Skip to main content

Vegetation cover, topography, and low-traffic roads influence Sonoran desert tortoise (Gopherus morafkai) movement and habitat selection

Abstract

Background

Anthropogenic activities occurring throughout the Sonoran Desert are replacing and fragmenting habitat and reducing landscape connectivity for the Sonoran desert tortoise (Gopherus morafkai). Understanding how the structure of the landscape influences tortoise habitat use and movement can help develop strategies for mitigating the impacts of these landscape alterations, which are conservation actions needed to support the species’ long-term persistence. However, how natural and anthropogenic features influence fine-scale habitat use and movement of Sonoran desert tortoises remains unclear.

Methods

The goals of this study were to (1) understand how characteristics of the landscape shape tortoise habitat use and movement in order to (2) identify factors that may reduce habitat use or threaten landscape connectivity for the species by discouraging or restricting movement. We collected GPS telemetry data from 17 adult tortoises tracked for two summer monsoon seasons, when tortoises are most active, in a U.S. National Monument along the international border between Arizona, USA and Sonora, Mexico. We used Hidden Markov Models (HMMs) to assign GPS locations to an encamped or a moving state. We used the moving state data in integrated Step Selection Analyses (iSSA) to examine how range-resident Sonoran desert tortoises select habitat and respond to landscape features while moving.

Results

Tortoises selected to move through areas of intermediate vegetation cover and terrain ruggedness and avoided areas far from desert washes and close to low-traffic roads. Tortoises increased their speed when approaching or crossing low-traffic roads but showed no detectable response to a highway.

Conclusion

Bare earth or high vegetation cover, flat or extremely rugged terrain, areas far from desert washes, and low-traffic roads may discourage or restrict tortoise movement. Therefore, preventing the development of roads, activities that degrade washes, and activities that thin, remove, or greatly increase vegetation cover may encourage tortoise habitat use and movement within those habitats.

Background

Resource selection studies enhance our understanding of how target species respond to features or processes occurring in the landscape [1,2,3,4]. These insights provide a foundation for conservation plans designed to identify important habitats, preserve landscape connectivity, and restore the ecological functions of degraded landscapes [1, 5]. Recent studies have demonstrated that we can further improve our understanding of how landscapes shape habitat use by incorporating animal movement behavior into these resource selection studies [1, 6].

Numerous studies have used GPS dataloggers to understand space use, habitat selection, and movement of North American tortoises (genus Gopherus; [7,8,9,10,11]). Preserving tortoise habitat and mitigating the negative impacts of anthropogenic disturbances on habitats have been identified as conservation priorities for Gopherus species like the gopher tortoise (Gopherus polyphemus) and the Mojave desert tortoise (Gopherus agassizii) [12, 13]. Studies examining tortoise movement and space use have demonstrated that both natural (e.g., terrain) and anthropogenic landscape features (e.g., fences and roads) can constrain or alter tortoise movement or habitat use [7,8,9,10]. For example, the use of GPS dataloggers has revealed that G. agassizii avoids areas of high slope, low vegetation coverage, and areas near roads, and that they alter their movement behavior when encountering anthropogenic features like fences and roads or disturbed areas [7,8,9]. These fine-scale insights into the influence of the landscape on tortoise habitat use and movement have provided valuable information applicable to developing or refining management actions intended to preserve or restore tortoise habitat such that it may facilitate landscape connectivity among tortoise habitats and populations. However, such studies have been limited to those tortoise species and populations protected under the Endangered Species Act (ESA), which excludes the Sonoran desert tortoise (Gopherus morafkai).

G. morafkai was denied listing under the ESA in 2022 based on findings suggesting their populations are not significantly declining throughout their range [14]. However, a recent IUCN Red List assessment found G. morafkai’s populations and geographic range to be shrinking, warranting a “Vulnerable” status [15]. G. morafkai is threatened by climate change, wildfire, the invasion of non-native grasses, development, barriers to movement, and habitat loss and fragmentation [16,17,18,19,20,21,22]. Development and the spread of transportation infrastructure are increasing throughout the species’ range [23, 24]. In Arizona, 70% of Sonoran desert tortoise habitat is within 1 km of development [20]. Furthermore, range-wide, urban development is projected to replace up to 10% of remaining tortoise habitat in the next 100 years [14]. Roads, which are proliferating throughout the species’ range, fragment habitat and isolate tortoise populations and may cause direct mortality, increase risk of illegal collection, and reduce landscape connectivity among populations [16, 25,26,27,28]. Unpaved roads and off-road vehicle (OHV) use can degrade tortoise habitat (e.g., by removing vegetation cover), increase the risk of direct mortality or illegal collection, and may discourage tortoise movement [9, 28,29,30]. In addition to the individual effects a single road may have, the increasing density of these features has been linked to declines in tortoise populations [31]. Ultimately, anthropogenic activities that degrade or replace tortoise habitat may reduce landscape connectivity among tortoise populations by making them less permeable to movement, which may subsequently reduce demographic rescue, gene flow, and other processes that support the species’ persistence [22, 32, 33]. For these reasons, protecting suitable tortoise habitat and preserving or restoring landscape connectivity have been identified as conservation priorities for G. morafkai [14, 15, 22].

Studies of G. morafkai’s space use, habitat use, and response to disturbance are relatively few compared to Gopherus species protected under the ESA (i.e., G. polyphemus and G. agassizii). Conservation plans for G. morafkai are thus partially dependent on knowledge of these protected species, particularly G. agassizii [16]. Although G. agassizii and G. morafkai are both desert tortoises and face similar threats [29], they differ in some aspects of their ecology, habitat use, and behavior [34]. For example, G. agassizii is most associated with flatter terrain whereas G. morafkai is most associated with steep and often rugged terrain [16, 35, 36]. It is plausible that behavioral differences may also extend to their response to disturbance or anthropogenic features, yet this remains unclear. Studies examining G. morafkai’s space use, habitat use, and response to disturbance may help understand which conservation actions developed for the well-studied G. agassizii may be applicable to G. morafkai or may help refine conservation strategies specific to the species.

Studies that have examined habitat use and space use of G. morafkai have found that topographic features like steep slopes and deeply incised desert washes are important habitat features [36,37,38], and that they make greater use of areas that are closer to washes and unpaved roads than what is available to them [28]. G. morafkai has been found to use microhabitat with more vegetation cover as both juveniles and adults [37, 38]. In contrast, another study examining habitat use at a larger spatial extent found vegetation to be less important than topographic and geomorphological features [36]. These studies have been limited to traditional survey methods (e.g., plot surveys and occupancy analysis) and radio-telemetry studies. While providing new and valuable information on the species’ habitat use, these studies lack the finer temporal resolution of GPS telemetry, which has elucidated patterns in space use for other North American tortoises and so are likely capturing only a fraction of the species’ daily activity and, possibly, habitat use. Furthermore, how landscape features, including anthropogenic features like roads, influence how and where G. morafkai moves remains unclear. The paucity of information on this subject specific to G. morafkai precludes the development of effective management strategies for the species.

In this study, we used GPS dataloggers to record fine-scale tortoise movements and combined Hidden Markov Models (HMMs) and integrated step-selection analyses (iSSA) to examine how G. morafkai selects habitat and responds to roads when moving. Generally, we expected G. morafkai to make selective decisions as they moved through the landscape. Hence, we expected that (1) tortoises would select to move through areas with landscape features associated with shelter (i.e., rugged terrain, vegetation cover, and incised washes); (2) tortoises would avoid flat terrain, bare earth, and areas near roads; and (3) tortoises would adjust their movement behavior when moving close to or crossing roads. Our findings provide insight into how landscape characteristics like rugged terrain, vegetation, or roads can alter space use and movement of range-resident tortoises. This information may be applied to the development of management plans focused on preserving or restoring habitat such that it maximizes use, increases permeability to movement, and minimizes the impacts of barriers on localized movement for G. morafkai.

