Content uploaded by Kyle Cutting
Author content
All content in this area was uploaded by Kyle Cutting on Nov 18, 2024
Content may be subject to copyright.
Large-scale fence modications increase nest survival in sage-grouse
Aaron W. O’Harra
a,b,*
, Kyle A. Cutting
a,1
, Michael R. Frisina
b,c
, Bok F. Sowell
b
a
U.S. Fish and Wildlife Service, Red Rock Lakes National Wildlife Refuge, 27650B S. Valley Road, Lima, MT 59739, USA
b
Montana State University, Department of Animal and Range Sciences, 100 Culbertson Hall, Bozeman, MT 59717, USA
c
August L. Hormay Wildlands Institute, Inc., PO Box 4712, Butte, MT 59701, USA
ARTICLE INFO
Keywords:
Management action
Fence ecology
Grazing infrastructure
Greater sage-grouse
Livestock
Nest predators
Nest survival
ABSTRACT
Livestock grazing is the most common land-use practice in aridland ecosystems of the American West. Wide-
spread and long-term declines of greater sage-grouse (Centrocercus urophasianus), an umbrella species for other
sagebrush-associated vertebrates, has challenged land managers to nd solutions to help bolster populations.
Livestock grazing infrastructure (e.g., fences) among western rangelands are one potential factor impacting sage-
grouse populations as nest survival –a key demographic rate that inuences population size –can be 4xhigher
when placed >100 m away from the nearest fence. We implemented the rst-ever long-term (8 years) and large-
scale (70 km in length) management action designed to evaluate changes in nest survival related to fence
modications in the Centennial Valley of Montana, USA. Fence modications included perch deterrents placed
atop fence posts to reduce perching by avian nest predators. Additionally, the bottom fence wire was raised to
45 cm from the ground to reduce the funneling of mammalian predators along fences who are in search of
carcasses from collisions and/or nests placed near fences. Using a Bayesian statistical framework, we found that
fence modications increased 28-day nest survival within 100 m of mitigated fences by an average of 10.6 %
(85 % CRI: 3.7–24.9 %). The difference in survival between fence types was most pronounced for modied fences
in low elevations, while modied fences resulted in higher survival rates throughout the nesting season. Our
ndings suggest that modifying fences designed to reduce the search efciency of sage-grouse nest predators
appear to benet nest survival in sage-grouse. The highlighted fence modication technique offers practitioners a
tool to advance grazing systems that reduce the impacts of grazing infrastructure on greater sage-grouse.
1. Introduction
Livestock grazing is the most widely applied land-use in the Amer-
ican West, with much of it occurring in sagebrush steppe ecosystems.
This creates high potential for overlap between livestock grazing and
wildlife populations. One such example is the greater sage-grouse
(Centrocercus urophasianus;hereafter sage-grouse), an important indica-
tor species for sagebrush (Artemisia spp.) ecosystems (Rowland et al.,
2006). One of the most ubiquitous features of livestock grazing pro-
grammes is the addition of infrastructure (Briske et al., 2011; Fleming
and Bateman, 2018). Permanent linear-and-point infrastructure–
including water tanks, roads, and fences –are critical tools used to
control the spatio-temporal patterns of livestock grazing. Yet, empirical
studies on grazing infrastructure, including fence designs and their ef-
fects on wildlife demography and behavior, are sparse (Hovick et al.,
2014).
Grazing systems are dependent upon infrastructure. There are mil-
lions of kilometers of fence used in grazing systems across the American
West, even exceeding total road distance in many areas (Jakes et al.,
2018). Fences are vertical linear features with posts that hold noncon-
tinuous structures (e.g., wire) that connect the interspace (McInturff
et al., 2020). Fence types in sagebrush ecosystems include smooth,
barbed, electric, woven wire, and wooden. Fences can be a valuable
management tool for controlling the distribution of livestock and for
protection of sensitive areas (e.g., riparian areas; Bailey, 2004).
Conversely, fences can be harmful as they can negatively inuence
wildlife populations either directly (e.g., collisions, mortality, hair loss;
Harrington and Conover, 2006;Stevens et al., 2012,) or indirectly (e.g.,
behavioral changes, obstructed movements; Jakes et al., 2018;Cutting
et al., 2019). Fences are spatially distributed, differ in density, condi-
tion, and type, and create varying vertical and horizontal obstacles for
the many wildlife species that encounter them. Anthropogenically
* Correspondence to: Montana Fish, Wildlife &Parks, 1 Airport Rd, Glasgow, MT 59230, USA.
E-mail address: aoharra5@gmail.com (A.W. O’Harra).
1
U.S. National Park Service, Wrangell-St. Elias National Park &Preserve, Copper Center, AK 99573, USA.
Contents lists available at ScienceDirect
Agriculture, Ecosystems and Environment
journal homepage: www.elsevier.com/locate/agee
https://doi.org/10.1016/j.agee.2024.109248
Received 16 April 2024; Received in revised form 12 August 2024; Accepted 14 August 2024
Agriculture, Ecosystems and Environment 377 (2025) 109248
Available online 24 August 2024
0167-8809/Published by Elsevier B.V. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ).
created grazing infrastructure have recently been shown to facilitate use
and movement patterns of nest predators including coyotes and corvids
(Coates et al., 2016;Harju et al., 2018)). Yet, understanding the impacts
of predator use of grazing-related infrastructure on subsequent prey
demography remains elusive.
