A combination of environmental and landscape variables drives movement and habitat use in two Anaxyrus toads Authors: Alexander M. Ferentinos, Courtney E. Check, Olivia Windorf, Matthias Leu Published: 2025 ABSTRACT BackgroundMovement patterns provide valuable information to guide habitat management strategies for declining amphibian populations. However, our understanding of movement ecology in amphibians is incomplete, as previous research has focused largely on environmental conditions driving movements rather than habitat drivers or landscape context. Here, we document the movement patterns of eastern American (Anaxyrus americanus americanus) and Fowler’s (Anaxyrus fowleri) toads as they transitioned from the breeding to the nonbreeding season. Specifically, we estimated the median distance migrated from breeding sites, identified the environmental and landscape level drivers of movement, estimated site fidelity to refugia on the basis of random walks, and evaluated refugia microhabitat use.MethodsWe tracked both taxa between late spring and summer in 2017, 2018, and 2023 in Virgina, USA. In total, we collected movement data from 46 eastern American and 57 Fowler’s toads on the basis of a combination of Harmonic Direction-Finding (HDF) and radiotelemetry relocation techniques.ResultsMedian distance moved from breeding grounds was similar for eastern American (63 m; interquartile range [IQR]: 43–122 m) and Fowler’s toads (64 m; IQR: 31–73 m), but the distribution of distances differed between the two taxa. Toads were more likely to move when minimum temperatures were high, after rain, in an area with a high proportion of conifers, and when far from water. Toads moved further when far away from trails. The probability and distance of a move differed among years. Both taxa exhibited site fidelity to daily refugia. Eastern American toads used woody structures and leaf litter as refugia more frequently than Fowler’s toad.ConclusionsThe distances which we tracked toads from their breeding grounds suggest that current regulations to protect amphibians are not biologically sound. To protect amphibians from decline, we suggest habitats should be protected at biologically meaningful scales based on data from tracking studies. Our results indicate that toads appear to use human trails as feeding substrates based on their activity and pattern of movement in relation to them. Trails likely improve foraging success by making invertebrate prey more obvious.Supplementary InformationThe online version contains supplementary material available at 10.1186/s40462-025-00613-x. Background Eastern deciduous forest ecosystem composition (i.e., tree species diversity) and configuration (i.e., forest patch size and arrangement) have changed drastically over the past 100 years. Economic shifts away from agriculture initially enabled regrowth [1, 2], but towards the end of the 20th century through the present day, forest cover declines have increased again [3]. Causes of this renewed decline include the conversion of forest patches into urban-rural gradients, as the amount of exurban area increased by five to seven times between 1950 and 2000 [4]. This has led to increasing fragmentation of forest land cover [5, 6]. A variety of species reliant on extensive forest patches experience challenges from this fragmentation, such as increased difficulty in moving among patches and greater exposure to edge effects [7–9]. Among the taxa being most affected are anurans (i.e., toads and frogs), which are increasingly facing hurdles to disperse and migrate in fragmented landscapes [8, 10]. Many anurans migrate between breeding and non-breeding habitats on an annual basis [11]. Within the eastern deciduous forest, these species commonly use isolated wetlands for breeding and the surrounding upland forest for non-breeding habitat [12]. Individuals can move large distances between their breeding and non-breeding sites, as average migration distances between 100 and 1000 m have been recorded [13]. Fragmentation and loss of habitat are consistently associated with declines in amphibian populations, as the creation of barriers to migration and dispersal decreases the proportion of successful movements [8, 14]. This is particularly problematic for amphibians due to their low dispersal capabilities relative to other taxa, leading to a greater number of local extinctions as source populations become isolated [14, 15]. When these limitations are removed, such as with the installation of road tunnels and fencing to direct migrating amphibians along safe routes [16, 17], or with the establishment of vegetated buffers around breeding sites, amphibian populations are adequately conserved and recover [18–20]. However, the effectiveness of vegetative buffers can vary, indicating that the buffer size required to conserve a population remains a key question [21, 22], with a call for research focusing for buffer sizes based on species-specific needs [23–25]. Here we investigate the movement ecology of two representative taxa of upland anurans: the eastern American toad (Anaxyrus americanus americanus) and the Fowler’s toad (Anaxyrus fowleri). Eastern American toads are a widespread subspecies found throughout northeastern North America [26], while Fowler’s toads occupy most of the eastern half of the U.S. [27]. While both taxa are considered secure at the range