See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/351579834

Human activity influences wildlife populations and activity patterns:

implications for spatial and temporal refuges

Article in Ecosphere · May 2021

DOI: 10.1002/ecs2.3487

CITATIONS

25

7 authors, including:

Jesse Lewis Ashley Gramza

Arizona State Univeristy (Mesa AZ) Arkansas Game & Fish Commission
86 PUBLICATIONS 2,881 CITATIONS 19 PUBLICATIONS 229 CITATIONS

SEE PROFILE SEE PROFILE

Sue Vandewoude
Colorado State University
426 PUBLICATIONS 4,232 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Wildlife connectivity View project

Machine learning photos View project

All content following this page was uploaded by Jesse Lewis on 17 May 2021.

The user has requested enhancement of the downloaded file.

 Human activity influences wildlife populations and activity
patterns: implications for spatial and temporal refuges
JESSE S. LEWIS ,1, SUSAN SPAULDING,2 HEATHER SWANSON,3 WILLIAM KEELEY,3 ASHLEY R. GRAMZA ,4
SUE VANDEWOUDE,5 AND KEVIN R. CROOKS4
1
College of Integrative Sciences and Arts, Arizona State University, Mesa, Arizona 85212 USA
2
Boulder County Parks and Open Space, Longmont, Colorado 80503 USA
3
City of Boulder Open Space and Mountain Parks, Boulder, Colorado 80303 USA
4
Department of Fish, Wildlife, and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA
5
Department of Microbiology, Immunology, and Pathology, Colorado State University, Fort Collins, Colorado 80523 USA

Citation: Lewis, J. S., S. Spaulding, H. Swanson, W. Keeley, A. R. Gramza, S. VandeWoude, and K. R. Crooks. 2021.
Human activity influences wildlife populations and activity patterns: implications for spatial and temporal refuges.
Ecosphere 12(5):e03487. 10.1002/ecs2.3487

Abstract. Human activity affects plant and animal populations across local to global scales, and the
management of recreation areas often aims to reduce such impacts. Specifically, by understanding patterns
of human activity and its influence on animal populations, parks and recreation areas can be managed to
provide spatial and temporal refuge to wildlife most sensitive to this type of human disturbance. However,
additional research is necessary to understand how human activity influences wildlife populations, habitat
use, and activity patterns for a diversity of wildlife species. We studied the potential impacts of human
activity (as measured by nonmotorized recreationists) on populations and activity patterns of 12 mammal
species, including herbivores and carnivores, from 63 motion-activated cameras that sampled game trails
and human trails with varying degrees of human activity along the Front Range of Colorado. Human
activity was greatest during the day and minimal or absent during the night. All wildlife species in our
study used human trails, although the extent to which human recreation altered the occupancy, relative
habitat use, and activity patterns of wildlife varied across species, where some animals appeared to be
more influenced by human activity than others. Some species (e.g., fox squirrel, red fox, and striped skunk)
did not demonstrate a response to human activity. Other species (e.g., black bear, coyote, and mule deer)
altered their activity patterns on recreation trails to be more active at night. Across all wildlife, the degree
to which animals altered activity patterns on human trails was related to their natural activity patterns and
how active they were during the day when human activity was greatest; species that exhibited greater
overlap in natural activity patterns with humans demonstrated the greatest shifts in their activity, often
exhibiting increased nocturnal activity. Further, some species (e.g., Abert’s squirrel, bobcat, and mountain
lion) exhibited reduced occupancy and/or habitat use in response to human recreation. Managing spatial
and temporal refuges for wildlife would likely reduce the impacts of human recreation on animals that use
habitat in proximity to trail networks.

Key words: carnivores; daily activity patterns; herbivores; hiking; natural areas; nighttime recreation; occupancy;
recreation; refuges; trails; wildlife cameras.

Received 13 December 2020; accepted 4 January 2021; final version received 20 February 2021. Corresponding Editor:
Elizabeth A. Flaherty.
Copyright: © 2021 The Authors. This is an open access article under the terms of the Creative Commons Attribution
License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
E-mail: [email protected]

v www.esajournals.org 1 May 2021 v Volume 12(5) v Article e03487

 LEWIS ET AL.

INTRODUCTION animal, with some species being more sensitive

to anthropogenic factors than others (Fig. 1;
Diverse human activities, including land-use Knight and Gutzwiller 1995, Papouchis et al.
change, vehicle traffic on roads, and resource 2001, Hammitt et al. 2015). In particular, distur-
extraction, have led to the displacement, bance characteristics (e.g., type, frequency, tim-
decreased fitness, and extirpation of plants and ing, and location of human activity) and
animals globally and continue to threaten popu- wildlife characteristics (e.g., species life history,
lations (Wilcove et al. 1998, Czech et al. 2000, age, and sex of species) are important consider-
Ceballos and Ehrlich 2002, Crooks et al. 2017). ations when understanding the effects of
Although considered less impactful, human human activities on wildlife. In addition, it is
recreation also can negatively affect the natural increasingly recognized that wildlife requires
environment (Knight and Gutzwiller 1995, Ham- refuges from human activity to avoid distur-
mitt and Cole 1998, Larson et al. 2019) and has bance from recreation. Spatial refuges have
been reported as the fourth leading cause of spe- been recommended to conserve some wildlife
cies endangerment in the United States (Czech species that share landscapes with human recre-
et al. 2000) and a threat to IUCN red-listed spe- ationists that are sensitive to human use (Reed
cies globally (Ballantyne and Pickering 2013). and Merenlender 2008, Switalski and Jones
Human recreational activities can directly impact 2012, Larson et al. 2019), but few studies and
wildlife through increased mortality, spatial and recommendations address the importance of
temporal avoidance of trails, altered behavior, temporal refuges for wildlife to avoid distur-
increased prevalence of nonnative species, and bance from human activities. For example, it is
reduced fitness (Garber and Burger 1995, Knight likely important to understand the time of day
and Gutzwiller 1995, Taylor and Knight 2003, that recreation occurs and how human activity
Reed and Merenlender 2008, Naylor et al. 2009, patterns overlap with daily activity patterns of
Steven et al. 2011, Larson et al. 2016). Given the different wildlife species to understand poten-
growing human population (Cohen 2003, Theo- tial effects on animal populations. Ultimately,
bald 2005, Seto et al. 2012) and increased the effective management of both spatial and
demand for recreational opportunities (Balmford temporal refuges will depend on the species
et al. 2009), this type of human activity in natural and levels of human activity present in an area.
systems is expected to increase and affect wildlife We studied the potential impacts of nonmotor-
on a broad scale. By understanding the patterns ized human recreation (i.e., hikers, bikers, and
of human recreation and associated impacts on equestrians) on 12 mammal species, including
animal populations, the negative effects of recre- herbivores and carnivores, using an extensive
ation can be mitigated through management data set collected along the Front Range of Color-
actions. ado across a gradient of recreational levels. We
Land managers of parks, recreation areas, also evaluated daily activity patterns for human
and public lands aim to balance conservation recreationists. This is an excellent study area to
with providing human recreational opportuni- evaluate the effects of human recreation on the
ties across human trails (Knight and Cole 1991, wildlife community due to the extensive network
Knight and Gutzwiller 1995, Hammitt et al. of open space and recreational trails, human visi-
2015). At the scale of a recreation area, the tation to portions of the study area rivaling that
effects of human recreation on wildlife can of the most popular National Parks with nearly
result in animals (1) avoiding or increasing use 5 million people visits annually (Vaske et al.
of an area, (2) reducing or increasing the fre- 2009), and some areas managed as wildlife refu-
quency of use of an area, or (3) changing daily gia where recreational activities are restricted.
activity pattern to avoid humans (Knight and Specifically, our objective was to evaluate
Gutzwiller 1995, George and Crooks 2006, Nay- whether recreation influenced if animals occu-
lor et al. 2009, Steven et al. 2011, Spaul and pied a site, reduced relative habitat use of a site,
Heath 2016). Each wildlife species responds to or altered their daily activity patterns. We pre-
human disturbance differently depending upon dicted that animal use would be negatively
the characteristics of the human activity and related to human recreation activity, but to

