The impacts of anthropogenic linear features on the space-use patterns of two sympatric ungulates

doi:10.21203/rs.3.rs-5154134/v1

Download PDF

Research Article

The impacts of anthropogenic linear features on the space-use patterns of two sympatric ungulates

https://doi.org/10.21203/rs.3.rs-5154134/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

1. As human activity increases worldwide, many ecologists have focused on how anthropogenic linear features (ALFs) such as roads and fences impact and disrupt animal space-use behavior and how this disruption could potentially affect population viability. The properties of an animal’s occurrence distribution (OD), namely its size, shape, and habitat associations, reflect the animal’s balance of costs and benefits and thus can act as indirect indicators of behavioral optimality. Measuring deviations from theoretical space-use optimality can provide insight into the non-lethal effects of ALFs on wildlife in different environmental contexts.

2. Here, we focused on the seasonal space-use patterns of two wide-ranging, highly mobile species of great cultural and economic value: mule deer (Odocoileus hemionus; n = 3105) and pronghorn (Antilocapra americana; n = 320). We calculated the average use of six environmental and three ALF attributes, weighted by their intensity of use within the OD, and contrasted those with their respective average availability within a 100-km² reference area centered on each animal’s OD.

3. We show that mule deer space-use is more impacted by roads, while pronghorn space-use is affected more by fences, specifically in the winter when snow depth may hinder their ability to cross fences.

4. Our results highlight the dynamic nature of the availability domain and the importance of properly accounting for this dynamism in habitat selection analyses. This research expands on the theoretical literature of animal space use and their response to ALFs in a rapidly changing world and further provides practical trajectories for wildlife managers to take when mitigating ALF impacts on their target species.

fence

habitat availability

habitat selection

home range

home-range size

home-range compactness

mule deer

occurrence distribution

pronghorn

resource selection

road ecology

As human populations increase worldwide, there are growing efforts to measure the impacts of anthropogenic activity on wildlife. In particular, anthropogenic linear features (ALFs; e.g. roads or fences) are constructed at high densities in regions with otherwise light human footprints [1, 2]. In these regions, ALFs may be the only anthropogenic feature many animals encounter regularly, making it critical to understand how ALFs affect individual space use and, consequently, their fitness. While some animals are attracted to ALFs such as roads because they are associated with specific resources [3] or because they allow for efficient movement over long distances [4, 5], there is consensus in the literature that ALFs have negative effects on many species of wildlife [6–8]. Besides lethal effects associated with, for example, vehicle collisions on roads [1], predation on seismic lines [4], or entanglement in fences [2, 9], ALFs can also exert non-lethal fitness costs on wildlife by hampering an animal’s ability to access resources [2, 9–13] and avoid risk [4, 14–17]. Measuring such non-lethal effects of ALFs, however, is challenging because fitness costs accumulate through time and are difficult to disentangle from other co-occurring causal factors.

Home ranges can be used as proxies for quantifying non-lethal effects of ALFs on wildlife fitness. The home range is the area an animal uses during its normal, daily activities [18] and manifests from the animal balancing risks and rewards while moving in space [19, 20]. Theory establishes a direct link between space-use decisions, home range properties (relative size and shape in geographic space and composition in environmental space), and an individual’s fitness [21–23]; thus, there is strong theoretical support for the use of home range properties as indirect indicators of behavioral optimality and, thus, fitness. Deviations from theoretical optimality in home range properties should correspond to a proportional decrease in fitness and can provide insight into the non-lethal effects of ALFs on wildlife.

Home-range size is a property emerging from balancing the energy required for an animal’s survival and available energy on the landscape [24]. Theoretically, and all else being equal, smaller home ranges arise in habitats with a high resource quantity and quality compared to habitats without [19, 25–28]. If animals behave optimally, home ranges should be as small as possible to encompass the minimum resources the animal needs and the costs associated with obtaining those [19]. Factors that affect an animal’s ability to move through the landscape and encounter resources can force animals away from optimal behaviors and inflate home-range size. Thus, when ALFs decrease habitat quality, animals may be forced to move more and farther to acquire the resources they need, and home-range size increases as a result. ALFs that are semi-permeable may also increase home-range size, as animals would be devoting time and energy attempting to cross the barrier, meaning they need to spend more time and move farther to acquire resources to make up for the energy spent [25]. However, when ALFs fully block movement, they constrain animals within smaller areas and may, conversely, decrease home-range size. Thus, the effects of ALFs on home-range size are context-dependent, and positive/negative relationships between ALF density and home-range size do not automatically indicate positive/negative (or vice versa) effects of ALFs; inferring on ALF effects based on home range properties requires multiple lines of evidence.

Home-range shapes are determined by the distribution of resources across the landscape and the balance of risk and reward. Theory predicts that home ranges would be completely circular and compact in a landscape with a homogeneous distribution of resources [29]. As animals use space within their home range non-randomly, certain parts of the home range are used at a higher intensity than others [30, 31]; a compact home range helps minimize movement costs between high-intensity use areas. Thus, a compact home range indicates a high intake o resources and a low expenditure of energy to access these resources. When the landscape becomes more heterogeneous or more difficult to move through, home-range shapes are expected to expand, elongate, and become less compact in response [29]. Partially or fully impermeable ALFs may force animals to move in an elongated and complex pattern and deviate from the optimal circular home-range shape, which increases movement costs and leads to less efficient access to core areas.

Characteristics of a home range include not only its geometry bu also its environmental composition, which reflects an animal’s habitat selection. Habitat selection is a functional response whereby the strength of selection for (or avoidance of) conditions, resources, and risks depends on their availability on the landscape [32–37]. Generally, animals are expected to avoid ALFs because ALFs fragment habitat and can pose direct mortality risks [1, 8, 11]. However, as ALF density increases, ALF avoidance is expected to decline because animals are unable to avoid features that are pervasive on the landscape [38]. The pervasiveness of ALFs on the landscape likely also impacts how animals select for other environmental characteristics [39]. For instance, according to the risk-disturbance hypothesis, animals may perceive anthropogenic disturbances as risky and react similarly as they would to an actual predator [15]; because animals balance trade-offs of risks and reward, an abundance of risks could compromise an animal’s normal habitat selection behaviors.

Building on the existing theory, this study aims to understand how the shape and size of and the habitat selection within the home range of two sympatric ungulate species change in response to ALF density–specifically roads and fences. We generated hypotheses based on previous literature on ALF effects on wildlife, and we developed predictions based on theoretical expectations of how deviations from behavioral optimality should manifest through home range characteristics. While ALFs have been shown to affect wildlife across taxa, responses to ALFs are taxon- or even species-specific [7]; thus, we developed our hypotheses based on the literature on ungulates. We tested four hypotheses: (H1) roads are permeable barriers that reduce access to resources by fragmenting and degrading habitat [10, 40]. Therefore, we predicted that roads force ungulates to move more and further in search of foraging opportunities, thus we hypothesized home range size to increase in response to increased road density. In contrast, because fences are less permeable and more likely to block movement [13, 41], we predicted home-range size to decrease in response to increasing fence density (Fig. 1a). (H2) Due to the semi-permeable to permeable nature of ALFs, we predicted home ranges to become more complex, elongated, and less compact as ALF density increases (Fig. 1b). (H3) Many animals, and most ungulates, avoid ALFs [10, 11, 13, 42], but may become habituated as they become more common (the ‘relaxed avoidance’ hypothesis, sensu [34, 43]). We predicted that ungulates would avoid ALFs at a low ALF density, but as ALF density increases on the landscape, avoidance behavior would erode because ALFs become so pervasive that they are impossible to avoid (Fig. 1c). (H4) As ALF density increases, the animal’s selection for or avoidance of other habitat characteristics should become compromised. We predicted that selection strength for habitat attributes would approach zero (i.e. no selection nor avoidance) in response to increasing ALF density (Fig. 1d), marking a deviation from natural behavioral patterns. Through the combination of these four hypotheses (multiple lines of evidence), rather than each hypothesis on its own, we will be able to draw conclusions of the impacts of ALFs on ungulate space-use and home-range behavior.

