Hibernation behavior of a federally threatened ground squirrel: climate change and habitat selection implications
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Journal of Mammalogy, 102(2):574–587, 2021
DOI:10.1093/jmammal/gyab021
Published online April 13, 2021
Hibernation behavior of a federally threatened ground squirrel:
climate change and habitat selection implications
Amanda R. Goldberg*, and Courtney J. Conway
Idaho Cooperative Fish and Wildlife Research Unit, Department of Fish & Wildlife Sciences, University of Idaho, 875 Perimeter
Drive, MS 1141, Moscow, ID 83844, USA (ARG)
U.S. Geological Survey, Idaho Cooperative Fish and Wildlife Research Unit, University of Idaho, 875 Perimeter Drive, MS 1141,
Downloaded from https://academic.oup.com/jmammal/article/102/2/574/6224539 by guest on 14 October 2021
Moscow, ID 83844, USA (CJC)
*Correspondent: goldberg.amandar@gmail.com
Hibernation is an adaptation to survive periods of stress, from food limitation or harsh thermal conditions.
A key question in contemporary ecology is whether rare, range-restricted species can change their behavior in
response to climate change (i.e., through behavioral plasticity). The northern Idaho ground squirrel, Urocitellus
brunneus (A. H. Howell, 1928), is a federally threatened species that hibernates for approximately 8 months per
year within the bounds of its small range in central Idaho, USA. Changes in temperature, snow accumulation,
and summer precipitation, all brought about as a result of climate change, may reduce survival or fecundity
of northern Idaho ground squirrels if they cannot adapt to these climate changes. Hibernating species can
respond to climate-change-induced thermal challenges in two ways: change their hibernation physiology and
behavior (i.e., emergence date or number of torpor bouts) or alter their environment (i.e., change hibernacula
depth or location). We explored a suite of intrinsic and extrinsic factors to document the extent to which they
influenced hibernation behavior of northern Idaho ground squirrels. Emergence date was positively associated
with snowpack and negatively associated with mean winter temperature. Mean minimum skin temperature was
negatively associated with canopy closure and slope of a squirrel’s hibernaculum. Duration of the heterothermal
period, number of euthermic bouts, and total time spent euthermic were positively associated with body mass.
Immergence date and duration of the longest torpor bout were negatively associated with body mass. Warmer
temperatures and less snow accumulation in the winter—caused by climate change—likely will cause altered
emergence dates. Our results suggest that any future climate-induced changes in snowfall, ambient temperature,
food availability, or habitat likely will impact survival of this rare ground squirrel, because such changes will
cause changes in hibernation behavior, percent mass loss during hibernation, and duration of the active season
when small mammals are more susceptible to predation.
Key words: behavioral plasticity, behavioral physiology, climate vulnerability, estivation, geolocators, habitat use, heterothermy,
phenological shifts, phenotypic plasticity, temperature loggers
Climate change is likely to impact plants and animals at many (2) temporally or phenologically (e.g., change timing of mi-
scales: individuals, populations, communities, and ecosystems. gration, daily activity, or reproduction to better match peak in
Animals can respond to changes in climate either through mi- resources or thermal tolerance zones—Humphries et al. 2004;
croevolution or behavioral plasticity (Parmesan 2006; Hoffmann Bellard et al. 2012). We need to better understand the ability of
and Sgrò 2011; Bellard et al. 2012). Species not capable of re- imperiled animals to adjust to climate change via either or both
sponding appropriately to changes in climate to meet their bi- of these mechanisms, because the frequency and magnitude
oenergetic or thermal needs may ultimately face extinction of extreme climatic conditions are projected to increase (Mote
(McCain and King 2014). Animals can potentially mitigate the et al. 2003; Klos et al. 2014; Lute et al. 2015; Gergel et al. 2017).
impacts of climate change via one of two behavioral responses: Many mammals use one of four strategies to survive chal-
(1) spatially (e.g., move to follow appropriate conditions); or lenging winter climate conditions: (1) migration to areas with
Published by Oxford University Press on behalf of American Society of Mammalogists 2021.
This work is written by (a) US Government employee(s) and is in the public domain in the US.
574GOLDBERG AND CONWAY—GROUND SQUIRREL HIBERNATION BEHAVIOR 575
more favorable conditions (Telfer and Kelsall 1996; Avgar should be maximized but periodic arousals from torpor will
et al. 2014); (2) behavioral changes (e.g., hoard food caches take more energy than during the summer torpor bouts—Wang
and increase body fat or huddle together—Merritt 1986; Post 1978), resulting in multiseason hibernation. Northern Idaho
et al. 2006); (3) morphological adaptations (e.g., grow thick ground squirrels are solo hibernators, hence do not share body
fur to stay warm during cold winters—Russell and Tumlison heat with other conspecifics, as is the case in most marmot spe-
1996); or (4) hibernation (Ruf et al. 2012). Western North cies (Patil et al. 2013).
America is expected to have warmer summers, earlier spring Juvenile survival and yearling female breeding propensity
snowmelt, and less snow accumulations (i.e., shallower snow of northern Idaho ground squirrels were lower than those of
depth), all of which could impact the relative effectiveness congeners, based on data from a 13-year study (Sherman and
of these four strategies if mammals do not adapt accord- Runge 2002). Sherman and Runge (2002) hypothesized that the
ingly (Stewart 2009). Mammals that hibernate must select or population they studied declined because animals were unable
Downloaded from https://academic.oup.com/jmammal/article/102/2/574/6224539 by guest on 14 October 2021
create a hibernaculum within the desired temperature range or to enter hibernation with sufficient body reserves, which re-
change their physiology to accommodate climate change ef- duced the ability for squirrels to survive hibernation. However,
fects. Some hibernators enter into hibernation for a large part squirrels could potentially mitigate the effects of changes in
of a year, which may include both warmer temperatures and nutrient availability by altering their hibernation behavior in
below-freezing temperatures. Hibernation may begin during ways that allow them to save energy and thereby survive the
the warmest part of the year and squirrels remain in hiberna- long hibernation period. The survival probability of northern
tion throughout the winter to avoid cold and food stress (Wang Idaho ground squirrels during the approximately 8-month
1978; Wilz and Heldmaier 2000; Staples 2016). Animals that heterothermal period (multiseason hibernation) is potentially
select a hibernaculum for long term (across multiple seasons) influenced by: (1) the availability of suitable hibernacula loca-
may be especially sensitive to changes in habitat suitability re- tions; (2) the squirrels’ ability to dig a burrow with the optimal
lated to climate conditions (e.g., earlier snowmelt, less snow, microclimate to allow them to meet their physiological needs;
and/or warmer summer temperatures). and (3) the squirrels’ ability to alter their behavior or metab-
We know relatively little about the plasticity of behaviors re- olism during suboptimal environmental conditions. Climate-
lated to hibernation and the capacity of hibernators to respond induced changes in environmental conditions over the past few
to climate change. However, there have been few studies on this decades may have changed the likelihood of one or more of
topic and they have reported both positive (Ozgul et al. 2010) these three requirements for survival.
and negative (Inouye et al. 2000; Lane et al. 2012; Sherwin This rare ground squirrel also faces new challenges because
et al. 2013; Tafani et al. 2013; Cordes et al. 2020) impacts the landscape that supports northern Idaho ground squirrels has
of climate change on survival and or juvenile recruitment. changed over the past few decades (Burak et al. 2018). Fire sup-
Notwithstanding, it does appear that these species are changing pression has allowed coniferous trees to encroach into some of
their behavior in response to a changing climate—although the the forest openings that support northern Idaho ground squirrels.
