Influence of seasonality and gestation on habitat selection by northern Mexican gartersnakes (Thamnophis eques megalops)
Influence of seasonality and gestation on habitat selection by northern Mexican gartersnakes (Thamnophis eques megalops)
Tiffany A. Sprague 0 1
Heather L. Bateman 0 1
0 College of Integrative Sciences and Arts, Arizona State University , Mesa, Arizona , United States of America
1 Editor: Christopher M. Somers, University of Regina , CANADA
Species conservation requires a thorough understanding of habitat requirements. The northern Mexican gartersnake (Thamnophis eques megalops) was listed as threatened under the U.S. Endangered Species Act in 2014. Natural resource managers are interested in understanding the ecology of this subspecies to guide management decisions and to determine what features are necessary for habitat creation and restoration. Our objective was to identify habitat selection of northern Mexican gartersnakes in a highly managed, constructed wetland hatchery. We deployed transmitters on 42 individual gartersnakes and documented use of habitat types and selection of specific habitat features. Habitat selection was similar between males and females and varied seasonally. During the active season (March±October), gartersnakes primarily selected wetland edge habitat with abundant cover. Gestating females selected similar locations but with less dense cover. During the inactive season (November±February), gartersnakes selected upland habitats, including rocky slopes with abundant vegetation. These results of this study can help inform management of the subspecies, particularly in human-influenced habitats. Conservation of this subspecies should incorporate a landscape-level approach that includes abundant wetland edge habitat with a mosaic of dense cover for protection and sparsely vegetated areas for basking connected to terrestrial uplands for overwintering.
Data Availability Statement: All relevant data are
within the paper and its Supporting Information
Funding: Funding was provided by Arizona Game
and Fish Department from the Heritage Fund,
IIAPM I15001, and the CAMP Fund (Conservation
and Mitigation Program) (to HLB), www.azgfd.
com/wildlife/heritagefund/program/. The funders
had no role in study design, data collection and
analysis, decision to publish, or preparation of the
One of the most important elements of species management and conservation is knowledge of
habitat requirements [
]. Understanding habitat selection at multiple spatial scales is vital for
effective species management [
]. Habitat selection in ectothermic animals can be driven by
intrinsic factors, such as body size and reproductive condition [5±7], and extrinsic factors,
such as distribution of resources, temperature, predators, and prey [8±10]. Selection of habitat
features is not static because animals can alter selection based on daily or seasonal variation
]. In snake species, behaviors such as hibernating, breeding, and foraging can strongly
influence habitat selection [
]. Habitat modification and loss can restrict the ability of
animal species to move about the landscape to preferentially select required resources [
Many species of snakes have experienced dramatic population declines because of habitat loss
and degradation [
], including species that rely on wetland areas [
]. Rivers and
wetlands around the world have been disrupted and fragmented, affecting the rich biodiversity
that depends on them [20±22]. Many species of aquatic and semi-aquatic snakes have declined
due to drought and human-caused impacts to their habitat [
In the semiarid southwestern United States, riparian areas are imperiled habitat [
occupy less than 3% of the total land area . However, riparian areas provide a mosaic of
productive habitats and support many species [
]. Numerous aquatic and semi-aquatic
animal species have declined due to damming and diversion of surface water and pumping of
]. One hallmark subspecies that has experienced such declines is the
northern Mexican gartersnake (Thamnophis eques megalops; S1 Fig). Substantial portions of the
historical range of the northern Mexican gartersnake have been dewatered, resulting in local
]. Many sites where this gartersnake persists have been reduced in size or
have become isolated [31±33]. Historically, the northern Mexican gartersnake ranged
throughout much of central and southern Arizona, into southwestern New Mexico and Mexico, and
may have occurred in California and Nevada along the Colorado River . The subspecies
now occurs at low densities and may be extirpated from as much as 90% of its historical range
in Arizona and New Mexico [31±33]. The northern Mexican gartersnake is state-listed and
federally listed as threatened [
The few studies on this subspecies have described home ranges and general habitat
associations (second and third-order habitat selection based on [
] and [
]). Third-order habitat
selection of gartersnakes has been described as protected backwaters, pools, cienegas, stock
tanks, and stream edges rich with emergent vegetation [
]. Gartersnakes have also been
documented in human-modified areas, such as fish hatcheries. Boyarski et al. [
gartersnakes spending active seasons on hatchery pond edges and in cattail-dominated areas
and overwintering in upland habitat composed of rocky, shady slopes. However, within these
larger-scale descriptions, little is known about selection of fine-scale structural features and
sites (e.g., ground cover, vegetation, and substrate of fourth-order selection [
]). Many species
of snakes select areas based on fourth-order habitat parameters, which are often more
important than third-order habitat features for thermoregulation, predator avoidance, and foraging
Similarly, little is known about seasonal variation in habitat selection of northern Mexican
gartersnakes during the active, gestation, and inactive seasons. Many animal studies focus on
habitat use during only one season and may exclude animals based on reproductive condition
or maturity. Such limitations provide an incomplete picture of the full habitat needs of a
species. Semi-aquatic species, such as the northern Mexican gartersnake, rely on both aquatic and
terrestrial habitats [39±41]. Understanding the spatial and temporal uses of habitat by
gartersnakes is critical for effective habitat conservation [
] by informing the size and type of
areas to be conserved. Knowledge of these habitat requirements can guide the timing and
location of management activities to minimize adverse effects or disturbance to the subspecies
The objective of our study was to identify habitat parameters selected by northern Mexican
gartersnakes. We examined habitat features selected by gartersnakes during the active,
gestation, and inactive seasons and compared habitat differences due to sex. We expected to
document seasonal differences in habitat due to a separation in overwintering and foraging
requirements and also expected to see some differences between genders due to the larger
body size and reproductive needs of females. Because northern Mexican gartersnakes are
], thermoregulation may be more important during gestation than foraging.
2 / 23
Therefore, we compared habitat requirements during the gestation period to active and
inactive periods. Results of this study will help resource managers understand specific habitat
features used by this subspecies during different seasons and across life-history stages. By
understanding the spatial and temporal ecology of this subspecies, managers can use this
information to maintain or construct features to provide suitable habitat.
Our study area was a 21.9-ha state-managed fish hatchery described as sustaining one of only
five viable populations of northern Mexican gartersnakes in the United States [
Ponds State Fish Hatchery (hereafter, the hatchery), located in Yavapai County, Arizona,
(UTM NAD83 0418091E 3847618N) raises warm water native fishes and introduced sportfish.
The Arizona Game and Fish Commission purchased the property in 1954 and has operated it
as a fish hatchery since 1955. Elevation ranges from 1052±1180 m. The hatchery includes lined
and unlined fish-rearing ponds, fallow ponds no longer used for fish production, meadows
dominated by sedges and grasses, mesquite (Prosopis velutina) and riparian woodlands, and
dense thickets of non-native blackberry (Rubus sp.; Fig 1). Other vegetation at the hatchery
includes a mix of native and non-native species. The hatchery is bordered by Oak Creek and
hills of semidesert grassland and mixed evergreen±deciduous shrubland [
We captured gartersnakes using a combination of Gee™ minnow traps [
], coverboards [
and visual encounters. We deployed 50±100 traps per day from May±October 2015 and April±
October 2016 for 4,966 trap days. Traps were checked and emptied twice a day. Some
nontarget species (anurans, salamanders, fish, and insects) were left in traps to serve as bait. We
marked, measured, and sexed each captured northern Mexican gartersnake. We marked all
gartersnakes using cautery branding [
] and microchipped gartersnakes >25g with PIT tags
We employed radio telemetry to monitor animal movement and habitat selection
]. We used a combination of internal and external transmitters on a total of 42
individual gartersnakes. A veterinarian surgically implanted temperature-sensing transmitters
(SB-2T [5.2g] or BD-2T [1.9g], Holohil Systems Ltd., Ontario, Canada) in 22 individual
gartersnakes. We followed recommended surgery and post-operative care sensu [
]. We fitted
external transmitters (BD-2 or temperature-sensing BD-2T units; 1.8g, Holohil Systems Ltd.,
Ontario, Canada) on an additional 20 gartersnakes using tape [
]. Eight gartersnakes received
more than one type of transmitter (internal/external) over the course of the study.
