Refining Reintroduction of Whooping Cranes with Habitat Use and Suitability Analysis

Abstract

A successful species reintroduction depends, in part, on the correct identification of suitable habitats. In cases where a species has been extirpated from a region for decades, however, this task is fraught with uncertainty. Uncertainty can be minimized and adjusted for by monitoring and adaptive management. The central goal of this study was to identify reintroduction sites that facilitate dispersion of whooping cranes (Grus americana), a federally listed endangered species, into optimal habitat as quickly as possible. First, we described the habitat selection of breeding home ranges for reintroduced whooping cranes in and around Necedah National Wildlife Refuge of central Wisconsin. We defined home ranges as 95% fixed spatial kernel density estimates from location data gathered from nesting whooping cranes from April through July 2005–2010. Whooping crane home ranges contained more emergent herbaceous wetlands than expected by chance and less developed or barren land, forest, and scrubland. Breeding whooping crane home ranges usually were composed of distinct nesting territories and off-territory elements; when moving off-territory, cranes decreased wetland selection and increased selection for open uplands. Second, we used habitat composition values and strength of selection (as determined by Jacob's index) to create a habitat suitability map to identify potential habitats that breeding whooping cranes could use in unoccupied eastern Wisconsin. With this method, we identified 2 large suitable wetland complexes within our study area associated with the Fox and Rock Rivers. Based on this analysis, the Whooping Crane Eastern Partnership began releasing whooping cranes into White River Marsh State Wildlife Area and Horicon National Wildlife Refuge in 2011. © 2014 The Wildlife Society.

Full text

Research Article
Refining Reintroduction of Whooping Cranes
with Habitat Use and Suitability Analysis
NATHAN D. VAN SCHMIDT,
1,2
International Crane Foundation, E-11376 Shady Lane Road, Baraboo, WI 53913, USA
JEB A. BARZEN, International Crane Foundation, E-11376 Shady Lane Road, Baraboo, WI 53913, USA
MIKE J. ENGELS, International Crane Foundation, E-11376 Shady Lane Road, Baraboo, WI 53913, USA
ANNE E. LACY, International Crane Foundation, E-11376 Shady Lane Road, Baraboo, WI 53913, USA
ABSTRACT A successful species reintroduction depends, in part, on the correct identification of suitable
habitats. In cases where a species has been extirpated from a region for decades, however, this task is fraught
with uncertainty. Uncertainty can be minimized and adjusted for by monitoring and adaptive management.
The central goal of this study was to identify reintroduction sites that facilitate dispersion of whooping cranes
(Grus americana), a federally listed endangered species, into optimal habitat as quickly as possible. First, we
described the habitat selection of breeding home ranges for reintroduced whooping cranes in and around
Necedah National Wildlife Refuge of central Wisconsin. We defined home ranges as 95% fixed spatial kernel
density estimates from location data gathered from nesting whooping cranes from April through July 2005–
2010. Whooping crane home ranges contained more emergent herbaceous wetlands than expected by chance
and less developed or barren land, forest, and scrubland. Breeding whooping crane home ranges usually were
composed of distinct nesting territories and off-territory elements; when moving off-territory, cranes
decreased wetland selection and increased selection for open uplands. Second, we used habitat composition
values and strength of selection (as determined by Jacob’s index) to create a habitat suitability map to identify
potential habitats that breeding whooping cranes could use in unoccupied eastern Wisconsin. With this
method, we identified 2 large suitable wetland complexes within our study area associated with the
Fox and Rock Rivers. Based on this analysis, the Whooping Crane Eastern Partnership began releasing
whooping cranes into White River Marsh State Wildlife Area and Horicon National Wildlife Refuge in
2011. Ó2014 The Wildlife Society.
KEY WORDS adaptive management, Grus americana, habitat selection, habitat suitability, home range,
reintroduction, whooping crane, Wisconsin.