overall effect of these changes differs. For instance, Inouye Canopy cover of overstory trees reduces snow depth but increases
et al. (2000) found that yellow-bellied marmots (Marmota the seasonal duration of snowpack (Davis et al. 1997; D’Eon
flaviventris) emerged earlier in response to warmer spring 2004; Varhola et al. 2010). Snow cover has an insulative effect
temperatures, resulting in a mismatch of snowmelt and access and likely helps maintain a constant burrow temperature within
to food resources. Columbian ground squirrels (Urocitellus hibernacula (Svendsen 1974; Young 1990a; Tafani et al. 2013).
columbianus) emerged later in years with late season snow fall Thus, snow cover likely influences optimal hibernacula depth.
but this led to a shortened active season and reduced survival As snow depth increases, the depth of frost in the soil column de-
(Lane et al. 2012). creases (Hardy et al. 2001; Decker et al. 2003), and hibernating
Choice of the hibernaculum may have enormous conse- squirrels should be better able to maintain constant body tem-
quences given that a hibernator’s fitness is influenced by the peratures above freezing for longer periods. Hence, temperature
conditions of that hibernaculum for a large portion of its annual of the hibernaculum, mediated by snowpack and hibernaculum
cycle. In addition, the optimal hibernaculum under one set of depth, may influence overwinter survival of northern Idaho
climatic conditions may not be the optimal hibernaculum under ground squirrels. Furthermore, canopy cover provides shade, re-
a different set of conditions (such as animals that undergo ducing direct effects of the sun on soil temperatures, resulting in
multiseason hibernation). For example, northern Idaho ground cooler soil temperatures with less temperature fluctuation prior
squirrels, Urocitellus brunneus, are a federally threatened spe- to snow cover (Breshears et al. 1998; Royer et al. 2012). Snow
cies and all individuals hibernate approximately 8 months per depth and duration of winter snowpack therefore may affect
year, some beginning as early as July and emerging the fol- northern Idaho ground squirrel survival by several mechanisms:
lowing March (Yensen and Sherman 1997; Sherman and Runge (1) snow's insulating quality during hibernation; (2) deep snow
2002). Thus, individuals begin torpor bouts during the warm may afford protection from terrestrial predators; and (3) the ef-
summer months (when body temperature will remain higher fect of snow on the date of vegetation green-up. The latter may
and total energy savings should be less than in the winter, i.e., affect reproduction and survival during the subsequent year.
more euthermic bouts—Geiser and Kenagy 1988; Hoelzl et al. To explore these relationships, we measured daily snow
2015) and continue throughout the winter (when energy savings depth, ambient air temperature, and light intensity, at hibernacula576 JOURNAL OF MAMMALOGY
locations of northern Idaho ground squirrels from 2016 through 205 g ( x̄ = 155.7 g) for females and 127 to 262 g ( x̄ = 196.0 g)
2019 and examined the relationships among hibernation behavior, for males. The weight of the collar therefore represented less
weather, and hibernacula habitat. Many climate change effects are than 3% of a squirrel’s body mass. We attached collars to 56
gradual, so examining their consequences on ground squirrels is females (seven in 2015, 16 in 2016, 19 in 2017, and 14 in
difficult in field studies. We therefore used space-for-time substi- 2018) and 43 males (three in 2015, 12 in 2016, 14 in 2017,
tution (Blois et al. 2013) in snow conditions (brought on by var- and 14 in 2018). We deployed two models of light loggers: five
iation in elevation) to evaluate the potential influence of climate Intigeo C65 in 2015 (older model) and 94 Intigeo F100. The
change on northern Idaho ground squirrel hibernation behavior Intigeo C65 models had ± 3.0°C accuracy and the newer F100
and survival. To examine the current relationship between abiotic models had ± 0.5°C accuracy. We used telemetry receivers and
and biotic conditions on hibernation behavior and hibernaculum handheld antennas to locate hibernacula and retrieve dropped
depth and inform the potential influence of climate change on collars. We deployed collars only at locations that supported
Downloaded from https://academic.oup.com/jmammal/article/102/2/574/6224539 by guest on 14 October 2021
northern Idaho ground squirrels, we addressed the following ex- ≥ 10 adult northern Idaho ground squirrels and we collared <
plicit questions: (1) are hibernation behaviors (torpor and arousal 33% of the adults at any one site (restrictions established by
bouts, duration of hibernation, and immergence/emergence dates) an interagency technical team to minimize potential negative
affected by weather, body condition, or habitat features? And (2) effects that collars may have on squirrel populations). Light
what is the relationship between hibernaculum depth and weather, loggers recorded both the light level (which allowed us to
body condition, and other habitat features? document when a northern Idaho ground squirrel immerged
and emerged from hibernation) and the skin temperature (Tsk)
at 15-min intervals (4-h intervals on the older C65 models).
Materials and Methods Skin temperature measurements during hibernation should be
Study sites.—We carried out field work at seven sites in similar to body temperature because light loggers produced
Adams County, Idaho, from 2015 to 2019. The seven sites were temperature readings during hibernation that were similar to
on both public and private land and ranged in elevation from those from implanted thermochrons in Arctic ground squirrels
1,200 to 1,700 m. During the spring and summer active season, (Urocitellus parryii—Williams et al. 2016). We used VHF col-
northern Idaho ground squirrels are associated with south- lars to track squirrels to their hibernacula. The Biotrack collars
facing aspects (U.S. Fish and Wildlife Service 2003) and open- were designed to emit a signal on a preselected schedule (we
canopy habitat (meadows and rocky scabs) patchily distributed programmed the collars to emit a signal 2 or 3 days per week
within a forest mosaic (Burak 2011). Active-season habitat con- to maximize the battery life). We also set the Biotrack collars
tains a wide variety of grasses and forbs with a few patches of to turn off during the winter, so they had enough power to turn
shrubs and trees (Goldberg et al. 2020b). The surrounding for- back on in the spring and still emit signals for 2–3 months,
ests (where northern Idaho ground squirrels often hibernate— thereby allowing us to locate and catch squirrels that retained
Goldberg et al. 2020a) are dominated by ponderosa pine (Pinus their collar over winter or to retrieve the collar for those that
ponderosa) and Douglas-fir (Pseudotsuga menziesii). had slipped from squirrels’ necks in the spring.
Animal handling and telemetry.—We trapped and collared Over 63% of the collared northern Idaho ground squirrels that
northern Idaho ground squirrels between 2 June and 19 July we recaptured in the spring had dropped their collars. If a squirrel
each year. We attached ear tags to both ears on each squirrel had slipped its collar and we could hear the signal of the collar in
for individual identification and recorded body mass, sex, and the spring (Biotrack collars only), we were able to: (1) dig collars
reproductive condition of each squirrel. We considered all fe- up (we recovered from underground 85% of the collars we know
males to be reproductively active if they had visual nipples were dropped); (2) recover collars on the ground if it was slipped
which are only visible above the fur when they are nursing or preyed upon in the spring after emergence (we recovered 15%
young (Kenagy et al. 1989b). All procedures were approved of the dropped collars aboveground); (3) measure the depth at
by the Institutional Animal Care and Use Committee at the which the collar was located in the hibernaculum (if the collar
University of Idaho and were in compliance with American was belowground and at the location of the hibernaculum); and
Society of Mammalogists guidelines (Sikes et al. 2016). (4) recover the light logger. We recaptured 19 squirrels the sub-
We deployed very high frequency (VHF) radiocollars (73 sequent spring that retained their collars throughout hibernation
Model CTx AG392, Biotrack [Biotrack Ltd., Wareham, United (both Holohil and Biotrack; 37% of all collared squirrels trapped
Kingdom] and 26 Holohil Model BD-2C [Holohil Systems the following spring retained their collars).