v www.esajournals.org 2 May 2021 v Volume 12(5) v Article e03487

 LEWIS ET AL.

Fig. 1. Conceptual figure outlining the three primary factors that can influence wildlife response to human
recreation activities. “Disturbance characteristics” and “wildlife characteristics” are summarized from Knight
and Gutzwiller (1995) and other sources, and we added “refuge characteristics in relation to disturbance and
wildlife.” For “spatial refuge,” important factors include the size and configuration of the spatial refuge, proxim-
ity to human trails and activity, the habitat quality of the area for the species, and other nearby human or land-
scape factors (e.g., additional sources of disturbance, potential attractants, landscape connectivity). For
“temporal refuge,” important factors include temporal overlap of human and wildlife activity patterns, the dura-
tion of the temporal refuge and time of day of recreation, seasonal considerations (e.g., human patterns or life his-
tory considerations of the species), and temporal overlap of activity patterns between wildlife species, which can
influence predatory and competitive interactions.

varying degrees depending on a species relative METHODS

sensitivity to human disturbance. On human
trails experiencing more recreation, we further Study area
predicted that wildlife activity patterns would We conducted our project along the Front
shift to nocturnal periods when recreational Range of Colorado, USA, which exhibited an
activity is low. In particular, we predicted that extensive network of open space properties with
animals that were naturally active during diurnal recreational trails experiencing varying degrees
and crepuscular time periods would shift their of nonmotorized human recreation, including
activity patterns the most if influenced by human hikers, bikers, and equestrians. Because hikers
recreation. Because the suite of wildlife that we comprised 98.2% of all recreation events in our
studied exhibited varying sensitivities to human study, and bikers and equestrians occurred at
activities and life history characteristics, we low levels at a limited number of locations, we
expected that our results could be used to catego- evaluated all forms of recreation collectively.
rize animals into groups based on their relative Overall, the study area experienced high recre-
sensitivity and response to human recreation. ation levels (5 million people visits annually;
Such information is critical to land managers Vaske et al. 2009) and was associated with the
designing and implementing management plans gateway community of Boulder (popula-
to conserve wildlife across landscapes experienc- tion = 97,385, U.S. Census Bureau 2010), which
ing human activities and recreation. facilitated access to public lands managed by the

v www.esajournals.org 3 May 2021 v Volume 12(5) v Article e03487

 LEWIS ET AL.

city of Boulder Open Space and Mountain Parks placed on game trails (n = 19). The second group
(OSMP), Boulder County Parks and Open Space of 20 cameras (the southern grid) sampled a
property (BCPOS), and the US Forest Service. greater number of human trails (n = 8) on OSMP
Across public lands, some areas were managed and BCPOS properties. These first two groups of
to promote recreation through an extensive net- cameras were systematically spaced approxi-
work of trails, whereas other areas were man- mately 1–2 km apart and operated from Septem-
aged to reduce disturbance to wildlife and ber 2010 to October 2011. The third group of 23
human trails were absent. Common vegetation cameras was associated with the southern grid,
included ponderosa pine (Pinus ponderosa), Dou- and cameras placed on human trails (n = 16)
glas fir (Pseudotsuga menziesii), juniper (Junipus sampled areas with relatively high recreational
osteosperma), aspen (Populus tremuloides), and activity in the study area. These cameras were
mountain mahogany (Cercocarpus montanus). See spaced approximately 250–1000 m apart and
Lewis et al. (2015b) for a more detailed descrip- occurred on OSMP property between September
tion of the study area. 2011 and October 2012 (Windell et al. 2019).

Wildlife camera data Occupancy and relative habitat use

To sample animals along game trails and We used two modeling approaches to evaluate
human recreation trails, we used Cuddeback (1) occupancy and (2) relative habitat use, which
(Non Typical, Green Bay, Wisconsin, USA) Cap- considered the camera location as the sampling
ture remote wildlife (RW) cameras with a white unit. First, we estimated (a) occupancy probabil-
flash and Cuddeback Attack Infrared cameras, ity (Ψ; the proportion of sites occupied or used
which were programmed to record a single pho- by the species) to evaluate whether recreational
tograph with a 30-s quiet period between pho- activity influenced the presence or absence of a
tographs. At each sampling location, we placed species at a site and (b) detection probability (p;
one RW camera at a site that we believed maxi- the probability of detecting the species given that
mized the opportunity to photograph wildlife, it used a site; MacKenzie et al. 2018).
which included both game and recreation trails. Second, we used the Royle-Nichols (RN) mod-
RW cameras were placed on trees approximately els to evaluate heterogeneity in abundance (k)
0.5 m above the ground and 4–5 m from the trail across sites considering heterogeneity in detec-
in a perpendicular orientation. Our sampling tion probability (Royle and Nichols 2003).
was passive in that we did not use attractants Research has demonstrated a positive relation-
(i.e., sight, sound, scent) to lure animals to the ship between RN models and population abun-
camera location. Our work focused on 12 mam- dance or density (Linden et al. 2017). However,
mal species, including small- to large-sized herbi- RN models might not be an unbiased surrogate
vores and carnivores. Eleven species were native for density in some systems due to the sensitivity
wildlife, including Abert’s squirrel (Sciurus of animal home range size relative to sampling
aberti), black bear (Ursus americanus), bobcat sites (Rogan et al. 2019, Nakashima 2020). There-
(Lynx rufus), cottontail rabbit (Sylvilagus spp.), fore, it is recommended that RN models be used
coyote (Canis latrans), gray fox (Urocyon cinereoar- as a relative measure of intensity of habitat use
genteus), mountain lion (Puma concolor), mule for a species (Nakashima 2020). We thus took the
deer (Odocoileus hemionus), raccoon (Procyon conservative approach and interpreted RN mod-
lotor), red fox (Vulpes vulpes), and striped skunk els as a measure of relative habitat use for this
(Mephitis mephitis). One species was nonnative, study, given that many of the species we evalu-
the fox squirrel (Sciurus niger). ated exhibit relatively large home ranges. For RN
We used data from 63 wildlife cameras that models, we assumed the spatial distribution of
sampled a range of human recreation levels from animals across sites followed a Poisson distribu-
no human recreation to relatively high levels tion and that the probability of detecting a spe-
(Fig. 2). Overall, wildlife cameras sampled 38 cies at a site was related to the species-specific
game trails and 25 human recreation trails. The detection probability (r) and the number of visits,
first group of 20 cameras (the northern grid) sam- or animals, of the species at that site (Donovan
pled primarily wildland with most cameras and Hines 2007).

v www.esajournals.org 4 May 2021 v Volume 12(5) v Article e03487

 LEWIS ET AL.