2.1 A note on home-range estimation

Various analytical methods have been developed for quantifying animal home ranges or ‘occurrence distribution’ (OD) based on positional data [44]. Recent developments (reviewed in [45]) highlight the distinction between ‘range distribution’ (the expected lifetime space use; an extrapolation of the positional data) and the ‘occurrence distribution’ (the transient space use during the observation period; an interpolation of the positional data). The range distribution is more compatible with the classical definition of the home range [18], but its quantification requires sufficient positional data to remove the effects of positional and velocity autocorrelations [46], and it does not correspond to where the animal has been but rather where it will be over its lifetime. Because our focus here is on the seasonal dynamics of home range and habitat utilization, we have decided to focus on the occurrence distribution which we quantify using Local Convex Hulls (LoCoH; [47, 48]). A LoCoH-based OD is a poor estimator of the range distribution [44], but it has the advantages of making few assumptions and having the capacity to capture hard boundaries (for example, where animals encounter impermeable barriers).

2.2 Study species and telemetry data

We focused on the space-use patterns of two wide-ranging, highly mobile ungulate species of great cultural and economic value in the western United States: mule deer (Odocoileus hemionus) and pronghorn (Antilocapra americana). Anthropogenic development in the western U.S. threatens both species’ habitats, migration routes, and stopover sites [13, 41, 49–53], causing population declines across species [54, 55]. The state of Utah is home to a wide environmental gradient (Fig. A1), giving us the opportunity to study how the space-use decisions of these two ungulate species in response to ALFs differ across different environmental contexts, and to tease apart the effects of ALFs from the effects of the environments they may be associated with.

GPS collar data for both pronghorn and mule deer were provided by the Utah Division of Wildlife Resources (UDWR). Experienced crews from two helicopter companies (Heliwild and Quicksilver) captured the animals using net-guns with no chemical immobilization. UDWR biologists performed processing (e.g. measurements, disease samples, collar deployment, etc.). Biologists either processed the animal at the capture location (for mule deer fawns and all pronghorn) or at a nearby staging site (for mule deer adults). Helicopter chase times did not exceed 5-min and processing times did not exceed 15-min to reduce stress to the animal. Mule deer captures began in December 2012 and pronghorn captures began in December 2017. All captures and processing followed standard UDWR big game capture protocols. Captured animals were fitted with GPS collars programmed to collect fixes at 2-hour intervals.

Both species are partial migrants that may or may not shift ranges in spring and autumn, like other migratory species in temperate regions. As the OD represents the space the animal uses in residency periods, we restricted our temporal window of observations to periods when we were confident the animals would remain resident and not exploring, dispersing, or migrating [37]. We thus chose to subset the data to include only GPS points in the height of winter (the month of February) and summer (the month of July), as neither mule deer nor pronghorn included in our study migrated during these months. Our average data fold (a unique individual-year-season combination) consistent of 280 GPS points (but could be as low as 20 or as high as 2,016), corresponding to an average fix rate of 2.5 hours.

2.3 Home range estimation

To delineate occurrence distributions for each individual-season-year, we fitted LoCoH ODs using the ‘amt’ package [56] in R Statistical Software (v4.1.1; [57]). Specifically, we used the fixed number or k-LoCoH method, which constructs kernels from each point and k-1 of its nearest neighbors [48]. We delineated ODs with nine isopleth levels from 15%-95%, where a 15% isopleth would indicate more intesnely used space (the area within which the top 15% of use intensity is concentrated) and a 95% isopleth would indicate less intensely used space [48]. We removed ODs from the analysis that had fewer than 20 GPS points, were tracked for only one day, were smaller than 900-m² (the size of a raster cell, described below), or were outside the extent of the study area. Because LoCoH-based ODs are expected to be sensitive to both the temporal grain and extent of sampling, we weighted each OD by the number of GPS points multiplied by the number of days the animal was tracked during the OD period, normalized. Consequently, in our statistical analysis (described below) we allow better informed ODs to be more influential by giving a higher weight to ODs who had longer tracking periods while also giving less weight to individual-ODs who had the same tracking period but fewer locations [58].

2.4 Environmental data

To evaluate habitat selection, we assembled remote-sensed habitat data—topographic, vegetation, and climate—that would be important predictors to mule deer and pronghorn space use (Table 1). Rasters were projected using a template raster with a 30-m cell resolution and WGS84 UTM12N coordinate reference system to maintain a consistent spatial scale.

Table 1

Remotely-sensed habitat data.
Topography		Climate	Vegetation				Environmental Characteristic
Roughness	Elevation	Snow depth	Shrub cover	Tree cover	Herbaceous cover (forage)	Vegetation phenology (forage)	Covariate
Digital Elevation Model (DEM)		Snow Data Assimilation System (SNODAS)	Rangeland Analysis Platform (RAP), v3			MOD09Q1 Terra Normalized	Product
Spatial: 30-m; Temporal: N/A		Spatial: 1-km Temporal: 1-day	Spatial: 30-m Temporal: 1-year			Spatial: 250-m Temporal: 8-day	Spatial-Temporal Grain
Increased predation risk and movement costs	Increased summer forage availability, increased snow depth, increased tree cover	Increased costs in movement and thermoregulation energy, decreased forage availability	Increased forage availability (especially in winter	Increased predation risk for pronghorn. Increased availability of protection from temperature, snow depth, and predation for mule deer. Decreased forage availability.	Increased forage quality		Relationship to pronghorn or mule deer ecology
Pronghorn: avoid Mule deer: avoid	Pronghorn: avoid Mule deer: select in summer	Pronghorn: avoid Mule deer: avoid	Pronghorn: select Mule deer: select	Pronghorn: avoid Mule deer: seasonal selections	Pronghorn: select Mule deer: select		Predictions (all else being equal)

We derived terrain roughness from a digital elevation model (DEM; [59]) using the ‘raster’ package [60] in R. We derived herbaceous cover (both species rely heavily on herbaceous vegetation for forage) by summing annual and perennial vegetation cover from the Rangeland Analysis Platform (RAP; [61]). We assigned winter vegetation cover values as those form the previous year, assuming no new vegetation growth over the winter. For environmental covariates with sub-monthly temporal scale—snow depth [62] and Normalized Difference Vegetation Index (NDVI; [63])—we used the median value to represent a monthly aggregate per cell. Finally, to quantity forage availability, we multiplied a month’s scaled and median NDVI values (native temporal grain: 16-day) by that month’s herbaceous cover values (native temporal grain: annual). This would attribute the proportion of vegetation greenness that corresponds to herbaceous vegetation, rather than that of tree or shrub cover. We term this product ‘forage’, as higher values correspond to a larger percent cover of green herbaceous vegetation [64].

2.5 Anthropogenic linear feature data

Roads and fences are the dominant ALFs with potential impacts on wildlife in Utah. We accessed GIS layers of road routes from the Utah Geospatial Resource Center (UGRC; [65]), categorizing roads into ‘paved’ (interstates, state and U.S. highways, and paved local roads) and ‘unpaved’ (rural, service access, and other unpaved local roads). Roads are easy to detect from satellite imagery and most are constructed and maintained by government agencies who often keep records of their various attributes, giving us high confidence in our road dataset.

Unlike roads, fences are more difficult to detect from remote imagery, are often constructed by private landowners, and are unregulated [2]. This detection issue is compounded by the dynamic nature of fences, which can be constructed, modified, removed, or deteriorate much faster than roads, necessitating a more frequent mapping effort than other ALFs [2, 41]. In 2020, the UDWR and the Bureau of Land Management (BLM) initiated fence surveys on Utah public lands (R.J., unpublished data); however, these data were spatially limited and not publicly available at the time of this study. The Utah Department of Transportation (UDOT) maintains a publicly available dataset of fences [66], but these data are biased towards fences near highways. Because there is not a readily available, complete dataset of fences in Utah, we created a composite dataset of known and assumed fences that were built on the incomplete fence databases from UDOT and the collaboration with the UDWR and BLM. For the “assumed” fences, we used publicly available GIS layers of land management boundaries (e.g. boundaries between federal, state, tribal, and private entities) [65] and grazing allotments [67, 68] to identify boundaries between rights-holders and assumed that such boundaries would be fenced (sensu [69]). We circle back to the implications of these assumptions in the Discussion.