Species reintroductions can be powerful tools for conserva-
tion biology, especially in cases of reestablishment where
populations have been extirpated. Key to the success of a
reintroduced population is often the presence of suitable
habitat (Armstrong and Seddon 2008). Identification of
suitable habitat prior to a reintroduction effort can be
difficult, especially for species whose historical distributions
were poorly known and/or geographically diverse. By
monitoring habitat use after a reintroduction, however,
the relative suitability of different habitats and landscapes
within a region can be assessed. Adaptive management can
be used to refine our understanding of what habitat is truly
suitable by placing individuals in different landscapes
or modifying habitat features during a reintroduction
and comparing individuals’ success in each environment
(Armstrong and Seddon 2008).
Whooping cranes (Grus americana) are a federally listed
endangered species with only 1 self-sustaining wild popula-
tion of whooping cranes, which breeds in Wood Buffalo
National Park in Saskatchewan, Canada and winters at
Aransas National Wildlife Refuge in Texas, USA. This
population has recovered from a nadir of 15 or 16 individuals
in the winter of 1941–1942 to just over 300 individuals in
early 2014 (Canadian Wildlife Service and U.S. Fish and
Wildlife Service 2005, Harrell 2014). Establishing a second
self-sustaining population is a primary goal of the whooping
crane recovery plan (Canadian Wildlife Service and U.S.
Fish and Wildlife Service 2005). The first reintroduction
effort, a migratory flock with breeding habitat in the Rocky
Mountains, was established in 1975 but discontinued in
1989. Currently, 3 established reintroduced populations
exist, but because none of these reintroduction efforts have
yet produced a self-sustaining population, reintroduction
techniques for this species remain unproven (Folk et al. 2010,
Urbanek et al. 2010, Moore et al. 2012, Stehn 2012). Two of
the surviving reintroduced populations are non-migratory: 1
in central Florida (established 1993) and 1 in Louisiana
(established 2011). The third surviving reintroduced
Received: 7 December 2013; Accepted: 9 July 2014
1
E-mail: nvanschmidt@gmail.com
2
Present address: University of California–Berkeley, 130 Mulford Hall
#3114, Berkeley, CA 94720, USA
The Journal of Wildlife Management; DOI: 10.1002/jwmg.789
Van Schmidt et al. Whooping Crane Habitat Use and Suitability 1

population, the Eastern Migratory Population (EMP),
migrates between Wisconsin and the southeastern United
States and is the focus of this study.
Reintroduction of the EMP began in 2001 and, through
the release of captive-raised young, has grown steadily to a
population of over 100 individuals by the end of 2010
(Garland 2012). Birds from this population migrate between
breeding areas centered on Necedah National Wildlife
Refuge (NWR) in central Wisconsin and wintering areas
ranging from southern Illinois to the gulf coast (Converse
et al. 2012). The EMP has a high annual survival rate of
0.877 for unpaired birds and 0.991 for paired birds (Converse
et al. 2012). Productivity, however, has been low because of
nest abandonment (Urbanek et al. 2010, King et al. 2013).
ThechoiceofNecedahNWRincentralWisconsinasthe
first release area for the EMP was based on best-available
knowledge of whooping crane habitat requirements as
estimated from historical distribution data and studies of
nesting sites at Wood Buffalo National Park in the Canadian
taiga (Cannon 1999). Allen (1952) described 3 historical
breeding ranges: the taiga (where whooping cranes still breed),
the upper tallgrass prairie region, and a non-migratory
population in coastal Louisiana. Estimated habitat require-
ments for the EMP were limited because of the absence of the
species from the upper tallgrass prairie region for 120 years.
Historical records suggest that core whooping crane summer
habitat occurred in southern Wisconsin along an ecological
tension zone (Fig. 1; Allen 1952) between the northern boreal
forest and the tallgrass prairie zones (Curtis 1959). However,
sighting records from Wisconsin at this time are sparse and
likely represent remnants of a population that may by then have
been biased towards habitats remote from human contact
(Allen 1952). Furthermore, small geographic deviations in
central Wisconsin can result in large habitat changes because
these habitats lay along a transition zone between strikingly
different biomes (Curtis 1959). Lastly, wetland drainage and
agricultural development have driven land-use change since
cranes were extirpated in the 1800s. Nesting habitats for the
wild population at Wood Buffalo National Park have been
described by Timoney (1999), but this landscape lies entirely in
the taiga and has no significant human land use. Conversely,
the upper tallgrass prairie region features no taiga and human
alterations are extensive. Habitat use by whooping cranes in
Wisconsin may therefore differ significantly from the habitats
used by the Wood Buffalo population, and what habitat is
suitable in this region remains unexamined.
