of pooled data for autumn-land and pond movements and autumn-land movement distances of pool frogs, with translocation, ambient temperaturesexclusively with more than one positive amplification, body mass and Bd occurrences as explanatory variables (a) on land and in ponds (autumn), (b) on land (autumn) and (c) in ponds (summer). Individual (random effect) were nested with body mass in the models. Deviations are based on the ML of the fixed effect variables. Table S6. Outcomes from the linear mixed models of autumn-land-voluntary movement distances of pool frogs with ambient temperatures, body mass and Bd occurrences as explanatory variables. Individual (random effect) were nested with body mass in the models. Deviations are based on the ML of the fixed effect variables. Table S7. Outcomes from the linear mixed models of autumn-land movement distances of translocated pool frogs with ambient temperatures, body mass and Bd occurrences as explanatory variables. Individual (random effect) were nested with body mass in the models. Deviations are based on the ML of the fixed effect variables. Table S8. Outcomes from the linear mixed models of summer movement distances of pool frogs with ambient temperatures, body mass and Bd loads (Genomic equivalents) as explanatory variables (a) on land and in ponds, and (b) in ponds. Land movements did not converge. Individual (random effect) were nested with body mass in the models. Table S9. Outcomes from the linear mixed models of autumn movement distances of pool frogs with ambient temperatures, body mass and Bd loads (Genomic equivalents) as explanatory variables (a) on land and in ponds, and (b) on land (no variation in Translocation and thus excluded and pond observation of infected individuals were too few). Individual (random effect) were nested with body mass in the models. Deviations are based on the ML of the fixed effect variables.