wide scale, Fowler’s toads have declined at local scales. Populations across the state of Virginia decreased by approximately 53% between 1999 and 2012 [28], and in Ontario, Canada, and several U.S. states at the periphery of its range, it is considered locally threatened [29, 30]. Previous studies have quantified a variety of migration distances in American toads [42 to 2,932 m; 31–33], but less is known about Fowler’s toads, as research has largely focused on a declining isolated population in Ontario, Canada [30]. The diet of both taxa is primarily composed of invertebrates [34, 35]. We tracked these two taxa to understand how their movement patterns relate to their core habitat size, drivers of movement, microhabitat use, and site fidelity. The distance that adult amphibians migrate to and from breeding sites is used to define a migratory breeding species’ core habitat, and therefore the minimum biologically relevant buffer, as this is the habitat that is necessary to support all life-history stages and activities [23]. Analyzing the variables associated with the movement of these species offers valuable information about interactions with the environment that can limit or initiate movement. A large body of work on movement ecology has identified amphibians’ dependence on appropriately warm and rainy weather conditions for movement, particularly in the breeding season [36–38]. Examining the effect of the landscape context in addition to the environment provides valuable information about habitat and resource use [39, 40]. Within these landscapes, microhabitats are a crucial component of habitat for amphibians, as they are utilized as refugia from adverse environmental conditions, and to hide from predators [41, 42]. Toads may also exhibit site fidelity to these microhabitats, as they do to their breeding grounds [43], and this can potentially be used to identify a state change from migration to a return to an established home range [13, 44]. Understanding movement patterns in frogs and toads is crucial to informing effective conservation strategies [23, 45]. Limited work has been done to evaluate amphibian movement on a multispecies basis. Amphibians are under-represented in the movement ecology literature, largely due to the difficulty of attaching transmitters to them [46–48]. However, the need to understand their movement is of increasing importance to protect their habitat and to ensure that migration and dispersal networks remain connected [13, 23, 49]. While prior work has focused largely on individual species [32, 38, 50–52], here we use a multitaxon approach with two con-generic taxa to more broadly apply our findings to other species of anurans which occupy forest habitat away from breeding grounds. We focus on identifying causes of movement to identify periods of anuran activity when their vulnerability increases [17, 53], and to identify movement distances from breeding grounds to provide the biologically relevant distances of core habitat areas requested by Semlitcsh and Bodie, 2003 [23]. Our work places amphibian movement within the spatial and environmental context in which it occurs, in order to understand how external factors drive decisions about when and how far to move [54]. This study is one of a few that uses a combination of harmonic direction-finding (HDF) and radiotelemetry relocation techniques to (1) examine how far toads travel from breeding sites to define their core habitat, (2) identify the environmental and landscape level drivers of movement, (3) evaluate site fidelity to refugia, and (4) examine differences in refugia microhabitat between eastern American toads and Fowler’s toads from late spring to late summer (i.e. the season in which these taxa are active). We hypothesized for objective two that amphibian movement patterns are governed by their dependence on moist environments to prevent desiccation and by ambient temperatures that increase their metabolism due to their poikilothermic nature. We predicted that toads would move larger distances on days with high humidity, after rainfall, and higher temperatures [55–57]. Because landscape-level effects of habitat composition and configuration are poorly understood for amphibian movement ecology, we also examined several landscape variables to explore the potential effects of habitat use on movement [39]. We hypothesized for objective three that utilizing known refugia confers survival benefits [58] by allowing toads to avoid predators [59], and to minimize exposure to inclement weather conditions [42, 60]. As it is well known that toads exhibit site fidelity to breeding locations [43, 49, 61], we predicted that both toads would also display site fidelity to daily refugia in the post-breeding season. We do not present hypotheses for objective one and four because their findings are descriptive in nature and are intended to inform management of these and similar taxa [62]. Methods Study areas We studied the movement of eastern American toads and Fowler’s toads at four study sites in Williamsburg, VA (Fig. 1), three of which were used in 2017, 2018, and 2023 (College Woods [CW], Greensprings Interpretive Trail [GIT], and Warhill Sports Complex [WSC]), and one (Greensprings Field [GF]) that was only used in 2023. Fig. 1– Map of the study areas on the Middle Peninsula east of Richmond, Virginia. Wide, light gray lines are