Ltd., Carp, Ontario, Canada]) and light loggers (geolocators; Hibernacula habitat characteristics.—We collected four
Migrate Technology Ltd., Cambridge, United Kingdom) on habitat measurements at each hibernaculum location to docu-
99 northern Idaho ground squirrels. We used a combination of ment the habitat conditions used by hibernating northern Idaho
epoxy and Amazing GOOP (Eclectic Products LLC, Eugene, ground squirrels. We measured: (1) slope; (2) percent canopy
Oregon) to attach the light loggers to the VHF collars. We ini- closure (measured using a convex spherical densiometer); (3)
tially selected squirrels for collaring that weighed > 120 g but diameter at breast height (DBH) of the closest tree (≥ 2 cm);
increased the minimum weight to 140 g beginning in 2016 due and (4) presence/absence of a log or a branch (any dead woody
to concerns about animal safety. In total, the collars weighed plant material on the ground; hereafter referred to as a log) ≥
~3.4 g. Body mass of collared squirrel’s ranged from 120 to 2 cm diameter within 5 m.GOLDBERG AND CONWAY—GROUND SQUIRREL HIBERNATION BEHAVIOR 577
We placed a graduated staff gauge and a trail camera (Bushnell two trapping sessions each summer with ~30 days between
Trophy Cam 119636C, Bushnell Corporation, Overland Park, the two sessions. The rate of seasonal increase in body mass
Kansas) at hibernacula each fall (we did not place snow gauges should differ between males and females due to changes in
at two hibernacula due to a limited number of snow gauges) mass brought on by pregnancy and lactation. We therefore re-
from 2016 to 2018 to document depth of winter snowpack. gressed adult body mass against capture date for each sex sep-
Gauges were placed ~2 m from the hibernating squirrel be- arately and also for each of the two trapping sessions (spring
cause we did not want to disturb them by placing them on top and summer). We also expected body mass to differ among
and driving stakes into the ground. We preprogrammed cam- our seven study sites (because they differed in elevation), so
eras to photograph the gauge two times per day throughout we included site in our regression models. These body mass
the winter. We estimated the snow depth daily and the dura- regressions were calculated using a generalized linear model
tion of snow cover throughout the winter at those six study in program R (v3.6.1—R Development Core Team 2017). We
Downloaded from https://academic.oup.com/jmammal/article/102/2/574/6224539 by guest on 14 October 2021
sites based on the photographs of the staff gauges. Cameras recorded the residual body mass from those regressions as a
occasionally died or failed due to disturbance (human or wild- seasonally adjusted metric of relative body mass for each adult
life), battery failure, or other malfunction (19.2% failure rate). squirrel. We combined data from all years for these regressions
Furthermore, snow occasionally was deeper than the height of because we had too few data for any 1 year at most of the seven
the snow gauge; maximum depth of snow was capped at 1.2 m study sites.
(the height of our shortest poles). Hibernation behavior analysis.—We used generalized
We placed two soil temperature probes in the ground at study linear models to evaluate the relationships between our
sites: (1) one within an open-canopy area (meadow) where 15 explanatory variables and each of seven hibernation be-
squirrels are active during the summer (near a previously used haviors: (1) immergence date; (2) emergence date; (3) total
summer burrow location); and (2) one in the adjacent forest length of the heterothermal period; (4) number of euthermic
(near a hibernaculum location from a previous year). Each soil bouts during the heterothermal period; (5) total sum of the
probe consisted of one 1.5-m long, 1.9-cm wide PVC pipe with time spent euthermic; (6) longest torpor bout duration; and
3–11 thermochron temperature loggers (Maximum Integrated (7) mean minimum skin temperature (Tsk). We did not eval-
Products, San Jose, California, USA) spaced at 15-cm intervals uate all combinations of variables but rather used a common
and separated by foam. Temperature probes were inserted into approach for model comparison and selection (Burnham and
the ground by digging a 1-inch-wide hole with a hand-auger. Anderson 2002; Grueber et al. 2011) whereby we selected a
Due to the high density of rocks and tree roots, we were un- priori those variables that were most likely to impact each hi-
able to dig a 1.5-m deep hole at all but one location (7% of all bernation behavior based on what we know about hibernation
probes). We cut the PVC pipe to fit the depth to which we were and behavior of ground squirrels (Table 1). However, some
able to dig at each location. As a result, soil temperature probes squirrels had missing values for ≥ 1 variable due to equipment
varied in maximum depth (and number of thermochrons) from failure, human error, or slipped collars. The number of squir-
0.3 to 1.5 m. We deployed the soil temperature probes in June rels for which we had missing values varied from zero to eight
2016 and downloaded the temperature data from the soil probes (of 36 total) for the 15 explanatory variables and varied from
after we retrieved the thermochrons the following spring. zero to 10 for the seven response variables. So that we could
We redeployed the temperature probes in the same locations use all 36 squirrels in our analyses, we used multiple imputa-
each spring. tion procedures to account for missing values. Specifically, we
We deployed Hobo temperature loggers (Onset Computer used predictive mean matching (PMM) implemented in the
Corporation, Bourne, Massachusetts) to measure light inten- mice package (Van Buuren and Groothuis-Oudshoorn 2011)
sity and air temperature during the active season: one in the in program R. The mice package uses conditional multiple
open-canopy area and one in the forest (under tree canopy). imputation, which is an iterative procedure that models the
All the Hobo loggers were placed 1–1.75 m above the ground. conditional distribution of a certain variable given the other
Loggers in the forest were placed on a ponderosa pine branch variables. Multiple imputation, including for both response
close to the trunk to reduce impact of light on air temperature. and explanatory variables, produces values that generally re-
Over the winter, we deployed the Hobo temperature loggers flect the true patterns in the data (Lang and Little 2018). One
at the hibernacula locations after the squirrel had entered into of the main advantages to PMM is that it is considered ro-
hibernation and at a snow gauge in the open-canopy areas. bust and able to preserve the distribution of the original data
Temperature and light were recorded every 30 min to 2 h well (Kleinke 2018). We assumed our data were missing at
(depending on the logger’s storage capacity). random (MAR). We implemented a two-step process (von
Animal mass.—We weighed all squirrels that we captured Hippel 2020) to calculate the number of imputations to use in
from 2013 to 2018 (we had body mass measurements from an R with the howManyImputations package (https://github.com/
ongoing companion study at the same study sites in years prior josherrickson/howManyImputations). As a result, we created
to collaring squirrels). Body mass increases across the summer 47 imputed data sets that we ran separately for each of the
active season, so we sought to standardize our body mass meas- seven response variables. We imputed both response and ex-
urements given that we caught animals throughout the 3-month planatory variables; this approach is unbiased when one uses
summer active season (May–July). We trapped squirrels during > 1 explanatory variable and many imputed data sets, as we did578 JOURNAL OF MAMMALOGY
Table 1.—Predicted relationship between each of 15 explanatory variables and seven response variables describing northern Idaho ground
squirrel hibernation behavior. If a cell is blank, we did not believe there is a relationship. “+” = predicted positive relationship, “−” = predicted
negative relationship, and “±” = predicted relationship may be positive or negative. Tsk refers to skin temperature and Ta refers to air temperature.