Transmitters were no more than 5% of the gartersnake's mass at the time of deployment. We released
gartersnakes at their capture locations, unless they had been captured inside a construction
area, in which case they were released into adjacent fallow ponds. We brought transmittered
gartersnakes that exhibited signs of illness to a veterinarian for care, and all functioning
transmitters were removed from gartersnakes by the end of the project. Our research was covered
under the following permits: US Fish and Wildlife Service (AGFD Section 6 and TE43322B-0),
and Arizona Game and Fish Department (SP733846 and SP744166). This study was carried
out in accordance with protocol approved by Institutional Animal Care and Use Committee
from Northern Arizona University (Permit NAU 14±010) during 2015 and from Arizona State
University (Permit ASU 15-1417R) during 2016. All surgery was performed under anesthesia
by experienced and licensed veterinarians, and all efforts were made to minimize suffering.
3 / 23
Fig 1. Study site: Bubbling ponds hatchery in Yavapai County, Arizona, looking north. Active fish-rearing ponds are the nine long oval ponds to the north and east.
Fallow ponds are the four vegetated blocks in the south middle. The four ponds to the southwest were drained during much of the study (June 2015 ±February 2016).
The pond in the far southwest was lined with black polypropylene liner and remained empty. To the south of the managed ponds are a rocky ridge covered by trees
and a wet meadow. Oak Creek borders the site on the east. Inset shows general location (star) of the study site in North America.
We located transmittered gartersnakes at least once per week from May 2015 through
August 2016. To ensure individuals were located during different diel periods, we assigned
gartersnakes to tracking cohorts, which we tracked weekly at different times on a rotating basis
(i.e., early day [0700±1100], midday [1100±1500], and late day [1500±1900]). Trained
observers located transmittered gartersnakes to within 30 cm, although snakes occasionally flushed
before their location could be pinpointed, and underground signals may have been distorted
by roots or rocks. We field tested position accuracy by locating 100% (n = 31) of shed external
transmitters. Because gartersnakes were frequently underground or relied on procrypsis when
aboveground, we were able to pinpoint locations without flushing snakes more than 97% of
the time. Each location, hereafter referred to as the snake point, was recorded using a global
positioning system (GPS) unit (Garmin Ltd., Schaffhausen, Switzerland) and was marked with
flagging tape to identify the exact location of the gartersnake. We also recorded GPS location
4 / 23
accuracy, whether or not the gartersnake was visible, snake behavior (if observable), and
transmitter pulse rate (used to calculate body temperature).
We divided sampling into three seasons: active (March±October), gestation (April±May for
females only), and inactive (November±February). Inactive season was determined for each
individual based on amount of movement and when each snake entered its overwintering
habitat. Gestation period was based on females known to be pregnant. Because we used a
handsoff approach to minimize influence on behaviors and habitat selection, we could not confirm
reproductive status for all females in 2016. However, all females captured during May 2015
(n = 7) and more than half (n = 6) of transmittered females in April±May 2016 were confirmed
to be pregnant. We determined the start of the gestation season based on enlarged ovaries
observed by a veterinarian during transmitter implant surgery in early April 2016 and
ultrasounds of embryos in May 2015. The end of gestation season was based on observation of
neonates during the first week of June in 2015 and 2016. Therefore, we determined the gestation
season to be April±May and included all females in the gestation habitat assessment.
We measured habitat where we found gartersnakes through tracking and visual observations.
Fourth-order habitat measurements included vegetative, environmental, and hydrologic
characteristics (Table 1) recorded at each snake point, in a 1-m-diameter plot, and along four
2.5-m transects sensu [
] (Fig 2). At each snake point, we measured aspect and slope,
water depth, distance to water, and canopy cover (>1m in height). Within a 1-m-diameter
plot centered on the snake point, we recorded number of plant stems ( 1 cm diameter) rooted
in the plot and percentages of ground cover type, submerged vegetation, and surface shade. In
plots, we also recorded percentage of low-height cover ( 1 m), which included vegetation
(living or dead), woody debris, deep loose litter, or human-made structures that a snake could use
for potential cover. We defined ground cover as anything a snake could be on top of when
aboveground (Fig 3). We used ocular estimates of cover classes  in the following
percentages: 0, <1, 1±5, 5±25, 25±50, 50±75, 75±95, >95. On four intersecting 2.5-m transects, we
quantified vegetation type (grass, forb, cattail, sedge/rush, shrub, tree, or none) at every 0.5m
mark. We measured habitat only at unique snake locations, which excluded points <3 m from
a previous location for that snake (to avoid overlap in measurements) that had been measured
in <4 weeks [
To compare used and available habitat, we quantified fourth-order habitat variables at
snake points and paired random points [
]. This matched-pairs design is more robust
than unmatched studies for assessing habitat selection because each random location
represents a true absence [59±61]. This technique also controls for variation in environmental
conditions and enables more accurate modeling of habitat selection by ensuring that each random
location is available to that individual at that time [
]. We randomized the distance
(between 5 and 155 m) and bearing of the paired point from each snake point using a random
number generator. If a random point occurred on private land or in an area not accessible to
snakes, a new location was determined.
Activity range size
We estimated 95% minimum convex polygon (MCP) activity range sizes (second-order
]) to better understand the amount of movement within each season (ArcGIS version
10.3, Esri, with ArcMET 10.3.1 v1 software extension). Although widely used, MCP provides a
rough estimate of second-order selection and might include large areas not used by the animal
]. Therefore, MCPs might not represent true activity ranges but are useful to understand
5 / 23
total range and relative movements . Because number of locations for each individual can
influence activity range size, we only included snakes for which activity range size plotted
against number of locations reached an asymptote [
]; we used all locations available for
these snakes to estimate activity range.
We tested data for normality and equal variance using R (version 3.1.2, The R Foundation for
Statistical Computing) package ªcar.º We calculated mean and standard error for each habitat
variable using R package ªplyrº and Oriana 4 (Kovach Computing Services, Anglesey, Wales).
For subsequent analyses, we converted aspect to a categorical variable (i.e., N, E, S, W) and
used the median of each ocular estimate class. To visualize habitat, we used a Principal
Component Analysis (PCA) to reduce variables into components using SPSS (version 23.0, IBM).
Because PCA is most suitable for datasets with a low number of zeros, we only included
variables for which <40% of values were equal to zero [
]. We scaled and centered data prior
to running the PCA to account for varying units of measurement among variables.
Components with an eigenvalue >1 were selected and plotted for visualization .
6 / 23
Fig 2. Diagram of plot and transect design used to measure habitat variables. One 1-m-diameter plot and four
randomly-oriented perpendicular 2.5-m transects placed with the snake/random location as the centerpoint.