Previously, mortality caused by direct collisions with fences was one
of the few identied risks fences pose to sage-grouse survival (Stevens
et al., 2012, Jones, 2014, Van Lanen et al., 2017). Recent research,
however, indicates a link between grazing related infrastructure, pred-
ator movement behavior, and survival of ground dwelling birds (Pedlar
et al., 1997; Andersson, et al., 2009; Coates et al., 2016; Vold, 2018;
Cutting et al., 2019). In grazing modied landscapes, predators may be
able to use vertical-linear features to increase their foraging efciency
through increased search efciency and handling time of prey (Harju
et al., 2018). Cutting et al. (2019) found that sage-grouse nests placed
close to fences display a 4-fold reduction in survival in comparison to
nests located further away (>100 m). The combination of increased
predation efciency and lower sage-grouse nest survival along fences is
signicant as predation is the leading cause of nest failure in ground
nesting birds (Donovan and Thompson, 2001; Moynahan et al., 2007;
Okada et al., 2019). However, increased experience through either age
of the nesting female or subsequent renesting attempts could compen-
sate the risks associated with nest predation near fences, as both age and
renesting provide individuals with additional knowledge of their nest
sites (Holloran and Anderson, 2005). Further research is now needed as
sage-grouse are a primary focal species that inuence the conservation
and management of sagebrush ecosystems across the American West
(Smith et al., 2020). Understanding changes in predator-prey in-
teractions in response to targeted management action, such as fence
modications, is an important rst step in testing the effectiveness of
alternative grazing strategies to advance sage-grouse conservation
practices. This management test also adds to the emerging eld of ‘fence
ecology’(Jakes et al., 2018; McInturff et al., 2020).
Furthering understanding of the indirect impacts fences have on nest
survival of sage-grouse can be achieved through implementing targeted
management actions that alter the search efciency of nest predators.
The three dominant nest predators of sage-grouse include ravens (Corvus
corax), coyotes (Canis latrans), and badgers (Taxidea taxus;Conover and
Roberts, 2017). Avian nest predators (e.g., ravens) use fences to increase
movements and foraging efciency (Hovick et al., 2014; Coates et al.,
2016). Ravens use vertical-linear features for perches that can improve
visibility (Andersson et al., 2009). Coyotes in the eastern United States
use habitat edges and fencerows between suburban and rural areas to
increase their movements (Atwood et al., 2004). In the western United
States, coyotes and other mesopredators use linear features, including
fences and roads, to facilitate occupancy and movement (Vold, 2018;
Jones et al., 2019; Rodriguez et al., 2021). Yet, no studies have initiated
a targeted management action focused on modifying fences with the
goal of reducing occupancy and movements of nest predators along
fences to improve survival in ground nesting birds, including
sage-grouse.
In this study, we use long-term data (8 years) collected using radio-
telemetry on a population of sage-grouse to evaluate the effectiveness of
a large-scale management action on sage-grouse nest survival by
modifying fences intended to reduce efciency of avian and mammalian
nest predators. We hypothesize that fence modications will increase
survival of nests that are close to fences, and this increase will be
observed across a range of elevations and throughout the nesting season.
Our large-scale management action aims to provide a novel design that
can easily be implemented by grazing practitioners to reduce impacts of
fences on nest survival in sage-grouse.
2. Materials and methods
2.1. Study area
The study was conducted in a high-elevation (base elevation
>2000 m) sagebrush landscape (1560 km
2
) located in southwestern
Montana, USA (Fig. 1). Land ownership is a mixture of private and
public. Multiple sagebrush species and sub-species are found across an
elevational range within the study area. The predominant land-use
practice of the area is livestock grazing, with most of the landscape
receiving grazing activity. Most of the study area is on a rest-rotation
grazing system, with pastures being grazed for 2–3 years followed by
a year of rest. Average stocking rate varied from 1.2 to 2.0 ha per adult
cow with an average weight of 454 kg (Cutting et al., 2019).
2.2. Fence modications
Approximately 70 km in length of fence were modied in the project
area between the years of 2015–2021. Modications to fences included
bundles of sharp metal spikes placed atop fence posts to preclude
landing by aerial nest predators (Nixalite®) and raising the bottom wire
to 45 cm above ground level to reduce funneling and increase perme-
ability through fences by mammalian nest predators (Fig. 2). We
considered traditional fences to comprise all fences that are not modied
with perch deterrents and with the bottom wire not raised to 45 cm.
2.3. Female sage-grouse capture and monitoring
Adult female sage-grouse were captured across 6 areas where sage-
grouse congregate in the spring to perform breeding displays, called a
“lek”, between 2014 and 2021. Spotlighting and net techniques were
used to capture female sage-grouse. We determined age (AHY: after
hatch year, ASY: after second year) using feather characteristics and
body mass (Braun and Schroeder, 2015). Once captured, each bird was
tted with a VHF necklace style radio transmitter. Females were tracked
2–3 times per week from mid-April through mid-July using intensive
ground and aerial surveys. A nest was conrmed when a female was in
the same location for 2 consecutive occasions. To avoid investigator
disturbance, we circled a suspected nesting individual at a minimum of a
15 m radius. Once a nest was located, we monitored survival 2 times per
week until the fate was determined as either successful (i.e., ≥1 egg
hatched) or unsuccessful. For individuals who failed nesting on the
initial attempt, we continued to track the individual until mid-June to
determine whether a second nest was initiated. We established telem-
etry points with a known bearing to the nest site location (placed
50–100 m away from a nest location) to determine if a female was still
incubating a nest and to monitor until nest fate was determined. All
failed nests monitored within the study were from predation, based on
either all eggs missing, or eggshell fragments found in or around the nest
site. Research was conducted in compliance with state (2014–050,
2015–23, 2016–049, 2017–043,2018–003, 2019–078, 2020–010-W,
2021–026-W) and animal care and use (2012–40, 2015–30, 2019–88)
permits.