Fig. 2. Study area where 63 wildlife cameras operated along the Front Range, Colorado, USA. The northern
grid was near the town of Lyons, Colorado, and sampled game and human trails characterized by lower levels of
human recreation. The southern grid and cameras west of Boulder, Colorado, sampled game trails and human
trails, which were typically characterized by higher levels of human recreation. White cells are 2 9 2 km in size.
Within each cell, yellow dots represent the placement of a wildlife camera.

v www.esajournals.org 5 May 2021 v Volume 12(5) v Article e03487

 LEWIS ET AL.

For our two modeling approaches (single-spe- game trails using the program R (R Development
cies single-season occupancy models and RN Core Team 2020) package Overlap (Meredith and
models), we evaluated data from spring to fall Ridout 2018), which fits kernel density functions
(May–October) with ten 18-d sampling occa- (using a von Mises kernel for the circular distri-
sions. Both modeling approaches used the same bution of time data) to animal observations and
input data of whether a species was detected (1) estimates a coefficient of overlap between activ-
or not detected (0) at a site during an occasion. ity patterns that range from 0 (no overlap) to 1
We used the program R (R Development Core (complete overlap). We followed their recom-
Team 2020) package RMark (Laake and Rexstad mendations for bandwidth selection (i.e.,
2018) to evaluate models in Program MARK smoothness of the activity curves), estimators for
(White and Burnham 1999). In occupancy and quantifying overlap between activity curves (i.e.,
RN models, we evaluated the covariate human estimator 1 was used if the smaller of the two
recreation (i.e., the total number of human recre- samples was n < 50; otherwise, estimator 4 was
ation events at a camera site that occurred >30 s used with larger sample sizes), and 10,000 boot-
apart). To control for variation in plant produc- strap simulations to estimate 95% confidence
tivity across the landscape, we evaluated the nor- intervals (Ridout and Linkie 2009, Meredith and
malized difference in vegetation index (NDVI) Ridout 2018). Activity patterns were considered
within a 1000 m radius buffer around each site. different from each other (i.e., wildlife shifted
If NDVI affected wildlife occupancy or habitat their activity patterns) if the upper bound of the
use, we expected animals to select for areas with 95% confidence interval for the coefficient of
higher plant productivity. We also considered overlap was <0.90. For activity analyses, samples
additional buffer sizes of NDVI, but they were for a wildlife species at a site were considered
highly correlated (e.g., Pearson’s correlation independent if photographs were taken >1 h
between 1000-m buffer and 500-m buffer = 0.97 apart. Due to unreliability of activity pattern
and Pearson’s correlation between 1000-m buffer results at small sample sizes, species were
and 1500-m buffer = 0.98), and we thus used the included in analyses if there were >20 indepen-
1000-m buffer for consistency across species. To dent observations within each trail category.
aid in model convergence, variables were stan- If humans influence daily activity for a spe-
dardized by subtracting the mean and dividing cies, we predicted that animals would be more
by the standard deviation (Schielzeth 2010). For likely to shift their temporal activity patterns if
occupancy and RN models, we evaluated all pos- they were more active during diurnal periods
sible model combinations and considered time- and thus had higher temporal overlap with
varying detection probabilities in occupancy recreational activity. To investigate whether
models, which resulted in eight model combina- activity pattern overlap between wildlife and
tions for occupancy models and four model com- humans was predictive of wildlife shifting their
binations for RN models. We compared models activity patterns in response to human recre-
using model selection methods and considered a ation, we evaluated the relationship between
variable as informing Ψ or k if they occurred in a “wildlife activity shift” and “wildlife vs.
model that outperformed the intercept-only human activity overlap”. To calculate wildlife
model (i.e., model without covariates), based on activity shift, we first estimated the overlap in
lower Akaike information criteria (AIC) values activity patterns for species on game trails (i.e.,
(Burnham and Anderson 2002). their expected natural activity pattern) vs. on
human trails (i.e., where wildlife would be
Daily activity patterns expected to alter their activity pattern in
To evaluate whether recreation influenced response to human recreation). Next, the esti-
daily activity patterns of animals, we compared mated overlap statistic was subtracted from 1
activity patterns of animals between human and to calculate wildlife activity shift. To calculate
game trails. If recreation altered wildlife activity wildlife vs. human activity overlap, we evalu-
patterns, we expected animal activity to differ on ated the overlap in activity patterns between
human vs. game trails. We evaluated the daily wildlife on game trails (i.e., their expected nat-
activity patterns of wildlife between human and ural activity pattern) and humans on human

v www.esajournals.org 6 May 2021 v Volume 12(5) v Article e03487

 LEWIS ET AL.

trails. Finally, we evaluated the correlation Table 1. Summary of responses to human recreation in
between wildlife activity shift and wildlife vs. relation to occupancy probability, relative habitat
human activity overlap, where we predicted a use (from the Royle-Nichols models), and daily
positive relationship (based on a positive slope activity patterns for 12 wildlife species on the Front
and high R2 value) for the 11 native wildlife Range, Colorado, USA.
species in our study. If human recreation
Relative habitat
altered wildlife activity patterns, we expected Species Occupancy use Activity
that as wildlife vs. human activity overlap
Fox squirrel
increased (i.e., more diurnal natural activity of
Red fox
wildlife), the amount of wildlife activity shift Striped skunk
would increase. Conversely, if human recre- Black bear
ation did not alter wildlife activity patterns as Gray fox
predicted, then the slope of the above relation- Mule deer
ship would equal 0. Bobcat
Coyote + +
Raccoon + +
Categorizing species in relation to human Cottontail
recreation rabbit
To categorize species based on their results of Mountain lion
occupancy, relative habitat use, and activity pat- Abert’s squirrel –
terns, we grouped species based on exhibiting Notes: For occupancy and relative habitat use, the general
negative, positive, or no relationship with human direction of the beta estimate relationship was included (ei-
ther positive + or negative ) when the model including the
recreation. This summary was used to broadly covariate human recreation was more supported than the
group species based on their relative sensitivities intercept-only model. For activity, a negative ( ) shift was
to human recreation. included for a species if the upper limit of the 95% confidence
interval was less than a value of 0.90 (see Methods).