Heterothermal Immergence Emergence Number of Total euthermic Max torpor Mean
period duration euthermic bouts duration duration min Tsk
Age ± + − + + + −
Canopy closure + ± + + + + −
Day soil increaseda + −
DBH of closest tree + ± − + + + −
Elevation + + + + + + −
Hours aboveground −
pre-immergenceb
Downloaded from https://academic.oup.com/jmammal/article/102/2/574/6224539 by guest on 14 October 2021
Log present (Y)c + ± − + + + −
Mean min Tsk − − −
Mean winter Ta − − + + − +
Min Tsk − − −
Residual body mass + − ± + + − −
Sex (M)d − − − − − − −
Slope + ± + + + + −
Snowmelt datee + + + + + −
Ta before immergence −
a
First date that Ts at 30 cm belowground (at the site-specific logger in the forest or meadow depending on location of hibernacula) began to steadily increase in the
spring.
b
Mean hours a squirrel was aboveground (> 300 lux) per day for the 5 days prior to immergence.
c
The presence of a log (Y) is the reference level.
d
Males (M) is the reference level.
e
Last day (Julian date) that there was < 2 cm of continuous snow on the ground.
here, and by accounting for any potentially nonlinearity in the
data by imputing by means of PMM (Young and Johnson 2010;
Kontopantelis et al. 2017; Lang and Little 2018; Van Ginkel
et al. 2020). We used methods described in Rubin (1987) to
pool all point and variance estimates for each response vari-
able based on our final model. We used the MAMI package
(Schomaker and Heumann 2017) in program R to imple-
ment model selection. MAMI carries out model selection and
model averaging on multiple imputed data sets and combines
the resulting estimates. We used a Gaussian distribution and
MA.criterion with AIC selection which uses stepAIC selection
from the MASS package (Schomaker and Heumman 2017).
In addition, we evaluated all the univariate relationships
between the 15 explanatory variables and the seven response
variables for a total of 77 univariate tests. We carried out uni-
variate analysis on hibernacula depth (we did not impute the
explanatory covariates like the other seven response variables
because we were missing depth measurements for > 50% of
the 36 squirrels). We also used univariate analyses to evaluate
the effects of pregnancy on the response variables; only fe- Fig. 1.—Example of the skin temperature of a hibernating northern
Idaho ground squirrel. The line represents the squirrel’s skin temper-
males were included in this analysis, reducing our data set to
ature over a 2.5-week interval during torpor. A torpor bout begins at
18 individuals. point “a” and ends at point “b.” A euthermic bout begins at point “b”
We assumed that a squirrel had entered its hibernaculum and ends at point “c.” We examined data from 36 squirrels for which
when all aboveground activity (light readings) ceased for more we recovered light loggers and identified the date and time when each
than 1 day (based on data from the light loggers). We assumed torpor bout ended and each euthermic bout began for the entire period
that a squirrel had entered the heterothermal period when the that each squirrel was belowground during hibernation.
first torpor bout was greater than 24 h. We assumed a torpor
bout had begun when a squirrel’s Tsk began to decrease and we
assumed a euthermic bout had begun when a squirrel’s Tsk began number of euthermic bouts were only those that occurred during
to increase steadily (Fig. 1). We assumed that the heterothermal the heterothermal period. We assumed a squirrel had emerged
period had ended when all subsequent torpor bouts were less when the light logger read 300 lux or greater. We used the 300-
than 24 h long. Total torpor length, euthermic bout length, and lux threshold to account for the fact that squirrels may pokeGOLDBERG AND CONWAY—GROUND SQUIRREL HIBERNATION BEHAVIOR 579
their head up and resume hibernation in the early spring and we ground squirrels, three of which were collared during two win-
did not want to dismiss any final torpor bouts. ters, thus a total of 33 individuals for 36 hibernation events.
We knew the exact age of 14 squirrels because we had trapped These data allowed us to estimate ≥ 1 of the hibernation behav-
them in a prior year as juveniles. We were confident that we could iors for these 36 squirrels (we could not document one or more
consistently discern the difference between juveniles (young hibernation traits for 10 of the 36 squirrels). We identified one
born that year) and adults based on visual inspection of their 2-year-old male in 2018 that briefly emerged from its burrow
size (mean mass of a juvenile = 110.0 g and mean mass of an and then entered back into torpor that lasted > 24 h. No other
adult = 168.5 g in the summer). We could not readily discern the individual squirrel exhibited this behavior (i.e., emerged and
difference between yearlings (those aged between 1 and 2 years then reentered torpor). Four other northern Idaho ground squir-
old) and those that were > 2 years old. However, if a squirrel was rels (three males and one female) emerged briefly and did not
initially trapped as an adult and subsequently trapped again the come aboveground again for 1.05–2.14 days. However, none of
Downloaded from https://academic.oup.com/jmammal/article/102/2/574/6224539 by guest on 14 October 2021
following year, we knew it was at least 2 years old and we used these four individuals resumed torpor bouts that lasted > 24 h
2.5 as its age in our analyses. Similarly, if we trapped a squirrel after their initial aboveground activity.
as an adult and collared it 2 years later, we knew it was at least We did not detect a difference in the number of days spent
3 years old and we used 3.5 as its age in our analyses. This ap- in their hibernacula between immergence and the start of the
proach allowed us to assign an age to as many squirrels as pos- heterothermal period between females and males (Table 2).
sible and to include age as a continuous variable in our models. Females emerged later than males after the snow melted from
Hibernation behavior can differ between yearlings and adults in the area above the hibernaculum (Table 2). Females spent
other species (Bailey and Davis 1965; Buck and Barnes 1999a; fewer days in their hibernacula after the heterothermal period
Bieber et al. 2014), so we wanted to examine age effects in our ended compared to males (Table 2; Fig. 2). Torpor bout length
analyses. For the 36 squirrels for which we had hibernation be- was longest when the minimum Tsk within each torpor bout
havior data, we were able to classify seven squirrels as yearlings, was coolest (F[1,716] = 391.3, R2 = 0.35, P < 0.001) for the entire
five as 2 years, one as 3 years, and one as 4 years. In addition, we heterothermal period. Torpor bout length prior to the longest
assigned 11 as 1.5 years, eight as 2.5 years, and three as 3.5 years torpor bout was longest when Tsk was lowest (F[1,483] = 484.7,
based on their prior capture history. R2 = 0.67, P < 0.001; Fig. 3A) and followed a second-degree
polynomial distribution, but torpor bout length had no rela-
tionship to Tsk following the longest torpor bout (F[1,161] = 0.01,
Results R2 = 0.00, P = 0.54; Fig. 3B). The lowest recorded Tsk tem-
Hibernacula habitat characteristics varied among squirrels, but perature during hibernation (0.20°C) was at the second lowest
we did not detect any significant differences between the sexes elevation study site. The minimum Tsk during hibernation did
(Table 2). We obtained light logger data from 36 northern Idaho not differ between sexes (t30 = −0.76, P = 0.456).
Table 2.—Explanatory variables and two summary metrics (means of raw data) used to explain variation in hibernation behavior of northern
Idaho ground squirrels (Urocitellus brunneus). Sample sizes in parentheses. We included t-test statistics to test differences between males and
females. Ta is air temperature, Tsk is skin temperature, and Ts is soil temperature.