As a result, the initial choice of the reintroduction site was
conjectural regarding what habitat was suitable. Historically,
whooping cranes may have used habitats that have never
been defined and may no longer be present in the Midwest.
Given these unknowns, whooping cranes were released in
wetlands near the tallgrass prairie-boreal forest tension zone
(Fig. 1) so that the cranes could disperse relatively short
distances to find appropriate habitat types in these widely
different ecological regions (Keys et al. 1995). Observations
of habitat use and selection, survivorship, and ultimately
productivity in these different regions can be used to refine
habitat models for future releases efforts.
The low productivity of whooping cranes in Necedah
NWR necessitates such a reassessment. Although Necedah’s
wetland vegetation communities are thought to be appro-
priate for whooping cranes (Maguire 2008), unproductive
soils and harassment by abundant avian-feeding black flies
(Simulium annulus) may be constraining crane productivity
there (Urbanek et al. 2010). Facilitated dispersion to other
nearby wetland complexes may improve the chances that
cranes will establish nests in productive regions and hasten
testing the suitability of a variety of possible environments.
By analyzing the habitat use and selection of nesting
whooping cranes in the Necedah region, we seek to better
estimate what nearby areas are appropriate for new
reintroduction sites.
Any analysis of habitat use and selection, however, must
consider constraints caused by the philopatric nature of
whooping cranes and the uneven distribution of habitats in
the landscape in and around Necedah NWR. Whooping
cranes tend to establish nests in geographic regions near to
where they fledged (Maguire 2008). From the perspective of
habitat use and selection analysis, philopatry means that
habitat use data are highly influenced by the place where
whooping crane chicks were released or fledged naturally and
do not represent a selection model where all suitable habitats
within the broader study area were equally available, skewing
estimates of habitat preference or avoidance. Though
whooping cranes were released at Necedah NWR because
wetlands within this refuge were predicted to be good habitat
(Cannon 1999), if some habitats were rare near the release
site but more suitable than those found at Necedah NWR,
Figure 1. The Central Wisconsin study area (left circle; extent of landscape
classified as available to whooping cranes in this study) including the
Necedah National Wildlife Refuge (NWR) and East Wisconsin study area
(right rectangle) as well as the southern vegetation community (mostly tall
grass prairie and oak opening) – northern vegetation community (mostly
northern hardwood and evergreen forest) tension zone from Curtis (1959).
2 The Journal of Wildlife Management 9999

they might not be chosen until dispersion from the area
occurred. Thus, philopatry and habitat rarity near the release
sites may combine to bias estimates of habitat selection.
We examined home range habitat use and selection of
reintroduced, territorial whooping cranes during nesting in
central Wisconsin. To address philopatric bias and the rarity
of certain habitat types near their release site, we created a
weighting system that modeled the landscape closer to the
whooping crane reintroduction release site as being more
available for habitat selection. From this analysis, we created a
habitat suitability model that identified the relative suitability
of areas in eastern Wisconsin for nesting whooping crane
home ranges and evaluated potential new release sites.
STUDY AREA
The central Wisconsin study area (hereafter, Central
Wisconsin), the current breeding area of the EMP, occurred
primarily within the 177 km
2
Necedah NWR (Fig. 1). Open
water, emergent herbaceous wetland, and woody wetland
predominated in the managed flowages, whereas dense
mixed forest, shrub communities, and graminoid communi-
ties dominated the uplands within Necedah NWR. Except
for the small towns of Necedah and Tomah, the area was
mostly isolated from high levels of development. Scattered
row croplands, cranberry bogs, and pastures were located to
the east, west, and north of the refuge. South of the refuge
had more extensive cropland and pastures. The 4 nesting
territories located outside of Necedah NWR were in areas
with a surrounding landscape of mixed wetlands and
agriculture areas. Individual whooping cranes moved
extensively over this heterogeneous landscape in the years
before creating their nesting territories (Whooping Crane
Eastern Partnership, unpublished data).