To assess habitat selection and to identify key environmental variables, we used matched
pairs logistic regression [
] to compare each snake point to its random location (R package
ªsurvivalº). We modeled habitat selection by gender and season. When generating habitat
models, the first step was to select variables for inclusion. We tested variables for
multicollinearity using pairwise comparisons (cutoff of r 0.6) and variance inflation factors (cutoff of
VIF 10). We then generated univariate matched-pairs logistic regression models to assess the
significance of each variable [
] using R package ªsurvival.º Variables were considered
significant at p<0.25 because some variables might not be significant on their own but are significant
in conjunction with other parameters [
]. Variables exhibiting complete separation (i.e., all
snake locations had zeros for that variable but some random points had non-zero values or
vice versa) were omitted to avoid statistical problems in the logistic regression [
]. We fitted a
multivariate model with all uncorrelated variables found to be significant during the univariate
]. If two or more highly-correlated variables were significant in univariate tests, we ran
separate multivariate models with one of those variables. Variables that were clearly
non-significant (p>0.25) were removed from the multivariate models. In a stepwise fashion, we then
added variables eliminated during preliminary univariate and multivariate tests back into the
models, one at a time, to test for significance [
]. Any non-significant variables (p>0.25) were
7 / 23
Fig 3. Examples of cover types. a) Canopy cover from trees, low-height cover from living vegetation (grass), and ground cover from
bare, rock, litter, woody debris, and small-diameter vegetation; b) low-height cover from living and dead vegetation, litter, and woody
debris and ground cover from litter and woody debris; c) low-height cover from living vegetation (forb and grass) and ground cover from
bare, rock, and small-diameter vegetation; d) low-height cover from vegetation, woody debris, and litter and ground cover from rock,
litter, and woody debris.
again removed. We repeated this process until we obtained final models in which all variables
were significant [
We used a ranked multiple-model inference approach to obtain unbiased coefficients for
variables determined by the final models [
]. All possible subsets were considered (R package
ªMuMInº). The top model had a ΔAIC = 0, but we also considered all models with a ΔAIC<2.
We calculated variable weights within each model and then summed across all models to
obtain weighted coefficients for each variable. Because a one-unit increase in an explanatory
variable is rarely practical for continuous data [
], we determined increases based on means
and ranges for each variable to calculate odds ratios.
We used a one-tailed t-test [
] to compare mass of females and males. We compared
gartersnake body temperature by season and month using two mixed-effects ANOVAs [
with season and sex as fixed effects and a second with month and sex as fixed effects; individual
8 / 23
snake was included as a random effect in both. ANOVAs were conducted using R package
ªlme4.º These data met assumptions of normality and equal variance. Tests were considered
significant at α 0.05.
Of the 42 transmittered gartersnakes, 25 were female and 17 were male (S1 Table). We located
transmittered gartersnakes 781 times to assess habitat and removed locations that were not
unique and one location from a female behaving abnormally due to illness. We quantified
habitat variables at 510 gartersnake locations and 510 random locations, including 486 telemetry
and 24 visual observations. Paired locations were grouped into three seasons: active (n = 348),
gestation (n = 57), and inactive (n = 105). Gartersnakes were visible 24.1% of times located for
fourth-order habitat assessment (23.0% during the active season, 56.1% during gestation, and
10.5% during the inactive season).
Ten of 23 variables were included in the PCA: canopy cover, low-height cover, shade, bare
ground cover, litter ground cover, small-vegetation ground cover, grass, forb, distance to
water, and slope. These variables were reduced to four components that, when combined,
explained 67.6% of variation in the data (Table 2). Component 1 described the most variation
in habitat (25.7%) and represented elements of vegetative cover. Biplots of these components
show high variability in habitat characteristics; however, gartersnakes displayed more narrow
fourth-order habitat selection during the inactive season (Fig 4).
Season influenced third- and fourth-order habitat selection (Tables 3 and 4). During the
active season, we primarily located gartersnakes on active or fallow pond banks or edges
(60.6% of female and 41.5% of male locations) or in marshy areas of the fallow ponds (20.2%
of female and 23.0% of male locations). Gartersnakes occasionally used other parts of the
hatchery, such as Oak Creek or a meadow south of the ponds. On separate occasions, we
located two females in semi-desert grassland habitat >100m from the ponds. On a
fourthorder habitat scale, both sexes selected sloping areas close to water with a high amount of
lowheight cover ( 1 m in height) and vegetation, specifically forbs, and generally avoided areas
Principal Component Analysis (PCA) included 10 variables with <40% occurrence of zero values. Variables with the highest loading for each component are in bold.
9 / 23
Fig 4. Biplots of four habitat components generated from PCA analyses. a) C1 (cover and litter) vs. C2 (slope, forb, and bare ground cover)
and b) C3 (small-diameter vegetation abundance) and C4 (distance to water). Percentages in parentheses show the amount of variation in the
data accounted for by that component.
PLOS ONE | https://doi.org/10.1371/journal.pone.0191829
10 / 23
Direction of snake selection shown as positive or negative relative to random points. Results from univariate matched-pairs logistic regression models and variables
included in multivariate models in bold.
with a high percentage of sedges or rushes and areas with deep water (Table 5). Females
selected areas with shrubs, and males selected areas away from trees.
During gestation, females were most often found on pond banks (78.9% of locations) or
other sloping areas near the ponds (7.0%). We frequently observed them basking aboveground
in mottled shade (56.1% of locations). Females selected sites close to water with a high
percentage of small-diameter (<1 cm) vegetation and litter and avoided areas with a high number of
large-diameter ( 1 cm) stems and a high percentage of canopy cover (Table 5).
During the inactive season, gartersnakes selected areas away from the ponds. Most
gartersnakes overwintered on a rocky slope south of the ponds (49.2% of female and 73.8% of male
locations) or other wooded sites (49.2% of female and 16.7% of male locations). One male
overwintered on the bank of Oak Creek (9.5% of male locations). On only one occasion was a
gartersnake (female) located in an area with water in the plot (1.5% of female inactive
locations; <1% of all inactive locations). On a fourth-order habitat scale, both sexes selected rocky
11 / 23
Direction of snake selection shown as positive or negative relative to random points. Results from univariate matched-pairs logistic regression models and variables
included in multivariate models in bold. Variables with a dash exhibited complete separation between snake and random locations so were omitted from multivariate
slopes with a high percentage of forbs (Table 5). Females selected areas with a high percentage
of canopy cover (>1 m in height) and avoided areas with a high amount of bare soil ground
cover. Males selected areas farther from water with a high amount of vegetation, especially
shrubs. Most transmittered gartersnakes went through a transition period just prior to and
after the inactive season, during which they moved between their overwintering areas and the
ponds multiple times before settling into their core overwintering areas.
Both sexes occupied larger areas during the active season than during other seasons, and males
generally had larger activity ranges than females. Females infrequently moved >10m in a week
during the gestation season, and we often found them in the same location as the previous
week. Most snakes went through a transition period just prior to and after the inactive season
12 / 23
Results from ranked multivariate matched-pairs logistic regression models showing weighted coefficients, odds ratios, and percent change in gartersnake selection
during each season. We used multiple-model inference to obtain weighted coefficients for significant variables.
(Table 6), during which they moved between their overwintering areas and the ponds multiple
times before settling into their core overwintering areas. The length of this transition period
varied by individual snake but generally occurred in September±November and February±
April. After settling into their core overwintering areas, females rarely moved >10m during a
week, but four of six males included in the activity-range analyses regularly moved >10m
during a week in the inactive season.