RESULTS
Occupancy models
Across our 63 camera sites, we obtained Occupancy and detection probability of wildlife
692,968 photographs, with human recreation ranged from 0.279 to 0.968 and from 0.209 to
comprising 676,905 photographs. On human 0.616, respectively (Table 2). The occupancy of
trails, the average number of annual human vis- some wildlife (i.e., black bear, bobcat, fox squirrel,
its = 26,980 (SE = 6,879, range = 615–123,861 gray fox, mule deer, red fox, and striped skunk)
across sites) and the range for the number of did not appear to be influenced by human recre-
daily human visits = 2–339 across sites. Human ation volumes (Table 3; Appendix S1). For other
use occurred on some game trails at low levels, species (i.e., Abert’s squirrel, cottontail rabbit, and
with the range for the number of daily human mountain lion), occupancy probability decreased
visits = 0–0.88. Across all sites, we obtained as the amount of human recreation increased. In
16,063 wildlife photographs, with adequate sam- contrast, coyote and raccoon were more likely to
ple sizes for 12 species. occupy a site as the volume of human recreation
Based on the results of occupancy, relative increased (Table 3; Appendix S1). The variable
habitat use, and activity patterns, the response of NDVI was not related to occupancy for most spe-
species to recreation was categorized as no cies; however, there was a positive relationship
response (i.e., fox squirrel, red fox, and striped for red fox (b = 0.59, SE = 0.34) with occupancy.
skunk), shifts in activity patterns only (i.e., black
bear, gray fox, and mule deer), or reduced occu- RN models (relative habitat use)
pancy and/or relative habitat use (i.e., bobcat, Relative habitat use as evaluated by the RN
cottontail rabbit, mountain lion, and Abert’s models often exhibited similar results to the
squirrel; Table 1). In addition, other species occupancy models explained above (Table 3;
exhibited a positive relationship with recreation Appendix S2). However, habitat use for bobcats
(i.e., coyote and raccoon; Table 1). decreased as the amount of human recreation

v www.esajournals.org 7 May 2021 v Volume 12(5) v Article e03487

 LEWIS ET AL.

Table 2. Estimates and standard errors (SE) of occupancy probability (Ψ) and detection probability (p) for 12
wildlife species across 63 sites along the Front Range, Colorado, USA.

Ψ p
Species Estimate SE Estimate SE

Abert’s squirrel 0.281 0.062 0.209 0.036

Black bear 0.780 0.053 0.448 0.023
Bobcat 0.674 0.060 0.363 0.024
Cottontail rabbit 0.524 0.063 0.521 0.028
Coyote 0.578 0.063 0.363 0.026
Fox squirrel 0.279 0.058 0.290 0.037
Gray fox 0.383 0.061 0.423 0.032
Mountain lion 0.894 0.051 0.247 0.020
Mule deer 0.968 0.022 0.616 0.020
Raccoon 0.451 0.066 0.261 0.029
Red fox 0.825 0.048 0.586 0.022
Striped skunk 0.667 0.062 0.312 0.024

Table 3. The relationship of the variable human recreation with estimates of occupancy (Ψ) from occupancy
models and relative habitat use (k) from the Royle-Nichols models for 12 wildlife species on the Front Range,
Colorado, USA.

Ψ k
95% CI 95% CI
Species b SE Lower Upper b SE Lower Upper

Abert’s squirrel 1.12 0.88 2.86 0.61 1.12 0.84 2.76 0.53
Black bear 0.10 0.34 0.57 0.77 0.10 0.13 0.35 0.15
Bobcat 0.30 0.27 0.82 0.23 0.28 0.19 0.64 0.09
Cottontail rabbit 0.65 0.41 1.45 0.15 0.60 0.32 1.23 0.03
Coyote 1.60 0.88 0.12 3.32 0.40 0.09 0.23 0.57
Fox squirrel 0.08 0.28 0.47 0.62 0.07 0.20 0.32 0.45
Gray fox 0.32 0.28 0.23 0.86 0.06 0.16 0.26 0.37
Mountain lion 0.64 0.33 1.29 0.00 0.18 0.14 0.45 0.09
Mule deer na na na na 0.04 0.09 0.13 0.21
Raccoon 0.86 0.61 0.33 2.04 0.25 0.11 0.03 0.47
Red fox 0.43 0.58 0.70 1.56 0.07 0.08 0.10 0.24
Striped skunk 0.01 0.28 0.55 0.56 0.04 0.14 0.31 0.22
Notes: Beta estimates (b), standard errors (SE), and lower and upper bounds for 95% confidence intervals (95% CI) are
reported for each method. Text is in bold for those species where the variable human recreation occurred in the top model,
which outperformed the intercept-only model and thus indicated a positive or negative relationship. Note that Ψ for mule deer
is not reported (i.e., na) because the real estimates of occupancy were very close to 1.0, and thus, there was no relationship with
the variable human recreation.

increased (Table 3; Appendix S2). There was a during the entire day, nighttime recreation
positive relationship for mule deer (b = 3.59, (30 min post-sundown to 30 min pre-sunrise)
SE = 0.86) with NDVI, where mule deer were was 1.11% of total recreation and late night
more likely to use a site as NDVI increased. (10:00 p.m. to 4:00 a.m.) recreation was only
0.13% of total recreation.
Activity patterns Human recreation appeared to have varying
Human recreation occurred primarily during effects on the activity patterns of animals
diurnal time periods and was relatively minimal (Table 4; Figs. 4, 5). Consistent with predictions,
or absent during nocturnal time periods (Fig. 3). wildlife that were active during crepuscular and
Demonstrating that human recreation at night diurnal time periods on game trails (i.e., bobcat,
occurred at relatively low levels compared with coyote, black bear, mule deer) tended to exhibit

v www.esajournals.org 8 May 2021 v Volume 12(5) v Article e03487

 LEWIS ET AL.

Table 4. Results for overlap in wildlife activity pat-

terns between game and human trails, including
total sample size (total n), smallest sample size
between game and human trails (min n), estimate of
activity pattern overlap (overlap estimate), and 95%
confidence intervals with lower and upper bounds
(95% CI), for 12 wildlife species on the Front Range,
Colorado, USA.