We rasterized these ALF layers into a road and fence raster where each 30-m cell was denoted by a 0 (meaning no road or no fence was within the cell) or a 1 (meaning a road or a fence was present in the cell). These two rasters were projected using the same template raster as the environmental attributes to maintain spatial consistency.

2.6 Response variables

We calculated the ‘home-range’ size as the area of the OD’s 95% isopleth. To quantify the compactness of ‘home-range’ shape, we used the perimeter (\(\:P\)) and area (\(\:A\)) of the 95% isopleth to calculate a shape index (\(\:SI\)):

\(\:SI=\:\frac{P}{2\pi\:\sqrt{\frac{\text{A}}{\pi\:}}}\) (Eq. 1)

where \(\:2\pi\:\sqrt{\frac{A}{\pi\:}}\) is the circumference of a circle, thus this equation compares the perimeter of the OD to the circumference of a circle of the same area. Thus, \(\:SI\approx\:1\) indicates a more circular, compact shape, and \(\:SI\gg\:1\)indicates a more elongated, complex shape (Fig. A2).

To quantify use intensity, we rasterized the OD isopleth hulls, assigning each raster cell a value between 1 (low-use intensity) and 9 (high-use intensity) depending on the corresponding isopleth; this would ensure that high-use cells were weighted high than low-use cells. We then normalized the raster cells so that they sum to 1. To make sure that our results were comparable between individuals, we delineated a standardized “availability” domain for each individual-season-year OD as a 100-km² (the maximum observed home-range area across all ODs) square extent centered around the closest GPS point to the centroid of the OD [37].

Selection or avoidance of a particular attribute is determined by contrasting what is used to what is available on the landscape. Our analytical approach was inspired by Resource Utilization Functions [70, 71]. To quantify the intensity of use of a given attribute, we multipled the attribute’s raster by the OD raster and summed the cells of the output raster. Because the values of the OD raster sum to 1, this would result in a weighted-average use-intensity of that attribute. To quantify the attribute’s availability, we calculated the average value of the attribute’s raster within each availability domain. To quantify the selection or avoidance of an attribute, we calculated a log-transformed selection ratio (\(\:logSR\)) for each attribute (\(\:j\)) for each OD:

\(\:logSR=\text{log}\left(\frac{use{d}_{j}}{availabl{e}_{j}}\right)\) (Eq. 2)

where \(\:logSR=0\) means that the attribute was used at the same intensity as expected based on its availability, indicating no selection nor avoidance. A positive \(\:logSR\) means that the attribute was used more intensely than expected based on its availability, indicating selection. A negative \(\:logSR\) means that the attribute was used less intensely than expected based on its availability, indicating avoidance. The \(\:logSR\) was calculated for each attribute and OD (individual-, season-, and year-specific).

2.7 Modeling

We first separated our data into subsets by randomly sampling 10% of observations in each species-season combination as ‘testing’ and the rest as ‘training’. The training sets were used in modeling and the testing sets were used in validation. We used the function weightedCorr() in the ‘wCorr’ package [72] in R to perform validation with observed compared to predicated values across models by calculating the Pearson’s correlation coefficient (Table S1).

We tested our four hypotheses with linear mixed models (LMM, Table 2) using the ‘lme4’ [73], ‘lmerTest’ [74], and ‘MuMIn’ [75] packages in R. To account for intra-species and -season behavioral variations, we fit every model for each species-season combination. We included an indicator for sex (two-level factor) to account for potential sexual differences in home range properties [21, 76]. We interacted the sex-indicator variable with ALF availabilities to measure how ALF effects vary between sexes. We included individual animal ID was a random effect and weighted the models by the number of GPS points multiplied by the number of days the animal was tracked during the OD period, normalized, allowing better informed ODs to better inform our inference. Besides roads and fences, other habitat characteristics (such as resource availability or landscape configuration) can also affect home range shape or size [19, 29, 76]. To account for this, and to avoid attributing to ALFs variation that is caused by other factors, we included all other environmental attributes as additional predictors in the models evaluating H1. For H2, we only included the environmental predictors that would make movement difficult, thus affecting the OD shape [76], thus we only included roughness and snow depth. For the models evaluating H3 and H4, where the \(\:logSR\) is the response variable, we included the \(\:logSR\) of all other attributes as predictor variables to control for the fact that selecting for one attribute may result in weakening the selection for another, as the animal cannot select for everything at once. The model for pronghorn-winter failed to converge as an LMM due to low sample size, so we removed the random effect and fit this model as a linear model.

Table 2

Full models for each hypothesis (all response variables were weighted by the normalized product of the sample size and the sampling duration).
Hypothesis Tested	Response Variable	Full Model
H1	\(\:\text{log}\left(Area\right)\)	\(\:avai{l}_{paved\:}+avai{l}_{unpaved\:}+avai{l}_{fence}+\) \(\:avai{l}_{elev}+{avai{l}_{rough}+avai{l}_{forage}+avai{l}_{shrub}+avai{l}_{tree}+avai{l}_{snow\:}}_{\:}+\) \(\:{I}_{male}+\) \(\:{I}_{male}:\left(avai{l}_{paved}+avai{l}_{unpaved}+avai{l}_{fence}\right)+\) \(\:nDay{s}^{*}\)
H2	\(\:\text{log}{\left(\text{log}\left(SI\right)\right)}^{†}\)	\(\:avai{l}_{paved}+avai{l}_{unpaved}+avai{l}_{fence}+\) \(\:\left(avai{l}_{rough}+avai{l}_{snow}\right)+\) \(\:{I}_{male}+\) \(\:{I}_{male}:\left(avai{l}_{paved}+avai{l}_{unpaved}+avai{l}_{fence}\right)+\) \(\:nDay{s}^{*}\)
H3	\(\:logS{R}_{j}\) (where \(\:j\) is either \(\:paved\:roads\), \(\:unpaved\:roads\), or \(\:fences\))	\(\:avai{l}_{j}+\) \(\:(logS{R}_{paved}+logS{R}_{unpaved}+logS{R}_{fence}{)}^{\text{\S\:}}+\) \(\:logS{R}_{elev}+logS{R}_{rough}+\:{logSR}_{forage}+{logSR}_{shrub}+{logSR}_{tree}+{logSR}_{snow}+\) \(\:{I}_{male}+{I}_{male}:\left(avai{l}_{j}\right)\)
H4	\(\:logS{R}_{j}\) (where \(\:j\) is either \(\:elevation\), \(\:roughness\), \(\:forage\), \(\:shrub\:cover\), \(\:tree\:cover\), or \(\:snow\:depth\))	\(\:avai{l}_{j}+\) \(\:logS{R}_{paved}+logS{R}_{unpaved}+logS{R}_{fence}+\) \(\:(logS{R}_{elev}+logS{R}_{rough}+\:{logSR}_{forage}+{logSR}_{shrub}+{logSR}_{tree}+{logSR}_{snow}{)}^{\text{\S\:}}+\) \(\:{I}_{male}+\) \(\:{I}_{male}:\left(avai{l}_{paved}+avai{l}_{unpaved}+avai{l}_{fence}\right)\)
*nDays is a predictor because the number of days an animal was tracked affects the size and shape of the OD

^†SI was log-transformed twice because its native range is (1, ∞], thus would be limited by (0, ∞], and so log-transforming it again would map SI to a range of [-∞, ∞].

^§ \(\:logSR\) would not be a predictor in the same model where it is a response variable.

We evaluated the relationships between home range characteristics and focal predictors by calculating model predictions across the range of observed values for each predictor one at a time, while holding all other predictors fixed at their mean value. We calculated 95% confidence intervals around mean estimates using the standard error method.