Female SE Male SE t d.f. P
Age 1.89 (18) 0.17 2.06 (18) 0.21 −0.605 34 0.549
Canopy closure (%) 43.06 (17) 8.14 43.29 (18) 7.79 −0.021 33 0.984
Ts increase datea 98.06 (16) 2.64 100.83 (12) 3.07 −0.685 26 0.5
DBH of closest tree (cm) 0.18 (18) 0.07 0.15 (17) 0.04 0.416 33 0.68
Elevation (m) 1466 (18) 34 1461 (18) 33 0.099 34 0.922
Activity pre-immergenceb (h) 8.35 (17) 0.53 8.94 (18) 0.53 −0.782 33 0.44
Percent with log presentc 33.33% (18) 33.33% (18)
Mean min Tskd (°C) 9.58 (16) 0.45 10.06 (16) 0.62 −0.625 30 0.537
Mean winter Ta (°C) −3.06 (12) 0.24 −2.71 (16) 0.2 −1.13 26 0.27
Min Tsk (°C) 1.66 (16) 0.19 1.92 (16) 0.27 −0.759 30 0.454
Residual body mass (g) 11.07 (18) 4.02 5.87 (18) 6.98 0.646 34 0.523
Slope (%) 12.36 (17) 1.91 8.72 (18) 1.25 1.61 33 0.117
Snowmelt datee (d) 93.46 (13) 3.62 98.44 (16) 3.41 −0.996 27 0.328
Ta pre-immerge (°C) 19.5 (18) 0.72 19.69 0.99 −0.151 34 0.881
Time between immergence and start of heterothermal period (days) 0.97 (18) 0.16 0.76 (18) 0.22 0.797 34 0.431
Time between emergence and snowmelt (days) 8.43 (7) 3.64 −3.31 (13) 2.8 2.517 18 0.02
Time between end of heterothermal period and emergence (light above 300 lux) 1.75 (12) 0.12 4.21 (14) 0.611 −3.73 24 0.001
a
First date that the soil temperature (Ts) at 30 cm belowground (at the site-specific logger in the forest or meadow depending on location of hibernacula) began to
steadily increase in the spring.
b
The mean number of hours an individual was aboveground (> 300 lux) per day for the 5 days prior to immergence.
c
Percent of hibernacula with a log present within 5 m. Downed wood had a minimum diameter of 2 cm to be considered a log.
d
The mean minimum skin temperature (Tsk) was calculated by taking the mean of the minimum Tsk during each torpor bout for the duration of the heterothermal
period.
e
Snow melt indicates the last day (Julian date) there was < 2 cm of continuous snow on the ground.580 JOURNAL OF MAMMALOGY
Downloaded from https://academic.oup.com/jmammal/article/102/2/574/6224539 by guest on 14 October 2021
Fig. 2.—Differences in hibernation phenology between male and female northern Idaho ground squirrels (Urocitellus brunneus) based on data
from seven study sites in Adams County, Idaho from 2015 to 2018. Error bars represent standard errors.
Mean hibernacula depth was 0.56 m ± 0.06 SE (range, 0.26– those from the full generalized linear models using the imputed
1.04 m). The closest sensors we had to that depth were 0.45 data. Mean immergence date was 9 July (± 1.66 days SE) for fe-
and 0.60 cm. However, we only had four probes at three sites males (N = 18) and 15 July (± 2.13 days SE) for males (N = 18);
that reached a depth of at least 0.60 m depth. Soil temperature heavier individuals immerged earlier (Table 3; Supplementary
(Ts) never dropped below 0°C either in the meadow or forest at Data SD1). Model selection of the imputed data set suggested
0.45 or 0.60 m depths despite Ts at 0 m depth (ground surface) that eight of the 10 explanatory variables be included, but only
dropping below 0°C between 15 and 109 days ( x̄ = 70.3 days ± residual body mass and canopy closure had a confidence in-
8.8 SE) at each of the four probes at the three sites that reached terval that did not include zero (Supplementary Data SD2).
a depth of at least 0.60 m. Soil temperatures were higher in the Mean number of euthermic bouts was similar between
meadow than in the forest for sensors at both 0.45 m (P < 0.001) sexes: 22.5 bouts for females (N = 16, ± 1.06 SE) and 22.0
and 0.60 m (P < 0.001). Ts also were higher at 0.60 m than at bouts for males (N = 16, ± 1.13 SE). Both heavier and older in-
0.45 m belowground (P < 0.001); the difference was greatest in dividuals had more euthermic bouts (but only according to the
the forest sensor compared to the meadow sensor (P < 0.001). univariate analysis; Table 3). Model selection of the imputed
The difference in mean Ts between meadow and forest was data set suggested that eight of the 10 explanatory variables
greatest during the summer and early fall and lowest during the be included, but only residual body mass had a confidence
winter (Fig. 4). Soil temperature was positively correlated with interval that did not include zero (Supplementary Data SD3).
elevation from 15 July through the end of March at 30 cm soil Similarly, heavier individuals and also males spent more total
depths for the four sites where we had sensors in both the forest time euthermic (Table 3; Supplementary Data SD4). Model
and the meadow in 2017 (P < 0.001). Overall soil temperature selection of the imputed data set suggested that nine of the
increased by 0.24°C for every 100 m increase in elevation. 10 explanatory variables be included, but only residual body
The majority of the univariate relationships between hiber- mass and sex had a confidence interval that did not include
nation behavior and our explanatory variables were similar to zero (Supplementary Data SD5).GOLDBERG AND CONWAY—GROUND SQUIRREL HIBERNATION BEHAVIOR 581
Downloaded from https://academic.oup.com/jmammal/article/102/2/574/6224539 by guest on 14 October 2021
Fig. 3.—Relationship between skin temperature and torpor bout duration for 33 northern Idaho ground squirrels (Urocitellus brunneus) in Adams
County, Idaho. There was a nonlinear negative relationship between torpor bout duration and skin temperature for bouts prior to the longest torpor
bout of the heterothermal period (A), but no such relationship for torpor bouts following the longest torpor bout (B). Gray shading in panel A rep-
resents the standard error.
The duration of the heterothermal period also was longer percent canopy closure and slope at the squirrel’s hibernaculum
for heavier individuals (Table 3; Supplementary Data SD6A). (Table 3). Eight of the 11 explanatory variables were included
Females (275.7 days ± 3.87 SE, N = 16) had longer heterothermal in the top model and none had confidence intervals that did not
periods than males (258.6 days ± 3.64 SE, N = 16; Table 3; include zero (Supplementary Data SD11).
Supplementary Data SD6B). Squirrels at higher elevation sites We retrieved 14 collars (seven from females and seven from
had longer heterothermal periods (Table 3). Model selection of males) from inside a hibernaculum and determined from them
the imputed data set suggested that all 12 explanatory variables hibernaculum depth for 14 squirrels. Depth of the hibernacula
be included, but only residual body mass, sex, canopy closure, was similar between sexes (0.51 m ± 0.06 SE for females and
and minimum Tb had confidence intervals that did not include 0.61 m ± 0.09 SE for males). We did not find any correlations
zero (Supplementary Data SD7). between hibernacula depth and the eight explanatory variables
The maximum torpor bout length was longest for those that that we evaluated. Fifty-six percent of the female squirrels that
had the lowest residual body mass (Table 3). Model selection we caught showed evidence of reproduction the summer prior
of the imputed data set suggested that nine of the 11 explana- to immergence into hibernation. However, we did not find any
tory variables be included, but only residual body mass had a correlations between reproductive condition and the seven re-
confidence interval that did not include zero (Supplementary sponse variables.