13 / 23
Descriptive statistics for activity ranges (m2 on top line, ha on second) calculated by season using 95% minimum convex polygons. Inactive season was further broken
down into inactive + transition period, which includes movements at the beginning and end of the inactive season. The core inactive season represented when snakes
settled into a small overwinter area. Number of individuals (n) per season includes individuals for which activity range size plotted against number of locations reached
an asymptote (active: minimum of 6±16 locations; gestation: 3±7 locations; inactive: 3±14 locations). During the core inactive season, some males continued to move,
and activity range size plotted against number of locations did not reach an asymptote; all locations were used in these cases.
Body size, temperature, and condition
Females were larger than males. Female mass (x 222:9 20:2 g) was greater than males
(x 92:9 4:9 g; F = 6.255, p<0.001). Snake body temperature, as calculated from
transmitter pulse rate, varied by season (season: F = 418.750, df = 2, 685, p<0.001; sex: F = 2.410,
df = 1, 31, p = 0.131) and by month (month: F = 96.048, df = 11, 671, p<0.001; sex: F = 0.445,
df = 1, 32, p = 0.510) but not by sex. Females were warmest during the gestation period
(x 31:6 C); both sexes were cooler during the inactive season (female and male x 18:9 C)
compared to the active season (female x 29:3 C, male x 27:5 C; Fig 5a). On a monthly
basis, gartersnakes were warmest from May±August and coolest from December±January
Ten transmittered gartersnakes (23.8%) exhibited short-term signs of illness, including
infection at the transmitter site, a herniated transmitter, or poor body condition. To determine
if these illnesses affected our results, we compared means and standard errors of habitat
variables from locations of sick animals with the remaining data and found that the overall pattern
of habitat selection did not vary. Therefore, we included all locations in habitat analyses.
Our approach of using radio telemetry to monitor gartersnakes across seasons provides an
assessment of habitat selection for a threatened subspecies occupying a highly-managed
environment. In this study, gartersnakes displayed distinct habitat selection during three seasons:
active (March±October), gestation (April±May), and inactive (November±February). We
found that gartersnakes displayed more precise habitat selection during winter and during the
period for female gestation. Gender did not substantially contribute to differences in habitat
selection. Notably, this hatchery can be used to identify spatial and temporal patterns
important for conservation of this subspecies across seasons.
During the active season, snakes must select areas that provide resources for growth and
]. In our study, northern Mexican gartersnakes primarily selected wetland edges
14 / 23
Fig 5. Mean gartersnake body temperatures calculated from temperature-sensing transmitters by a) season and b) month. Bars show standard error and letters
represent significant differences between seasons/months from mixed-effects ANOVAs. Sex was not a significant factor.
during the active season, including active and fallow pond banks and edges. These areas
provided access to foraging opportunities and basking sites while also providing cover and
abundant rodent burrows for thermoregulation and protection from predators. Both females and
males selected sloping sites close to water with dense vegetation and low-height cover. Females
were more often found near shrubs, which might provide important cover, and males were
rarely found near trees. These results appear to be consistent with preliminary findings from
more-natural habitats in central Arizona. Our colleagues [
] found that gartersnakes selected
sloping areas at aquatic edges with dense emergent vegetation.
In our study, gartersnakes used marshy habitats in the fallow ponds to a lesser extent than
pond edges. These marshy habitats offered abundant cover and access to prey, including
amphibians. Studies of other species of gartersnakes have documented use of marshy habitat
for cover and prey [
]. Our colleagues  suggested fallow ponds at the hatchery might
be most important following amphibian breeding in spring and during the monsoon (July to
early September [
]). However, in our study, gartersnakes consistently used fallow ponds
throughout the active season, perhaps selecting more for cover characteristics than for foraging
Selection of dense cover has been documented for several species of gartersnakes [
], vipers [
], and pythons [
]. Low-height cover might be especially important
at the hatchery because it supports high numbers of predators, including raptors, herons, and
15 / 23
]. Cover along pond banks was not static during our study because hatchery
personnel occasionally trimmed or removed vegetation along banks of fish-rearing ponds. After
vegetation removal, gartersnakes responded to this habitat alteration by relocating to more
vegetated banks or to unaffected areas of the hatchery close to water.
Our data were consistent with studies that found proximity to water is important for other
species of snakes [
]. Proximity to water provides gartersnakes with foraging
opportunities and an escape from terrestrial predators. We observed both of these behaviors at the
hatchery, and gartersnakes would occasionally flee from observers into ponds. Although both
sexes generally used pond shallows, gartersnakes occasionally used deeper sections of ponds
for foraging and possibly for thermoregulation. Some studies have documented snakes using
water to regulate body temperatures [
During gestation, females exhibited similar third-order habitat selection as during the active
season but selected different fourth-order habitat features. The most notable difference was
cover. Females avoided canopy cover, and cover 1 m in height was not important. We often
observed females basking or located them underground in sites exposed to sun. Elevated body
temperatures calculated from transmitter pulse rates indicated that gestating females selected
areas for thermal qualities. Pregnant females thermoregulate more precisely and typically at
higher temperatures than non-pregnant snakes [
] and often select sites with optimal sun
exposure and heat . For example, we commonly located two females under black pond
liners where temperatures were generally warmer than the surrounding area. In addition to
thermoregulation needs, viviparous snakes also experience reduced locomotor ability due to
developing embryos [
], which presents a trade-off between thermoregulation and predator
avoidance. At the hatchery, females selected sites that appeared to satisfy both needs±close to
open areas for basking but near dense vegetation or rodent burrows for escape from predators.
During gestation, females continued to select sloping areas close to water, primarily pond
banks. Although females of many species of snakes cease foraging during the latter part of
], pregnant females often choose sites close to water. Partially because of this
lateterm feeding avoidance, post-parturient snakes often appear emaciated [
] and might
select areas close to foraging opportunities for after they give birth . Females also might
require increased water intake during gestation [
]. Another possibility is that selection of
sites near water could be important for neonates. For example, perhaps females of aquatic
species give birth near water to facilitate shedding of neonatal skin with minimal water loss or to
place neonates in close proximity to habitats with suitable prey. Postpartum females in our
study resumed activities in hatchery ponds.
During the inactive season, gartersnakes selected rocky slopes or woodlands more distant
from ponds. In comparison, our colleagues [
] found that northern Mexican gartersnakes
in more-natural areas used a variety of overwinter sites, including upland habitats, meadows,
and aquatic edges. These studies and ours indicate that northern Mexican gartersnakes
commonly overwinter in upland habitats, although riparian floodplains and water edges are also
used to a lesser extent. Use of terrestrial, upland habitats has been documented for a variety of
semi-aquatic herpetofauna [
], perhaps due to thermoregulatory benefits or to avoid
potential flooding events during the winter. However, importance of upland sites is often
overlooked for semi-aquatic species . Habitat modifications and soil compaction that occur in
these upland areas could have negative effects on overwintering success of gartersnakes.