In total, our analysis consisted of 11,464 ODs: 10,479 from mule deer (54.5% winter and 45.5% summer) and 985 from pronghorn (51.9% winter and 48.1% summer; Fig. A4). Each OD was composed of, on average, 280 GPS points (ranged from 20–2.016). These ODs were derived from 3,425 individuals who had, on average, three ODs each. Of these 3,454 individuals, 3,105 were mule deer (73% female) and 320 were pronghorn (72% female). Because there were no significant differences in the results between sexes, here we present results on females for simplicity, as females made up most of our sample size. Similarly, while there were differences in effects from unpaved roads compared to paved roads, these differences were minor, so here we present on paved roads and fences only. Results from males and unpaved roads can be found in Appendix B).

3.1 Home range size

Pronghorn tended to have larger ODs (0.528 km²; SE: 0.50) than mule deer (0.298 km²; SE: 0.061). Winter ODs were larger (0.36 km²; SE: 0.0607 for deer and 0.67 km²; SE: 0.596 for pronghorn) than summer ODs (0.24 km²; SE: 0.0604 for deer and 0.38 km²; SE: 0.405 for pronghorn).

Higher paved road density was significantly correlated with smaller mule deer ODs across seasons (winter: \(\:\beta\:\) = -0.14, SE: 0.0115; summer: \(\:\beta\:\) = -0.082, SE: 0.0265; Fig. 2a). Fences, however, had little effect on deer OD size (winter: \(\:\beta\:\) = -0.0073, SE: 0.0136; summer: \(\:\beta\:\) = -0.013, SE: 0.0197; Fig. 2a). Pronghorn OD sizes decreased in the winter as paved road density increased (\(\:\beta\:\) = -0.23, SE: 0.096; Fig. 2b), but otherwise pronghorn OD size did not significantly respond to increasing ALF density (paved roads-summer: \(\:\beta\:\) = -0.0048, SE: 0.103; fences-winter: \(\:\beta\:\) = -0.12, SE: 0.0749; -summer: \(\:\beta\:\) = 0.077, SE: 0.0645).

3.2 Home range shape

OD SIs ranged from 1.07–3.63 and 1.28–3.32 for mule deer and pronghorn, respectively. Mule deer tended to have more compact ODs (SI = 1.9, SE: 5.47) than pronghorn (SI = 2.4, SE: 8.27). Summer ODs tended to be slightly more compact (SI = 1.7, SE: 4.94 and SI = 2.2, SE: 7.78 for deer and pronghorn respectively) than winter ODs for both species (SI = 2.1, SE: 5.99 and SI = 2.6, SE: 8.76 for deer and pronghorn respectively).

Mule deer OD shapes, overall, did not respond to increasing ALF density. Paved roads and fences did not significantly impact deer OD shapes (paved roads-winter: \(\:\beta\:\) = -0.0019, SE: 0.00337; -summer: \(\:\beta\:\) = 0.0023, SE: 0.006237; fences-winter: \(\:\beta\:\) = 0.0055, SE: 0.00394; -summer: \(\:\beta\:\) = 0.009, SE: 0.00466; Fig. 3a). Pronghorn OD shapes, however, showed variable responses to ALFs. In the summer, pronghorn ODs tended to become more compact in response to paved roads (\(\:\beta\:\) = -0.062, SE: 0.0220) while winter OD shapes showed no significant response (\(\:\beta\:\) = -0.019, SE: 0.0228; Fig. 3b). In contrast, increased fence density was highly correlated with pronghorn OD shapes, where they become more elongated in the winter (\(\:\beta\:\) = 0.052, SE: 0.0166) and more compact in the summer (\(\:\beta\:\) = -0.034, SE: 0.0137; Fig. 3b).

3.3 Anthropogenic linear feature selection patterns

Across species and season, individuals tended to increase their avoidance of paved roads as this ALF density increased, particularly in the winter (deer-winter: \(\:\beta\:\) = -0.371, SE: 0.0142; deer-summer: \(\:\beta\:\) = -0.219, SE: 0.0270; pronghorn-winter: \(\:\beta\:\) = -0.42, SE: 0.0609; pronghorn-summer: \(\:\beta\:\) = -0.26, SE: 0.0667). Mule deer across seasons and pronghorn in summer did not significantly select nor avoid fences and this response did not significantly change in response to fence density (deer-winter: \(\:\beta\:\) = 0.036, SE: 0.0162; deer-summer: \(\:\beta\:\) = -0.0027, SE: 0.0225; pronghorn-summer: \(\:\beta\:\) = -0.02, SE: 0.0283; Fig. 4a,b). However, pronghorn in winter significantly increased their avoidance of fences as fence density increased (\(\:\beta\:\) = -0.24, SE: 0.0347; Fig. 4b).

3.4 Habitat selection patterns

Increasing ALF density had variable impacts on habitat selection patterns across ALF types, species, and season. Here, we highlight results from forage and snow depth to demonstrate how ALF density affected selection a habitat attribute we expected animals to select and one we expected animals to avoid, respectively (see Appendix B for other habitat attributes). As expected mule deer tended to select for forage, and this selection was more significant in the summer (winter: \(\:\beta\:\) = 0.013, SE: 0.0197; summer: \(\:\beta\:\) = 0.27, SE: 0.0483). Increased paved road availability had minor effects on mule deer selection for forage in all seasons (winter: \(\:\beta\:\) = -0.002, SE: 0.00925; summer: \(\:\beta\:\) = 0.017, SE: 0.0343). Mule deer selection for winter forage decreased as fence density increased (\(\:\beta\:\) = -0.063, SE: 0.00998), but their summer forage selection patterns did not respond to fences (\(\:\beta\:\) = -0.0085, SE: 0.02620; Fig. 5a). Against expectations, pronghorn significantly avoided forage in the summer (\(\:\beta\:\) = -0.25, SE: 0.0835) and did not significantly select nor avoid forage in the winter (\(\:\beta\:\) = -0.021, SE: 0.07). Pronghorn forage patterns across season did not significantly respond to increased paved roads (winter: \(\:\beta\:\) = -0.041, SE: 0.0739; summer: \(\:\beta\:\) = 0.11, SE: 0.0102). Fences, however, significantly reduced pronghorn selection for forage in all season, particularly in winter (winter: \(\:\beta\:\) = -0.34, SE: 0.0563; summer: \(\:\beta\:\) = -0.14, SE: 0.0628; Fig. 5b).

Mule deer did not significantly select nor avoid snow depth (\(\:\beta\:\) = 0.025, SE: 0.0213), however, pronghorn, against expectations, slightly tended to select snow depth (\(\:\beta\:\) = 0.13, SE: 0.0347). Changes in snow depth selection patterns were significantly correlated with increased ALF density for both species, but there were opposite responses to paved roads compared to fences. As paved road density increased, snow depth avoidance increased (mule deer: \(\:\beta\:\) = -0.16, SE: 0.0176; pronghorn: \(\:\beta\:\) = -0.18, SE: 0.0563). In contrast, as fence density increased, snow depth selection increased (mule deer: \(\:\beta\:\) = 0.084, SE: 0.0189; pronghorn: \(\:\beta\:\) = 0.16, SE: 0.0444; Fig. 5c,d).

Our study examined the non-lethal effects of anthropogenic linear features on wildlife by comparing seasonal home ranges for two sympatric ungulate species in western North America. Using empirical data, we quantified pronghorn and mule deer home-range properties in response to increased road and fence density, which allowed us to capture deviations from their theoretical optimum. By examining results from multiple hypotheses, we were able to understand how these two ungulate species were impacted by roads and fences in and around their home ranges. We hypothesized that H1) home-range size would increase in response to road density but decrease in response to fence density; H2) home-range shape would become more complex, elongated, and less compact in response to increased ALF density; H3) mule deer and pronghorn would avoid ALFs at a low ALF density, but this avoidance would erode as ALF density increased; H4) selection strength for surrounding habitat attributes would approach zero in response to ALF density (Fig. 1). By looking at the results of each hypothesis as pieces of a whole, our overall findings are: 1) mule deer space-use patterns were heavily affected by paved roads, which reduced their ability to move, and 2) seasonality and snow altered the permeability of fences for pronghorn. These results indicate that roads and fences prevent optimal space-use behavior, possibly resulting in non-lethal but detrimental effects on individual fitness.