Data SD8).
Northern Idaho ground squirrels emerged later at higher ele-
vations (Table 3; Supplementary Data SD9A) and at colder sites Discussion
(Table 3; Supplementary Data SD9B); average emergence date A changing climate may impact the effectiveness of behavioral
was ~12 days later at 1,700 m compared to 1,300 m. Females choices that hibernators make to survive periods of extreme
(17 April ± 1.76 days SE, N = 12) emerged later than males (4 temperature. The northern Idaho ground squirrel hibernates
April ± 2.52 days SE, N = 14; Table 3; Supplementary Data across a ~1,200 m elevational gradient and must deal both with
SD9C). Squirrel age, snowmelt date, minimum Tsk, slope, and temporal (among years) and spatial (across sites) variation in
soil temperature also may influence emergence date (Table 3). snowpack. Our results suggest that snow cover affects emer-
Model selection of the imputed data set suggested that all 13 gence date: squirrels emerged later at hibernacula in areas with
explanatory variables be included in the top model, but deeper snow and later snowmelt. Snowfall totals and duration
only sex had a confidence interval that did not include zero of snowpack are predicted to decline (Mote et al. 2003; Klos
(Supplementary Data SD10). et al. 2014; Lute et al. 2015; Gergel et al. 2017; Marshall et al.
Mean minimum Tsk during hibernation was 9.58°C ± 0.45 SE 2019) and winter temperatures are predicted to rise, resulting
for females (N = 16) and 10.06°C ± 0.62 SE for males (N = 16). in more rain and less snow during winter (Mote et al. 2003;
Mean Tsk during hibernation was negatively correlated with Gergel et al. 2017) and therefore less insulation for hibernacula582 JOURNAL OF MAMMALOGY
Downloaded from https://academic.oup.com/jmammal/article/102/2/574/6224539 by guest on 14 October 2021
Fig. 4.—Soil temperature 30 cm belowground during the hibernation season (15 July to 1 April) differed between hibernacula site in the forest
and those in the meadow (open canopy) for northern Idaho ground squirrel (Urocitellus brunneus) hibernacula locations. The four panels repre-
sent data from four study sites that differed in elevation: 1,288 m (A), 1,359 m (B), 1,591 m (C), and 1,696 m (D). The greatest difference in soil
temperature between meadow and forest hibernacula were recorded in the first few months of hibernation. Standard errors are too small to be
visually noticeable.
(Young 1990a; Tafani et al. 2013). Northern Idaho ground may reflect local selection regimes (Gavin et al. 1999; Garner
squirrels only are active for ~4 months per year and likely time et al. 2005).
their emergence date to match seasonal forage phenology. If an Females emerged later than males, similar to other ground
individual emerges too early, they may find suboptimal forage squirrels (Kenagy et al. 1989a; Young 1990b; Michener 1992;
conditions. In contrast, emerging too late reduces the time an Buck and Barnes 1999a; Millesi et al. 1999). Males exit
individual has available to consume food while it is most nu- hibernacula prior to female emergence to establish and defend
tritious and easily digested compared to later in the summer territories and gain access to females (Michener 1992). Female
when plants dry out and senesce. It would be worthwhile to northern Idaho ground squirrels are sexually receptive for just
determine whether these behaviors are phenotypically plastic a few hours on the first or second afternoon after they emerge
or fixed but vary among populations. Northern Idaho ground (Yensen and Sherman 1997). Emerging prior to females en-
squirrel populations have moderate genetic differences that sures that a male does not miss this short time window.GOLDBERG AND CONWAY—GROUND SQUIRREL HIBERNATION BEHAVIOR 583
Table 3.—Univariate generalized linear model coefficients (Est.) based on the relationship between seven hibernation behaviors and 15 explan-
atory variables for northern Idaho ground squirrels (Urocitellus brunneus). The reference level for log is “yes” (Y) and the reference level for sex
is “males” (M).
Heterothermal Immergence Emergence date Number of Total Max torpor Mean min Tsk
period duration date euthermic euthermic du- duration
bouts ration
Est. P Est. P Est. P Est. P Est. P Est. P Est. P
Age 5.33 0.163 −1.83 0.281 5.94 0.044 2.21 0.035 2.12 0.005 −0.29 0.720 −0.27 0.616
Canopy closure −0.11 0.233 0.12 0.003 0.01 0.829 0.01 0.814 0.01 0.621 0.01 0.782 −0.03 0.006
Snowmelt datea 0.35 0.088 0.41 0.003 0.02 0.793 0.00 0.977 −0.04 0.215
Day Ts increasedb 0.51 0.041 −0.07 0.146
DBH of closest tree −30.32 0.129 6.36 0.265 −19.94 0.173 1.79 0.753 2.34 0.579 1.52 0.722 −1.21 0.608
Downloaded from https://academic.oup.com/jmammal/article/102/2/574/6224539 by guest on 14 October 2021
Elevation 0.04 0.027 0.00 0.944 0.03 0.043 0.01 0.135 0.01 0.150 0.00 0.997 0.00 0.282
Activity pre-immergencec (h) −1.1 0.068
Log present (Y/N) 1.69 0.787 2.83 0.331 3.11 0.503 0.91 0.601 1.25 0.330 −0.65 0.620 −1.3 0.126
Mean min Tskd −0.26 0.845 −1.38 0.148 −0.17 0.552
Mean winter Tae −4.44 0.267 −4.71 0.049 −0.40 0.728 −0.35 0.665 0.88 0.255 0.79 0.131
Min Tsk 0.66 0.834 0.14 0.954 0.70 0.280
Residual body mass 0.25 0.022 −0.20 0.000 0.09 0.332 0.08 0.007 0.07 0.002 −0.06 0.014 −0.01 0.600
Sex (M/F) −20.27 0.000 3.39 0.215 −12.20 0.002 −2.94 0.054 1.89 0.097 1.67 0.152 0.48 0.537
Slope 0.47 0.259 0.22 0.281 0.84 0.002 0.11 0.362 0.02 0.796 −0.01 0.874 −0.14 0.009
Ta before immergencef 0.56 0.14
a
Last day (Julian date) with < 2 cm of continuous snow on the ground.
b
First date that the soil temperature at 30 cm belowground began to steadily increase in the spring.
c
Mean hours that a squirrel was aboveground (> 300 lux) per day for 1 week prior to immergence.
d
Mean of the minimum Tsk during each torpor bout during the heterothermal period.
e
Mean air temperature (Ta) between November and March.
f
Mean air temperature for 1 week prior to immergence.