2.4. Statistical analysis
The fence modication area (FMA) encompasses 125 km
2
(Fig. 1).
We dened the FMA to be within 1.5 km of a modied fence, creating an
area where both fence types are represented at roughly equal lengths
(70 km of modied fence and 67 km of traditional fence). Within the
FMA, we selected nests located within 1 km of either fence type for
analyses as this distance captured the maximum distance a nest could be
from one fence type without overlapping the other. This spatial
constraint also focuses our analyses to those nests that are close enough
for potential inuence of the management action allowing us to evaluate
the effectiveness of fence modications on nest survival in sage-grouse.
A.W. O’Harra et al. Agriculture, Ecosystems and Environment 377 (2025) 109248
2
2.4.1. Model covariates
Predictor variables used to evaluate variation in nest survival
included fence type (i.e., modied or traditional) and distance, eleva-
tion, initiation of nest incubation date (hereafter initiation date), female
age, and renesting attempt (Table 1). Information on the locations and
structure types of fences throughout the FMA were compiled using
existing data and veried through eld reconnaissance surveys. Due to
the ongoing nature of the fence modications, nests from each year were
assigned a fence type based on that specic year’s conguration of either
modied or traditional. We used ArcGIS10.6.1 (ESRI, Redlands, Cali-
fornia, USA) to measure linear distances from nests to the nearest fence
type.
We calculated elevation using a 30-m digital elevation model (EROS,
2016). Managers implemented most of the fence modications in lower
elevations within the FMA, which precluded us from evaluating the ef-
fect of fence modications on nest survival across different sagebrush
types. We instead use continuous elevation as a proxy for sagebrush type
to capture this predictable and known gradient.
We considered initiation date (i.e., date the last egg was laid, and
incubation started) to test for temporal variation in nest survival across
the nesting season. Female sage-grouse that nest early experience colder
and wetter conditions than individuals nesting later (Klett and Johnson,
1982). Sage-grouse also nest earlier than other birds in the sagebrush
environment and before many mammalian prey species are active
following winter hibernation. This, in turn, may focus predators on
sage-grouse until alternative food sources become abundant later in the
season. Initiation date was calculated by subtracting the average incu-
bation period (28-days; Moynahan et al., 2007) for successful in-
dividuals from the hatch date. Since we could not determine the exact
initiation date for unsuccessful individuals, we subtracted 2.5 days from
the date when we rst discovered a nest, which was the average number
of days from nest initiation to investigator discovery in our sample of
successful nests. For individual sage-grouse who did not initiate a nest,
we continued to track them twice per week until June 19th, which was
the last day a nest was discovered from the years of 2014–2018.
2.4.2. Nest survival model
A Bayesian statistical modeling framework was used to model daily
survival rates (hereafter DSR) of nests, which also allows for the inclusion
of random effect terms. Given the long-time span of the current study (8
years), we were less interested in annual variation on DSR. We therefore
accounted for annual variation in DSR by adding a Year term as a
random effect with varying intercepts. We used a Bayesian statistical
model similar to Schmidt et al. (2010), with binomially distributed er-
rors and an inverse logit link function to model DSR of nests. To evaluate
the outcome of the fence modication management action on nest sur-
vival, we modeled DSR (ϕ
i,t
) as a function of fence type, fence distance,
elevation, initiation date, female age (i.e., AHY or ASY), and nest
attempt (i.e., 1st or renest). We used a Bayesian variable selection
technique that uses an indicator variable to quantify support for effects
of each covariate on DSR (Converse et al., 2013). We derived support for
effects of regression coefcients (β) by multiplying each with a binary
indicator variable (w) such that:
ilogit(ϕi,t) = β0+w1β1×fence.typei+w2β2×fence.distancei+w3β3
×elevationi+w4β4×initiation.datei+w5β5×agei+w6β6
×nest numberi+
α
YEARt
α
t∼N(0,
τ
)
We scaled the variance of parameter prior distributions to create
uninformative priors by creating a mean zero normal prior with variance
Fig. 1. Fence modication area and nest site locations of female sage-grouse, 2014–2021, Centennial Valley, Montana, USA. Inset shows years when fence modi-
cation occurred, along with proximity to traditional fences. Sage-grouse nests utilize the same color scheme as fence modications to indicate the year a nest was
monitored (e.g., light blue nests were monitored in 2015). Dark blue nest points indicate a nest from 2014, prior to fence modications.