major roads, while narrow, dark gray lines are streams. The four major study areas were the College Woods (CW, black outline), Greensprings Interpretive Trail (GIT, purple outline), Greensprings Field (GF, orange outline), and Warhill Sports Complex (WSC, green outline). The insert map in the upper-left corner depicts the location of the study areas in the Commonwealth of Virginia, USA – Map of the study areas on the Middle Peninsula east of Richmond, Virginia. Wide, light gray lines are major roads, while narrow, dark gray lines are streams. The four major study areas were the College Woods (CW, black outline), Greensprings Interpretive Trail (GIT, purple outline), Greensprings Field (GF, orange outline), and Warhill Sports Complex (WSC, green outline). The insert map in the upper-left corner depicts the location of the study areas in the Commonwealth of Virginia, USA CW (37.2752 oN, 76.7234 oW) was composed primarily of a mixed beech (Fagus grandifolia) and maple (Acer spp.) forest with several shallow wetlands surrounding a large reservoir lake and included several graveled hiking trails. GIT (37.2499 oN, 76.7913 oW) consisted of mixed hardwood and conifer forests surrounding a beaver-maintained wetland and included several well-maintained hiking and biking trails. GF (37.2397 oN, 76.7868 oW) was located on a portion of the paved Virginia Capital bike trail to the southeast of GIT and bordered on one side by a mixed-deciduous forest and on the other side by rotated crop farm fields. The terrain on both sides of the bike trail had depressions and ditches, which filled with water during rain events and were used as breeding sites by both toad taxa. WC (37.324 oN, 76.7603oW) consisted of a large reservoir, surrounded by mixed deciduous-conifer forest, along with a well-maintained set of graveled hiking trails. Tracking methods We captured toads from sunset to midnight at all study sites from late March to early October and tagged them with either an HDF tag or a radiotransmitter. In total, we tagged 103 toads. We outfitted 40 toads (eastern American toad, n = 23; Fowler’s toad, n = 17) with HDF in 2017 and 2018, and an additional 36 (eastern American toads, n = 12, Fowler’s toads, n = 24) in 2023. In 2023 only, we outfitted 27 with radiotransmitters (eastern American toads, n = 11, Fowler’s toads, n = 16). Toads were manually relocated every one to three days until they could no longer be found, or their tag fell off. For a full description of our tagging and relocation methods see supplementary material 1. Movement The distances between subsequent relocations of toads were calculated using the R package sf [63]. Distances between relocations were divided by the number of days between them to calculate average daily movements. We did not include the capture location and first relocation for each toad in this or subsequent analyses to avoid the impacts of potential fear-driven movements due to tagging. We evaluated variation in average daily movements on the basis of time of year, landscape data, and environmental data. We included day of year as anuran movement varies in relation to time of year [64, 65], and because the movement data were exclusively post breeding for the eastern American toad (breeding season in North and South Carolina: February – March; NC and SC are the closest known breeding period locations), but included movement during the breeding season for the Fowler’s toad (breeding season in Virginia: April – July) [66]. Landscape predictor variables We examined terrain ruggedness because rugged terrain is less favorable to amphibian movements [67, 68]. We examined deciduous and coniferous forest cover because studies have shown that eastern American toads frequently make use of woody debris in forests for shelter, while Fowler’s toads generally favor more open habitat types [28, 69, 70]. We also examined the anthropogenic feature of distance to trail (either gravel or paved as described in the Study Site section), because we observed that toads were frequently on trails at night, and experimentation with cane toads (Rhinella marina) found that they made more attempts to capture prey on trail-like substrates [71]. We examined distance from the nearest breeding grounds and water bodies because we were interested if proximity to them related to movements of toads, even outside of the breeding season. We identified breeding areas by mapping vocalizing individuals over several nights of observation. We developed a terrain ruggedness index (TRI) raster from the national Digital Elevation Model at the 1-m resolution [72] on the basis of the “Ruggedness Index” tool in QGIS [73]. A TRI is created via a 3 × 3 moving-window analysis where the new value of the central cell is calculated as the square root of the sum of squares of the difference between the central cell and the surrounding cells in each window. A high value indicates a location with large differences in elevation between adjacent cells (i.e., steep terrain), while a low value indicates a location with small differences in elevation (i.e., flat areas). Landcover data were obtained from the Virginia Land Cover Dataset, available at 1-m resolution [74]. We reclassified landcover by combining fine-resolution categories, such as levels of imperviousness, into coarse-level categories (supplementary material 2, Table S1). Using images from NASA