16 / 23
Gartersnakes in our study also exhibited more precise selection of habitat parameters during
the inactive season compared to active season. Others [
] suggest that precise selection of
overwintering sites can be more important than site selection during the active season because
overwinter sites that do not provide adequate resources might result in reduced fitness or
mortality. Because of this specific selection of habitat, individuals might repeatedly use the same
overwintering sites, which has occasionally been observed in other studies with this subspecies
During the inactive season, females and males selected areas in close proximity to each
other with some differences in specific habitat features, possibly due to variation in thermal
] and subterranean characteristics [
]. Females selected areas with a high
percentage of canopy cover, whereas this variable was not as important for males. Perhaps due to
their smaller body size, male gartersnakes might have selected sites with more sun exposure
and warmth in order to maintain body temperature [
]. Body temperatures were similar
between sexes during the inactive season. Because body size and temperature are closely
linked, larger individuals maintain heat longer [
] and females might have selected sites
protected from daily temperature changes. The smaller body size of males might also have
enabled them to inhabit a wider variety of subterranean sites, whereas females might have
made use of burrow systems provided by tree roots or by rodents associated with those roots
Studies of rare species or species of conservation concern often involve conducting research
using wild-caught animals. Researchers can be faced with ethical dilemmas when balancing
the potential harm of research with the benefits gained from understanding how to conserve
species and their habitats [
]. Animal conservation research is designed to understand aspects
of an animal's biology or ecology with the hope of identifying how to safeguard these species.
Some conservation efforts have made great strides (such as understanding and mitigating
white-nose syndrome in bats [
]) by studying animals to understand their ecology and
ecosystems. Researchers must consider effects of their work on study animals and populations
Compared to other northern Mexican gartersnake research [39±41], we observed a high
occurrence of illness, mortality, and premature failure of telemetry units in transmittered
snakes, even though implant and tracking methods were similar. However, limited
information is available on the effect of transmitters on snake morbidity and survival. Telemetry could
have negatively affected survival in black ratsnakes (Elaphe obsoleta) [
]. Others [
increased infection and inflammatory response in eastern massasauga rattlesnakes (Sistrurus
catenatus catenatus) implanted with transmitters. Methods such as telemetry represent an
important trade-off between possible negative effects to individual animals or populations and
understanding species ecology and management needs [
The use of telemetry can provide researchers with data unattainable using other methods,
such as visual encounter-type surveys. Telemetry has been found to be an effective way to
track some species of snakes without impacting their movement patterns [
]. It is important
to determine which monitoring techniques are best to achieve a specific conservation goal
], such as identifying critical habitat for species recovery. Regardless of technique,
researchers should consider how robust a population is and what proportion of individuals will be
included in the study. Researchers should monitor indicators of stress during studies .
We also suggest monitoring for external signs of illness and infection by assessing body
condition periodically during research. Animals that exhibit abnormal levels of stress, signs of
17 / 23
illness, or poor body condition should, if appropriate, receive medical attention and have
monitoring devices removed. Every project is unique, and it is incumbent upon researchers to
minimize negative outcomes for populations and individuals to the best of their ability, especially
when working with imperiled species.
Conclusions and implications
Incorporating habitat needs of northern Mexican gartersnakes into development and resource
management plans is an essential component of ensuring that populations of this subspecies
are maintained or restored [
]. As the human population continues to grow, demand for
land and water will also increase , causing profound effects on riparian habitats and the
species that depend on these areas [
]. Management decisions occurring within the
range of northern Mexican gartersnakes must take into account the full range of third- and
fourth-order habitat parameters required for this subspecies, which includes needs during
different seasons and physiological periods. Conservation of this subspecies requires a
landscapelevel approach that incorporates protection of wetlands, including abundant wetland edge
habitat, and connected terrestrial upland both adjacent to and more distant from these wetlands
]. Connectivity of these areas is vital. Just prior to and after the inactive season
(October±November and February±March), many gartersnakes moved between their
overwinter sites and the ponds several times before settling into their overwinter habitat. Managers
should be aware of such movements and should stage activities (such as moving heavy
equipment or vehicles on roads around ponds and overwintering sites) to avoid disturbing
gartersnakes that may be active during these times.
Managers should maintain structural diversity of the habitat, including varying degrees of
cover density and height. For example, sites close to water with dense vegetative cover should
be provided for thermoregulation and predator avoidance during the active season. Adjacent
open or less-densely vegetated areas for basking are beneficial during the active and gestation
seasons. Rocky slopes that offer a mix of open and closed tree or shrub canopy are necessary
for the inactive season. Habitat modifications that occur in these areas, including when
gartersnakes are not currently using them, could have profound effects on individuals or the
population. Large-scale vegetation removal or soil compaction activities should be avoided.
Changes that occur in overwintering habitat may be especially detrimental, considering the
precise fourth-order habitat selection we observed. Further research to determine if this
subspecies exhibits overwinter site fidelity over multiple years would benefit management
Bubbling Ponds Hatchery is one of five known viable populations of northern Mexican
gartersnakes in the United States [
]. Therefore, management of this site for northern Mexican
gartersnake conservation is important, and suitable resources for the population should be
maintained. As habitat at the hatchery changes resulting from human activities and ecological
succession (such as invasion of trees in fallow ponds, altering the marsh-like characteristics of
this habitat), understanding how this population responds to changes in habitat can help
inform management decisions across this subspecies' range. The hatchery study site could
serve as a model for restoration or management of similar areas, especially where
human-constructed ponds provide prey and adequate cover along pond edges. However, some habitat
features provided by the hatchery area may not translate into more natural riparian and wetland
areas with lentic or intermittent flows. Further research in more natural systems can help
determine appropriate management techniques. Interestingly, this subspecies appears to be
thriving in a highly-modified and heavily-used area that also supports abundant predators.
This population provides an example of where, even as development and human activities
18 / 23
continue, the gartersnake is able to persist. We recommend careful land use planning to
maintain areas where appropriate resources provide habitat for the northern Mexican gartersnake.
S1 Fig. Northern Mexican gartersnake. A female northern Mexican gartersnake (Thamnophis
eques megalops) during the gestation season.
S1 Table. Transmittered gartersnakes included in the study. Mass was averaged for snakes
captured more than once. Snakes received internal (I), external (E), or both (I/E) types of
transmitters. Number of locations includes all locations for that gartersnake. Months tracked
not continuous for all snakes due to shed transmitters. Mean (±SE) mass for females and
males are shown in the bottom rows. A one-way t-test was used to test if female mass was
greater than male mass.
S1 Dataset. Habitat measurements collected at snake and random locations, body
We thank T. R. Jones, E. M. Nowak, and S. Cunningham for assistance with project design
and for reviewing previous versions of the manuscript. We thank T. Cotten, S. Hedwall, J.
Carter, I. Emmons, J. Edwards, and Bubbling Ponds and Page Springs hatchery staff for
providing logistical and technical support. S. Sprague, K. Dunn, A. Stromecki, C. Hartson, and
more than 100 people assisted with fieldwork. Supplies were furnished through Arizona State
University, Arizona Game and Fish Department, Bureau of Reclamation, Northern Arizona
University, and J. C. Flamand and M. Lundquist. Alpine Animal Clinic, Arizona Exotic
Animal Hospital, and Northern Arizona University's animal care staff provided veterinary care.
Conceptualization: Tiffany A. Sprague, Heather L. Bateman.
Data curation: Tiffany A. Sprague.
Formal analysis: Tiffany A. Sprague.
Funding acquisition: Heather L. Bateman.
Methodology: Tiffany A. Sprague, Heather L. Bateman.
Project administration: Heather L. Bateman.
Supervision: Tiffany A. Sprague, Heather L. Bateman.
Visualization: Tiffany A. Sprague.
Writing ± original draft: Tiffany A. Sprague.
Writing ± review & editing: Heather L. Bateman.
19 / 23
20 / 23
21 / 23
22 / 23
1. Brito JC . Seasonal variation in movements, home range, and habitat use by male Vipera latastei in northern Portugal . J Herpetol . 2003 ; 37 ( 1 ): 155 ± 160 .