A.W. O’Harra et al. Agriculture, Ecosystems and Environment 377 (2025) 109248
3
V/K, where K represents the number of effects present in a model at a
particular sample (from 0 to 7). We placed a Gamma distributed prior on
the total variance of the linear predictor, V, with parameters 3.29 and
7.8 as this prior results in a marginal distribution that approximates nest
fate outcomes (0,1; Link and Barker, 2006). A Bernoulli (0.5) prior in-
clusion probability was placed on the indicator variables, which repre-
sents no prior knowledge on individual variable importance. The
posterior inclusion probability (w
p
;hereafter PIP) score represents the
frequency a given variable was included across all the Markov Chain
Monte Carlo (hereafter MCMC) simulations. PIP scores >0.5 were used to
determine which variables from the full model containing all the cova-
riates should be carried forward into the nal model. To account for
annual effects, we included the covariate Year (t) as a random effect (
α
)
from a Normal(0,
τ
) prior, where
τ
is a precision parameter from the
scaled Gamma(1,2)
2
distribution. We make inferences based on the nal
model, and evaluated parameter estimates and whether a given coef-
cient in the model overlapped zero.
We simulated the model using MCMC sampling with 3 independent
chains, each with 200,000 iterations and discarding the initial 20,000 as
burn-in. We assessed model convergence by visual inspection and the
Gelman-Rubin statistic ensuring Rˆwas <1.1. The models were t using
JAGS (version 4.3.0, mcmc-jags.sourceforge.net, accessed 31 October
2019) via the R2jags package (Su and Yajima, 2015) in program R
(version 3.6.1, www.r-project.org, accessed 5 July 2019). Predictions of
survival were made across covariates found in the nal model by
holding other variables constant at their median value. Since the chance
of stochastic variation increases with analytical scale resolution (Levin,
1992), we evaluated parameter estimates from the posterior distribu-
tions at the 85 % credible interval. Prior to running models, we evalu-
ated pairwise correlations among covariates and did not nd evidence
for high collinearity (Pearson’sr<0.70).
Fig. 2. Example of the eld-tested management action designed to modify fences to reduce predation on sage-grouse nests. Fence modications include raising the
lower wire to 45 cm off the ground (top-left image) along with attaching anti-perching devices consisting of small bundles of sharp pointed metal spikes placed atop
each fence post (top-right image). A motion detection camera is mounted to the vertical wood fence post (top-left image) and is placed 11 cm from the ground for
reference. The diagram below the photos indicates the two modications that were needed to consider a fence “modied”in comparison to two examples of a
“traditional”fence. Photograph credits: A. O’Harra.
Table 1
Covariates used in daily survival analysis of greater sage-grouse nests,
2014–2021, Centennial Valley, MT, USA.
Variable Description Type
Fence Type Modied or traditional Binary
Fence
Distance
Distance to fence Continuous
Elevation Proxy for sagebrush type Continuous
Initiation
Date
Ordinal date when nest incubation was initiated Continuous
Female Age Age of individual either after-hatch-year or after-
second-year
Binary
Nest Number Whether a nest number was from rst or renest
attempt
Binary
Year Year nest was discovered Random
Effect
A.W. O’Harra et al. Agriculture, Ecosystems and Environment 377 (2025) 109248
4
3. Results
We discovered 177 nests located within 1 km of a fence in the FMA
from 2014 to 2021. Fifty-two nests were located nearest a modied
fence while 125 nests were located nearest a traditional fence. Twenty-
three (44.2 %) of the 52 nests nearest modied fences and 44 (35.2 %) of
the 125 nests nearest traditional fences were successful. MCMC chains
for all estimated parameters in the models converged (Rˆ<1.1), and
posterior predictive checks suggest the model t the data adequately. All
unsuccessful nests were caused by predation.
3.1. Survival of nests across fence type and predictor variables
From the full model, both hen age and nest number had the same PIP
score of 0.48, indicating less than half of the iterations ran contained
either of these variables. Additionally, the estimated effect size of each
of these variables strongly overlapped zero (Hen age:
β= − 0.034,
85 % CRIs: -0.305 –-0.239; Nest Number:
β=0.117, 85 % CRIs: -0.112 –
-0.351). As a result, both hen age and nest number were removed from
our nal model. The nal model contained fence type, fence distance,
elevation, and initiation date (Table 2). The random effect of year had
standard deviations of 0.33, 0.29, 0.23, 0.27, 0.23, 0.23, 0.67, 0.28 for
years 2014 through 2021 respectively. Based on these values and esti-
mates centered around zero (Table 2), there is only minor evidence in
support of yearly variation in DSR. The high estimate and standard
deviation seen in 2020 (0.56 ±0.67) is explained due to an abnormally
high nest success rate (100 %) in concert with a small sample size (n=9).
Both fence type and fence distance had strong positive effects on DSR
(Table 2). For example, when considering the combined, additive effects
of fence type and fence distance along modied fences, DSR changed
from 94.0 % to 97.6 % (85 % CRIs: 93.6–94.1 % and 96.0–98.8 %,
respectively) as covariate fence distance went from the 1 % quantile
(5 m) to 99 % quantile (900 m) in the observed dataset. The change in
DSR equates to nest survival changing from 17.7 % to 50.7 % (85 %
CRIs: 15.7–18.2 % and 31.9–71.3 %, respectively) as distance from a
modied fence increase. In contrast, considering the same combined
additive effects for traditional fences, DSR changed from 91.8 % to
96.8 % (85 % CRIs: 91.2–91.9 % and 94.4–98.2 %, respectively) across
the same quantiles in the observed dataset (Fig. 3a). These DSR rates
equate to nest survival changing from 9.1 % to 40.2 % (85 % CRIs:
7.6–9.4 % and 19.9–60.1 %, respectively) with increasing distance from
a traditional fence. These results imply that the fence modications in-
crease the median overall 28-day nest survival by an average of 10.6 %
(85 % CRIs: 3.7–24.9 %). DSR of nests in lower elevations (1 % quantile
in observed dataset =2028 m) was higher for nests with the additive
effect of a modied fence type (96.0 %; 85 % CRIs: 96.0–96.1 %) as
compared to a traditional fence type (95.0 %; 85 % CRIs: 94.9–95.2 %),
but the additive effect of modied fences on DSR became less pro-
nounced in higher elevations (99 % quantile =2260 m; DSR: modied
=97.6 %; 85 % CRIs: 97.0–98.1 % vs. traditional =97.2 %; 85 % CRIs:
96.3–97.8 %; Fig. 3b). For initiation date, DSR was higher for those nests
initiated later in the nesting season (8 June; 98.0 %; 85 % CRIs:
95.8–99.0 %) as compared to those initiated earlier (8 April; 94.7 %;
85 % CRIs: 90.9–96.2 %), with the additive effect of a modied fence
improving DSR by an average of 1.2 % (85 % CRIs: 0.7–1.8 %) across
the range of initiation dates observed (Fig. 3c).