4.1 Effects of roads

Mule deer intensely avoided paved roads, particularly as paved road density increased. We further found that mule deer home-range sizes significantly decreased in response to paved road density (12.7% and 7.9% decrease in winter and summer respectively). These finding combined suggest that paved roads are not forcing deer to move more, as we had predicted, but rather are boxing them in and preventing them from moving, thus restricting and shrinking their home-range size. Home range theory predicts that animals should use the smallest space possible to maximize resource intake and minimize energy expenditure [19]. However, as other researchers have noted, it can be difficult to predict how the home range will respond to anthropogenic disturbances and how these responses impact animal fitness [25]. Our results of paved roads decreasing home-range size demonstrates this complexity, as the home range decreasing in this context does not necessarily mean the animal is more efficiently exploiting its environment, but rather, it likely indicates that roads are fully restricting their movement and preventing them from accessing resources. An alternative explanation is that vegetation growing alongside paved roads provides sufficient food subsidies for mule deer to fulfill energy requirements while maintaining a smaller home range; the exploitation of roadside vegetation has been documented in other ungulate species (e.g. [77]). However, in the arid environment of Utah, it is unlikely that mule deer would rely on roadside vegetation to the point of substantially shrinking their space use. A more likely interpretation of the pattern that emerged from our results is that roads are constraining their movement. Even though it did not match the expected deviation from theoretical optimality [19], the pattern we observed suggests that roads have a negative effect on mule deer space use.

While pronghorn also intensely avoided paved roads, this avoidance did not seem to impact pronghorn home ranges in the same way that it did for mule deer. Pronghorn home-range size and shape, overall, did not respond significantly to increased paved road and neither did their forage selection patterns. Perhaps a reason for this lack of strong responses is that there were so few paved roads within pronghorn seasonal ranges to begin with—while some mule deer seasonal ranges had a paved road density as high as 20%, the highest paved road density in a pronghorn seasonal range was 7%. This could indicate that pronghorn avoid paved roads at a coarser scale (2nd-order habitat selection, or where an animal in its geographic range places its home range) than this analysis, leaving their home ranges largely unaffected by road density. Another possibility is that there is a decreased need for people to develop roads in the habitats where pronghorn live that they do not need to avoid them at a coarse scale. This suggests that pronghorn space use at the scale of this analysis (3rd-order habitat selection, or where and what the animal selects within its home range) is less sensitive to paved road development than mule deer.

4.2 Effects of fences

In contrast to paved roads, fences hardly impacted mule deer home ranges. Deer did not strongly avoid fences nor did this selection pattern change as fence density increased, their home-range size and shape did not respond to increased fence density, and their habitat selection patterns were generally unaffected by increased fence density. This indicates that fences may not be a movement impediment for mule deer as paved roads are.

Pronghorn responses to fences were more intense than from mule deer. In the winter, pronghorn significantly avoided fence density, and this avoidance intensely increased with increasing fence density. While this strong avoidance did not seem to affect pronghorn home-range size, their winter home-range shape became more elongated as fence density increased, corroborating our prediction.

The differing responses of deer and pronghorn to increased fence density indicate that the permeability of fences is species-specific. Pronghorn tend to crawl under fences rather than jump over [78,though see 79]. As pronghorn rely on their fast speed to avoid predation [80], corssing fences in this way can impair their ability to escape. There is some evidence in the literature to suggest that fences led to increased pronghorn predation, such as the case with [80], who observed coyotes using fences to trap adult pronghorn. In contrast to pronghorn, mule deer can jump over fences. This more agile fence crossing behavior, combined with their general escape behavior—a “stot,” where all four feet touch and leave the ground at the same time and allows mule deer to traverse rough terrain more easility [81]—suggests that mule deer are more able to escape risk and predation through areas with high fence density and other rough terrain. Our findings support those of [79], who found that fences likely pose a lower mortality risk to deer than pronghorn, even though they observed deer crossing fences more frequently. Overall, our findings match our understanding of these two species’ ecology, providing additional evidence that fences constrain pronghorn movement more so than mule deer.

Not only is fence permeability different between species but also between seasons. Pronghorn did not significantly avoid fences in the summer but intensely did so in the winter, and pronghorn home-range composition was only impacted in the winter. This seasonal permeability combined with the result that pronghorn were increasingly unable to avoid snow depth in areas of high fence density suggests that snow accumulation may be altering fence permeability by reducing the space between the ground and the fence and preventing pronghorn from crawling underneath. Compared to other ungulates, pronghorn are more sensitive to cold temperatures and snow [82] and will move hundreds of kilometers to avoid severe winter weather [78, 80]. The ability to avoid deep snow cover is critical for pronghorn winter survival, as demonstrated by a case in 1983 when hundreds of pronghorn starved or froze to death due to snow piling up near fences and preventing them from crossing [83]. As pronghorn in our system also increasingly avoided green forage as fence density increased, areas with high fence density and high snow accumulation could be detrimental for these animals to find resources in the winter when survival is critical.

4.3 Concluding remarks

Our results demonstrate the use of home range characteristics to detect deviations from behavioral optimality that are theoretically associated with reduced fitness. There has been growing interest in linking animal space-use with its fitness consequences at the individual level and its demographic consequences at the population level. Establishing these links has been challenging, first, due to the lack of availability of simulatneous information on fitness and space use for the same individuals; second, due to the diffuse and cumulative nature of non-lethal effects, which are difficult to attribute to any single driver. The development of multiple hypotheses and the testing of those hypotheses, while controlling for alternative drivers, has recently provided important breakthroughs—for instance, in measuring non-consumptive effects of predators on prey [84]. Similarly, our study shows that non-lethal effects of anthropocentric structures can be detected and quantified by simulatenously evaluating multiple hypotheses pertaining to different metrics of space use while controlling for other factors.

While we did not measure fitness, we relied on a large body of theoretical literature that ties the properties of animal home ranges to their fitness consequences. Despite the importance of this theoretical connection, home ranges are often used as a purely descriptive tool to quantify the patterns of animal space use. Our study provides an example of how the theoretical underpinnings of the home range concept can be leveraged to make inferences beyond description and begin to answer elusive questions about the effects of chronic stressors on wildlife populations. In this study, we focused on non-lethal effects of roads and fences; however, a similar approach can be used to evaluate non-lethal effects of other factors that are expected to affect animal home-ranging behavior—such as drought or altered fire regimes—so long as concurrent drivers of animal space use are identified and measured.

Our study highlights the importance of the availability domain in ecological analyses. The logSR of many of the ALF and habitat attributes switched from positive (selection) to negative (avoidance) or vice-versa as ALF density increased. This means that we could draw very different conclusions about these animals’ habitat selection patterns if we had observed populations that happened to live in areas with only high or low ALF densities; and if we did not include ALF density, habitat selection responses may cancel out and we may incorrectly conclude that these animals’ home ranges do not respond to the focal habitat feature at all.

While we presented responses at the population level, our results showed wide individual variability, even among groups of the same species and sex and in the same season. These wide variations in space-use responses could lead us to conclude that, at the population level, these animals respond to roads and fences in a certain way, while individuals within different sub-populations may respond differently. Further research should focus on disentangling drivers of this variability. One study, for example, found that migrating pronghorn avoided all ALFs except for fenced rods, while resident pronghorn selected areas closer to fenced pastures but avoided all roads [50], indicating that migration status can affect habitat selection. Another possibility for this variation could be because of the wide variation in climate and topography in Utah. While we did not separate individuals by geographic or ecological regions in our study, it is reasonable to expect that a pronghorn in the sagebrush steppe of the northeast of our study area would use space and respond to its environment differently than one in the southwest desert and basins. These additional behavioral, intrinsic, and extrinsic factors could provide more nuance to the wide individual variation we observed.