A 20,218-km
2
area of eastern Wisconsin was considered for
new reintroduction locations (Fig. 1). The eastern Wisconsin
study area (hereafter, East Wisconsin) was chosen for
evaluation because it existed outside the range of avian
feeding black files (1 of the hypothetical constraints on
whooping crane reproduction; Adler 2010), was already used
by non-breeding EMP birds in summer (Whooping Crane
Eastern Partnership, unpublished data), had more productive
soils than the sandy soils of Necedah NWR, was close enough
to allow for interaction with the Central Wisconsin birds, and
contained numerous emergent wetlands but few forests. The
landscape within East Wisconsin was primarily agricultural,
dominated by row crops or pasture, and interspersed with
wetlands, lakes, and grasslands. A number of large towns and
cities were within East Wisconsin. Though the majority of
the land was privately owned, some sizable publicly owned
lands that encompassed extensive wetlands were also present.
METHODS
Radiotracking of Nesting Whooping Cranes
We identified whooping crane home ranges from 2005 to
2010 by monitoring color and radio marked birds. All
whooping cranes in the EMP had unique colored leg bands
and leg-mounted very high frequency (VHF) transmitters
for individual identification (Melvin et al. 1983). Band mass
accounted for at most 2.8% lean body mass, as the lightest
bird measured was 3.7 kg and maximum transmitter mass
was 0.105 kg.
We tracked all nesting crane pairs during the breeding
season (Apr through Jul) 2005–2010. No pairs nested before
April or after July. We defined nesting pairs as those with a
nest and at least 1 egg. For each pair, we used all locations
gathered within these 4 months regardless of whether they
were actively nesting. We did not include territorial pairs that
built a nest mound but did not lay eggs because they could
not be distinguished from temporary pairs that did not
persist in a well-defined territory. Average time between
determinations of a pair’s location was 3.3 11.1 (mean 
SD; mode ¼1) days. We radiotracked birds from vehicles
with roof-mounted Yagi antennas and receivers (Advanced
Telemetry Systems, Isanti, MN) as well as with aircraft
carrying H antennas mounted on wing struts. We confirmed
most crane locations visually and plotted locations on aerial
photos (within 200 m). When we did not see tracked birds,
we determined bird locations via triangulation, obtaining 3
separate bearings for the bird within an hour to reduce error
caused by a moving bird. We discarded triangulations if error
polygons were >3.5 ha. Under the assumption that only 1
member of a pair would chose a location, we used 1
coordinate to describe where both members of a pair were
sighted together. We knew nest locations for all pairs and
included them as location data.
Kernel Density Estimation of Home Ranges
Of the 115 cranes in the EMP, only 18 crane pairs nested
over the course of our study (Table 1). Pairs that nested in
multiple years (n¼13) reused approximately the same home
range each year. We therefore refer to a pair’s home range in a
single year as a home range-year. We calculated home range-
years as 95% confidence kernel density contours. We
calculated final estimates of each pair’s home range size,
habitat use, and habitat selection as weighted averages from
all years, with each year weighted by the number of location
data in that home range-year. We refer to this weighted
average as the pair’s home range. Two pairs (3a and 4a;
Table 1) split up during the study and 1 member of the
original pair formed a new pair with a different mate (3b and
4b) but reused approximately the same home range. Because
these 2 home ranges were not independent, we averaged the
home range-years from both the a and b pairings together for
final estimates, treating them as a single home range. Thus,
our final number of unique home ranges was 16.
Whooping cranes often flew large distances (sometimes
>20 km) away from their nests to forage without using
habitats between these remote foraging areas and the area
where the nest was located. This resulted in many home
ranges being composed of multiple spatially discontinuous
95% kernel density contours. To test if habitat use and
selection differed between nesting territories and remote
foraging locations, we further defined the nesting territory as
the 95% kernel density contour surrounding that pair’s nest
location for that year (Fig. 2). We also included any parts of
Van Schmidt et al. Whooping Crane Habitat Use and Suitability 3

the home range-year that were not spatially continuous
with the nest but were within 2,660 m from the nest (the
maximum observed distance in territories continuous with
the nest). We refer to parts of the home ranges located
outside of the nesting territory as off-territory elements.