Heavier ground squirrels immerged earlier. Small-bodied hi- heat stress (Vispo and Bakken 1993; Sharpe and Van Horne
bernators should enter into torpor as soon as they are heavy 1999) and may not be able to forage during the night. Reduced
enough to survive hibernation to reduce their risk of predation forage time may lead to delays in immergence date due to dif-
by spending additional days foraging aboveground (Turbill ficulties in gaining the mass needed to survive the hibernation
et al. 2011; Bieber et al. 2014). In contrast to several studies on season. Examination is required to evaluate the effects of cli-
other ground squirrels (Fagerstone 1998; Young 1990b; Millesi mate on timing, diversity, and quality of northern Idaho ground
et al. 1999; Williams et al. 2014; Gür and Gür 2015), we did squirrel forage.
not detect a difference between sexes in immergence date. Male Body condition also affected northern Idaho ground squirrel
ground squirrels immerge prior to females in many but not all euthermic bouts; body mass was positively correlated with the
species and the variation may reflect interspecific differences in time spent euthermic and number of euthermic bouts. Similarly,
the need to protect food caches for spring (Millesi et al 1999; heavier woodchucks (Marmota monax) had warmer body tem-
Williams et al. 2014). We also did not detect a relationship be- peratures and spent less time in torpor (Zervanos et al. 2013).
tween immergence date and elevation, but we did not carry out However, we did not find relationships between Tsk and maximum
field work on populations at the highest elevation sites that sup- torpor length. Animals with extra energy may prefer to stay in
port northern Idaho ground squirrels (> 2,000 m) which are torpor for less time because prolonged torpor can have negative
expected to have the latest snowmelt and greatest snow accu- consequences (Humphries et al. 2003). Furthermore, males had
mulations of any populations. If immergence date does not shift longer euthermic bouts than females. Males are heavier and have
along with emergence date, earlier snow melt dates may lead shorter heterothermal periods. Extra energy may be needed prior
to a prolonged active season (shorter hibernation period) and to female emergence while males defend their territory at a time
longer period of time exposed to higher predation rates (Turbill when forage is suboptimal. Euthermic bouts are energetically
et al. 2011; Bieber et al. 2014). Future climate scenarios pre- expensive and account for ~70% of the energy spent during hi-
dict that summers in central Idaho will be warmer with less bernation (Wang 1978), but all individuals need to arouse period-
rainfall (Mote and Salathe 2010), resulting in changes in plant ically from torpor (euthermic bouts). The reasons why euthermic
phenology and diversity (Bertin 2008; Khanduri et al. 2008). bouts are needed are still debated (Barnes and Buck 2000;
Climate-induced changes in hibernation behavior may result in Körtner and Geiser 2000; Dark 2005). Potential functions of these
phenological mismatches with preferred food items or changes euthermic bouts include: (1) reduce immune system suppression
in competitive interactions with other herbivores. Furthermore, (Prendergast et al. 2002); (2) reduce sleep deprivation (Daan et al.
warmer temperatures during the peak active season (lactation 1991); and (3) increase renal function (Humphries et al. 2003).
and peak mass gain) may reduce the number of hours a squirrel Hibernacula location may impact hibernation behavior.
can forage aboveground because many squirrels are subject to Both canopy cover and slope at a squirrel’s hibernaculum were584 JOURNAL OF MAMMALOGY
negatively related to minimum Tsk. Higher canopy closure was among species (Inouye et al. 2000; Lane et al. 2012; Tafani
positively related to immergence date, and slope was positively et al. 2013; Rézouki et al. 2016; Falvo et al. 2019). To avoid
related to emergence date. Both canopy closure and slope extinction, northern Idaho ground squirrels may need to adjust
may impact soil temperature, snow cover, and snowmelt date their hibernacula location, depth, and immergence/emergence
(Clinton 2003; D’Eon 2004; Varhola et al. 2010). Further work timing in response to a changing climate.
is needed to understand the mechanisms underlying these rela-
tionships. Northern Idaho ground squirrel Tsk (while in torpor)
Acknowledgments
tracks soil temperature and likely does so until a fatal lower
threshold (Geiser et al. 1990; Michener 1992). Arctic ground Many crew leaders and technicians helped collect the data.
squirrels use relatively little thermogenesis when the soil tem- A. Allison oversaw collar deployment and collection during
perature is > 0°C (Barnes and Buck 2000) and 2°C was the the 2018–2019 field season. The Hixon family and their ranch
Downloaded from https://academic.oup.com/jmammal/article/102/2/574/6224539 by guest on 14 October 2021
ideal Tb during hibernation for the little brown bat (Myotis employees provided housing and access to their land on the
lucifugus—Humphries et al. 2002). Body temperature during OX Ranch. Funding was provided by the U.S. Forest Service
hibernation correlates with torpor duration (Geiser and Kenagy (Payette National Forest) (grant number GCK159), the Idaho
1988; Barnes and Buck 2000) and selection likely favors squir- Chapter of The Wildlife Society, Curt Berklund Graduate
rels that select hibernacula depths where soil temperature re- Research Scholarship, J. Michael & Sharon Scott Scholarship,
mains > 0°C but stays cool enough for the squirrel to stay in and Craig Terry Kvale Scholarship. J. Galloway, A. Egnew,
torpor without raising metabolic processes (Barnes and Buck J. Almack, and R. Richards provided logistical support while
2000). Shallower hibernacula likely experience more fluctu- working in Payette National Forest. D. Evans Mack and
ation in soil temperature before snowfall (Kay and Whitford G. Burak provided logistical support. E. Harrity, D. Biggins,
1978; Baker and Baker 2002) and likely have lower soil tem- R. Long, J. Rachlow, and E. Lehmer provided constructive
peratures after snow has fallen (Hardy et al. 2001; Zhang comments that improved this paper. This study was carried
2005). Soil temperatures during the hibernation season (15 July out under the auspices of University of Idaho IACUC protocol
to 1 March) in our system were warmer in the meadows (open #2015-53. Any use of trade, firm, or product names is for de-
canopy) compared to the forest (Fig. 3). If cooler Ts is preferred scriptive purposes only and does not imply endorsement by the
during the start of hibernation (summer and early fall), squirrels U.S. Government.
either should select deeper hibernacula or areas with greater
canopy closure. Northern Idaho ground squirrels may select
hibernacula in forests rather than in open areas, especially at
Supplementary Data
higher elevations (Goldberg et al. 2020a), because these areas Supplementary data are available at Journal of Mammalogy
have higher Ts in the winter and lower Ts in the summer, re- online.
ducing metabolic needs in the summer when its warm and in the Supplementary Data SD1.—Relationship between residual
winter when they need to avoid freezing. Shallower hibernacula body mass and date of immergence into hibernation. Residual
at higher elevations are more likely to remain > 0°C due to body mass represents the squirrel’s body mass (corrected for
snow insulation. Further research is needed to better under- capture date) in the summer prior to hibernation.
stand how snow depth and duration impact soil temperature Supplementary Data SD2.—Generalized linear model
and how both interact with canopy closure to influence optimal coefficient estimates (Est.) evaluating the relationships be-
thermal conditions for hibernation of this rare ground squirrel. tween 10 explanatory variables and the northern Idaho ground
Furthermore, we also need more information regarding the squirrel immergence date. We evaluated the data three ways: a)
consequences of variation in hibernaculum depth. Hibernacula univariate analysis using only the data collected, b) multiple re-
depths varied between 0.26 and 1.04 m, corroborating a pre- gression using imputed data, and c) AIC model selection using
vious study (0.24 to 0.75 m—Yensen and Dyni 2020). Depth imputed data.
of hibernacula may be regulated by: (1) predator avoidance; (2) Supplementary Data SD3.—Generalized linear model co-
ideal thermal conditions (Young 1990b; Michener 1992; Buck efficient estimates (Est.) evaluating the relationships between
and Barnes 1999b); (3) energetics (the amount of energy nec- 10 explanatory variables and the number of euthermic bouts
essary to dig a deeper burrow); or (4) some combination of the during the northern Idaho ground squirrel hibernation period.
three mechanisms. We evaluated the data three ways: a) univariate analysis using
Climate change not only may impact vegetation phenology only the data collected, b) multiple regression using imputed
but also may alter availability of key nutrients necessary for data, and c) AIC model selection using imputed data.