95% CI
Total Min Overlap
Species n n estimate Lower Upper

Abert’s 155 24 0.57 0.38 0.73

squirrel
Black bear 926 349 0.68 0.61 0.73
Bobcat 650 277 0.81 0.76 0.87
Cottontail 1830 672 0.90 0.87 0.94
rabbit
Fig. 3. Daily activity pattern of human recreation Coyote 1347 177 0.76 0.70 0.83
along the Front Range, Colorado. Time of day is pre- Fox squirrel 517 64 0.90 0.92 0.98
Gray fox 577 209 0.82 0.74 0.88
sented on the x-axis, and activity as measured by ker-
Mule deer 6217 1452 0.68 0.64 0.70
nel density is presented on the y-axis.
Mountain 627 261 0.88 0.83 0.94
lion
Raccoon 223 34 0.81 0.68 0.92
the least overlap in activity patterns between Red fox 3138 722 0.94 0.93 0.97
human recreation and game trails, and thus Striped 574 177 0.89 0.84 0.95
skunk
shifted their activity patterns to be less active dur-
ing day and more active at night on human recre-
ation trails. Animals that were mostly active
during nocturnal time periods on game trails (i.e., (Fig. 5), indicating that although it exhibited
cottontail rabbit, mountain lion, raccoon, red fox, high overlap between their natural activity pat-
striped skunk) tended to exhibit similar activity tern and human activity patterns, it did not alter
patterns between game and human recreation its activity pattern in response to recreation.
trails. Lastly, species that were primarily diurnal
on game trails either shifted their activity patterns DISCUSSION
to be less active mid-day and more active during
the morning and afternoon (i.e., Abert’s squirrel) All wildlife in our study used human and
or did not shift their activity patterns (i.e., fox game trails, although each species responded
squirrel) when using human trails. differently to the effects of human recreation
As predicted for 11 native wildlife species, as considering the three factors in our study (occu-
overlap in activity patterns for animals on game pancy, relative habitat use, and activity pat-
trails (considered their natural activity pattern) terns). First, some wildlife species (i.e., fox
and humans on human trails (i.e., wildlife vs. squirrel, red fox, and striped skunk) did not
human activity overlap) increased, wildlife avoid using sites, did not exhibit reduced habi-
shifted their activity patterns to avoid human tat use, and did not alter activity patterns in
recreation (i.e., wildlife activity shift), where this response to human recreation. These species
correlation exhibited a strong positive relation- might be classified as relatively tolerant of
ship (b = 0.465, SE = 0.088, r2 = 0.73; Fig. 5). In human recreation observed in our study. Sec-
other words, species whose natural activity pat- ond, some medium- and large-sized mammals
terns occurred during crepuscular and diurnal (i.e., black bear, coyote, and mule deer) reduced
periods more strongly shifted their activity in the amount of time that they were active during
response to human recreation. However, the diurnal time periods to be more active during
nonnative fox squirrel did not follow this pattern nocturnal periods, presumably to use recreation

v www.esajournals.org 9 May 2021 v Volume 12(5) v Article e03487

 LEWIS ET AL.

Fig. 4. Daily activity patterns of 12 wildlife species on game trails (solid line) and human trails (dashed line)
along the Front Range, Colorado. Wildlife species include Abert’s squirrel (a), black bear (b), bobcat (c), cottontail
rabbit (d), coyote (e), fox squirrel (f), gray fox (g), mule deer (h), mountain lion (i), raccoon (j), red fox (k), and
striped skunk (l). Time of day is presented on the x-axis, and activity as measured by kernel density is presented
on the y-axis. Shaded regions are areas of overlap between activity patterns on game and human trails, and used
to estimate the coefficient of overlap between activity patterns.

v www.esajournals.org 10 May 2021 v Volume 12(5) v Article e03487

 LEWIS ET AL.

Consistent with our results, several studies

have reported that wildlife will shift their daily
activity patterns in response to human distur-
bance and recreation, where animals avoid times
of day when humans are most active (George
and Crooks 2006, Barrueto et al. 2014, Reilly et al.
2017, Gaynor et al. 2018). To further understand
these patterns, it is useful to evaluate the rela-
tionship between shifts in wildlife activity and
the degree of activity overlap with humans. In
our study, wildlife that demonstrated the great-
est shift in their activity patterns in response to
recreation (i.e., Abert’s squirrel, mule deer, black
bear, coyote, and bobcat) were species with natu-
ral activity patterns that occurred at least par-
Fig. 5. Relationship between “wildlife activity shift” tially during diurnal or crepuscular periods, and
(i.e., 1 minus the overlap in activity patterns for wildlife thus overlapped with human activity on recre-
species on a game trail vs. human recreation trail) and ation trails (Fig. 5). Consistent with our predic-
“wildlife vs. human activity overlap” (i.e., overlap in tions, as overlap in activity between wildlife and
activity pattern of wildlife on game trail vs. humans on humans increased, wildlife shifted their activity
human recreation trail) for 11 native wildlife species (ab- patterns, often to be more active at night. Thus,
breviations are AS, Abert’s squirrel; BB, black bear; BC, species with higher natural diurnality more
bobcat; CR, cottontail rabbit; CO, coyote; GF, gray fox; strongly shifted their temporal activity in
ML, mountain lion; MD, mule deer; RA, raccoon; RF, response to human recreation. Conversely, spe-
red fox; and SS, striped skunk) on the Front Range, Col- cies that exhibited low activity overlap with
orado. Note that the nonnative fox squirrel ( FS) was humans (i.e., wildlife active mostly at night) were
included for comparison, but this species was excluded less likely to shift activity patterns in response to
from the regression analysis. As overlap in activity pat- human recreation. Because nearly all recreation
terns between wildlife and humans increased, wildlife occurred during the day, there was little opportu-
increasingly shifted their activity patterns (i.e., exhibited nity for recreation to alter the activity patterns of
less overlap in activity patterns between wildlife and these nocturnal species. An exception to our pre-
human recreation trails). Thus, species with higher natu- diction was for the nonnative fox squirrel, which
ral diurnality more strongly shifted their temporal activ- exhibited high overlap in activity with humans
ity in response to human recreation. The regression during diurnal periods, but did not appear to
equation is y = 0.0155 + 0.465 9 b1, where the coeffi- avoid human recreation spatially or temporally,
cient for b1 (i.e., 0.465) has a SE = 0.088. demonstrating that some animals that are com-
mensal with humans might not substantially
areas when human disturbance was minimized alter their behavior in relation to human recre-
(George and Crooks 2006, Barrueto et al. 2014). ation. Overall, these results are useful for evalu-
However, these species did not avoid human ating and predicting the degree to which wildlife
trails or reduce their use of sites as human recre- activity will be influenced by human recreation.
ation increased. These species might be classi- Some studies have suggested that body size and
fied as moderately sensitive to human demographics of animals could potentially
recreation. Lastly, some species used sites less explain which species will alter their activity pat-
often or avoided them all together as human terns in response to humans (Gaynor et al. 2018).
disturbance increased on a trail, and in some Although our study did not allow us to evaluate
cases also shifted their activity patterns to avoid these relationships, our results indicate that the
periods of high recreation on human trails (e.g., measure of overlap in activity patterns between
Abert’s squirrel, bobcat, and mountain lion). wildlife and humans is a strong predictor for
These species appeared to be most sensitive to how species will shift their activity patterns in
disturbance associated with human recreation. response to human disturbance (Fig. 5). Potential

v www.esajournals.org 11 May 2021 v Volume 12(5) v Article e03487

 LEWIS ET AL.