Planet Labs at 3-m resolution, we classified coniferous landcover on the basis of winter scenes [75], using the Supervised Classification tool in ENVI version 5.7 [76]. If not available online, we digitized trails from aerial imagery and with a GPS unit (Garmin GPSMAP 64s Olathe, KS) while walking trails. Reclassified landcover data, including the proportion of forest and coniferous forest, were sampled with moving-window analyses at three extents reflecting observed non-zero movement distances of both toad taxa: 1st quartile (3 m), median (7 m), and 3rd quartile (17 m). Exponential decay functions ([Euclidean distance/negative scalar] where scalars were the three aforementioned extents), were also fitted to the distances of toads from trails, water, and breeding sites to investigate non-linear associations [77]. Different biologically derived scales were used for each variable to reflect the potential scales at which toads perceive and interact with their environment [69]. We scaled predictor variables when assessed in linear associations. Delineations of bodies of water were taken from a combined dataset of both open water based on the Virginia Land Cover Dataset [74], as well as small streams based on the National Hydrography Dataset [78]. Several bodies of water which were absent from these data sets were also hand digitized. All spatial analyses were conducted in ArcGIS Pro version 3.2 unless noted otherwise [79]. Environmental predictor variables We examined daily temperature, rainfall, and relative humidity because increases in these variables are associated with greater movements by toads [36–38]. Daily maximum, mean, and minimum temperatures, rainfall, and mean dew point were obtained from the PRISM data set, which is available at the 4-km resolution [80]. Relative humidity was calculated on the basis of the mean temperature and mean dewpoint temperature [81], and rainfall was transformed into a one day total as well as seven cumulative totals ranging from 2 to 7 days and 14 days, as our pilot studies found that cumulative totals were more informative than daily totals (C.E. Check and O. Windorf, unpublished observations). Weather data were joined to dates with a lag of 1 day, as toads primarily move at night, indicating that the previous day’s weather would be a more accurate predictor of movement [64, 82]. Data analyses Migration distance We compared the distributions of the distances that the two taxa of toads were observed from their known breeding sites with a two-sample Kolmogorov-Smirnov Test given that distance data for both taxa were not normally distributed. Breeding site locations were known from several years of prior breeding call surveys. This statistical analysis, and all others, were conducted in R, version 4.3.1 [83]. Movement We used generalized linear mixed-effects models (GLMM)from the glmmTMB package [83] to relate variation in distances moved to species or subspecies (binary variable with eastern American toad set to 1), tracking method (binary variable with radiotelemetry set to 1), Julian date, landscape, and environmental predictor variables. Site and individual toad were included as a nested random intercept as we expected distance moved to differ among sites and individuals. To improve precision among parameter estimates, we only included individuals with six or more observations. We used a hurdle model with a binomial and gamma distributed component because the data were continuous and had a large proportion of zeroes (846/1454, 58.2%). This allowed us to both model if a variable caused a movement to occur and if it affected the distance of a movement. Each level of a given predictor variable was independently examined against the null model (i.e., intercept model). We used the Akaike Information Criterion corrected for small sample size (AICc) [84] to select which level of a candidate variable, if any, to include in a final model. We did not carry a candidate variable forward to the final model if the null model was the best model or was within 2 AICc units of the top model in the univariate analysis. We checked the final set of top candidate variables for multicollinearity on the basis of the variance inflation factor (VIF), considering lack of multicollinearity if VIF < 10 [85]. All combinations of top candidate variables that were carried forward were evaluated with AICc with the dredge function of the MuMin package [86]. The effect sizes and standard errors of all models with a better performance than the null model were model averaged to derive the final model in order to reduce model selection bias [87]. The residuals for the final model were evaluated with the DHARMa package [88]. Site fidelity We used the fidelity package [44] to generate a null model of site fidelity for each individual on the basis of the Blischke-Scheuer method-of-moments estimator, which calculates the step length parameters of shape and scale for a comparison with simulated toad movements according to a Correlated Random Walk (CRW). We included every individual with three or more relocations (i.e., the minimum number of observations to count returns) after the removal of the initial capture points in this analysis. The Blischke-Scheuer method-of-moments estimator was used due to the presence of zeros in the movement between observations [89]. Movement distances between observations were