Although we lacked data on territorial defense behaviors to
confirm the exact area that constituted a defended nesting
territory, breeding cranes were more frequently observed in
family-only groups within nesting territories and in larger,
multi-family groups within off-territory elements (x
2
¼96.3,
P<0.001), strongly suggesting this was a valid approxima-
tion of the true nesting territory.
We defined home range-years as 95% confidence fixed
spatial kernel density estimates calculated with Hawth’s
Tools (Hawth’s Tools v3.27, www.spatialecology.com/
htools/, accessed 18 Jun 2012) in ArcGIS 9.31 (Environ-
mental Systems Research Institute, Inc., Redlands, CA). We
calculated smoothing parameters (h) individually for each
home range-year (n¼46) with Animal Space Use (Animal
Space Use v1.3, http://www.cnr.uidaho.edu/population_e-
cology/animal_space_use, accessed 18 Jun 2012) and then
created the kernel density estimates using the mean value
(

h¼612 m) of all home range-years. We used likelihood-
cross-validation (CVh) to calculate smoothing parameters,
as this performs better than least-squares cross-validation
with <50 locations per animal, as was the case for most of our
home ranges (Horne and Garton 2006).
We graphed the frequency of locations against home
range-year size to determine the number of locations needed
to accurately delineate a home range-year. A reasonable
asymptote was reached after 10 locations: mean (SD) sizes
for home range-years estimated with 3–9 locations (n¼6),
10–30 locations (n¼20), and 30þlocations (n¼20) were
2.55 km
2
1.37, 3.62 km
2
1.35, and 3.99 km
2
1.18,
respectively. Although this minimum number of points is
lower than that often recommended (30) for precise
estimates of home range size, our use of CVh for bandwidth
estimation should correct for some of the size overestimation
introduced by using 10–30 points (Horne and Garton 2006),
and variation in home range size is often more driven by
individual-to-individual variation than by estimation meth-
Table 1. Pairs, years, and number of locations sampled for breeding
whooping crane home ranges in central Wisconsin during the breeding
season (Apr–Jul), 2005–2010.
Home
range Female
a
Male
a
Years
nested
Mean no. annual
locations
1 2–02 1–01 2005–2006 21.0
2 17–02 11–02 2005–2009 42.6
3A
b
3–02 17–03 2006 48.4
3B
b
3–03 17–03 2007–2010
4A
b
9–02 2–03 2006 10.0
4B
b
9–02 16–04 2007–2008
c
5 18–02 13–02 2006–2010 59.2
6 15–04 5–05 2008–2010
c
9.67
7 9–03 3–04 2008–2010 63.7
8 19–05 8–04 2008–2010 8.7
9 12–03 11–03 2008–2010 14.7
10 8–05 1–04 2008–2010 17.7
11 19–04 12–02 2008–2010 18.7
12 13–03 18–03 2008–2010 27.7
13 W1–06 10–03 2009–2010 64.0
14 1–05 5–01 2009 15.0
15 27–05 12–04 2010 28.0
16 46–07 2–04 2010 77.0
a
Individual identification codes describe birth order-year born; W
indicates wild hatched.
b
Individual had multiple mates but was considered to represent 1 home
range because they renested in approximately the same area with the new
mates.
c
One year excluded from analysis because of small number of location data
(<10) and estimated home range size (<2km
2
).
Figure 2. Example of a whooping crane home range, territory, and off-
territory element as delineated in this study. We defined nesting territories as
the part of the home range that contained the nests (some pairs, like this one,
re-nested). This nesting territory was located within Necedah National
Wildlife Refuge wetlands in central Wisconsin, whereas the off-territory
elements of the home range were located outside of the refuge in croplands to
the south.
4 The Journal of Wildlife Management 9999

od (Bo
¨rger et al. 2006). Nine of the 16 pairs had 30
locations for at least 1 home range-year, and our approach of
using weighted averages over multiple years for final home
range estimation should further reduce the influence of any
substandard estimation in any 1 year. Thus, we felt the
benefits of including additional home range-years estimated
with <30 location data outweighed the costs. All 6 home
range-years that had <10 location data were from pairs that
had other home range-years with at least 10 location data, so
their inclusion or exclusion did not change the final sample
size for the number of home ranges. To maximize the
amount of data included on this rare species, we kept in our
dataset 4 of these 6 home range-years that had <10 locations
but were within the size range (>2km
2
) of home range-years
with 10 locations. Conducting the analysis without these 4
home range-years did not change the results. Overall, we
excluded only 2 of the 46 home range-years from the
remaining analysis because both number of locations and
estimated home range size were too small (Table 1).