ecological implications of species shifting their with human recreation, providing spatial refuge
activity patterns include higher temporal overlap is also important to wildlife persisting in proxim-
between predator and prey or increased inter- ity to recreational trails (Knight and Gutzwiller
specific competition, which could influence 1995, Larson et al. 2019). Wildlife may use spatial
trophic interactions and ecological communities refuges during diurnal time periods and access
(Lewis et al. 2015a, Gaynor et al. 2018). habitat associated with recreational trails at
Because some wildlife species demonstrate night. However, for wildlife that were sensitive
flexibility in their activity patterns and can find to human recreation, the distance from human
temporal refuge from human disturbance at trails that animals altered their behavior in
night, it might be assumed that wildlife and response to human disturbance was unknown.
humans can coexist in areas experiencing recre- The “area of influence” that animals are affected
ation through temporal avoidance. However, by a disturbance depends upon the species and
although nocturnal species altered their daily warrants additional study for most wildlife (Tay-
activity patterns the least, some animals exhibited lor and Knight 2003). Indeed, the indirect effect
reduced occupancy or habitat use (e.g., mountain of behavioral avoidance of a larger area can be
lion), indicating that human disturbance can much greater than the direct effect of habitat loss
potentially displace wildlife from areas near and human disturbance from human develop-
human recreation trails. In addition, it is impor- ment (Sawyer et al. 2006, Dwinnell et al. 2019).
tant to consider that other recreation patterns Conversely, although some wildlife might avoid
could potentially have a greater impact on wild- or reduce use of trails, they might use habitat in
life that are active at night. Nighttime recreation relatively close proximity to these areas. To more
is increasing in popularity and being promoted effectively manage parks and public lands gov-
as a potential opportunity for recreationists to erned under a dual mandate of protecting habitat
exploit and uniquely experience nature (Metcalfe and providing recreational opportunities, it is
1974, Beeco et al. 2011, Smith and Hallo 2013). It imperative to better understand the area of influ-
is unclear, however, how wildlife would respond ence caused by recreational trails and distur-
to increased human activity along trails during bance-free habitat requirements for animals
nocturnal time periods. If human recreation throughout the year.
increased during nocturnal time periods, our In our study, some species exhibited increased
results suggest that animals may shift their activ- occupancy or relative habitat use on recreation
ity patterns during nocturnal time periods to find trails. This was likely due to some wildlife, espe-
temporal refuge from human activities. This cially carnivores, preferring to use well-defined
could lead animals to be active for an overall human recreation trails as travel corridors (Lewis
shorter amount of time each day and even later at et al. 2015a, Kays et al. 2017), especially at night
night, or to avoid the area altogether if there is when human activity was low. Wild canids (e.g.,
insufficient disturbance-free time. Although we coyotes) potentially increased the use of human
observed a relatively low percent of human recre- recreation trails due to the presence of domestic
ation occurring at night, considering that nearly 5 dogs accompanying hikers (Kellner et al. 2017),
million recreation visits occur on these trails because coyotes are territorial and might increase
annually (Vaske et al. 2009), this can result in a marking behavior in the presence of potential
relatively large amount of human activity. Recre- competitors and their scent (Lenth et al. 2008). In
ationists are increasingly enthusiastic about addition, it is possible that some wildlife species
nighttime recreational activities and although the were accessing these areas to take advantage of
expansion of nighttime hiking is being promoted natural and anthropogenic food resources
by some segments of the public, the potential (Knight and Gutzwiller 1995).
negative impacts on wildlife (e.g., behavioral or The effects of human disturbance and recre-
fitness effects on animal populations) need to be ation on wildlife activity patterns and popula-
evaluated by land managers (Staine and Burger tions vary across studies and regions, which can
1994, Burger 1995, Beeco et al. 2011). be related to a suite of landscape and wildlife
Although some animals can potentially find characteristics (Gaynor et al. 2018, Larson et al.
temporal refuge from disturbance associated 2019). Although our study demonstrated that

v www.esajournals.org 12 May 2021 v Volume 12(5) v Article e03487

 LEWIS ET AL.

some wildlife species are influenced by human recreation (i.e., disturbance characteristics and
recreation, there are several considerations when wildlife characteristics; Knight and Gutzwiller
interpreting our results. First, we did not evalu- 1995), our work proposes that two additional fac-
ate how different age and sex classes of wildlife tors should be included in our conceptual under-
were affected by human recreation (Ladle et al. standing of human recreation impacts on
2019). Some animals might exhibit bolder or wildlife: characteristics of both spatial and tem-
more na€ıve behavior (e.g., younger animals and poral refuges in relation to disturbance and wild-
males), whereas other animals might be more life (Fig. 1). Indeed, the importance in providing
wary of human disturbance (e.g., adult females, spatial refuge from human activities has been
especially with offspring) (Knight and Gutzwiller emphasized in mitigating the effects of human
1995). Second, wildlife might exhibit different disturbance on wildlife populations (Reed and
responses to recreation depending upon the type Merenlender 2008, Stankowich 2008, Switalski
of activity (e.g., motorized recreation, hiking, and Jones 2012, Fortin et al. 2016) and both spa-
biking), which is an important consideration for tial and temporal refuges warrant further evalua-
understanding the effects of recreation on wild- tion and consideration in recreation ecology and
life and creating management plans for recre- public land management.
ation areas (Knight and Gutzwiller 1995, Larson Our study provides novel information that can
et al. 2019). Third, wildlife might exhibit different be used to inform management decisions related
responses to recreation across regional biomes to human recreation to minimize impacts to
(e.g., dense forested areas vs. open woodlands) wildlife. Recreation areas provide important out-
and degrees of habituation to people (Knight and door opportunities to people, as well as conserv-
Gutzwiller 1995, Kays et al. 2017), which we ing habitat for a suite of wildlife species. High
were unable to evaluate in this study. Fourth, levels of human recreation can cause wildlife to
when comparing wildlife activity on game vs. avoid using areas or reduce the amount of use
human trails, we assumed that wildlife occurring near trails. Land managers might consider strate-
on game trails exhibited “natural” activity pat- gies to reduce recreation levels in areas where
terns. However, on both game and human trails sensitive species occur. This might also entail
wildlife could be influenced by additional maintaining spatial refuges for wildlife without
human and environmental factors (Larson et al. human recreation that animals can use during
2019, Suraci et al. 2019). Although plant produc- the day and night (Reed and Merenlender 2008,
tivity as measured by NDVI generally had little Stankowich 2008, Switalski and Jones 2012, Lar-
effect on wildlife use relative to human recre- son et al. 2019). In addition, because wildlife will
ation, wildlife distributions and activity could be shift their daily activity patterns to be more
affected by additional biotic and abiotic charac- active during the night when human activity is
teristics. Lastly, we focused on three key metrics lowest, maintaining adequate temporal refuges
of how recreation affects wildlife populations: for wildlife is likely important to mitigate the
avoidance of an area, relative use of an area, and impacts of human recreation for some species.
altering daily activity patterns. There are, how- Indeed, the interaction between spatial and tem-
ever, other important factors to consider, which poral refuges likely requires a broader considera-
we were unable to evaluate with wildlife cam- tion of species-specific spatiotemporal
eras, such as animal flight response and distance, requirements. To ensure an intact ecological com-
energy expenditure, physiological response, munity, land managers of parks and public lands
reproductive success, and fitness (Papouchis might consider the varying sensitivities of wild-
et al. 2001, Stankowich 2008, Naylor et al. 2009, life, especially for the most sensitive species,
Steven et al. 2011, Spaul and Heath 2016). when creating and implementing management
Our work builds upon prior conceptual frame- plans.
works about the effects of human recreation on
wildlife populations (Knight and Gutzwiller ACKNOWLEDGMENTS
1995). When understanding the important factors
that influence whether a particular species or Funding and support were provided by Colorado
population will be influenced by human State University, Arizona State University, Colorado

v www.esajournals.org 13 May 2021 v Volume 12(5) v Article e03487

 LEWIS ET AL.