Methods

Study species

Gopherus morafkai is a terrestrial, herbivorous tortoise found throughout most of the Sonoran Desert. It was described as a distinct species from G. agassizii in 2011 based on differences in genetics, habitat use, ecology, and behavior [39]. G. morafkai has an activity pattern that coincides with seasonal rains [40]. They may exhibit some activity in late winter or spring, but activity peaks following monsoon rains that arrive in late summer and give rise to food resources like annual plants [40]. During the winter and arid months of late spring and early summer, G. morafkai is largely inactive in shelters like boulder piles, caliche caves, packrat middens, and occasionally burrows of other animals, in which they may spend up to 90% of their lives [16]. G. morafkai is typically associated with rugged terrain, inhabiting rocky mountainsides, coalescing alluvial slopes (i.e., bajadas), and deeply incised ephemeral stream beds (i.e., washes) that provide resources like forage and shelter [16, 39, 41]. They are also known to occur lower on bajada slopes and, in lower densities, in valley floors, although tortoises in these habitats have been less thoroughly studied [41].

Study site

We studied tortoise movement and habitat selection between June and October 2021 and 2022 in Organ Pipe Cactus National Monument (OPCNM; Fig. 1). OPCNM is a 1,338 km2 United Nations Educational, Scientific, and Cultural Organization (UNESCO) biosphere reserve managed by the National Park Service (NPS) in Pima County, Arizona (USA) along the United States-Mexico border. The study area experiences an annual average minimum temperature of 12.3 °C, an annual average maximum temperature of 29.9 °C, and annually receives 237.2 mm of precipitation [42]. OPCNM has a bimodal rainfall pattern of light rains occurring in winter and spring and heavy monsoon rains that typically occur from mid-June to late September and account for most of the annual precipitation. The landscape of the study area is characterized by two subdivisions of the Sonoran Desert biotic community: the Arizona Upland and Lower Colorado River Valley subdivisions [43]. The topography of OPCNM is comprised of rugged mountains, bajadas, and valleys with shallow or deeply incised washes. Elevation within the park ranges from approximately 297 m to 1,469 m above sea level. The study area is bounded to the north and west by US Fish and Wildlife Service and Bureau of Land Management land, to the east by the Tohono O’odham Nation, and to the south by the international border between Arizona, USA and Sonora, Mexico. Tall (~ 10 m) fencing, likely impermeable to tortoise movement [14], spans the entire southern boundary of the study area. Approximately 311 km of public roads, active and inactive service roads, and roads designed to facilitate access to remote areas of the park by United States Customs and Border Protection (USCBP) agents and law enforcement officers exist within the boundaries of the study area (hereafter referred to as low-traffic roads). The study area is also bisected by State Route 85, which is a high-traffic, high-speed highway that spans ~ 60 km from the international border of Arizona, USA and Sonora, Mexico north through the study area. OPCNM is one of the main protected areas for G. morafkai [15] and provides an opportunity to examine how tortoises select habitat in intact desert and areas where anthropogenic features occur, and how those features influence their movement.

Fig. 1
figure 1

Map of the study area, biotic communities, and roads within the study area. The study area is approximately central to the species’ range. The predominant vegetation types are characterized by Arizona Upland and Lower Colorado subdivisions of the Sonoran Desert, and the terrain ranges from flat, intermountain valleys to rugged mountain peaks. Low-traffic roads in the study area are typically dirt, gravel, or partially paved, and a highway bisects the study area. Geospatial data for the Sonoran desert tortoise range boundary was downloaded from https://www.iucnredlist.org/species/97246109/97246177 and all other data used in the map were provided by the NPS

GPS telemetry data

Tortoises used in the study were found by actively searching potential habitat. Since we were interested in understanding how tortoises generally select for habitat or respond to roads, we attempted to balance the number of males and females in our study. We also included tortoises found lower on mountain slopes or in valleys where less is known about Sonoran desert tortoise habitat use [41, 44]. We did this by selecting approximately half of the study animals from areas greater than 1 km from the nearest 5-degree slope.

Tortoises were outfitted with very high frequency (VHF) radio-transmitters (RI-2B; Holohil Systems Inc.) and GPS dataloggers (i-gotU GT-600; Mobile Action Technology Inc.) affixed to the carapace using putty epoxy and positioned to minimize change in the animal’s vertical profile. VHF and GPS equipment combined weighed less than 5% of the tortoises’ body mass. All tortoises were adults with a midline carapace length greater than 190 mm and were handled following protocols from State of Arizona Scientific Activity License no. SP847069, National Park Service (NPS) Scientific Research and Collecting Permit no. ORPI-2022-SCI_0003, NPS Animal Care and Use Committee (IACUC) protocol no. 21036-04, and Texas Tech University IACUC protocol no. 21036-04. Tortoises were tracked to their location approximately monthly using VHF telemetry equipment to recover GPS dataloggers. The GPS dataloggers must be recovered for data to be manually downloaded. As purchased, they have a battery life of approximately 30 days when programmed to record a position every 30 min, 24 h per day. In 2021, we used custom-made 3D printed mounts that held the GPS logger following methods similar to Hromada et al. [8], which allowed dataloggers to be replaced monthly with fully charged dataloggers. In 2022, tortoises were outfitted with GPS dataloggers modified to maximize battery life and reduce recapture events, using methods following Paden et al. [45]. Tortoises tracked in 2022 were then typically captured once in June to deploy fully charged equipment and again in October to recover and download data. Upon downloading data from each GPS datalogger, inaccurate locations were identified and removed. We used several criteria to identify potentially erroneous locations: an elevational difference greater than 25 m between the elevation recorded by the GPS datalogger and a 30-meter digital elevation model (DEM), or locations that had both a high speed (600 m/hour or greater) and a high turning angle (> 180° turn) between two consecutive locations, which reflect improbable tortoise movements.

After removing potentially erroneous positions, we recorded the minimum distance between the closest GPS position and the nearest road for each tortoise and the number of times each tortoise crossed any low-traffic road or the highway. To account for the accuracy of the GPS dataloggers, we only considered a tortoise to have crossed a road if it had moved a distance greater than 20 m (twice the approximate location error of the GPS dataloggers) from one side of the road and remained there for a minimum of one hour (twice the sampling interval), or two recorded GPS locations. The GPS dataloggers used in the study are reported to have an approximate location error of 10 m prior to screening for erroneous positions, and the location error should be minimally impacted by vegetation cover [46]. To confirm this, we performed four stationary GPS tests by placing GPS dataloggers in places tortoises may take refuge (e.g., under a bush or within a boulder pile), which we further describe in the Supplementary Materials.

Both the HMM and iSSA approaches we used to examine tortoise movement and selection assume temporal regularity (e.g., a regular sampling interval) in location data. Occasionally, scheduled data points (locations) were missing or delayed. This may have arisen from the tortoises’ use of shelters (e.g., rock piles and caves), which can delay or preclude the collection of GPS locations [9, 45], or other instances where satellite signal is insufficient to record a location. To temporally regularize the data, we first split each study animal’s movement trajectories into separate trajectories where gaps of 1.5 h (i.e., three locations) or greater existed. This was done to avoid imputing locations over long periods of time. Missing fixes resulting from tortoise shelter use, the removal of erroneous locations, and splitting trajectories at long gaps resulted in many short-duration tracks with occasional missing fixes. As we were interested in bouts of movement, we removed any trajectories under 6 h long, which represented sheltering tortoises. We then used a continuous-time correlated random walk model to impute missing locations into the movement trajectories where any gaps remained (i.e., one or two locations) and regularize the data to 30-minute sampling intervals. Imputation was performed using the ‘crawlWrap’ function in the momentuHMM package [47] in R version 4.3.0 [48]. This resulted in gapless, temporally regular movement trajectories that were 6 h or longer for use in our behavioral state assignments.