4. Discussion
Fences have been shown to directly inuence sage-grouse through
collision caused mortalities (Stevens et al., 2012; Jones, 2014; Van
Lanen et al., 2017), yet fences also appear to indirectly reduce nest
survival (Cutting et al., 2019). Our modeling approach highlights the
Table 2
Parameter estimates and credible intervals (CRIs) from the nal model,
describing xed and random effect terms on daily nest survival of greater sage-
grouse, 2014–2021, Centennial Valley, Montana, USA. Posterior Inclusion
Probability (PIP) scores listed in the table were derived from the full model
where a threshold of 0.5 was used to determine which variables were carried
forward into the nal model. Random effects are marked by an rin front of a
given year.
Parameter PIP
Score
Estimate Lower 85 %
CRIs
Upper 85 %
CRIs
b
0
(intercept) - 0.11 −0.30 0.55
b
1
(fence type) 0.64 0.34 0.03 0.66
b
2
(fence
distance)
0.74 0.17 0.06 0.27
b
3
(elevation) 0.99 0.28 0.12 0.44
b
4
(initiation
date)
0.90 0.02 0.01 0.02
r(2014) −0.290 −0.827 0.036
r(2015) - −0.176 −0.654 0.126
r(2016) - −0.031 −0.364 0.265
r(2017) - 0.062 −0.270 0.461
r(2018) - 0.002 −0.309 0.315
r(2019) - 0.008 −0.304 0.335
r(2020) - 0.555 −0.016 1.613
r(2021) - 0.134 −0.164 0.573
Fig. 3. Predicted daily survival rates for sage-grouse nests across different
fence types including modied (blue) and traditional (red) in relation to dis-
tance to fence (a), elevation (b), and initiation date (c). Each covariate was
plotted over the range of values in the observed dataset with other covariates
held at their median value. The solid lines represent the median effect of each
fence type, and the colored ribbons represent 85 % Bayesian credible intervals.
A.W. O’Harra et al. Agriculture, Ecosystems and Environment 377 (2025) 109248
5
importance of a management-led action designed to increase nest sur-
vival in sage-grouse when a nest is placed near a fence (<100 m) by
reducing perches for avian and funneling for mammalian nest predators.
We also show that fence modications appear to have an additive
benet to nest survival in lower elevation habitats ranging from ~2025
to 2100 m, but this additive effect appears to lessen with increasing
elevations. Nest survival also appeared to increase throughout the
nesting season, with the additive effect of modied fences improving
nest survival more so than traditional fence types for nests placed near a
fence (100 m).
The change in nest survival across elevations is likely related to
different assemblages of aerial and ground nest predators. O’Harra
(2023) demonstrates that occupancy rates of two of the three most
common sage-grouse nest predators (i.e., badgers and avian; Conover
and Roberts, 2017) had the highest occupancy in high elevation sage-
brush, while the other predator (i.e., coyotes) was most common in low
elevation sagebrush. Higher elevation areas oftentimes have more trees
which increase the number of perches for aerial nest predators. More
perches in turn can increase sightability of nests (Andersson et al., 2009;
Coates et al., 2016; Harju et al., 2018), a possible reason why the
combined additive effects of fence type and elevation on nest survival
did not vary between the two fence types at higher elevations. Coyotes
on the other hand have lower successful prey capture attempts in more
densely vegetated forested areas (Gese et al., 1996). To counter the
lower success rate of capture, coyotes in turn have shown to specialize
on edge habitat or patch boundaries (Atwood et al., 2004).
Vertical-linear features like fences found in open shrubland landscapes
likely provide ideal straight-line travel corridors to navigate quickly
through dense stands of shrubs which otherwise would require slow and
meandering movements in search of nests. Coyote movement patterns in
relation to vertical-linear fence features, along with other
grazing-related infrastructure, deserves further attention.
We also found support that fence modications help increase nest
survival of sage-grouse across the nesting season. Improved nest survival
as the nesting season progresses suggests that predator-prey dynamics
are also likely to shift. Later in the nesting season more alternative prey
sources are present, such as rodents and other avian species (Wegge and
Storaas, 1990). Vegetation phenology also progresses across the nesting
season and provides greater concealment of nests initiated later (Smith
et al., 2018). Alternative prey and/or greater amounts of hiding cover
are both possible explanations as to why DSR increases across the
nesting season.