these squirrels to reproduce and survive the hibernation season. Supplementary Data SD4.—Relationship between residual
Our results suggest that changes in climate likely will result body mass and the number of days that a northern Idaho ground
in the need for northern Idaho ground squirrels to alter their squirrel was euthermic. Residual body mass represents the
hibernation behavior. Other hibernating mammals have al- squirrel’s body mass (corrected for capture date) in the summer
tered immergence and emergence dates in response to cli- prior to hibernation.
mate change (Lane et al. 2012; Williams et al. 2014), but the Supplementary Data SD5.—Generalized linear model co-
effects of these changes on demographic traits have varied efficient estimates (Est.) evaluating the relationships betweenGOLDBERG AND CONWAY—GROUND SQUIRREL HIBERNATION BEHAVIOR 585
10 explanatory variables and the total time spent euthermic Bertin, R. I. 2008. Plant phenology and distribution in relation to
during the northern Idaho ground squirrel hibernation period. recent climate change. The Journal of the Torrey Botanical Society
We evaluated the data three ways: a) univariate analysis using 135:126–146.
only the data collected, b) multiple regression using imputed Blois, J. L., J. W. Williams, M. C. Fitzpatrick, S. T. Jackson,
and S. Ferrier. 2013. Space can substitute for time in predicting
data, and c) AIC model selection using imputed data.
climate-change effects on biodiversity. Proceedings of the National
Supplementary Data SD6.—The relationship between the
Academy of Sciences 110:9374–9379.
total heterothermal period (time spent in hibernation from the Bieber, C., K. Lebl, G. Stalder, F. Geiser, and T. Ruf. 2014.
start of the first torpor period greater than 24 h to the end of the Body mass dependent use of hibernation: why not prolong the ac-
last torpor period greater than 24 h) and residual body mass tive season, if they can? Functional Ecology 28:167–177.
(A), and sex (B) of northern Idaho ground squirrels. Breshears, D. D., J. W. Nyhan, C. E. Heil, and B. P. Wilcox.
Supplementary Data SD7.—Generalized linear model co- 1998. Effects of woody plants on microclimate in a semiarid wood-
Downloaded from https://academic.oup.com/jmammal/article/102/2/574/6224539 by guest on 14 October 2021
efficient estimates (Est.) evaluating the relationships between land: soil temperature and evaporation in canopy and intercanopy
12 explanatory variables and the duration of the northern Idaho patches. International Journal of Plant Sciences 159:1010–1017.
ground squirrel heterothermal period. We evaluated the data Buck, C. L., and B. M. Barnes. 1999a. Annual cycle of body com-
three ways: a) univariate analysis using only the data collected, position and hibernation in free-living arctic ground squirrels.
Journal of Mammalogy 80:430–442.
b) multiple regression using imputed data, and c) AIC model
Buck, C. L., and B. M. Barnes. 1999b. Temperatures of hibernacula
selection using imputed data.
and changes in body composition of arctic ground squirrels over
Supplementary Data SD8.—Generalized linear model co- winter. Journal of Mammalogy 80:1264–1276.
efficient estimates (Est.) evaluating the relationships between Burak, G. S. 2011. Status review for northern Idaho ground squirrel
11 explanatory variables and the maximum torpor length (Spermophilus brunneus brunneus). U.S. Fish and Wildlife Service,
during the northern Idaho ground squirrel hibernation period. Region 1, Idaho Fish and Wildlife Office. Boise, Idaho.
We evaluated the data three ways: a) univariate analysis using Burak, G. S., A. R. Goldberg, J. M. Galloway, D. Evans Mack,
only the data collected, b) multiple regression using imputed and C. J. Conway. 2018. Collaborating to save a tiny threatened
data, and c) AIC model selection using imputed data. species: what does the northern Idaho ground squirrel need to sur-
Supplementary Data SD9.—Relationship between the vive? The Wildlife Professional 12:39–42.
northern Idaho ground squirrel emergence date and elevation Burnham, K. P., and D. R. Anderson. 2002. Model selection and
multimodel interence. 2nd ed. Springer. New York, New York.
(A), mean winter air temperature (B), and sex (C).
Clinton, B. D. 2003. Light, temperature, and soil moisture re-
Supplementary Data SD10.—Generalized linear model co-
sponses to elevation, evergreen understory, and small canopy gaps
efficient estimates (Est.) evaluating the relationships between in the southern Appalachians. Forest Ecology and Management
13 explanatory variables and northern Idaho ground squirrel 186:243–255.
emergence date. We evaluated the data using a) univariate anal- Cordes, L. S., et al. 2020. Contrasting effects of climate change
ysis using only the data collected, b) multiple regression using on seasonal survival of a hibernating mammal. Proceedings of the
imputed data, and c) AIC model selection using imputed data. National Academy of Sciences of the United States of America
Supplementary Data SD11.—Generalized linear model 117:18119–18126.
coefficient estimates (Est.) evaluating the relationships be- Daan, S., B. M. Barnes, and A. M. Strijkstra. 1991. Warming up
tween 11 explanatory variables and the mean minimum skin for sleep? Ground squirrels sleep during arousals from hibernation.
temperature (Tsk) during the northern Idaho ground squirrel Neuroscience Letters 128:265–268.
Dark, J. 2005. Annual lipid cycles in hibernators: integration of
hibernation period.
physiology and behavior. Annual Review of Nutrition 25:469–497.
Davis, R. E., et al. 1997. Variation of snow cover ablation
in the boreal forest: a sensitivity study on the effects of co-
Literature Cited nifer canopy. Journal of Geophysical Research: Atmospheres
Avgar, T., G. Street, and J. M. Fryxell. 2014. On the adaptive 102:29389–29395.
benefits of mammal migration. Canadian Journal of Zoology Decker, K. L. M., D. Wang, C. Waite, and T. Scherbatskoy.
92:481–490. 2003. Snow removal and ambient air temperature effects on forest
Bailey, E. D., and D. E. Davis. 1965. The utilization of body fat soil temperatures in northern Vermont. Soil Science Society of
during hibernation in woodchucks. Canadian Journal of Zoology America Journal 67:1234–1242.
43:701–707. D’Eon, R. G. 2004. Snow depth as a function of canopy cover and
Baker, J. M., and D. G. Baker. 2002. Long-term ground heat other site attributes in a forested ungulate winter range in south-
flux and heat storage at a mid-latitude site. Climatic Change east British Columbia. Journal of Ecosystems and Management
54:295–303. 3:1–9.
Barnes, B. M., and C. L. Buck. 2000. Hibernation in the ex- Fagerstone, K. A. 1998. The annual cycle of Wyoming ground
treme: burrow and body temperatures, metabolism, and limits to squirrels in Colorado. Journal of Mammalogy 69:678–687.
torpor bout length in arctic ground squirrels. Pp. 65–72 in Life in Falvo, C. A., D. N. Koons, and L. M. Aubry. 2019. Seasonal cli-
the cold: Eleventh International Symposium (G. Heldmaier and mate effects on the survival of a hibernating mammal. Ecology and
M. Klingenspor, eds.). Springer. Berlin, Germany. Evolution 9:3756–3769.
Bellard, C., C. Bertelsmeier, P. Leadley, W. Thuiller, and Garner, A., J. L. Rachlow, and L. P. Waits. 2005. Genetic diver-
F. Courchamp. 2012. Impacts of climate change on the future of sity and population divergence in fragmented habitats: conserva-
biodiversity. Ecology Letters 15:365–377. tion of Idaho ground squirrels. Conservation Genetics 6:759–774.You can also read