Behavioral state assignments

During their most active period, the summer monsoon season, G. morafkai is typically active in morning and evening, sheltering through potentially lethal mid-day heat and through the night [40]. Since our objective was to examine habitat selection and movement when tortoises are active (i.e., not sheltering), we isolated GPS locations reflecting movement from the overall GPS dataset. To accomplish this, we fit HMMs to assign GPS locations into different behavioral states. We predicted that a two-state model would differentiate between bouts of tortoise movement and when the animals are “encamped” (not moving or sedentary) which, as described in other studies using GPS dataloggers to study tortoise movement (e.g [10]), are readily apparent when visually inspecting GPS locations. We modeled transition probabilities as a function of time of day (hour) and air temperature, which we expected to influence whether a tortoise would be in the encamped state or moving. Air temperatures recorded at 5-minute intervals were downloaded from weather stations positioned throughout the study area (https://wrcc.dri.edu/organpipe/) and a rolling join was used to assign each GPS location the most recent air temperature from the closest station with available weather data. We then used the Viterbi algorithm to assign each GPS location to its most likely state [49]. Our behavioral state assignments were performed using R package momentuHMM [47]. After assigning behavioral states to the GPS locations, we visually inspected them in ArcGIS Pro 3.2.1 to ensure the behavioral state assignment captured the two states (i.e., “encamped” and “moving”) we observed during our initial inspection of the GPS locations.

Predictors of selection and movement

We selected natural and anthropogenic features that we expected to influence tortoise habitat selection or movement as predictor variables in our selection and movement analysis. All predictor variables were rasters calculated at or resampled to 30-meter resolution, which we expected to be appropriate for the measurement error associated with our GPS dataloggers. For landscape characteristics (i.e., natural characteristics of the landscape), we included measurements of vegetation, terrain ruggedness, and distance from the nearest wash. To measure vegetation on the landscape, we calculated the mean modified soil-adjusted vegetation index (MSAVI2; [50]) for the 2021 and 2022 monsoon season independently. We chose this index because it is well-suited to quantifying vegetation greenness (i.e., cover of green vegetation) while compensating for the high reflectance of desert soils. Our calculations of mean MSAVI2 were derived from Sentinel II imagery using Google Earth Engine [51]. To quantify the ruggedness of terrain associated with tortoise habitat, we calculated the terrain ruggedness index (TRI) as described by Wilson et al. (2007; [52]) from a 30-meter USGS Digital Elevation Model using the ‘terrain’ function and an 8-cell neighborhood rule in the R package terra [53]. We used the ‘Euclidean Distance’ tool in ArcGIS Pro to calculate the distance from incised desert washes, which required us to build a custom model to map incised washes using a high-resolution (0.5 m) DEM provided from the NPS (described in Supplementary Materials). We compared our measures of terrain ruggedness, vegetation, and our wash layer with 60-centimeter aerial images from the National Agricultural Imagery Program (provided by the NPS) to ensure the variables were adequately capturing the landscape characteristics that we expected would influence tortoise habitat use and movement. This was necessary because our measure of terrain ruggedness (a first-generation ruggedness metric) can be correlated with slope, which can complicate its interpretation [54]. We used TRI because it has been widely used in resource selection studies and adequately captured the landscape features that we hypothesized to influence tortoise movement; we recommend Dilts et al. (2023; [54]) for alternative measures of terrain complexity or ruggedness. For anthropogenic features, we considered all roads other than the highway in the study area to be low-traffic roads. We expected all road types to influence tortoise movement but that the effect would differ based on road type (low-traffic or highway). We generated a raster reflecting the distance in meters from the nearest low-traffic road and from the highway using Open Street Maps (https://www.openstreetmap.org) and road data provided by the NPS, again using the ‘Euclidean Distance’ tool in ArcGIS Pro 3.2.1. We took the natural logarithm of the distance to washes and distance to roads variables as we expected the effect of these features on habitat section and movement would be close ranged. All predictor variables were scaled and centered.

Selection and movement analyses

To examine how tortoises select habitat and respond to roads while moving through the landscape, we used GPS locations assigned to the moving state, or movement trajectories, in integrated step-selection analyses (iSSA; [55]). In step-selection analyses, a “step” refers to a pair of sequential GPS locations. “Step length” refers to the distance in meters between paired points, and “turning angle” corresponds to the deviation in degrees between one step and the next. Step-selection analyses pair observed steps (e.g., derived from GPS data) with randomized available steps. This allows for a statistical comparison between what the animal selected on the landscape and what was available to it but not selected. We used the R package amt [56] to pair each observed step with 10 available steps. The step lengths of available steps were modeled from a gamma distribution and turn angles were modeled from a von Mises distribution, which we assumed would best represent the respective distribution of step lengths and turning angles [56]. We extracted values for each predictor variable at the start and end of each observed and available step. Values of predictor variables at the end of each step were used to test hypotheses about habitat selection. Values at the start of each step were used in interactions with predictor variables to test hypotheses about how that predictor variable influences movement parameters (e.g., step lengths and turning angles). We followed the approach of Muff et al. (2020; [57]) and used a Poisson formulation of a conditional logistic regression, included the unique ID of each step as a random intercept and fixed its variance to a large number, and we used individual-specific random coefficients for all predictor variables and interactions in our models. This allowed us to account for individual variation in selection and differences in habitat availability for each individual tortoise. All models were fit using R package glmmTMB [58].

We built a single model to quantify habitat selection of natural landscape characteristics by moving tortoises. We included quadratic terms for TRI and MSAVI2 as we expected selection responses may be non-linear for these variables. We built separate models, using a subset of tortoises, to isolate and test the effects of low-traffic roads and the highway on tortoise movement and habitat selection. Other tortoise species have been shown to adjust their movement and selection behavior near or when crossing dirt, unpaved, and low-traffic roads [7, 9]. We included movement data only from tortoises that came within 50 m of a low-traffic road at any point during the study (N = 7) for our low-traffic roads model, expecting we would detect an effect within this range. There have been no studies explicitly addressing how highways influence movement and selection in this species; however, tortoise populations are less dense, and signs of their presence are often scarce within 1 km of roads [20, 26]. For the highway model, we included movement data only from tortoises that were found within 1 km of the highway (N = 6). We included the logarithm of the step length and cosine of the turning angle in all models, which is recommended to account for the movement process underlying selection [55].

We then calculated the relative selection strength (RSS; [59]) for any significant effects of natural landscape characteristics or roads on selection. RSS is an estimate of how likely an animal is to select one location versus another based on the values of the predictor variable at those locations, while keeping the value of all other predictor variables constant. This allowed us to compare the relative probability of tortoises choosing locations across the range of values experienced by our sampled tortoises to the minimum value of that predictor variable. To calculate RSS for each significant predictor variable, we used the calculations described in Avgar et al. [59].

Results

GPS telemetry

We tracked a total of 17 adult tortoises (8 male, 9 female) over 2021 and 2022. This resulted in an average tracking period of 93 days per individual, ranging from 66 to 105 days each year between the months of June to October, during which tortoise activity peaks. The final dataset used in our behavioral assignment included 809 temporally regularized movement trajectories, comprising 54,513 locations from the 17 tortoises. The mean location error of the GPS data recovered from our stationary tests was under 10 m (Table S1). Tortoises crossed low-traffic paved roads a minimum of zero times and a maximum of 31 times during the study (Table S2). We observed only a single highway crossing during the study (Table S3). Although we recorded a single GPS location where a tortoise could conceivably have crossed a low-traffic road within the 30-minute sampling interval, we recorded no other evidence of road crossings nor locations that could have been confused with crossings (within the approximate location error of the GPS dataloggers and a road) for this individual.

Behavioral assignment

Using HMMs allowed us to assign GPS locations to one of two behavioral states: an encamped state and a moving state (Fig. 2). Step lengths were longer and turning angles were lower when tortoises were moving compared to when encamped. In the encamped state, step lengths were short and turning angles were high. (Fig. 2). Tortoises were more likely to be moving during the day (Fig. S1) and in cooler temperatures (Fig. S2).