used instead of the averaged daily distance to avoid producing underestimates of the movements of the simulated toads in the event of irregular sampling intervals. We defined the distance threshold for considering a movement to be a return to a previous site as 3.65 m, which was based on the margin of error of the Garmin GPSMAP 64s we used [90]. An individual had to be absent from a previously occupied site for at least one observation day before returning to be considered a return. We compared the number of returns, standardized by the number of steps of each individual, between the observed and simulated movements within the taxa derived from a CRW with paired t-tests. We then compared the observed returns of the two taxa with a Mann-Whitney U test due to a lack of normality in the observed data. We evaluated the data used in all tests for the assumption of normality with Shapiro-Wilk tests and the visual inspection of quantile-quantile (QQ) plots and histograms. Microhabitat analysis We compared microhabitat use between taxa with a binomial GLMM in the lme4 package, with site set as a random intercept [91]. The response variable was a binary variable with eastern American toad set as 1, and the predictor variable was the microhabitat type. Microhabitat variables were collated into major categories (supplementary material 2 Table S2). We used the same statistical approach to estimate the final model as outlined for the final distance-moved model. Results Distance migrated from breeding sites The median distance of eastern American toads from their breeding sites was 63 m (IQR: 43–122 m, n = 632) and for Fowler’s toads it was 64 m (IQR: 31–73 m, n = 865). The distributions of the distance migrated from their breeding sites were significantly different (D = 0.25, p < 0.0001, Fig. 2). We found that both taxa did not move between observations over 57.5% of the time (861/1497 observations). The stationary data comprised 58.5% of eastern American toad observations (370/632) and 56.8% for Fowler’s toads (491/865 observations). Non-zero movements for both toad taxa combined had a median of 7 m (3–17 m). Fig. 2– Histogram of the distribution of Anaxyrus americanus americanus (light-green, n = 632) and Anaxyrus fowleri (blue, n = 865) distances that they were observed from their respective breeding sites. Colors are transparent with dark green indicating overlaps between the two taxa – Histogram of the distribution of Anaxyrus americanus americanus (light-green, n = 632) and Anaxyrus fowleri (blue, n = 865) distances that they were observed from their respective breeding sites. Colors are transparent with dark green indicating overlaps between the two taxa Environmental and landscape level drivers of movement We examined 13 candidate predictor variables as drivers of movement events and average daily movement (supplementary material 2 Table S3-6) but included only 11 in our final analyses, with the remaining variables having higher AICc values than or were within 2 ΔAICc from the null model. The final model predicted that a movement event (Fig. 3a) was more likely to occur when minimum temperatures and rainfall of the previous day were higher, in areas with elevated proportions of coniferous landcover within 17 m, and when individuals were farther from water (decay distance with scalar of 17 m where high values indicate close locations and low values indicate distant locations; Fig. 3a). We also found a year effect, where toads were less likely to move in 2023. The residual analysis of the final model had a dispersion value of 1.01 (p = 0.77). Based on the low, non-significant dispersion value, along with the appearance of the QQ plot and the residual vs. predicted graph (see supplementary materials 3, Fig. S1), we concluded this to be a good model [88]. We also found no evidence of multicollinearity in the non-quadratic terms given that VIF values were ≤ 1.86. For the average daily distance moved (Fig. 3b), we found that toads moved longer distances when further from a trail. We also found a year effect, as they moved shorter distances in 2018. The residuals appeared acceptable (see supplementary materials 3, Fig. S2), the dispersion value of the model was low (0.81, p = 0.81) and there was no evidence of multicollinearity (all VIF ≤ 1.67). The type of tagging method did not affect the occurrence of a movement or the distance moved. We present a comparison of various tracking metric models for the effects of tag type in supplementary materials 2, Tables S7 - S9. Fig. 3– Model-averaged effect sizes and 95% confidence interval for all variables included in the (a) binomial (i.e., moved yes or no) and (b) gamma components (i.e., distance moved) of the hurdle model for toad movement. Variables are sorted from greatest to least AICc weight, which is reported after variable name. For species, Anaxyrus americanus americanus was set to 1. For distance to water, the 3Q represents the third quantile of daily movement which was used as a scalar in the exponential decay distance to water (decay distance = e (Euclidian distance to water/−3Q; where 3Q = 17 m)). For percent conifer forest and TRI, the 3Q and 2Q represents the size of the moving window at 17 m and 7 m respectively – Model-averaged effect sizes and 95% confidence interval for all variables included in the (a) binomial (i.e., moved yes or no) and (b) gamma components (i.e., distance moved) of the hurdle model for toad movement. Variables are sorted from greatest to least AICc weight, which is reported after variable name. For species, Anaxyrus americanus americanus was set to 1. For distance to water, the 3Q represents the third quantile of daily movement which was used as a scalar in the exponential decay distance to water (decay distance = e (Euclidian distance to water/−3Q; where 3Q = 17 m)). For percent conifer forest and TRI, the 3Q and 2Q represents the size of the moving window at 17 m and 7 m respectively Site fidelity Both eastern American toads (t = 3.04, p = 0.004, n = 36) and Fowler’s toads (t = 3.98, p = 0.0002, n = 48) significantly exhibited site fidelity compared to CRW simulated site fidelity data (Fig. 4). Site fidelity did not differ between the taxa (W = 953.5, p = 0.41). Fig. 4– Strip plots of a) Anaxyrus americanus americanus and b) Anaxyrus fowleri site fidelity as measured by returns per observations in simulated (null) versus observed toad movements. Grey dots represent individuals, while the black dots and bars represent mean ± 1 standard deviation. Asterisks indicate significance (p < 0.05) – Strip plots of a) Anaxyrus americanus americanus and b) Anaxyrus fowleri site fidelity as measured by returns per observations in simulated (null) versus observed toad movements. Grey dots represent individuals, while the black dots and bars represent mean ± 1 standard deviation. Asterisks indicate significance (p < 0.05) Microhabitat We began with eight candidate predictor variables related to refugia substrate, and after testing them individually, we were left with four in our final model (supplementary materials 2, Table S10). After model averaging, our analysis of microhabitat choice indicated that the top predictors of taxa by refugia microhabitat were woody structures and leaf litter (supplementary materials 2, Table S11; Fig. 5). The positive slopes of these two predictor variables indicate that both were used more frequently by eastern American toads (n = 632). The other two predictor variables had negative effect sizes; however, their 95% CI overlapped with 0. This indicates that no microhabitats were used more frequently by Fowler’s toads (n = 865) in comparison to eastern American toads. This model was evaluated for fit via residual analysis and had a dispersion value of 1.07 (p = 0.85). An examination of the QQ plot and residual vs. predicted graph (supplementary materials 3, Fig. S3) revealed some deviation, but this was still considered acceptable given the large sample size and low dispersion value [88]. We also found no evidence of multicollinearity given that the VIF values were ≤ 2.5. Fig. 5- Model-averaged effect sizes and 95% confidence interval for all variables included in the final model of refugia microhabitat choice by toads. Variables are sorted from greatest to least AICc weight, which is reported after variable name. For the dependent variable of taxa, Anaxyrus americanus americanus was set to 1 - Model-averaged effect sizes and 95% confidence interval for all variables included in the final model of refugia microhabitat choice by toads. Variables are sorted from greatest to least AICc weight, which is reported after variable name. For the dependent variable of taxa, Anaxyrus americanus americanus was set to 1 Discussion Very few studies have evaluated the movement ecology of anurans or amphibians in general [46]. Those that have tended to focus on a single species [51, 64, 69, 92, 93]. Our study is one of a few that evaluated movement ecology, refugia fidelity, and microhabitat use in two toad species over three years. We found that both eastern American and Fowler’s toads move farther from breeding sites than the recommended buffer of 30 m in Virginia, though the distributions of these movements differed between taxa [94]. Rain, high temperatures, the amount of the surrounding landscape that was composed of coniferous trees, and proximity to trails and water were important in explaining variability in movement for both taxa. We also found that both taxa exhibit site fidelity to refugia. Our microhabitat use study revealed that the eastern American toad is a forest specialist compared to the Fowler’s toad which appears to be a generalist occupying a broader array of microhabitat. Consistent with other amphibian movement studies, the toad movement we observed was partially driven by environmental conditions, as the probability of a movement event was positively related to minimum temperature and rainfall. The poikilothermic nature of amphibians drives their dependence on warmer temperatures for movement, and this strong relationship was reflected in minimum temperature’s high variable weight [38, 51, 95]. Rainfall from the previous day was almost as heavily weighted as temperature, and the positive nature of this relationship was also expected given the known dependence of amphibians on wet conditions for movement [36, 96, 97]. This emphasizes that even toads, a group considered to be more desiccation resistant than other amphibians due to a combination of less permeable skin [98] and the use of their bladders for water storage [99], are still influenced by rainfall [100]. Alongside these expected environmental results, we also found that toad movements were driven by their local landscape conditions. Toads were more likely