2. Noss RF , O'Connell MA , Murphy DD . The science of conservation planning: habitat conservation under the Endangered Species Act . Washington, D.C.: Island Press; 1997 .
3. Morrison ML . A proposed research emphasis to overcome the limits of wildlife-habitat relationship studies . J Wildl Manage . 2001 ; 65 ( 4 ): 613 ± 623 .
4. Miller JR , Hobbs RJ . Habitat restoration±do we know what we're doing? Restor Ecol . 2007 ; 15 ( 3 ): 382 ± 390 .
5. Reinert HK . Habitat selection in snakes . In: Seigel RA , Collins JT , editors. Snakes: ecology and behavior . Caldwell: Blackburn Press; 1993 . pp. 201 ± 204 .
6. Charland MB , Gregory PT . Movements and habitat use in gravid and nongravid female garter snakes (Colubridae: Thamnophis) . J Zool . 1995 ; 236 ( 4 ): 543 ± 561 .
7. Harvey DS , Weatherhead PJ . Habitat selection as the mechanism for thermoregulation in a northern population of massasauga rattlesnakes (Sistrurus catenatus) . Ecoscience . 2010 ; 17 ( 4 ): 411 ± 419 .
8. Gibbons JW , Semlitsch RD . Activity patterns . In: Seigel RA , Collins JT , Novak SS , editors. Snakes ecology and evolutionary biology . New York: MacMillan Publishing Company; 1987 . pp. 396 ± 421 .
9. Blouin-Demers G , Weatherhead PJ . Habitat use by black rat snakes (Elaphe obsoleta obsoleta) in fragmented forests . Ecology . 2001 ; 82 ( 10 ): 2882 ± 2896 .
10. Row JR , Blouin-Demers G . Thermal quality influences habitat selection at multiple spatial scales in milksnakes . Ecoscience . 2006 ; 13 ( 4 ): 443 ± 450 .
11. Shine R . Intraspecific variation in thermoregulation, movements, and habitat use by Australian blacksnakes, Pseudechis porphyriacus (Elapidae) . J Herpetol . 1987 ; 21 ( 3 ): 165 ± 177 .
12. Burger J , Jeitner C , Jensen H , Fitzgerald M , Carlucci S , Shukla S , et al. Habitat use in basking northern water (Nerodia sipedon) and eastern garter (Thamnophis sirtalis) snakes in urban New Jersey . Urban Ecosyst . 2004 ; 7 : 17 ± 27 .
13. Halstead BJ , Wylie GD , Casazza ML . Habitat suitability and conservation of the giant gartersnake (Thamnophis gigas) in the Sacramento Valley of California . Copeia. 2010 ; 4 : 591 ± 599 .
14. Halstead BJ , Valcarcel P , Wylie GD , Coates PS , Casazza ML , Rosenberg DK . Active season microhabitat and vegetation selection by giant gartersnakes associated with a restored marsh in California . J Fish Wildl Manag . 2016 ; 7 ( 2 ): 397 ± 407 .
15. Shine R , Webb JK , Fitzgerald M , Sumner J. The impact of bush-rock removal on an endangered snake species, Hoplocephalus bungaroides . Wildlife Res . 1998 ; 25 : 285 ± 295 .
16. Santos X , Brito JC , Sillero N , Pleguezuelos JM , Llorente GA , Fahd S , Parellada X . Inferring habitatsuitability areas with ecological modeling techniques and GIS: a contribution to assess the conservation status of Vipera latastei . Biol Conserv . 2006 ; 130 ( 3 ): 416 ± 425 .
17. Dodd CK Jr. Status, conservation, and management . In: Seigel RA , Collins JT , Novak SS , editors. Snakes ecology and evolutionary biology . New York: MacMillan Publishing Company; 1987 . pp. 478 ± 514 .
18. Gibbons JW , Scott DE , Ryan TJ , Buhlmann KA , Tuberville TD , Metts BS , et al. The global decline of reptiles, deÂjà vu amphibians . Bioscience . 2000 ; 50 ( 8 ): 653 ± 666 .
19. Brode JM , Bury RB . The importance of riparian systems to amphibians and reptiles . In: Warner RE , Hendrix KM , editors. California riparian systems: ecology, conservation, and productive management . Berkeley: University of California Press; 1984 . pp. 30 ± 36 .
20. Ward JV , Tockner K , Schiemer F. Biodiversity of floodplain river ecosystems: ecotones and connectivity . Regul River . 1999 ; 15 : 125 ± 139 .
21. Zedler JB , Callaway JC , Sullivan G . Declining biodiversity: why species matter and how their functions might be restored in Californian tidal marshes . Bioscience . 2001 ; 51 ( 12 ): 1005 ± 1017 .
22. Zedler JB , Kercher S. Wetland resources: status, trends, ecosystem services, and restorability . Annu Rev Env Resour . 2005 ; 30 : 39 ± 74 .
23. Matthews KR , Knapp RA , Pope KL . Garter snake distributions in high-elevation aquatic ecosystems: is there a link with declining amphibian populations and nonnative trout introductions? J Herpetol. 2002 ; 36 : 16 ± 22 .
24. Willson JD , Winne CT , Dorcas ME , Gibbons JW . Post-drought responses of semi-aquatic snakes inhabiting an isolated wetland: insights on different strategies for persistence in a dynamic habitat . Wetlands . 2006 ; 26 ( 4 ): 1071 ± 1078 .
25. Hendrickson DA , Minckley WL . Cienegas±vanishing climax communities of the American Southwest . Desert Plants . 1984 ; 6 ( 3 ): 131 ± 175 .
26. Marshall RM , Robles MD , Majka DR , Haney JA . Sustainable water management in the southwestern United States: reality or rhetoric? PLoS One . 2010 ; 5 ( 7 ):1± 7 .
27. Naiman RJ , Decamps H. The ecology of interfaces: riparian zones . Annu Rev Ecol Syst . 1997 ; 28 : 621 ± 658 .
28. Latterell JJ , Bechtold JS , O'Keefe TC , Van Pelt R , Naiman RJ . Dynamic patch mosaics and channel movement in an unconfined river valley of the Olympic Mountains . Freshwater Biol . 2006 ; 51 : 523 ± 544 .
29. Minckley WL , Deacon JE . Battle against extinction: native fish management in the American West . Tucson: University of Arizona Press; 1991 .
30. Fagan WF , Unmack PJ , Burgess C , Minckley WL . Rarity, fragmentation, and extinction risk in desert fishes . Ecology . 2002 ; 83 : 3250 ± 3256 .
31. Holycross AT , Burger WP , Nigro EJ , Brennan TC . Surveys for Thamnophis eques and Thamnophis rufipunctatus in the Gila River Watershed of Arizona and New Mexico . Tempe (AZ): Arizona State University, School of Life Sciences; 2006 Aug. Final report . Sponsored by Arizona Game and Fish Department .
32. Endangered and threatened wildlife and plants: threatened status for the northern Mexican gartersnake and narrow-headed gartersnake . Final rule. Fed Regist . 2014 ; 79 ( 130 ): 38677 ± 38746 .
33. Rosen PC , Schwalbe CR . Status of the Mexican and narrow-headed garter snakes (Thamnophis eques megalops and Thamnophis rufipunctatus ) in Arizona. Phoenix (AZ): Arizona Game and Fish Department, Wildlife Management Division; Feb 1988 . Unpublished report to US Fish and Wildlife Service .
34. Arizona 's state wildlife action plan: 2012±2022 . Phoenix (AZ): Arizona Game and Fish Department; 2012 May. [ cited 2017 Feb 27 ]. https://www.azgfd.com/PortalImages/files/wildlife/2012-2022_ Arizona_State_Wildlife_Action_Plan.pdf.