Previous knowledge gained through either age or experience can
increase an individual’s familiarity of the risks associated with a given
nest site location (Holloran and Anderson, 2005). Surprisingly, we found
that neither female age nor within-year renesting increased DSR of nests.
The lack of relationship between age and experience on DSR may be
more related to philopatry of nest sites regardless of whether they were
successful or not on previous attempts (Fischer et al., 1993).
4.1. Advancing grazing systems for sage-grouse conservation
The positive effect of fence modications on nest survival designed
herein appears to advance sustainable grazing practices in high-
elevation sagebrush systems in southwest Montana. We implemented
the rst-ever, large-scale, 8 year management action to test the effec-
tiveness of fence modications on nest survival in sage-grouse. The most
prominent effect of modied fences on DSR is seen in areas near a fence
(i.e. <100 m; Fig. 3a). For the rst time, results support the management
action of modifying fences to support increased sage-grouse nest sur-
vival by reducing predation by avian and mammalian predators. Un-
derstanding the relationship between nest predators and fence
modications across a topographically diverse landscape could be key to
improving our understanding of nest survival in sage-grouse. From an
accompanying project that is evaluating impacts of fence modication
on the nest predator community, badger occupancy was shown to be
lower along modied fences immediately following management action
while the response in coyote occupancy appears lagged, decreasing in
the years following fence modication (O’Harra 2023). Indirect effects
of these fence modications are generally more difcult to conceptu-
alize and measure, but doing so is critical as they can be just as sub-
stantial as direct effects on wildlife (Kristov, 2008).
The current predator assemblage seen today has evolved with the
presence of fences. Their utilization of this vertical-linear feature has
likely changed from simply co-occurring in areas with fences to now
choosing to use areas with fences. While fences appear to increase pre-
dation rates on sage-grouse nests, future work should consider whether
this conclusion can be generalized to other ground nesting bird species
and geographic sites used by nesting grouse. Two major limitations to
the current study are the limited geographic scope of the study area and
lack of experimental design through randomization and interspersion.
Our study area included a geography that is largely considered to be
high elevation and intact relative to the sagebrush biome, and man-
agement treatments were based on logistical and nancial constraints of
where and when these treatments occurred. A simplication to the
current fence modication design, where only one modication tech-
nique is utilized, may also prove adequate. However, knowledge of the
suite of predators and unique landscape features in an area would be
required to guide such a decision. We managed to achieve a similar
structure for modied fences considered in the current study, but
traditional fence structures were dependent on who built the fence, its
intended purpose, and the materials used in construction. The high de-
gree of variation in traditional fence structures likely holds inuence
over predator behavior and subsequent nest survival. Variation in fence
structure was not able to be assessed in our modeling efforts but future
studies should consider different fence structure types. For example, nest
predators may use wooden fence posts with wide landing platforms
differently than metal t-posts with narrow platform surfaces for perch-
ing. An in depth understanding of the relationship between predators,
their prey, and fences is needed to advance the eld of fence ecology and
advance grazing systems.
Fencing for livestock is one of the most ubiquitous grazing practices
used throughout the world, yet proportionately understudied (Jakes
et al., 2018;McInturff et al., 2020). With most fence construction being
unregulated, owned, and maintained mostly by private landowners
(Jakes et al., 2018), it is no wonder that new fences are often constructed
with minimal specicity in terms of their role to conservation, with
short-term functions prioritized over the long-term outcomes (Sun et al.,
2020). With roughly one third of sage-grouse nests occurring on private
lands in this study alone, public-private community-based collabora-
tions centered around fence design and construction, and where
appropriate –modications –are necessary to mitigate the negative
effects already documented of fences on wildlife. We strongly encourage
collaborations of this type that could expand this management tactic
across large sagebrush landscapes, thereby improving a key vital rate
driving sage-grouse populations. Fences are continuing to proliferate
across the American West. There is currently over an estimated 1 million
kilometers of fence already present (McInturff et al., 2020), highlighting
the potential for population level impacts from both the well-studied
direct and emerging indirect effects of fences. The relationship be-
tween sage-grouse populations and fences should be a current research
priority of utmost urgency. A powerful conservation recipe to improve
sage-grouse nest survival could be a duel management approach that
directly and indirectly reduces the effect of fences on sage-grouse nest
survival through fence marking and modifying fences as described in the
current study. Novel yet simple fence modications like these should
make this management action achievable across various stakeholders,
given fences are the primary technique for managing the
spatio-temporal patterns of livestock in agroecosystems.
A.W. O’Harra et al. Agriculture, Ecosystems and Environment 377 (2025) 109248
6
CRediT authorship contribution statement
Aaron W. O′Harra: Writing –review &editing, Writing –original
draft, Visualization, Validation, Supervision, Software, Resources,
Project administration, Methodology, Investigation, Formal analysis,
Data curation, Conceptualization. Michael R. Frisina: Writing –review
&editing, Supervision, Methodology, Investigation, Funding acquisi-
tion. Kyle A. Cutting: Writing –review &editing, Validation, Super-
vision, Software, Resources, Project administration, Methodology,
Investigation, Funding acquisition, Formal analysis, Data curation,
Conceptualization. Bok F. Sowell: Writing –review &editing, Writing –
original draft, Visualization, Validation, Supervision, Software, Re-
sources, Project administration, Methodology, Investigation, Funding
acquisition, Formal analysis, Data curation, Conceptualization.