to move when in areas with a higher proportion of coniferous vegetation. Coniferous stands are considered unfavorable habitat for amphibians due to their acidic soils and suppression of undergrowth necessary for large invertebrate prey communities [101–103]. The greater probability of movement that we detected represent responses that would be expected given these factors, as toads would be expected to move around more in search of sparser prey or to find less acidic refugia [101, 104–106]. The decay distance to water was negatively associated with the probability of a movement event, indicating that toads were less likely to move when close to a body of water. This is likely because the moister conditions around bodies of water were attractive to toads and discouraged forays into drier conditions [39, 107]. These areas around bodies of water may also represent higher quality habitat that is more desirable for toads to occupy [108]. Only a toad’s distance to a trail influenced how far it moved, as they moved more when far from trails. Previous work on the interactions between anurans and man-made trails, has focused on their ability to aid invasive species dispersal or threaten connectivity [8, 109]. However, short movements near a resource are good indicators of foraging in many animal species [39]. Additionally, our paths were largely made up of light-colored substrates, which can help toads visually identify dark-colored insects, leading us to hypothesize that trails are used as feeding areas due to the improvements they provide for foraging success of insects based on the movement patterns and behavior we observed [71]. This finding potentially identifies a novel manner is which wildlife can use anthropogenic habitat modifications or environments to their advantage [110, 111]. Key differences exist in the frequency of microhabitat use between the two taxa. As eastern American toads have frequently been described as forest specialists, it is not surprising that they more frequently used leaf litter and woody structures consistent with forested habitat [32, 112]. For Fowler’s toads, we found that none of the microhabitat variables measured in this study were used more frequently compared to American toads, indicating that this taxon is a habitat generalist. The lack of associations may be due to Fowler’s toad preferred soil type (sandy) being abundant in the coastal plain region where our study took place [113, 114]. Previous research on Fowler’s toad habitat has focused on a declining population in Ontario, where they are largely associated with sand dunes, and consequently have a limited regional distribution [30]. However, the relative abundance of sandy soils in our study area may allow them to have greater microhabitat flexibility and a broader distribution compared to restricted northern populations. This research adds to the growing body of support for the protection of adjacent non-breeding habitat at biologically relevant distances as core habitat for anurans [23]. We found that toads moved large distances from their breeding ponds, but that once they left, they remained in small areas for the rest of the year, as demonstrated by the relatively small distances of non-zero movement (median 7 m, IQR 3–17 m) and their significant site fidelity. This indicates that for eastern American and Fowler’s toads, the recommended buffer of 30 m in Virginia surrounding wetlands and ponds that drain into the Chesapeake Bay is too narrow to accommodate their movements away from breeding sites [94]. Both taxa’s median distance from breeding grounds was more than double this value. This suggests that breeding sites for these taxa need to be surrounded by substantial patches of forest to ensure that these sites do not become abandoned oases in an otherwise inhospitable desert for amphibians. Our research demonstrates a study design that allows researchers to estimate the size of vegetative buffers needed around breeding habitat to maintain sensitive amphibian populations in human-altered landscapes [23]. Future research is needed to address juvenile movement after emergence from their natal pond [115–117]. Juveniles provide crucial population connectivity via dispersal among amphibian populations as their survival and propensity to travel greater distances than adults across patches prevents deleterious inbreeding effects [23, 118]. These critical amphibian movements of migration and dispersal are both affected by the external environment [7, 49, 50, 116]. Conclusion Our study identified key distances moved by the adults of two closely related taxa of toads from their breeding habitats in a multitaxa tracking approach using two different technologies. The potential use of human trails by toads as feeding areas is particularly interesting as it adds to the diversity of responses of wildlife to anthropogenic habitat modifications [110, 111]. The detection of significant amounts of site fidelity to refugia in both taxa using a novel method indicates the importance of stable habitat to anurans. To increase the resilience of amphibians to the current threats leveled at them, we encourage a prioritization of protection for large and connected areas of habitat for amphibians of both their breeding and upland habitats at scales that are biologically relevant [119–122]. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3