Fig. 2
figure 2

Distribution of step lengths and turning angles for the encamped and moving states. Step length distributions (A) and turning angle distributions (B) for the encamped and moving states for the 17 tortoises tracked over the duration of the study. The encamped state is characterized by shorter step lengths and high turning angles, whereas the moving state is characterized by longer step lengths and lower turning angles. An example segment of a state-assigned movement trajectory from a single study animal is provided (C). This reflects GPS locations of a single tortoise collected over 40 h in their assigned behavioral states: clusters of orange points indicate when the tortoise is in the encamped state. Blue points and line segments indicate when the tortoise is in the moving state

Step selection models

After removing the encamped locations, we used the remaining 7,612 steps to examine tortoise habitat selection for natural landscape characteristics and roads, and to determine how roads influence tortoise movement. We found a significant effect of all natural landscape covariates (Fig. 3A) and low-traffic roads (Fig. 3B) on habitat selection. We also found a significant interaction between tortoise step length and distance from low-traffic roads indicating faster movements in proximity to those roads (Fig. 3B). We found no significant effect of low-traffic roads on turning angle nor any significant effect of the highway on selection or interactions with movement parameters (i.e., the log of step length and cosine of turning angle; Fig. 3C).

Fig. 3
figure 3

Beta coefficients and 95% confidence intervals for all iSSA models. Beta coefficient estimates and 95% confidence intervals for selection (x-axis) of predictor variables (y-axis; natural landscape characteristics, low-traffic roads, and the highway) and interactions between movement parameters (the logarithm of step length and the cosine of turning angle) and predictor variables. Tortoises select intermediate values of terrain ruggedness (TRI) and vegetation index (MSAVI2) (A) and areas farther from low-traffic roads (B). They avoid extreme values of terrain ruggedness and vegetation index, and areas far from incised washes (A). Positive coefficients (black triangles) indicate selection and negative coefficients (grey circles) indicate avoidance. Coefficient estimates with 95% confidence intervals overlapping zero likely have little effect on selection. We performed one model for all natural landscape covariates (A) and separate models to isolate the effects of low-traffic roads (B) and the highway (C) on selection and movement

The positive coefficient for TRI indicates that tortoises selected rugged terrain, while the negative coefficient for the quadratic term (TRI2) indicates that selection for terrain ruggedness was non-linear. Tortoises were more likely to select areas of higher terrain ruggedness compared to areas of lower terrain ruggedness (e.g., desert flats) until values exceeded approximately 31 where selection reversed (Figs. 3A and 4A). For example, fewer than 7% of tortoise steps (i.e., GPS locations) ended in TRI values above 31, despite ~ 3,000 steps being available to them within that range. Similarly, selection for the vegetation index (MSAVI2) was non-linear. Tortoises strongly selected to move through vegetated areas compared to bare earth (MSAVI2 values near 0). For example, fewer than 8% of tortoise steps ended in MSAVI2 values under 0.15, while ~ 23% of all steps available to them fell within that range of values. Selection strength began to decrease as vegetation index values approached the maximum values available to the tortoises (Figs. 3A and 4B). A description of what TRI and MSAVI2 values correspond to on the landscape can be found in the Supplementary Materials (Fig. S3 and Fig. S4, respectively) and are described in the Discussion. We found that tortoises were less likely to move through areas further from incised washes then closer to incised washes (Fig. 3A), though the effect was weaker than other predictor variables (Fig. 3A) and became less pronounced as tortoises moved through areas greater than 50 m from the nearest wash (Fig. 4C). These results suggest that rugged terrain, vegetation, and incised washes encourage tortoise movement (Fig. 3A), but that very rugged terrain and higher vegetation cover may discourage or limit tortoise movement (Fig. 4A and B).

Fig. 4
figure 4

Tortoises selected areas closer to incised washes and of intermediate terrain ruggedness and vegetation index. Relative selection strength (RSS) and 95% confidence intervals for landscape covariates and roads included in the selection models for which significant relationships were found (Fig. 3). The RSS calculated in this study reflects the likelihood of moving into a location with a value of the landscape covariate exceeding that of the minimum value of the covariate. RSS values greater than 1, indicated by the dashed line, indicate selection. RSS values less than 1 indicate avoidance. An RSS value of 5 for a certain value of a covariate indicates that a tortoise is 5 times more likely to select a location with that value compared to a location with the minimum value of that covariate

Tortoises also avoided moving close to low-traffic roads (Figs. 3B and 4D) and increased their step length (i.e., speed) when moving near low-traffic roads, evidenced by a significant negative coefficient for the interaction between the log of the step length and distance to the nearest low-traffic road (Fig. 3B). These findings suggest that areas near low-traffic roads may be lower-quality movement habitat. Tortoises appear more likely to move faster when near or crossing low-traffic roads. Although demonstrating a similar trend, the lack of a significant effect of the highway on habitat selection or movement (Fig. 3C) suggests that tortoise interactions with the highway were too few to detect a significant effect or selection for areas far from the highway (e.g., home range selection) may have already occurred prior to our study.

Discussion

We demonstrated that vegetation, moderately rugged terrain, and areas with incised desert washes may encourage tortoise movement, whereas bare earth, high vegetation cover, flat and extremely rugged terrain, areas far from incised washes, and low-traffic roads may restrict or discourage movement typical of range-resident tortoises. Desert tortoises are referred to as “corridor-dwellers”, meaning a corridor that supports connectivity among habitats or populations must be sufficiently wide to allow for multiple tortoises’ home ranges to overlap and have sufficient resources to support those individuals [60, 61]. Our results highlight important habitat features and characteristics that may promote or discourage range-resident movement behaviors of adult tortoises, like foraging and mate-seeking. Additionally, our results are reflective of tortoises from steep slopes and tortoises from lower bajadas and valleys where comparatively little is known about tortoise habitat use. So, although our results are most directly applicable to managing intra-patch connectivity, our study provides insights that may be applicable to managing the matrix surrounding high-quality tortoise habitats, thus increasing the permeability of the matrix between tortoise habitats or populations.

We found that tortoises strongly select areas with vegetation coverage compared to areas with little to no vegetation. Vegetation is an important food and shelter resource for desert tortoises [17, 62], and our results corroborate studies that have found vegetation cover to be an important predictor of their habitat use at different spatial scales [36,37,38]. One such study, which examined G. morafkai’s microhabitat selection, suggested that management actions should maintain or possibly increase vegetation cover in the landscape [37]. Although our results partially support this prescription, we found that tortoises are less likely to select areas of higher vegetation cover compared to areas of intermediate vegetation cover available to them. Areas of higher vegetation index values in the study area (> 0.3 MSAVI2; Fig. S4) are typically characteristic of shaded reaches of large washes, forested areas, or roadside thickets that are denser than the surrounding landscape. Weaker selection of these areas may suggest that tortoises avoid moving through such dense vegetation when traversing their home range, possibly because dense vegetation is more resistant to movement.

Understanding the mechanisms underlying this nonlinear relationship may be of growing importance as processes that can change the structure or availability of vegetation, like wildfire, the invasion of non-native grasses, OHV use, and development, continue to occur in the region. Management practices that support native vegetation (e.g., preventing OHV use) may help ensure the conservation of habitat that provides forage, shelter, and that facilitates or encourages tortoise habitat use and movement. Based on our finding that selection weakens as vegetation cover increases, we hypothesize that dense stands of vegetation, including those created by non-native grasses, could conceivably discourage movement and form a physical barrier (i.e., physically costly to move through) or a perceptual barrier (i.e., perceived risk), as has been shown for other vertebrates [63]. We recommend that future studies seek to understand tortoise habitat use and movement in areas where disturbances or non-native grasses have changed the native vegetation.