35. Threatened and endangered species of New Mexico: 2016 biennial review . Santa Fe (NM): New Mexico Department of Game and Fish; 2016 Oct. [cited 2017 Feb 27 ]. http://www.wildlife.state.nm.us/ download/conservation/threatened-endangered-species/biennial-reviews/2016-Biennial-ReviewFINAL.pdf.
36. Johnson DH . The comparison of usage and availability measurements for evaluating resource preference . Ecology . 1980 ; 61 ( 1 ): 65 ± 71 .
37. Heisler LM , Poulin RG , Somers CM . Stop using dichotomous terms to reference observations of scale-dependents habitat selection . Landsc Ecol . 2017 ; 32 : 1531 ± 1542 .
38. Rossman DA , Ford NB , Seigel RA . The garter snakes . Norman: University of Oklahoma Press; 1996 .
39. Emmons I , Nowak E. Northern Mexican gartersnake (Thamnophis eques megalops) habitat use and ecology; monitoring surveys and radiotelemetry in the Verde Valley, Arizona . Flagstaff (AZ): Northern Arizona University, Colorado Plateau Research Station; 2016 Mar. Grant No. : I12028. Sponsored by Arizona Game and Fish Department Heritage Fund .
40. Boyarski VL , Young ME , Cotten TB . Home range and habitat use of northern Mexican gartersnakes (Thamnophis eques megalops) in a highly modified habitat . Phoenix (AZ): Arizona Game and Fish Department, Nongame Wildlife Program; 2015 Nov. Technical Report No.: 291 . Sponsored by US Fish and Wildlife Service and Arizona Game and Fish Department Heritage Fund .
41. Nowak E , Bergamini R , Emmons I. 2015 surveys and over-winter telemetry of northern Mexican gartersnakes in Lower Tonto Creek . Flagstaff (AZ): Northern Arizona University, Colorado Plateau Research Station; 2016 Feb. Interim report. Sponsored by Salt River Project.
42. Roe JH , Kingsbury BA , Herbert NR . Wetland and upland use patterns in semi-aquatic snakes: implications for wetland conservation . Wetlands . 2003 ; 23 ( 4 ): 1003 ± 1014 .
43. Camper JD . Core terrestrial habitat around wetlands: contributions from the spatial ecology of the redbelly watersnake (Nerodia erythrogaster erythrogaster) . Copeia . 2009 ; 3 : 556 ± 562 .
44. Ciucci P , Boitani L , Francisci F , Andreoli G . Home range, activity and movements of a wolf pack in central Italy . J Zool . 1997 ; 243 : 803 ± 819 .
45. Lee HJ , Lee JH , Park D. 2011 . Habitat use and movement patterns of the viviparous aquatic snake, Oocatochus rufodorsatus, from northeast Asia . Zool Sci . 2011 ; 28 ( 8 ): 593 ± 599 . https://doi.org/10. 2108/zsj.28.593 PMID: 21801000
46. Brown D. Biotic communities: southwestern United States and northwestern Mexico . Salt Lake City: University of Utah Press; 1994 .
47. LANDFIRE 1.1.0 Existing vegetation type layer [Internet]. Reston (VA): US Geological Survey . [cited 2017 Mar 21 ]. http://landfire.cr.usgs.gov/viewer.
48. Fellers GM , Drost CA. Sampling with artificial cover . In: Heyer WR , Donnelly MA , McDiarmid RW , Hayek LC , Foster MS , editors. Measuring and monitoring biological diversity: standard methods for amphibians . Washington, D.C.: Smithsonian Institution Press; 1994 . pp. 146 ± 150 .
49. Winne CT , Willson JD , Andrews KM , Reed RN . Efficacy of marking snakes with disposable medical cautery units . Herpetol Rev . 2006 ; 37 ( 1 ): 52 ± 54 .
50. Keck MB . Test for detrimental effects of PIT tags in neonatal snakes . Copeia . 1994 ; 1 : 226 ± 228 .
51. Gibbons WJ , Andrews KM . PIT tagging: simple technology at its best . BioScience . 2004 ; 54 ( 5 ): 447 ± 454 .
52. Ciofi C , Chelazzi G. 1991 . Radiotracking of Coluber viridiflavus using external transmitters . J Herpetol . 1991 ; 25 ( 1 ): 37 ± 40 .
53. Keck MB . Habitat use by semi-aquatic snakes at ponds on a reclaimed strip mine . Southwest Nat . 1998 ; 43 ( 1 ): 13 ± 19 .
54. Wylie GD , Smith JJ , Amarello M , Casazza ML . A taping method for external transmitter attachment on aquatic snakes . Herpetol Rev . 2011 ; 42 ( 2 ): 187 ± 191 .
55. Tuttle K. Natural history of the plains garter snake (Thamnophis radix) at the northern limit of its range in Alberta , Canada [thesis]. Victoria (BC): University of Victoria. 2007 .
56. Mosher K , Bateman H. The effects of riparian restoration following saltcedar (Tamarix spp.) biocontrol on habitat and herpetofauna along a desert stream . Restor Ecol . 2016 ; 24 ( 1 ): 71 ± 80 .
57. Hatton TJ , West NE , Johnson PS . Relationships of the error associated with ocular estimation and actual total cover . J Range Manage . 1986 ; 39 ( 1 ): 91 ± 92 .
58. Watson JW , McAllister KR , Pierce DJ . Home ranges, movements, and habitat selection of Oregon spotted frogs (Rana pretiosa) . J Herpetol . 2003 ; 37 ( 2 ): 292 ± 300 .
59. Keating KA , Cherry S . Use and interpretation of logistic regression in habitat-selection studies . J Wildl Manage . 2004 ; 68 ( 4 ): 774 ± 789 .
60. Johnson CJ , Nielsen SE , Merrill EH , McDonald TL , Boyce MS . Resource selection functions based on use±availability data: theoretical motivation and evaluation methods . J Wildl Manage . 2006 ; 70 ( 2 ): 347 ± 357 .
61. Duchesne T , Fortin D , Courbin N. Mixed conditional logistic regression for habitat selection studies . J Anim Ecol . 2010 ; 79 : 548 ± 555 . https://doi.org/10.1111/j.1365- 2656 . 2010 . 01670 . x PMID : 20202010
62. Compton BW , Rhymer JM , McCollough M. Habitat selection by wood turtles (Clemmys insculpta): an application of paired logistic regression . Ecology . 2002 ; 83 ( 3 ): 833 ± 843 .
63. Gorman TA , Haas CA. Seasonal microhabitat selection and use of syntopic populations of Lithobates okaloosae and Lithobates clamitans clamitans . J Herpetol . 2011 ; 45 ( 3 ): 313 ± 318 .
64. Powell RA . Animal home ranges and territories and home range estimators . In: Boitani L, Fuller T , editors. Research techniques in animal ecology: controversies and consequences . New York City: Columbia University Press; 2000 . pp. 65 ± 110 .
65. Burgman MA , Fox JC . Bias in species range estimates from minimum convex polygons: implications for conservation and options for improving planning . Anim Conserv . 2003 ; 6 : 19 ± 28 .
66. Rogers KB , White GC . Analysis of movement and habitat use from telemetry data . In: Guy SC , Brown ML , editors. Analysis and interpretation of freshwater fisheries data . Bethesda: American Fisheries Society; 2007 . pp. 625 ± 676 .