CRediT authorship contribution statement
A.W.O., K.A.C., and B.F.S. conceived the idea, study design, and data
collection. A.W.O. and K.A.C. performed the eld study. A.W.O. and K.
A.C. analyzed the data. A.W.O. and K.A.C. with critical inputs from all
co-authors wrote the paper.
Declaration of Generative AI and AI-assisted technologies in the
writing process
The authors have declared no use of AI related technologies.
Declaration of Competing Interest
The authors declare no known competing nancial interests or per-
sonal relationships that could appear to be a conict of interest to in-
uence the ndings in this paper.
Data availability
Data will be made available on request.
Acknowledgements
We thank many public land managers and private landowners who
granted access to their land for this study. We thank J. Waxe, A. and H.
Anderson, E. Nunlist, T. Martin, B. Sparklin, K. Doherty, M. Bryant, B.
West, and K. Benzel for logistical support and various contributions
throughout the study. R. Levine and J.A. Dullum assisted with database
design and support, A. Litt for advice on study design, J. Tack for
assistance with statistical analyses, and comments provided by two
anonymous reviewers greatly improved the study. This project could not
have been possible without the dedication of eld crews, including: E.
Grusing, D. Driscoll-Roach, E. Fullerton, A. Gilpin, A. Hardigg, M.
Hockett, L. Landau, M. Odden, P. Platis, and J. Rees. Findings and
conclusions in this article are those of the authors and do not necessarily
represent the views of the U.S. Fish and Wildlife Service. This research
was funded by the Western Sustainable Agriculture Research and Edu-
cation (WSARE; Project #SW13-056) program, Montana Agricultural
Experiment Station, Hormay Institute, Bureau of Land Management, U.
S. Forest Service, Montana Department of Natural Resources and Con-
servation, U.S. Fish and Wildlife Service Inventory and Monitoring
Program, The Nature Conservancy –Montana, U.S. Fish and Wildlife
Service Science Applications, Red Rock Lakes National Wildlife Refuge,
and Montana State University Animal and Range Sciences.
References
Andersson, M., Wallander, J., Isaksson, D., 2009. Predator perches: a visual search
perspective. Funct. Ecol. 23, 373–379.
Atwood, T.C., Weeks, H.P., Gehring, T.M., 2004. Spatial ecology of coyotes along a
suburban-to-rural gradient. J. Wild. Manag. 68, 1000–1009.
Bailey, D.W., 2004. Management strategies for optimal grazing distribution and use of
arid rangelands. J. Anim. Sci. 82, 147–153.
Braun, C.E., Schroeder, M.A., 2015. Age and sex identication from wings of sage-grouse.
Wild. Soc. Bull. 39, 182–187.
Briske, D.D., Derner, J.D., Milchunas, D.G., Tate, K.W., 2011. An evidence-based
assessment of prescribed grazing practices. Conserv. Benets Rangel. Pract.: Assess.
Recomm. Knowl. Gaps 21–74.
Coates, P.S., Brussee, B.E., Howe, K.B., Gustafson, K.B., Casazza, M.L., Delehanty, D.J.,
2016. Landscape characteristics and livestock presence inuence common ravens:
relevance to greater sage-grouse conservation. Ecosphere 7, e01203.
Conover, M.R., Roberts, A.J., 2017. Predators, predator removal, and sage-grouse: a
review. J. Wild. Manag. 81, 7–15.
Converse, S.J., Royle, A., Adler, P.H., Urbanek, R.P., Barzen, J.A., 2013. A hierarchical
nest survival model integrating incomplete temporally varying covariates. Ecol.
Evol. 3, 4439–4447.
Cutting, K.A., Rotella, J.J., Schroff, S.R., Frisina, M.R., Waxe, J.A., Nunlist, E., Sowell, B.
F., 2019. Maladaptive nest-site selection by a sagebrush dependent species in a
grazing-modied landscape. J. Environ. Manag. 236, 622–630.
Donovan, T.M., Thompson, F.R., 2001. Modeling the ecological trap hypothesis: a habitat
and demographic analysis for migrant songbirds. Ecol. Appli. 11, 871–882.
EROS [USGS Earth Resources Observation and Science Center]. 2016. “The National
Map.”
Fischer, R.A., Apa, A.D., Wakkinen, W.L., Reese, K.P., Connelly, J.W., 1993. Nesting-area
delity of sage grouse in southeastern Idaho. Condor 95, 1038–1041.
Fleming, P.A., Bateman, P.W., 2018. Novel Predation Opportunities in Anthropogenic
Landscapes. Anim. Behav. 138, 145–155.
Gese, E.M., Ruff, R.L., Crabtree, R.L., 1996. Foraging Ecology of Coyotes (Canis latrans):
The Inuence of Extrinsic Factors and a Dominance Hierarchy. Canad. J. Zool. 74,
769–783.
Harju, S.M., Olson, C.V., Hess, J.E., Bedrosian, B., 2018. Common raven movement and
space use: inuence of anthropogenic subsidies within greater sage-grouse nesting
habitat. Ecosphere 9, e02348.
Harrington, J.L., Conover, M.R., 2006. Characteristics of ungulate behavior and
mortality associated with wire fences. Wild. Soci. Bull. 34, 1295–1305.
Holloran, M.J., Anderson, S.H., 2005. Spatial distribution of greater sage-grouse nests in
relatively contiguous sagebrush habitats. Condor 107, 742–752.