As anticipated, tortoises were more likely to select areas of intermediate terrain ruggedness and areas close to incised washes than areas that had extreme values of terrain ruggedness (i.e., flat or extremely rugged terrain) or areas farther away from incised washes. Rocky mountain slopes and deeply incised washes are associated with tortoise shelter [36, 37, 40]. Compared to flat terrain, selection strength was higher for values of TRI that correspond to bajada slopes, rocky mountainsides, and areas with a high density of smaller washes or adjacent to deeply incised washes (Fig. S3). The similarities between previous studies and this study indicate that rugged terrain and incised washes are important to desert tortoises at multiple scales of selection [36, 37, 41, 44]. However, our findings highlight a threshold ( > ~ 31 TRI) at which tortoises select areas of intermediate terrain ruggedness compared to more rugged slopes when moving. In the study area, TRI values above this threshold are characteristic of very steep slopes above bajadas and mountainsides tortoises inhabit (Fig. S3). We expect that, despite being associated with rugged terrain, very steep and rugged terrain is costly for G. morafkai to traverse. Tortoises in the study area can be found kilometers from the nearest slope in areas where complex terrain is surrounded by otherwise flat terrain (e.g., valleys), especially where deeply incised washes provide tortoises shelter (e.g., caliche caves). However, selection strength for incised washes was relatively weak compared to other predictor variables. We expect this is the result of a combination of the conservative definition of availability used in our study and because most of the tortoises tracked in the study inhabit areas with a high density of incised washes: the difference in the distance from the nearest incised wash for each observed tortoise location and its paired available points were relatively little. Approximately 85% of tortoise locations (i.e., GPS locations) were within 100 m of an incised wash, which reinforces our findings that incised washes are important landscape features for G. morafkai as has been found in other studies (e.g. [41, 44]).

Compared to vegetation, topographical or geomorphological features like rugged slopes and incised washes are not typically regarded as manageable aspects of the landscape. However, areas where these features occur may be important targets for management actions focused on preserving tortoise habitat and facilitating range-resident movement behaviors. For example, tortoises may be more likely to attempt crossing sections of roads that intersect incised washes, so areas where roads intersect important habitat features like incised washes may be priority areas for the installation of “tortoise-friendly” road crossing structures (e.g., road culverts) or road-side fencing that may reduce tortoise mortality and facilitate tortoise movements within their home ranges and habitats. Washes in the study area and surrounding region may serve as corridors for human foot-traffic [64] and vehicular traffic from border patrol agents, both of which may reduce tortoise habitat quality. Management efforts (e.g., trash removal or vegetation restoration) targeting washes that receive intensive foot traffic or vehicle traffic may benefit tortoises that use those washes or the surrounding areas.

We found that tortoises were more likely to select areas farther from low-traffic roads than close to them, and that they increased their speed (i.e., made longer steps) when moving near or crossing low-traffic roads. Previous studies that used GPS telemetry to examine fine-scale movement and habitat selection of G. agassizii have found similar results to ours: tortoises typically avoided moving near unpaved roads and low-traffic roads and increased their movement speed when near them [7, 8, 30]. These studies posited that an increase in speed near low-traffic roads may indicate avoidance behavior; tortoises may be increasing their speed near roads because they offer less refuge from sun exposure, predation, and other risks [8, 9]. Peaden et al. [7] found that tortoises moved faster and had higher carapace temperatures when near a road than when farther from it, suggesting roads may cause tortoises to increase the energy they spend to access resources and may experience a greater risk of thermal stress [7]. Roads in our study area may have a similar effect on G. morafkai and may increase the distance tortoises need to move to access resources. In contrast to our findings, Grandmaison et al. [37] found G. morafkai selected microhabitat closer to gravel roads than areas available to them, which could be explained by lower traffic volumes, structural similarities to desert washes, and greater availability of forage on the roadside that may attract desert tortoises [37, 65]. This contrasting pattern may be reflective of site- or context-specific differences between the roads in each of the studies (e.g., differences in traffic volume). For example, a study using GPS telemetry found tortoises alter their movement behavior in different ways near dirt roads that experience different levels of local anthropogenic disturbance [9]. Alternatively, the contrast between the findings of Grandmaison et al. [37] and our findings may be reflective of the contrast in temporal resolutions used: the use of VHF telemetry in their study may not have captured the range of activity that is possible with the use of GPS telemetry.

While 6 tortoises tracked in our study crossed low-traffic roads, the number of road crossings in our study were ultimately rare. With the exception of a single individual that regularly crossed a paved section of a low-traffic road (31 times over two years), the remaining 5 individuals that crossed low-traffic roads did so only one to 5 times over two years (Table S2). The scarcity of road crossings in our study reinforces our findings that suggest that tortoises are likely reluctant to move near or across roads. These roads and adjacent habitats may serve as a perceptual barrier to movement; tortoises may perceive these roads as a potential risk of exposure or predation and may thus be reluctant to approach or cross them. The proliferation of roads (i.e., spreading in extent and increasing in density) has the potential to cause negative individual- and population-level consequences for tortoises, as has been found in other studies [26, 31, 66]. These roads continue to be created through illegal OHV use and border patrol activities by the USCBP in the study area and surrounding region. Consequently, we recommend the closure of unused or unnecessary roads, which is indicated as a suggested management action in the recovery plan for the closely related G. agassizii [13] and the Candidate Conservation Agreement (CCA) for G. morafkai [22]. We also recommend the restoration of habitats degraded by road development or OHV use as a potential strategy to increase landscape connectivity for G. morafkai and support the persistence of their populations.

In contrast to our expectation, we found no significant effect of the high-traffic highway on tortoise movement behavior or habitat selection. This lack of avoidance was unexpected given that high-traffic paved roads have been shown to have a variety of negative impacts on tortoises and tortoise populations [26, 27] and that, during our study, the highway bisecting the study area received approximately 10 to 570 times as many vehicles per year than low-traffic roads in the park [67]. Although several tortoises in our study frequently moved or sheltered within meters of the highway, we only observed a single crossing event during the study (Table S3). Similarly, Hromada et al. [9] found that only four of 28 tracked tortoises found within a kilometer of a highway crossed it and attributed the scarcity of crossing events to potential avoidance behavior, or possibly because tortoises established home ranges near but not including the highway [9]. Previous studies have demonstrated that G. morafkai displays high fidelity to their home ranges (e.g., [68]). Tortoises tracked in our study were all adults, and it is likely that they may have established home ranges in areas adjacent to the highway and, since establishing their home range, have become indifferent to it or are willing to take advantage of resources (e.g., incised washes or vegetation) adjacent to the highway. However, our finding that tortoises selected habitat farther from low-traffic roads may indicate that avoidance could be more severe for roads with higher traffic volumes, resulting in too few interactions between tortoises and the highway to detect a significant effect.

Despite the potential for road avoidance, tortoises do attempt to cross high-traffic roads that present a higher risk of mortality than low-traffic roads. Tortoise highway-crossings and road mortalities are occasionally observed in the study area (pers. obs. 2022). Numerous studies have demonstrated population declines related to road mortality and reduced tortoise densities or reduced tortoise abundance up to four kilometers from highways [26, 66, 69]. Tortoise fencing (i.e., fencing designed to reduce tortoise mortalities) is commonplace along highways throughout the protected G. agassizii’s habitat and has been found to effectively reduce tortoise road mortalities [66]. Although recommended as a conservation action [22], tortoise fencing is not widely used throughout the range of G. morafkai. Previously existing structures like drainage culverts (designed to allow runoff to pass under the road) may facilitate safe passage for tortoises attempting to cross barriers like highways, which has been documented, when tortoise fencing is present (e.g., [10]). In the study area, the highway (SR-85) and the border wall are likely the most significant anthropogenic barriers to tortoise movement and landscape connectivity. Drainage culverts along SR-85 and flood gates and small openings designed for wildlife along the border wall may allow tortoises to cross these barriers. We recommend future research evaluating the efficacy of previously existing structures (e.g., drainage culverts or wildlife passages developed for other species), or structures designed to facilitate the movement of other wildlife as crossing structures for tortoises. Furthermore, identifying cost-effective strategies for increasing the permeability of barriers to tortoise movement may help facilitate gene flow, demographic rescue, and other processes and should be a priority area for future research.