Parks and Wildlife (CPW), City of Boulder Open Space Czech, B., P. R. Krausman, and P. K. Devers. 2000. Eco-
and Mountain Parks (OSMP), Boulder County Parks nomic associations among causes of species endan-
and Open Space (BCPOS), US Forest Service, and a germent in the United States. BioScience 50:593–
grant from the National Science Foundation-Ecology 601.
of Infectious Diseases Program (NSF EF-0723676; EF- Donovan, T. M., and J. Hines. 2007. Exercises in occu-
1413925). We greatly thank R. Alonso, M. Durant, J. pancy modeling and estimation. https://www.uvm.
Feltner, L. Heck, A. Kellner, R. Larson, and D. Morin edu/rsenr/vtcfwru/spreadsheets/?Page=occupancy/
for their invaluable assistance in the field. In addition, occupancy.htm
we thank the numerous landowners who allowed us Dwinnell, S. P., H. Sawyer, J. E. Randall, J. L. Beck, J. S.
access to their properties for our research. We thank Forbey, G. L. Fralick, and K. L. Monteith. 2019.
CPW for the use and modification of their Access pho- Where to forage when afraid: Does perceived risk
tograph database. An earlier draft of this research pro- impair use of the foodscape? Ecological Applica-
ject was presented as an public report with BCPOS tions 29:e01972.
and OSMP in 2014. We greatly appreciate the insight Fortin, J. K., K. D. Rode, G. V. Hilderbrand, J. Wilder,
from M. Durant, D. Hoerath, M. Kobza, C. Larson, S. Farley, C. Jorgensen, and B. G. Marcot. 2016.
and anonymous reviewers that improved the paper. Impacts of human recreation on brown bears
(Ursus arctos): a review and new management tool.
LITERATURE CITED PLOS ONE 11:e0141983.
Garber, S. D., and J. Burger. 1995. A 20-yr study docu-
Ballantyne, M., and C. M. Pickering. 2013. Tourism menting the relationship between turtle decline
and recreation: a common threat to IUCN red- and human recreation. Ecological Applications
listed vascular plants in Europe. Biodiversity and 5:1151–1162.
Conservation 22, 3027–3044. Gaynor, K. M., C. E. Hojnowski, N. H. Carter, and J. S.
Balmford, A., J. Beresford, J. Green, R. Naidoo, M. Wal- Brashares. 2018. The influence of human distur-
pole, and A. Manica. 2009. A global perspective on bance on wildlife nocturnality. Science 360:1232–
trends in nature-based tourism. PLoS Biology 7: 1235.
e1000144. George, S. L., and K. R. Crooks. 2006. Recreation and
Barrueto, M., A. T. Ford, and A. P. Clevenger. 2014. large mammal activity in an urban nature reserve.
Anthropogenic effects on activity patterns of wild- Biological Conservation 133:107–117.
life at crossing structures. Ecosphere 5:1–19. Hammitt, W. E., and D. N. Cole. 1998. Wildland recre-
Beeco, J. A., J. C. Hallo, E. D. Baldwin, and F. A. ation: ecology and management. John Wiley &
McQuire. 2011. An examination of the guided Sons, New York, New York, USA.
night hiking experience in parks and protected Hammitt, W. E., D. N. Cole, and C. A. Monz. 2015.
areas. Journal of Park and Recreation Administra- Wildland recreation: ecology and management.
tion 29:72–88. John Wiley & Sons, Oxford, UK.
Burger, J. 1995. Beach recreation and nesting birds. Kays, R., A. W. Parsons, M. C. Baker, E. L.
Pages 281–295 in R. L. Knight and K. J. Gutzwiller, Kalies, T. Forrester, R. Costello, C. T. Rota, J.
editors. Wildlife and recreationists: coexistence J. Millspaugh, and W. J. McShea. 2017. Does
through management and research. Island Press, hunting or hiking affect wildlife communities in
Washington, D.C., USA. protected areas? Journal of Applied Ecology
Burnham, K. P., and D. R. Anderson. 2002. Model 54:242–252.
selection and multimodel inference: a practical Kellner, A., S. Carver, A. Gramza, J. S. Lewis, S. Vande-
information-theoretic approach. Second edition. Woude, and K. R. Crooks. 2017. Outdoor recre-
Springer Verlag, New York, New York, USA. ation at the wildland—urban interface: examining
Ceballos, G., and P. R. Ehrlich. 2002. Mammal popula- human activity patterns and compliance with dog
tion losses and the extinction crisis. Science management policies. Natural Areas Journal
296:904–907. 37:515–529.
Cohen, J. E. 2003. Human population: the next half Knight, R. L., and D. N. Cole. 1991. Effects of recre-
century. Science 302:1172–1175. ational activity on wildlife in wildlands. Transac-
Crooks, K. R., C. L. Burdett, D. M. Theobald, S. R. tions of the North American Wildlife and Natural
King, M. Di Marco, C. Rondinini, and L. Boitani. Resources Conference 56:238–247.
2017. Quantification of habitat fragmentation Knight, R. L. and K. J. Gutzwiller, editors. 1995. Wild-
reveals extinction risk in terrestrial mammals. Pro- life and recreationists: coexistence through man-
ceedings of the National Academy of Sciences of agement and research. Island Press, Washington,
the United States of America 114:7635–7640. D.C., USA.

v www.esajournals.org 14 May 2021 v Volume 12(5) v Article e03487

 LEWIS ET AL.