A promising direction for future research is to quantify responses of animals to ALFs depending on their specific attributes, beyond coarse characterizations such as paved versus unpaved roads, e.g. lane width, traffic volume, fence height, fence type (e.g. barbed wire versus PVC-wrapped). These attributes are likely to modulate behavioral responses to ALFs, and a fine-scale examination of ALF effects on animal movement and space use would benefit from adding these attributes to an analysis [50]. Furthermore, our fence data, as it was a combination of known and assumed fences, might not have fully represented (or possibly misrepresented) actual fences on the landscape. Fences are one of the most expansive anthropogenic features on the landscape, often at higher densities than roads [2, 9, 49], and yet comprehensive and accurate data on their locations is lacking; having a comprehensive census of known fences would provide a more accurate understanding of wildlife responses to fences.

This research not only expands the theoretical literature of animal home ranges and their response to ALFs but also offers some practical trajectories for wildlife managers to take when mitigating ALF impacts on their target species. Our findings highlight the species- and season-specific nature of ALF impacts, showing that mitigation strategies on one linear feature for one species may be ineffective for another species. The result that pronghorn are more sensitive to fences in the winter than the summer shows the seasonal affects of ALFs and demonstrates that regions with deep snow cover may need to make different mitigation efforts—such as lay-down fences [85]—than regions that do not receive as much snow. More broadly, simultaneously looking at how anthropogenic structures affect multiple aspects of an animal’s space use can provide detailed information in support of management decisions that we could not get from looking at various aspects in isolation.

Acknowledgements: We would like to thank the Utah Division of Wildlife Resources (UDWR) for funding this project and collecting tracking data. We thank Dr. Brian Smith for his assistance with data analysis and members of the Avgar and Jones lab for their help in reviewing this paper.

Funding Declaration: This project was funded by the Utah Division of Wildlife Resources, grant ID 204370

Conflict of Interest: The authors have no conflicts of interest to declare.

Author Contributions: Following the CRediT’s nomenclature: R.H.’s contributions include conceptualization, data curation, formal analysis, investigation, methodology, visualization, writing—original draft. S.P.’s contributions include project administration, supervision, writing—review and editing. T.A’s contributions include conceptualization, funding acquisition, methodology, project administration, resources, supervision, writing—review and editing.

Consent to Participate declarations: not applicable

Animal Ethics declaration: all animal handling, captures, and processing followed standard Utah Division of Wildlife Resources big game capture protocols