In this study, we demonstrated how natural and anthropogenic landscape characteristics can influence tortoise movement and habitat selection. However, the applicability of our study to populations throughout the species’ range and to landscape connectivity among those populations may be limited by the number and age of individuals tracked, the limited distribution of these individuals in respect to the species’ range, and the duration of the study. Tortoises could display different patterns in habitat selection and movement at different ages or when making non-resident movements, like dispersal events or temporary exploratory movements or sallies. For example, a study examining space use of G. morafkai found that immature tortoises are more likely to disperse or make temporary exploratory movements outside of their home range than adults [70]. Previous studies have also demonstrated that G. morafkai is capable of long-distance, interpopulation movements (at least in the absence of impermeable barriers to movement), which may be important for processes like gene flow and demographic rescue among populations that manifest through landscape-scale connectivity [21]. It is plausible that tortoises may select habitat differently during exploratory movements or dispersal events compared the normal (i.e., day-to-day) range-resident movement behavior of tortoises in our study. Therefore, our results are best interpreted as trends in habitat selection and movement relevant to adult, range-resident tortoises. We recommend future studies attempt to capture a greater variety in age classes, which may be limited by the technology available (i.e., size and weight relative to an immature tortoise), and attempt to capture different movement behaviors, which may be more likely with a longer study. Such studies may reveal differences in habitat selection and movement response to landscape characteristics than demonstrated by our results.

Although the responses of our study animals to natural and anthropogenic landscape characteristics generally followed similar trends in habitat selection (Fig. S5), tortoises living in different habitat contexts (e.g., more developed or degraded areas) or experiencing different environmental conditions (e.g., drought) may show different responses to landscape characteristics. For example, one study animal inhabiting particularly rugged terrain displayed the opposite response in selection of rugged terrain. While this opposite response was likely due to the available terrain (i.e., most steps used by the tortoise were less rugged than the surrounding terrain), it demonstrates that individual tortoises may display different selection responses and that studies examining tortoises in different habitat types or landscape contexts may yield different results. Our study took place during two monsoon seasons with above-average rainfall [42]. The relationship between G. morafkai space use and drought remains unclear [70], despite the documented negative impacts of short- and long-term drought on desert tortoises [14, 71, 72], but it is possible that tortoises were moving more frequently and farther during our study than they may have in a drier year. We recommend that future studies examine fine-scale habitat selection for G. morafkai during drier years, which may reveal a lesser or greater reliance on certain landscape characteristics that may help tortoises to persist during drought.

While iSSA allowed us to relax certain assumptions that have been shown to produce bias in resource selection studies [55, 57], our approach used a conservative definition of availability. Studies examining resource selection for G. morafkai at any scale are still few. We recommend that future studies examine resource selection for G. morafkai at different spatial and temporal scales throughout the species’ range and seek to discern patterns between use of space and resources incorporating different movement behaviors and age classes, and in light of drought, and natural and anthropogenic disturbances.

Conclusions

In this study, we elucidated patterns in habitat use and response to roads for adult, range-resident Sonoran desert tortoises as they move through the landscape. We highlighted how characteristics of tortoise habitat transition from facilitating or encouraging tortoise movement to restricting or discouraging it. We demonstrated that low-traffic roads may degrade or reduce tortoise movement habitat, and that management actions focused on maintaining vegetation at natural densities (e.g., preventing off-road vehicle use, wildfire, or the invasion of non-native grasses), preserving moderately rugged areas and areas with incised desert washes, and restoring unused roads may encourage habitat use and facilitate localized movements. Applying such management actions to degraded areas between tortoise habitats may enhance landscape connectivity for G. morafkai. Maintaining landscape connectivity through connected habitats is critical for this long-lived species, especially as roads and other barriers continue to proliferate throughout the Sonoran Desert [15]. Our work helps refine the collective understanding of this species’ habitat use and movement behavior and which provides valuable information for management strategies and conservation plans focused on supporting G. morafkai’s long-term persistence.

Data availability

GPS tracking data used in the study are not publicly available due to the sensitivity of tortoise locations and other resources within the study area and are restricted by data use agreements with the NPS.

Abbreviations

GPS:

Global positioning system

HMM:

Hidden Markov model

iSSA:

Integrated step selection analysis

ESA:

Endangered Species Act

OHV:

Off-Highway Vehicle

OPCNM:

Organ Pipe Cactus National Monument

UNESCO:

United Nations Education Scientific, and Cultural Organization

NPS:

National Park Service

USCBP:

United States Customs and Border Protection

VHF:

Very-High Frequency

IACUC:

Institutional Animal Care and Use Committee

DEM:

Digital Elevation Model

MSAVI2:

Modified Soil Vegetation Index II

TRI:

Terrain Ruggedness Index

RSS:

Relative Selection Strength

CCA:

Candidate Conservation Agreement

References

  1. Abrahms B, Jordan NR, Golabek KA, McNutt JW, Wilson AM, Brashares JS. Lessons from integrating behaviour and resource selection: activity-specific responses of African wild dogs to roads. Anim Conserv. 2016;19:247–55.

    Article  Google Scholar 

  2. Manly BF, McDonald LL, Thomas DL, McDonald TL, Erickson WP. Resource selection by animals: statistical design and analysis for field studies. Dordrecht: Kluwer Academic; 2002.

    Google Scholar 

  3. Prugh LR, Cunningham CX, Windell RM, Kertson BN, Ganz TR, Walker SL, et al. Fear of large carnivores amplifies human-caused mortality for mesopredators. Science. 2021;380:754–8.

    Article  Google Scholar 

  4. Brown L, Zedrosser A, Arnemo JM, Fuchs B, Kindberg J, Pelletier F. Landscape of fear or landscape of food? Moose hunting triggers an antipredator response in brown bears. Ecol Appl. 2023;33:1–15.

    Article  Google Scholar 

  5. Nathan R, Getz WM, Revilla E, Holyoak M, Kadmon R, Saltz D, et al. A movement ecology paradigm for unifying organismal movement research. Proc Natl Acad Sci USA. 2008;105:19052–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  6. Londe DW, Elmore RD, Davis CA, Hovick TJ, Fuhlendorf SD, Rutledge J. Why did the chicken not cross the road? Anthropogenic development influences the movement of a grassland bird. Ecol Appl. 2020;32:1–17.

    Google Scholar 

  7. Peaden JM, Nowakowski AJ, Tuberville TD, Buhlmann KA, Todd BD. Effects of roads and roadside fencing on movements, space use, and carapace temperatures of a threatened tortoise. Biol Conserv. 2017;214:13–22.

    Article  Google Scholar 

  8. Hromada SJ, Esque TC, Vandergast AG, Dutcher KE, Mitchell CI, Gray ME, et al. Using movement to inform conservation corridor design for Mojave desert tortoise. Mov Ecol. 2020;8:1–18.

    Article  Google Scholar 

  9. Hromada SJ, Esque TC, Vandergast AG, Drake KK, Chen F, Gottsacker B, et al. Linear and landscape disturbances alter Mojave desert tortoise movement behavior. Front Ecol Environ. 2023;11:1–18.

    Google Scholar 

  10. Harju S, Cambrin S, Berg J. Indirect impacts of a highway on movement behavioral states of a threatened tortoise and implications for landscape connectivity. Sci Rep. 2024;14:1–12.

    Article  Google Scholar 

  11. Stemle L, Rothermel BB, Searcy CA. GPS technology reveals larger home ranges for immature gopher tortoises. J Herpetol. 2022;2:172–79.

  12. Florida Fish and Wildlife Conservation Commission (FWC). Gopher Tortoise management plan. Gopherus polyphemus); 2012.

  13. U. S. Fish and Wildlife Service. Revised recovery plan for the Mojave population of the desert tortoise (Gopherus agassizii); 2011.

  14. U. S. Fish and Wildlife Service. Species status assessment for the Sonoran desert tortoise (Gopherus morafkai). Version 2.1. Albuquerque, US Fish and Wildlife Service, Southwest Region. 2021.