67. McCune B , Grace JB , Urban DL . Analysis of ecological communities . Gleneden Beach: MjM Software Design; 2002 .
68. Ramette A . Multivariate analyses in microbial ecology . FEMS Microbiol Ecol . 2007 ; 62 ( 2 ): 142 ± 160 . https://doi.org/10.1111/j.1574- 6941 . 2007 . 00375 . x PMID : 17892477
69. Kaiser HF . The application of electronic computers to factor analysis . Educ Psychol Meas . 1960 ; 20 ( 1 ): 141 ± 151 .
70. Hosmer DW , Lemeshow S , Sturdivant RX . Applied logistic regression . 3rd ed. Hoboken: John Wiley and Sons; 2013 .
71. Altman M , Gill J , McDonald MP . Convergence problems in logistic regression . In: Shewhart WA , Wilks SS , editors. Numerical issues in statistical computing for the social scientist . Hoboken: John Wiley and Sons, Inc.; 2004 . pp. 238 ± 252 .
72. Burnham KP , Anderson DR . Multimodel inference: understanding AIC and BIC in model selection . Sociol Method Res . 2004 ; 33 ( 2 ): 261 ± 304 .
73. Zar JH . Biostatistical analysis . 4th ed. Upper Saddle River: Prentice Hall; 1999 .
74. Grantz K , Rajagopalan B , Clark M , Zagona E. Seasonal shifts in the North American monsoon . J Clim . 2007 ; 20 : 1923 ± 1935 .
75. Slip DJ , Shine R . Habitat use, movements and activity patterns of free-ranging diamond pythons, Morelia spilota spilota (Serpentes: Boidae): a radiotelemetric study . Aust Wildlife Res . 1988 ; 15 : 515 ± 531 .
76. Osgood DW . Thermoregulation in water snakes studied by telemetry . Copeia . 1970 ; 3 : 568 ± 571 .
77. Nelson KJ , Gregory PT . Activity patterns of gartersnakes, Thamnophis sirtalis, in relation to weather conditions at a fish hatchery on Vancouver Island, British Columbia . J Herpetol . 2000 ; 34 ( 1 ): 32 ± 40 .
78. Gier PJ , Wallace RL , Ingermann RL . Influence of pregnancy on behavioral thermoregulation in northern Pacific rattlesnake Crotalus viridis oreganus . J Exp Biol . 1989 ; 145 : 465 ± 469 .
79. Reinert HK , Zappalorti RT . Timber rattlesnakes (Crotalus horridus) of the Pine Barrens: their movement patterns and habitat preference . Copeia . 1988 ; 4 : 964 ± 978 .
80. Seigel RA , Huggins MM , Ford NB . Reduction in locomotor ability as a cost of reproduction in gravid snakes . Oecologia . 1987 ; 73 ( 4 ): 481 ± 485 . https://doi.org/10.1007/BF00379404 PMID: 28311962
81. Gregory PT , Crampton LH , Skebo KM . Conflicts and interactions among reproduction, thermoregulation and feeding in viviparous reptiles: are gravid snakes anorexic? J Zool. 1999 ; 248 : 231 ± 241 .
82. Gregory PT , Skebo KM . Trade-offs between reproductive traits and the influence of food intake during pregnancy in the gartersnake, Thamnophis elegans . Am Nat . 1998 ; 151 ( 5 ): 477 ± 486 . https://doi.org/ 10.1086/286134 PMID: 18811321
83. Shine R. Reproduction in Australian elapid snakes II: female reproductive cycles . Aust J Zool . 1977 ; 25 : 655 ± 666 .
84. Semlitsch RD , Bodie JR . Biological criteria for buffer zones around wetlands and riparian habitats for amphibians and reptiles . Conserv Biol . 2003 ; 17 ( 5 ): 1219 ± 1228 .
85. Huey RB . Physiological consequences of habitat selection . Am Nat . 1991 ; 137 :S91± S115 .
86. Burger J , Zappalorti RT , Gochfeld M , Boarman WI , Caffrey M , Doig V , et al. Hibernacula and summer den sites of pine snakes (Pituophis melanoleucus) in the New Jersey Pine Barrens . J Herpetol . 1988 ; 22 ( 4 ): 425 ± 433 .
87. Rudolph DC , Schaefer RR , Burgdorf SJ , Duran M , Conner RN . Pine snake (Pituophis ruthveni and Pituophis melanoleucus lodigni) hibernacula . J Herpetol . 2007 ; 41 ( 4 ): 560 ± 565 .
88. Huey RB , Peterson CR , Arnold SJ , Porter WP . Hot rock and not-so-hot rocks: retreat-site selection by garter snakes and its thermal consequences . Ecology . 1989 ; 70 ( 4 ): 931 ± 944 .
89. Stevenson RD . Body size and limits to the daily range of body temperature in terrestrial ectotherms . Am Nat . 1985 ; 125 ( 1 ): 102 ± 117 .
90. Shine R , Olsson MM , Lemaster MP , Moore IT , Mason RT . Effects of sex, body size, temperature, and location on the antipredator tactics of free-ranging gartersnakes (Thamnophis sirtalis, Colubridae) . Behav Ecol . 2000 ; 11 ( 3 ): 239 ± 245 .
91. Panuska JA , Wade NJ . The burrow of Tamias striatus . J Mammal . 1956 ; 37 ( 1 ): 23 ± 31 .
92. Kinlaw A. A review of burrowing by semi-fossorial vertebrates in arid environments . J Arid Environ . 1999 ; 41 ( 2 ): 127 ± 145 .
93. Reeder DM , Field KA , Slater MH . Balancing the costs of wildlife research with the benefits of understanding a panzootic disease, white-nose syndrome . ILAR J . 2016 ; 56 ( 3 ): 275 ± 282 . https://doi.org/10. 1093/ilar/ilv035 PMID: 26912714
94. White GC , Garrott RA . Analysis of wildlife radio-tracking data . San Diego: Academic Press; 1990 .
95. Langkilde T , Shine R . How much stress do researchers inflict on their study animals? A case study using a scincid lizard, Eulamprus heatwolei . J Exp Biol . 2006 ; 209 : 1035 ± 1043 . https://doi.org/10.1242/ jeb.02112 PMID: 16513929
96. Weatherhead PJ , Blouin-Demers G . Long-term effects of radiotelemetry on black ratsnakes . Wildl Soc Bull . 2004 ; 32 ( 3 ): 900 ± 906 .
97. Lentini AM , Crawshaw GJ , Licht LE , McLelland DJ . Pathologic and hematologic responses to surgically implanted transmitters in eastern massasauga rattlesnakes (Sistrurus catenatus catenatus) . J Wildl Dis . 2011 ; 47 ( 1 ): 107 ± 125 . https://doi.org/10.7589/ 0090 -3558- 47 .1.107 PMID: 21270001
98. Cooke SJ . Biotelemetry and biologging in endangered species research and animal conservation: relevance to regional, national, and IUCN Red List threat assessments . Endanger Species Res . 2008 ; 4 : 16 ± 185 .
99. Hale VL , MacGowan B , Corriveau L , Huse DC , Currylow AFT , Thompson S. Radio transmitter implantation and movement in the wild timber rattlesnake (Crotalus horridus) . J Wildl Dis . 2017 ; 53 ( 3 ): 591 ± 595 . https://doi.org/10.7589/2016-08-186 PMID: 28192045 100 . Wilson RP , McMahon CR . Measuring devices on wild animals: what constitutes acceptable practice? Front Ecol Environ . 2006 ; 4 ( 3 ): 147 ± 154 .