Forman RTT, Alexander LE. Roads and their major ecological effects. Annual Review of Ecology and Systematics [Internet]. 1998;29:207–31. https://www.annualreviews.org/doi/10.1146/annurev.ecolsys.29.1.207
McInturff A, Xu W, Wilkinson CE, Dejid N, Brashares JS. Fence ecology: Frameworks for understanding the ecological effects of fences. BioScience [Internet]. 2020;70:971–85. https://doi.org/10.1093/biosci/biaa103
Grosman PD, Jaeger JAG, Biron PM, Dussault C, Ouellet J-P. Trade-off between road avoidance and attraction by roadside salt pools in moose: An agent-based model to assess measures for reducing moose-vehicle collisions. Ecological Modelling [Internet]. 2011;222:1423–35. https://www.sciencedirect.com/science/article/pii/S0304380011000548
Dickie M, Serrouya R, McNay RS, Boutin S. Faster and farther: wolf movement on linear features and implications for hunting behaviour. Journal of Applied Ecology [Internet]. 2017;54:253–63. https://onlinelibrary.wiley.com/doi/abs/10.1111/1365-2664.12732
Latham ADM, Latham MC, Boyce MS, Boutin S. Movement responses by wolves to industrial linear features and their effect on woodland caribou in northeastern Alberta. Ecological Applications [Internet]. 2011;21:2854–65. https://onlinelibrary.wiley.com/doi/abs/10.1890/11-0666.1
Benítez-López A, Alkemade R, Verweij PA. The impacts of roads and other infrastructure on mammal and bird populations: A meta-analysis. Biological Conservation [Internet]. 2010;143:1307–16. https://www.sciencedirect.com/science/article/pii/S0006320710000480
Fahrig L, Rytwinski T. Effects of roads on animal abundance: An empirical review and synthesis. Ecology and Society [Internet]. 2009;14. https://www.jstor.org/stable/26268057
Shepard DB, Kuhns AR, Dreslik MJ, Phillips CA. Roads as barriers to animal movement in fragmented landscapes. Animal Conservation [Internet]. 2008;11:288–96. https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1469-1795.2008.00183.x
Jakes AF, Jones PF, Paige LC, Seidler RG, Huijser MP. A fence runs through it: A call for greater attention to the influence of fences on wildlife and ecosystems. Biological Conservation [Internet]. 2018;227:310–8. https://www.sciencedirect.com/science/article/pii/S0006320718308449
Beyer HL, Gurarie E, Börger L, Panzacchi M, Basille M, Herfindal I et al. ‘You shall not pass!’: quantifying barrier permeability and proximity avoidance by animals. Journal of Animal Ecology [Internet]. 2016;85:43–53. https://onlinelibrary.wiley.com/doi/abs/10.1111/1365-2656.12275
Jalkotzy MG, Ross I, Nasserden EM. The effects of linear developments on wildlife. Arc Wildl Serv. 1997.
Scrafford MA, Avgar T, Heeres R, Boyce MS. Roads elicit negative movement and habitat-selection responses by wolverines (Gulo gulo luscus). Behavioral Ecology [Internet]. 2018;29:534–42. https://academic.oup.com/beheco/article/29/3/534/4844878
Xu W, Dejid N, Herrmann V, Sawyer H, Middleton AD. Barrier Behaviour Analysis (BaBA) reveals extensive effects of fencing on wide-ranging ungulates. Journal of Applied Ecology [Internet]. 2021;58:690–8. https://onlinelibrary.wiley.com/doi/abs/10.1111/1365-2664.13806
Berger J. Fear, human shields and the redistribution of prey and predators in protected areas. Biology Letters [Internet]. 2007; https://royalsocietypublishing.org/doi/10.1098/rsbl.2007.0415
Frid A, Dill L. Human-caused disturbance stimuli as a form of predation risk. Conservation Ecology [Internet]. 2002;6. https://www.jstor.org/stable/26271862
McKenzie HW, Merrill EH, Spiteri RJ, Lewis MA. How linear features alter predator movement and the functional response. Interface Focus [Internet]. 2012;2:205–16. https://royalsocietypublishing.org/doi/abs/10.1098/rsfs.2011.0086
Riva F, Acorn JH, Nielsen SE. Narrow anthropogenic corridors direct the movement of a generalist boreal butterfly. Biology Letters [Internet]. 2018;14:20170770. https://royalsocietypublishing.org/doi/10.1098/rsbl.2017.0770
Burt WH. Territoriality and home range concepts as applied to mammals. Journal of Mammalogy [Internet]. 1943;24:346–52. https://doi.org/10.2307/1374834
Mitchell MS, Powell RA. Foraging optimally for home ranges. Journal of Mammalogy [Internet]. 2012;93:917–28. https://academic.oup.com/jmammal/article-lookup/doi/10.1644/11-MAMM-S-157.1
Powell RA, Mitchell MS. What is a home range? Journal of Mammalogy [Internet]. 2012;93:948–58. https://academic.oup.com/jmammal/article-lookup/doi/10.1644/11-MAMM-S-177.1
Boratyński Z. Energetic constraints on mammalian home-range size. Functional Ecology [Internet]. 2020;34:468–74. https://onlinelibrary.wiley.com/doi/abs/10.1111/1365-2435.13480
Froy H, Börger L, Regan CE, Morris A, Morris S, Pilkington JG et al. Declining home range area predicts reduced late-life survival in two wild ungulate populations. Ecology Letters [Internet]. 2018;21:1001–9. https://onlinelibrary.wiley.com/doi/abs/10.1111/ele.12965
Leblond M, Dussault C, Ouellet J-P. Impacts of Human Disturbance on Large Prey Species: Do Behavioral Reactions Translate to Fitness Consequences? PLOS ONE [Internet]. 2013;8:e73695. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0073695
McNab BK. Bioenergetics and the determination of home range size. The American Naturalist [Internet]. 1963;97:133–40. https://www.journals.uchicago.edu/doi/10.1086/282264
Dickie M, Serrouya R, Avgar T, McLoughlin P, McNay RS, DeMars C et al. Resource exploitation efficiency collapses the home range of an apex predator. Ecology [Internet]. 2022;103:e3642. https://onlinelibrary.wiley.com/doi/abs/10.1002/ecy.3642
Mk FST, Ew B, Rosing-Asvid A, Messier F. Determinants of Home Range Size for Polar Bears (Ursus maritimus). Ecology Letters [Internet]. 1999;2:311–8. https://onlinelibrary.wiley.com/doi/abs/10.1046/j.1461-0248.1999.00090.x
Morellet N, Bonenfant C, Börger L, Ossi F, Cagnacci F, Heurich M et al. Seasonality, weather and climate affect home range size in roe deer across a wide latitudinal gradient within Europe. Journal of Animal Ecology [Internet]. 2013;82:1326–39. https://onlinelibrary.wiley.com/doi/abs/10.1111/1365-2656.12105
Viana DS, Granados JE, Fandos P, Pérez JM, Cano-Manuel FJ, Burón D et al. Linking seasonal home range size with habitat selection and movement in a mountain ungulate. Movement Ecology [Internet]. 2018;6:1. https://doi.org/10.1186/s40462-017-0119-8
Ford RG. Home range in a patchy environment: Optimal foraging Predictions1. American Zoologist [Internet]. 1983;23:315–26. https://doi.org/10.1093/icb/23.2.315
Powell RA, Zimmerman JW, Seaman DE. Ecology and Behaviour of North American Black Bears: Home Ranges, Habitat, and Social Organization. Springer Science & Business Media; 1997.
Samuel MD, Pierce DJ, Garton EO. Identifying areas of concentrated use within the home range. Journal of Animal Ecology [Internet]. 1985;54:711–9. https://www.jstor.org/stable/4373
Avgar T, Betini GS, Fryxell JM. Habitat selection patterns are density dependent under the ideal free distribution. Journal of Animal Ecology [Internet]. 2020;89:2777–87. https://onlinelibrary.wiley.com/doi/abs/10.1111/1365-2656.13352
Duparc A, Garel M, Marchand P, Dubray D, Maillard D, Loison A. Revisiting the functional response in habitat selection for large herbivores: a matter of spatial variation in resource distribution? Quinn J, editor. Behavioral Ecology [Internet]. 2019;30:1725–33. https://academic.oup.com/beheco/article/30/6/1725/5552079
Dupke C, Peters A, Morellet N, Heurich M. Holling meets habitat selection: Functional response of large herbivores revisited. Movement Ecology [Internet]. 2021;9:45. https://doi.org/10.1186/s40462-021-00282-6
Matthiopoulos J, Hebblewhite M, Aarts G, Fieberg J. Generalized functional responses for species distributions. Ecology [Internet]. 2011;92:583–9. https://onlinelibrary.wiley.com/doi/abs/10.1890/10-0751.1
Mysterud A, Ims RA. Functional Responses in Habitat Use: Availability Influences Relative Use in Trade-Off Situations. Ecology [Internet]. 1998;79:1435–41. https://onlinelibrary.wiley.com/doi/abs/10.1890/0012-9658%281998%29079%5B1435%3AFRIHUA%5D2.0.CO%3B2
Winter VA, Smith BJ, Berger DJ, Hart RB, Huang J, Manlove K et al. Forecasting animal distribution through individual habitat selection: insights for population inference and transferable predictions. 2024; https://onlinelibrary.wiley.com/doi/abs/10.1111/ecog.07225
Seigle-Ferrand J, Marchand P, Morellet N, Gaillard J-M, Hewison AJM, Saïd S et al. On this side of the fence: Functional responses to linear landscape features shape the home range of large herbivores. Journal of Animal Ecology [Internet]. 2022;91:443–57. https://onlinelibrary.wiley.com/doi/abs/10.1111/1365-2656.13633
Houle M, Fortin D, Dussault C, Courtois R, Ouellet J-P. Cumulative effects of forestry on habitat use by gray wolf (canis lupus) in the boreal forest. Landscape Ecology [Internet]. 2010;25:419–33. https://doi.org/10.1007/s10980-009-9420-2
Eigenbrod F, Hecnar SJ, Fahrig L. Accessible habitat: an improved measure of the effects of habitat loss and roads on wildlife populations. Landscape Ecology [Internet]. 2008;23:159–68. https://doi.org/10.1007/s10980-007-9174-7
Jones PF, Jakes AF, Telander AC, Sawyer H, Martin BH, Hebblewhite M. Fences reduce habitat for a partially migratory ungulate in the Northern Sagebrush Steppe. Ecosphere [Internet]. 2019;10:e02782. https://onlinelibrary.wiley.com/doi/abs/10.1002/ecs2.2782
Dickie M, McNay SR, Sutherland GD, Cody M, Avgar T. Corridors or risk? Movement along, and use of, linear features varies predictably among large mammal predator and prey species. Journal of Animal Ecology [Internet]. 2020;89:623–34. https://onlinelibrary.wiley.com/doi/abs/10.1111/1365-2656.13130