  15. Averill-Murray RC, Rosen PC, Jones CA, Jones TR, Lara-Resendiz RA, Edwards T et al. Gopherus morafkai. The IUCN Red List of Threatened Species. 2023. e.T97246109A97246177. https://doi.org/10.2305/IUCN.UK.20231.RLTS.T97246109A97246177.en

  16. Howland JM, Rorabaugh JC. Conservation and protection of the desert tortoise in Arizona. In: Van Devender TR, editor. The Sonoran desert tortoise: natural history, biology, and conservation. Tucson: The University of Arizona Press and the Arizona-Sonora Desert Museum; 2002. pp. 334–54.

    Google Scholar 

  17. Esque TC, Burquez A, Schwalbe CR, Van Devender TR, Annig PJ, Nijhuis MJ. Fire ecology of the Sonoran desert tortoise. In: Van Devender TR, editor. The Sonoran desert tortoise: natural history, biology, and conservation. Tucson: The University of Arizona Press and the Arizona-Sonora Desert Museum; 2002. pp. 312–33.

    Google Scholar 

  18. Shryock DF, Esque TC, Chen FC. A 30-year Chronosequence of burned areas in Arizona: effects of wildfires on Vegetation in Sonoran Desert Tortoise (Gopherus morafkai) habitats. US Department of the Interior, US Geological Survey; 2015.

  19. Gray KM, Steidl RJ. A plant invasion affects condition but not density or population structure of a vulnerable reptile. Biol Invasions. 2015;17:1979–88.

    Article  Google Scholar 

  20. Carter SK, Nussear KE, Esque TC, Leinwand II, Masters E, Inman RD, et al. Quantifying development to inform management of Mojave and Sonoran desert tortoise habitat in the American southwest. Endanger Species Res. 2020;42:167–84.

    Article  Google Scholar 

  21. Edwards T, Schwalbe CR, Swann DE, Goldberg CS. Implications of anthropogenic landscape change on inter-population movements of the desert tortoise (Gopherus agassizii). Conserv Genet. 2004;5:485–99.

    Article  Google Scholar 

  22. Arizona Interagency Desert Tortoise Team (AIDTT). Candidate conservation agreement for the Sonoran desert tortoise (Gopherus morafkai) in Arizona; 2015.

  23. Gammage G, Hall JS, Lang R, Welch N, Melnick R, Megapolitan. Arizona’s sun corridor. Morrison Institute for Public Policy, Arizona State University.; 2008.

  24. Rosen PC. Conservation status, ecology, and distribution of desert tortoises in Mexico. The Mexico tortoise project (2001–2013). United States Fish and Wildlife Service; 2014.

  25. Sutor S, McIntyre NE, Griffis-Kyle K. Characterizing range-wide impacts of anthropogenic barriers on structural landscape connectivity for the Sonoran desert tortoise (Gopherus morafkai). Landsc Ecol. 2023;38:1729–46.

    Article  Google Scholar 

  26. Nafus MG, Tuberville TD, Buhlmann KM, Todd BD. Relative abundance and demographic structure of Agassiz’s desert tortoise (Gopherus agassizii) along roads of varying size and traffic volume. Biol Conserv. 2013;162:100–6.

    Article  Google Scholar 

  27. Dutcher KE, Vandergast AG, Esque TC, Mitelberg A, Matocq MD, Heaton JS, et al. Genes in space: what Mojave desert tortoise genetics can tell us about landscape connectivity. Conserv Genet. 2020;21:289–303.

    Article  Google Scholar 

  28. Grandmaison DD, Frary VJ. Estimating the probability of illegal desert tortoise collection in the Sonoran Desert. J Wildl Manag. 2012;76:262–8.

    Article  Google Scholar 

  29. Berry KM, Aresco MJ. Threats and conservation needs for north American tortoises. In: Rostal DC, McCoy ED, Mushinsky HR, editors. Biology and Conservation of North American Tortoises. Baltimore: Johns Hopkins University; 2014. pp. 149–58.

    Google Scholar 

  30. Sadoti G, Gray ME, Farnsworth ML, Dickson BG. Discriminating patterns and drivers of multiscale movement in herpetofauna: the dynamic and changing environment of the Mojave desert tortoise. Ecol Evol. 2017;7:7010–22.

    Article  PubMed  PubMed Central  Google Scholar 

  31. Averill-Murray RC, Allison LJ. Travel management planning for wildlife with a case study on the Mojave desert tortoise. J Fish Wildl Manag. 2023;14:269–81.

    Article  Google Scholar 

  32. Taylor PD, Fahrig L, Henein K, Merriam G. Connectivity is a vital element of landscape structure. Oikos. 1993;68:571–3.

    Article  Google Scholar 

  33. Tewksbury JJ, Levey DJ, Haddad NM, Sargent S, Orrock JL, Weldon A, et al. Corridors affect plants, animals, and their interactions in fragmented landscapes. Proc Natl Acad Sci USA. 2002;99:12923–6.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  34. Berry K, Morafka DJ, Murphy RW. Defining the Desert Tortoise(s): our First Priority for a coherent conservation strategy. Chelonian Conserv Biol. 2002;4:249–62.

    Google Scholar 

  35. Nussear KE, Tuberville TD. Habitat characteristics of north American tortoises. In: Rostal DC, McCoy ED, Mushinsky HR, editors. Biology and Conservation of North American Tortoises. Baltimore: Johns Hopkins University; 2014. pp. 77–84.

    Google Scholar 

  36. Zylstra ER, Steidl RJ. Habitat use by Sonoran desert tortoises. J Wildl Manag. 2009;73:747–54.

    Article  Google Scholar 

  37. Grandmaison DD, Ingraldi MF, Peck FR. Desert tortoise microhabitat selection on the Florence Military Reservation, south-central Arizona. J Herpetol. 2010;44:581–90.

    Article  Google Scholar 

  38. Bridges A, Bateman HL, Owens AK, Jones CA, Miller W. Microhabitat selection of juvenile Sonoran Desert tortoises (Gopherus morafkai) in central Arizona. Chelonian Conserv Biol. 2016;15:219–30.

    Article  Google Scholar 

  39. Murphy RW, Berry KH, Edwards T, Leviton AE, Lathrop A, Riedle JD. The dazed and confused identity of Agassiz’s land tortoise, Gopherus agassizii (Testudines, Testudinidae) with the description of a new species, and its consequences for conservation. ZooKeys. 2011;113:39–71.

    Article  Google Scholar 

  40. Averill-Murray RC, Martin BE, Bailey SJ, Wirt EB. Activity and behavior of the Sonoran Desert Tortoise. In: Van Devender TR, editor. The Sonoran desert tortoise: natural history, biology, and conservation. Tucson: The University of Arizona Press and the Arizona-Sonora Desert Museum; 2002. pp. 135–58.

    Google Scholar 

  41. Riedle JD, Averill-Murray RC, Lutz CL, Bolen DK. Habitat use by desert tortoises (Gopherus agassizii) on alluvial fans in the Sonoran Desert, south-central Arizona. Copeia. 2008;2:414–20.

    Article  Google Scholar 

  42. Western Regional Climate Center. 2016. https://wrcc.dri.edu/Climate/west_coop_summaries.php. Accessed 10/28/2023.

  43. Brown DE. Biotic communities: southwestern United States and northwestern Mexico. Salt Lake: University of Utah; 1994.

    Google Scholar 

  44. Averill-Murray RC, Averill-Murray A. Regional-scale estimation of density and habitat use of the desert tortoise (Gopherus agassizii) in Arizona. J Herpetol. 2005;39:65–72.

    Article  Google Scholar 

  45. Paden LM, Andrews KM. Modification and validation of low-cost recreational GPS loggers for tortoises. Wildl Soc Bull. 2020;44:773–81.

    Article  Google Scholar 

  46. Morris G, Conner LM. Assessment of accuracy, fix success rate, and use of estimated horizontal position error (EHPE) to filter inaccurate data collected by a common commercially available GPS logger. PLoS ONE. 2017;12:e0189020.

    Article  PubMed  PubMed Central  Google Scholar 

  47. McClintock BT, Michelot T, momentuHMM. R package for generalized hidden Markov models of animal movement. Methods Ecol Evol. 2018;9:1518–30.