Laake, J., and E. Rexstad. 2018. RMark—an alternative human recreation. Journal of Wildlife Management
approach to building linear models in MARK. In E. 65:573–582.
G. Cooch, and G. C. White, editors. Program R. Development Core Team. 2020. R: a language and
MARK: a gentle introduction. http://www.phidot. environment for statistical computing. R Founda-
org/software/mark/docs/book/ tion for Statistical Computing, Vienna, Austria.
Ladle, A., T. Avgar, M. Wheatley, G. B. Stenhouse, S. E. Reed, S. E., and A. M. Merenlender. 2008. Quiet, non-
Nielsen, and M. S. Boyce. 2019. Grizzly bear consumptive recreation reduces protected area
response to spatio-temporal variability in human effectiveness. Conservation Letters 1:146–154.
recreational activity. Journal of Applied Ecology Reilly, M., M. W. Tobler, D. L. Sonderegger, and P.
56:375–386. Beier. 2017. Spatial and temporal response of wild-
Larson, C. L., S. E. Reed, A. M. Merenlender, and K. R. life to recreational activities in the San Francisco
Crooks. 2016. Effects of recreation on animals Bay ecoregion. Biological Conservation 207:117–
revealed as widespread through a global system- 126.
atic review. PLOS ONE 11:e0167259. Ridout, M., and M. Linkie. 2009. Estimating overlap of
Larson, C. L., S. E. Reed, A. M. Merenlender, and K. R. daily activity patterns from camera trap data. Jour-
Crooks. 2019. A meta-analysis of recreation effects nal of Agricultural, Biological, and Environmental
on vertebrate species richness and abundance. Statistics 14:322–337.
Conservation Science and Practice 1:1–9. Rogan, M. S., G. A. Balme, G. Distiller, R. T. Pitman, J.
Lenth, B. E., R. L. Knight, and M. E. Brennan. 2008. Broadfield, G. K. Mann, G. M. Whittington-Jones,
The effects of dogs on wildlife communities. Natu- L. H. Thomas, and M. J. O’Riain. 2019. The influ-
ral Areas Journal 28:218–228. ence of movement on the occupancy–density rela-
Lewis, J. S., L. L. Bailey, S. VandeWoude, and K. R. tionship at small spatial scales. Ecosphere 10:
Crooks. 2015a. Interspecific interactions between e02807.
wild felids vary across scales and levels of urban- Royle, J. A., and J. D. Nichols. 2003. Estimating abun-
ization. Ecology and Evolution 5:5946–5961. dance from repeated presence–absence data or
Lewis, J. S., K. A. Logan, M. W. Alldredge, L. L. Bailey, point counts. Ecology 84:777–790.
S. VandeWoude, and K. R. Crooks. 2015b. The Sawyer, H., R. M. Nielson, F. Lindzey, and L. L.
effects of urbanization on population density, occu- McDonald. 2006. Winter habitat selection of mule
pancy, and detection probability of wild felids. Eco- deer before and during development of a natural
logical Applications 25:1880–1895. gas field. Journal of Wildlife Management 70:396–
Linden, D. W., A. K. Fuller, J. A. Royle, and M. P. Hare. 403.
2017. Examining the occupancy–density relation- Schielzeth, H. 2010. Simple means to improve the
ship for a low-density carnivore. Journal of interpretability of regression coefficients. Methods
Applied Ecology 54:2043–2052. in Ecology and Evolution 1:103–113.
MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pol- Seto, K. C., B. G u € neralp, and L. R. Hutyra.
lock, L. Bailey, and J. E. Hines. 2018. Occupancy 2012. Global forecasts of urban expansion to 2030
estimation and modeling: inferring patterns and and direct impacts on biodiversity and carbon
dynamics of species occurrence. Elsevier, San pools. Proceedings of the National Academy of
Diego, California, USA. Sciences of the United States of America
Meredith, M., and M. Ridout. 2018. Overlap. R Core 109:16083–16088.
Development Team. https://cran.r-project.org/web/ Smith, B., and J. Hallo. 2013. A system-wide assess-
packages/overlap/overlap.pdf ment of night resources and night recreation in the
Metcalfe, J. A. 1974. The wonders of Night Hiking. US National Parks: a case for expanded definitions.
Journal of Health, Physical Education, Recreation Park Science 29:54–59.
45:8. Spaul, R. J., and J. A. Heath. 2016. Nonmotorized
Nakashima, Y. 2020. Potentiality and limitations of N- recreation and motorized recreation in shrub-
mixture and Royle-Nichols models to estimate ani- steppe habitats affects behavior and reproduction
mal abundance based on noninstantaneous point of golden eagles (Aquila chrysaetos). Ecology and
surveys. Population Ecology 62:151–157. Evolution 6:8037–8049.
Naylor, L. M., M. J. Wisdom, and R. G. Anthony. 2009. Staine, K. J., and J. Burger. 1994. Nocturnal foraging
Behavioral responses of North American elk to behavior of breeding Piping Plovers (Charadrius
recreational activity. Journal of Wildlife Manage- melodus) in New Jersey. Auk 111:579–587.
ment 73:328–338. Stankowich, T. 2008. Ungulate flight responses to
Papouchis, C. M., F. J. Singer, and W. B. Sloan. 2001. human disturbance: a review and meta-analysis.
Responses of desert bighorn sheep to increased Biological Conservation 141:2159–2173.

v www.esajournals.org 15 May 2021 v Volume 12(5) v Article e03487

 LEWIS ET AL.

Steven, R., C. Pickering, and J. Guy Castley. 2011. A U.S. Census Bureau. 2010. Census demographic pro-
review of the impacts of nature based recreation on files interactive population map. https://www.
birds. Journal of Environmental Management census.gov/2010census/popmap/
92:2287–2294. Vaske, J., L. Shelby, and M. Donnelly. 2009. Estimating
Suraci, J. P., M. Clinchy, L. Y. Zanette, and C. C. Wil- visitation to boulder open space and mountain
mers. 2019. Fear of humans as apex predators has parks. Colorado State University, Colorado State
landscape-scale impacts from mountain lions to University, Human Dimensions of Natural
mice. Ecology Letters 22:1578–1586. Resources, Fort Collins, Colorado, USA.
Switalski, T. A., and A. Jones. 2012. Off-road vehicle best White, G. C., and K. P. Burnham. 1999. Program
management practices for forestlands: a review of MARK: survival estimation from populations of
scientific literature and guidance for managers. marked animals. Bird Study 46:120–139.
Journal of Conservation Planning 8:12–24. Wilcove, D. S., D. Rothstein, J. Dubow, A. Phillips, and
Taylor, A. R., and R. L. Knight. 2003. Wildlife E. Losos. 1998. Quantifying threats to imperiled
responses to recreation and associated visitor per- species in the United States. BioScience 48:607–615.
ceptions. Ecological Applications 13:951–963. Windell, R. M., J. S. Lewis, A. R. Gramza, and K. R.
Theobald, D. M. 2005. Landscape patterns of exurban Crooks. 2019. Carnivore carrying behavior as doc-
growth in the USA from 1980 to 2020. Ecology and umented with wildlife camera traps. Western
Society 10:1–34. North American Naturalist 79:471–480.

SUPPORTING INFORMATION
Additional Supporting Information may be found online at: http://onlinelibrary.wiley.com/doi/10.1002/ecs2.
3487/full

v www.esajournals.org 16 May 2021 v Volume 12(5) v Article e03487

View publication stats