Holbrook JD, Olson LE, DeCesare NJ, Hebblewhite M, Squires JR, Steenweg R. Functional responses in habitat selection: clarifying hypotheses and interpretations. Ecological Applications [Internet]. 2019;29:e01852. https://onlinelibrary.wiley.com/doi/abs/10.1002/eap.1852
Noonan MJ, Tucker MA, Fleming CH, Akre TS, Alberts SC, Ali AH et al. A comprehensive analysis of autocorrelation and bias in home range estimation. Ecological Monographs [Internet]. 2019;89:e01344. https://onlinelibrary.wiley.com/doi/abs/10.1002/ecm.1344
Alston JM, Fleming CH, Noonan MJ, Tucker MA, Silva I, Folta C et al. Clarifying space use concepts in ecology: Range vs. Occurrence distributions. 2022; https://www.biorxiv.org/content/early/2022/09/30/2022.09.29.509951
Silva I, Fleming CH, Noonan MJ, Alston J, Folta C, Fagan WF et al. Autocorrelation-informed home range estimation: A review and practical guide. Methods in Ecology and Evolution [Internet]. 2022;13:534–44. https://onlinelibrary.wiley.com/doi/abs/10.1111/2041-210X.13786
Getz WM, Wilmers CC. A local nearest-neighbor convex-hull construction of home ranges and utilization distributions. Ecography [Internet]. 2004;27:489–505. https://onlinelibrary.wiley.com/doi/abs/10.1111/j.0906-7590.2004.03835.x
Getz WM, Fortmann-Roe S, Cross PC, Lyons AJ, Ryan SJ, Wilmers CC. LoCoH: Nonparameteric kernel methods for constructing home ranges and utilization distributions. PLOS ONE [Internet]. 2007;2:1–11. https://doi.org/10.1371/journal.pone.0000207
Berger J. The Last Mile: How to Sustain Long-Distance Migration in Mammals. Conservation Biology [Internet]. 2004;18:320–31. https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1523-1739.2004.00548.x
Jones PF, Jakes AF, Vegter SE, Verhage MS. Is it the road or the fence? Influence of linear anthropogenic features on the movement and distribution of a partially migratory ungulate. Movement Ecology [Internet]. 2022;10:37. https://doi.org/10.1186/s40462-022-00336-3
Sandoval Lambert M, Sawyer H, Merkle JA. Responses to natural gas development differ by season for two migratory ungulates. Ecological Applications [Internet]. 2022;32:e2652. https://onlinelibrary.wiley.com/doi/abs/10.1002/eap.2652
Sawyer H, Kauffman MJ, Nielson RM. Influence of Well Pad Activity on Winter Habitat Selection Patterns of Mule Deer. The Journal of Wildlife Management [Internet]. 2009;73:1052–61. https://onlinelibrary.wiley.com/doi/abs/10.2193/2008-478
Sawyer H, Kauffman MJ, Middleton AD, Morrison TA, Nielson RM, Wyckoff TB. A framework for understanding semi-permeable barrier effects on migratory ungulates. Journal of Applied Ecology [Internet]. 2013;50:68–78. https://onlinelibrary.wiley.com/doi/abs/10.1111/1365-2664.12013
Gedir JV, Cain JW III, Harris G, Turnbull TT. Effects of climate change on long-term population growth of pronghorn in an arid environment. Ecosphere [Internet]. 2015;6:art189. https://onlinelibrary.wiley.com/doi/abs/10.1890/ES15-00266.1
Sawyer H, Korfanta NM, Nielson RM, Monteith KL, Strickland D. Mule deer and energy developmentLong-term trends of habituation and abundance. Global Change Biology [Internet]. 2017;23:4521–9. https://onlinelibrary.wiley.com/doi/abs/10.1111/gcb.13711
Signer J, Fieberg J, Avgar T. Animal movement tools (amt): R package for managing tracking data and conducting habitat selection analyses. Ecology and Evolution [Internet]. 2019;9:880–90. https://onlinelibrary.wiley.com/doi/abs/10.1002/ece3.4823
R Core Team. R: The r project for statistical computing [Internet]. 2021. https://www.r-project.org/
Börger L, Franconi N, De Michele G, Gantz A, Meschi F, Manica A et al. Effects of sampling regime on the mean and variance of home range size estimates. Journal of Animal Ecology [Internet]. 2006;75:1393–405. https://www.jstor.org/stable/4125081
U. S. Geological Survey. 3D elevation program 30-meter resolution digital elevation model. 2019; https://apps.nationalmap.gov/downloader/
Hijmans RJ, van Etten J, Sumner M, Cheng J, Baston D, Bevan A et al. Raster: Geographic data analysis and modeling [Internet]. 2023. https://cran.r-project.org/web/packages/raster/index.html
Allred BW, Bestelmeyer BT, Boyd CS, Brown C, Davies KW, Duniway MC et al. Improving Landsat predictions of rangeland fractional cover with multitask learning and uncertainty. Methods in Ecology and Evolution [Internet]. 2021;12:841–9. https://onlinelibrary.wiley.com/doi/abs/10.1111/2041-210X.13564
National Operational Hydrologic Remote Sensing Center. Snow data assimilation system (SNODAS) data products at NSIDC, version 1. 2004; https://nsidc.org/data/G02158/versions/1
Vermote E. MODIS/terra surface reflectance 8-day L3 global 250m SIN grid V061. 2021; https://lpdaac.usgs.gov/products/mod09q1v061/
Smith BJ, MacNulty DR, Stahler DR, Smith DW, Avgar T. Density-dependent habitat selection alters drivers of population distribution in northern Yellowstone elk. Ecology Letters [Internet]. 2023;26:245–56. https://onlinelibrary.wiley.com/doi/abs/10.1111/ele.14155
Center UGR. SITLA, Bureau of Land Management. UGRC - Utah Land Ownership. 2021; https://gis.utah.gov/products/sgid/cadastre/land-ownership/
Utah Department of Transportation. Barriers. 2022; https://data-uplan.opendata.arcgis.com/datasets/barriers/explore
Bureau of Land Management. Utah geospatial data. 2021; https://gbp-blm-egis.hub.arcgis.com/pages/utah
Forest Service USDA. US Geological Survey. Monitoring trends in burn severity (MTBS) thematic burn severity. 2022.
Poor EE, Jakes A, Loucks C, Suitor M. Modeling Fence Location and Density at a Regional Scale for Use in Wildlife Management. PLOS ONE [Internet]. 2014;9:e83912. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0083912
Hooten MB, Hanks EM, Johnson DS, Alldredge MW. Reconciling resource utilization and resource selection functions. Journal of Animal Ecology [Internet]. 2013;82:1146–54. https://onlinelibrary.wiley.com/doi/abs/10.1111/1365-2656.12080
Marzluff JM, Millspaugh JJ, Hurvitz P, Handcock MS. Relating Resources to a Probabilistic Measure of Space Use: Forest Fragments and S℡ler’s Jays. Ecology [Internet]. 2004;85:1411–27. https://onlinelibrary.wiley.com/doi/abs/10.1890/03-0114
Sikali E, Bailey P, Emad A, Buehler E. wCorr: Weighted correlations [Internet]. 2023. 10.32614/CRAN.package.wCorr
Bates D, Mächler M, Bolker B, Walker S. Fitting Linear Mixed-Effects Models Using lme4. Journal of Statistical Software [Internet]. 2015;67:1–48. https://doi.org/10.18637/jss.v067.i01
Kuznetsova A, Brockhoff PB, Christensen RHB. lmerTest Package: Tests in Linear Mixed Effects Models. Journal of Statistical Software [Internet]. 2017;82:1–26. https://doi.org/10.18637/jss.v082.i13
Bartoń K. MuMIn: Multi-model inference [Internet]. 2023. https://cran.r-project.org/web/packages/MuMIn/index.html
Hiller TL, Beringer J, Belant JL. Shape complexity of space used by american black bears influenced by sex and intensity of use. Basic and Applied Ecology [Internet]. 2017;18:67–74. https://www.sciencedirect.com/science/article/pii/S1439179116301062
Rea RV, Child KN, Spata DP, MacDonald D. Road and Rail Side Vegetation Management Implications of Habitat Use by Moose Relative to Brush Cutting Season. Environmental Management [Internet]. 2010;46:101–9. https://doi.org/10.1007/s00267-010-9502-6
O’Gara BW, Yoakum JD, McCabe RARA, Fichter E, Metz DP. Pronghorn : Ecology and management [Internet]. Wildlife Management Institute; University Press of Colorado; 2004. https://cir.nii.ac.jp/crid/1130282270429166720
Harrington JL, Conover MR. Characteristics of Ungulate Behavior and Mortality Associated with Wire Fences. Wildlife Society Bulletin [Internet]. 2006;34:1295–305. https://onlinelibrary.wiley.com/doi/abs/10.2193/0091-7648%282006%2934%5B1295%3ACOUBAM%5D2.0.CO%3B2
Richardson C. Pronghorn habitat requirements. 2006. p. 512.
Lingle S, Pellis S. Fight or flight? Antipredator behavior and the escalation of coyote encounters with deer. Oecologia [Internet]. 2002;131:154–64. http://link.springer.com/10.1007/s00442-001-0858-4
Telfer ES, Kelsall JP. Adaptation of Some Large North American Mammals for Survival In Snow. Ecology [Internet]. 1984;65:1828–34. https://onlinelibrary.wiley.com/doi/abs/10.2307/1937779
Johnson D. Cheyenne journal; when antelope don’t roam free. The New York Times [Internet]. 1988; https://www.nytimes.com/1988/11/18/us/cheyenne-journal-when-antelope-don-t-roam-free.html
Ford AT, Goheen JR. Trophic cascades by large carnivores: A case for strong inference and mechanism. Trends in Ecology & Evolution [Internet]. 2015;30:725–35. https://www.sciencedirect.com/science/article/pii/S0169534715002487
Paige C, Stevensville M. A landowner’s guide to wildlife friendly fences. Landowner/Wildlife Resource Program, Montana Fish, Wildlife and Parks, Helena, MT. 2008.

No competing interests reported.

02Appendices.docx

Download PDF

Reviewers invited by journal
14 Oct, 2024
Editor assigned by journal
09 Oct, 2024
Submission checks completed at journal
09 Oct, 2024
First submitted to journal
25 Sep, 2024

You are reading this latest preprint version

The impacts of anthropogenic linear features on the space-use patterns of two sympatric ungulates

Status:

Version 1

Abstract

Figures

1. Introduction

2. Materials and Methods

2.1 A note on home-range estimation

2.2 Study species and telemetry data

2.3 Home range estimation

2.4 Environmental data

2.5 Anthropogenic linear feature data

2.6 Response variables

2.7 Modeling

3. Results

3.1 Home range size

3.2 Home range shape

3.3 Anthropogenic linear feature selection patterns

3.4 Habitat selection patterns

4. Discussion

4.1 Effects of roads

4.2 Effects of fences

4.3 Concluding remarks

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1