    Article  Google Scholar 

  48. R Core Team. R: a Language and Environment for Statistical Computing. Vienn: R Foundation for Statistical Computing; 2023.

    Google Scholar 

  49. Zucchini W, MacDonald IL. Hidden Markov models for time series: an introduction using R. CRC; 2017.

  50. Qi K, Chehbouni A, Huete AR, Kerr YH, Sorooshian S. A modified soil adjusted vegetation index. Remote Sens Environ. 1994;48:119–26.

    Article  Google Scholar 

  51. Gorelick N, Hancher M, Dixon M, Ilyushchenko S, Thau D, Moore R. Google Earth Engine: planetary-scale geospatial analysis for everyone. Remote Sens Environ. 2017;202:18–27.

    Article  Google Scholar 

  52. Wilson MF, O’Connell B, Brown C, Guinan JC, Grehan AJ. Multiscale terrain analysis of multibeam bathymetry data for habitat mapping on the continental slope. Mar Geod. 2007;30:3–35.

    Article  Google Scholar 

  53. Hijmans RJ, Bivand R, Forner K, Ooms J, Pebesma E, Sumner MD. Package ‘terra’ https://rspatial.org/index.html

  54. Dilts TE, Blum ME, Shoemaker KT, Weisberg PJ, Stewart KM. Improved topographic ruggedness indices more accurately model fine-scale ecological patterns. Landsc Ecol. 2023;38:1395–410.

    Article  Google Scholar 

  55. Avgar T, Potts JR, Lewis MA, Boyce MS. Integrated step selection analysis: bridging the gap between resource selection and animal movement. Methods Ecol Evol. 2016;7:619–30.

    Article  Google Scholar 

  56. Signer J, Fieberg J, Avgar T. Animal movement tools (amt): R package for managing tracking data and conducting habitat selection analyses. Ecol Evol. 2019;9:880–90.

    Article  PubMed  PubMed Central  Google Scholar 

  57. Muff S, Signer J, Fieberg J. Accounting for individual-specific variation in habitat‐selection studies: efficient estimation of mixed‐effects models using bayesian or frequentist computation. J Anim Ecol. 2020;89:80–92.

    Article  PubMed  Google Scholar 

  58. Brooks ME, Kristensen K, Van Benthem KJ, Magnusson A, Berg CW, Nielsen A, et al. glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. R J. 2017;9:378–400.

    Article  Google Scholar 

  59. Avgar T, Lele SR, Keim JL, Boyce MS. Relative selection strength: quantifying effect size in habitat-and step‐selection inference. Ecol Evol. 2017;7:5322–30.

    Article  PubMed  PubMed Central  Google Scholar 

  60. Beier P, Majka DR, Spencer WD. Forks in the road: choices in procedures for designing wildland linkages. Conserv Biol. 2008;22:836–51.

    Article  PubMed  Google Scholar 

  61. Averill-Murray RC, Esque TC, Allison LJ, Bassett S, Carter SK, Dutcher KE, et al. Connectivity of Mojave Desert tortoise populations—management implications for maintaining a viable recovery network. US Geological Survey; 2021.

  62. Van Devender TR. A natural history of the Sonoran Tortoise in Arizona. In: Van Devender TR, editor. The Sonoran desert tortoise: natural history, biology, and conservation. Tucson: The University of Arizona Press and the Arizona-Sonora Desert Museum; 2002. pp. 3–28.

    Google Scholar 

  63. Rieder JP, Newbold TAS, Ostoja SM. Structural changes in vegetation coincident with annual grass invasion negatively impacts sprint velocity of small vertebrates. Biol Invasions. 2010;12:2429–39.

    Article  Google Scholar 

  64. Rubke CA, Hoffman HA, Leavitt DJ, Branch WC. Sonoran desert tortoise (Gopherus morafkai) occupancy monitoring at the Organ Pipe Cactus National Monument. Arizona Game and Fish Department; 2016.

  65. Boarman WI, Sazaki M, Jennings BW. (1997). The effect of roads, barrier fences, and culverts on desert tortoise populations in California, USA. In: Proceedings: conservation, restoration, and management of tortoises and turtles-an international conference. New York: New York Turtle and Tortoise Society. 1997. pp. 54–58.

  66. Boarman WI, Sazaki M. A highway’s road-effect zone for desert tortoises (Gopherus agassizii). J Arid Environ. 2006;65:94–101.

    Article  Google Scholar 

  67. National Park Service. Traffic Counts by Location. https://irma.nps.gov/Stats/Reports/Park/ORPI. Accessed 10 December 2023.

  68. Sullivan BK, Owens AK, Sullivan KO, Sullivan EA. Spatial ecology of Sonoran desert tortoises (Gopherus morafkai): I. Fidelity in home range, refuge use and foraging behavior. J Herpetol. 2016;50:509–19.

    Article  Google Scholar 

  69. von Seckendorff Hoff K, Marlow RW. Impacts of vehicle road traffic on desert tortoise populations with consideration of conservation of tortoise habitat in southern Nevada. Chelonian Conserv Biol. 2002;4:449–56.

    Google Scholar 

  70. Averill-Murray RC, Fleming CH, Riedle J. Reptile home ranges revisited: a case study of space use of Sonoran Desert tortoises (Gopherus morafkai). Herpetol Conserv Biol. 2020;15:253–71.

    Google Scholar 

  71. Lovich JE, Puffer SR, Cummings K, Arundel TR, Vamstad MS, Brundige KD. High female desert tortoise mortality in the western Sonoran Desert during California’s epic 2012 2016 drought. Endanger Species Res. 2023;50:1–16.

    Article  Google Scholar 

  72. Zylstra ER, Steidl RJ, Jones CA, Averill-Murray RC. Spatial and temporal variation in survival of a rare reptile: a 22-year study of Sonoran desert tortoises. Oecologia. 2013;173:107–16.

    Article  PubMed  Google Scholar 

Download references

Acknowledgements

We thank the SWBRPP for providing funding, the NPS for providing housing and support for field work, and the Arizona Fish and Game Department for providing equipment. We especially appreciate Danny Martin, Tyler Coleman, Ami Pate, and Peter Holm of the NPS for coordinating field research and providing valuable information about Sonoran desert tortoises, the study system, and field work in the study area. We thank Tom Jones, Chad Rubke, and Gad Perry for providing field equipment. Finally, we thank Madison McGinnis, Sadie Roth, and Brice Martin for their assistance in the field and Carl Campbell for designing the custom-made 3D printed mounts. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Funding

Funding was provided by US Department of the Interior NPS Southwest Border Resource Protection Program (SWBRPP; Grant Nos. P20AC00294, P21AC10672), the Texas Tech Association of Biologists (TTUAB) 2022 Grand-in-Aid, the Charles H. Lowe, Jr. Herpetology Research Fund 2021 through the Tucson Herpetological Society (THS), and a private crowdfunding effort. Equipment donations came from the Arizona Game and Fish Department.

Author information

Authors and Affiliations

Authors

Contributions

SWS developed the methods, performed all analyses, and drafted the manuscript. NEM and KGK wrote the proposals that funded the work, financially supported the research, provided valuable feedback on the manuscript, and edited the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Sean Sutor.

Ethics declarations

Ethics approval and consent to participate

All tortoises were handled in accordance with State of Arizona Scientific Activity License no. SP847069, National Park Service (NPS) Scientific Research and Collecting Permit no. ORPI-2022-SCI_0003, NPS Institutional Animal Care and Use Committee (IACUC) protocol no. 21036-04 and Texas Tech University IACUC protocol no. 21036-04.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary Material 1

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Sutor, S., McIntyre, N.E. & Griffis-Kyle, K.L. Vegetation cover, topography, and low-traffic roads influence Sonoran desert tortoise (Gopherus morafkai) movement and habitat selection. Mov Ecol 12, 68 (2024). https://doi.org/10.1186/s40462-024-00503-8

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s40462-024-00503-8

Keywords