Hovick, T.J., Elmore, D., Dahlgren, D.K., Fuhlendorf, S.D., Engle, D.M., 2014. Evidence of
Negative Effects of Anthropogenic Structures on Wildlife: a Review of Grouse
Survival and Behaviour. J. Appl. Ecol. Blackwell Publishing Ltd.
Jakes, A.F., Jones, P.F., Christine Paige, L., Seidler, R.G., Huijser, M.P., 2018. A fence
runs through it: a call for greater attention to the inuence of fences on wildlife and
ecosystems. Biol. Conserv. 227, 310–318.
Jones, P.F., 2014. Scarred for life: the other side of the fence debate. Hum. -Wildl.
Interact. 8, 150–154.
Jones, P.F., Jakes, A.F., Telander, A.C., Sawyer, H., Martin, B.H., Hebblewhite, M., 2019.
Fences reduce habitat for a partially migratory ungulate in the Northern Sagebrush
Steppe. Ecosphere 10 (7), e02782.
Klett, A.T., Johnson, D.H., 1982. Variability in nest survival rates and implications to
nesting studies. Auk 99, 77–87.
Kristov, V., 2008. Indirect effects in ecology. In: Sven Erik Jorgensen (Ed.), Ecosystem
Ecology. Elsevier B.V, pp. 81–91.
Levin, S.A., 1992. The problem of pattern and scale in ecology. Ecol 73 (6), 1943–1967.
Link, W.A., Barker, R.J., 2006. Model weights and the foundations of multimodel
inference. Ecol 87, 2626–2635.
McInturff, A., Xu, W., Wilkinson, C.E., Dejid, N., Brashares, J.S., 2020. Fence ecology:
frameworks for understanding the ecological effects of Fences. BioSci 70, 971–985.
Moynahan, B.J., Lindberg, M.S., Rotella, J.J., Thomas, J.W., 2007. Factors Affecting nest
survival of greater Sage-Grouse in Northcentral Montana. J. Wild. Manag. 71,
1773–1783.
O’Harra, A.W., 2023. Greater Sage-grouse (Centrocercus urophasianus) Nest Survival and
Nest Predator Response to Fence Modications in a Grazing Modied Landscape.
Master’s thesis. Montana State University, Bozeman, USA.
Okada, S., Lindenmayer, D.B., Wood, J.T., 2019. Does Land Use Change Inuence
Predation of Bird Nests? Aust. Ecol. 44, 768–776.
Pedlar, J.H., Fahrig, L., Merriam, H.G., 1997. Raccoon habitat use at 2 spatial scales.
J. Wild. Manag. 102–112.
Rodriguez, J.T., Lesmeister, D.B., Levi, T., 2021. Mesocarnivore landscape use along a
gradient of urban, rural, and forest cover. PeerJ 9, e11083.
Rowland, M.M., Wisdom, M.J., Suring, L.H., Meinke, C.W., 2006. Greater sage-grouse as
an umbrella species for sagebrush-associated vertebrates. Biol. Conserv. 129,
323–335.
Schmidt, J.H., Walker, J.A., Lindberg, M.S., Johnson, D.S., Stephens, S.E., 2010.
A General Bayesian hierarchical model for estimating survival of nests and young.
Auk 127, 379–386.
Smith, J.T., Allred, B.W., Boyd, C.S., Carlson, J.C., Davies, K.W., Hagen, C.A., Naugle, D.
E., Olsen, A.C., Tack, J.D., 2020. Are sage-grouse ne-scale specialists or shrub-
steppe generalists? J. Wild. Manag. 84, 759–774.
Smith, J.T., Tack, J.D., Doherty, K.E., Allred, B.W., Maestas, J.D., Berkeley, L.I.,
Dettenmaier, S.J., Messmer, T.A., Naugle, D.E., 2018. Phenology largely explains
taller grass at successful nests in greater sage-grouse. Ecol. Evol. 8, 356–364.
Stevens, B.S., Connelly, J.W., Reese, K.P., 2012. Multi-scale assessment of greater sage-
grouse fence collision as a function of site and broad scale factors. J. Wild Manag. 76,
1370–1380.
Su, Y.S., Yajima, M., 2015. R2jags: Using R to Run ‘JAGS’. R. Package Version 0, 5–7.
A.W. O’Harra et al. Agriculture, Ecosystems and Environment 377 (2025) 109248
7
Sun, J., Liu, M., Fu, B., Kemp, D.R., Zhao, W., Liu, G., Han, G., et al., 2020. Reconsidering
the efciency of grazing exclusion using fences on the Tibetan Plateau. Sci. Bull. 65,
1405–1414.
Van Lanen, N.J., Green, A.W., Gorman, T.R., Quattrini, L.A., Pavlacky, D.C., 2017.
Evaluating efcacy of fence markers in reducing greater sage-grouse collisions with
fencing. Biol. Cons. 213, 70–83.
Vold, S.T., 2018. Effects of Livestock Grazing Management on the Ecology of Grassland
Birds and their Predators in a Northern Mixed-grass Prairie Ecosystem. Master’s
thesis. Montana State University, Bozeman, USA.
Wegge, P., Storaas, T., 1990. Nest loss in capercaillie and black grouse in relation to the
small rodent cycle in southeast Norway. Oecol 82, 527–530.
A.W. O’Harra et al. Agriculture, Ecosystems and Environment 377 (2025) 109248
8