Analysis of Habitat Use and Selection
We derived habitat characteristics from the National Land
Cover Database 2006, a 30-m resolution raster dataset with
15 land cover classes in Wisconsin: open water, emergent
herbaceous wetland, woody wetland, grassland/herbaceous,
shrub/scrub, deciduous forest, evergreen forest, mixed forest,
pasture/hay, cultivated crops, barren land, developed open
space, developed low intensity, developed medium intensity,
and developed high intensity (Fry et al. 2011). In the Great
Lakes region, accuracy for this dataset was assessed at 87.6%
for Level I land cover classes (e.g., forest) and 81.6% for
Level II land cover classes (e.g., deciduous forest; Wickham
et al. 2010). To increase accuracy further, we examined the
land cover data and manually corrected errors we saw in the
Necedah region based on our knowledge of the area.
We consolidated these 15 land use classes into 9 classes based
on habitat affinities of whooping cranes to increase statistical
power for habitat selection. We grouped developed open space
and developed low intensity into developed low and developed
medium intensity and developed high intensity into developed
high, because these land cover types were relatively rare in the
Necedah region. We also grouped deciduous forest, evergreen
forest, and mixed forest into forest, because we did not
hypothesize upland forests would be used extensively by
whooping cranes (Timoney 1999). Finally, we reclassified
grassland/herbaceous, pasture/hay, and cultivated crops together
into upland open. Combining these land uses was based on
extensive field observations during the course of this study that
suggested whooping cranes used grouped land use classes in
similar ways. To analyze habitat use, we calculated the total and
proportional cover of land classes for each home range-year with
FragStats (FragStats v4.0, www.umass.edu/landeco/research/
fragstats/fragstats.html, accessed 4 Jul 2012).
We quantified habitat selection with Jacob’s index:
D¼(rp)/(rþp2rp), where rwas the percent a given
habitat class was found within the home range and pwas the
percent of this habitat available within the broader landscape.
Delineating what habitat in the Central Wisconsin landscape
was available to whooping cranes was difficult because the
EMP is a small and expanding population and because
whooping cranes are philopatric. Thus, habitat on the edge
of the population may be high quality but simply not used
because there were not enough birds to occupy all available
habitats. To adjust for unsaturated occupancy and philopatry,
we created a function that weighted the landscape closer to
the whooping crane release sites as more available. We first
created a spatial kernel density estimate (as above) for all
observed breeding whooping cranes in the EMP. We then
created contours at 5% spatial confidence intervals from 5%
to 95% and transformed these into minimum concentric
circular buffers centered on the spatial mean of the Necedah
NWR release sites. We then multiplied the amount of
available habitat in each ring by 1–ring’s confidence (e.g., the
second to last ring, representing 90% confidence, was
weighted at 10%). We excluded areas beyond the 95% spatial
confidence interval ring. We then calculated the overall
amount of available habitat in a given class on the landscape
as the sum of these weighted rings.
Following the definition of Aarts et al. (2008), we refer to
habitats receiving more usage than expected by chance
(taking into account accessibility) as preferred and areas
receiving less as avoided. A Jacob’s index value of 1 indicated
complete preference and 1 indicated complete avoidance
(Jacobs 1974). To test if cranes exhibited habitat selectivity,
we constructed 95% 2-tailed confidence intervals to
determine if the home range Jacob’s indices differed
significantly from 0 (no selection). For land cover classes
where Jacob’s index was not normally distributed, we used
nonparametric Wilcoxon signed-rank tests. We used paired
t-tests to test for differences in selection between nesting
territory and off-territory elements of the home ranges; we
used nonparametric paired Wilcoxon signed-rank tests when
differences in means were not normally distributed. We
excluded 1 home range that did not have off-territory
elements from these pairwise comparisons but included it in
estimation of average territory size.
Analysis of Habitat Suitability
We derived suitability of habitats for nesting home ranges in
East Wisconsin based on their similarity to preferred habitats
and dissimilarity to avoided habitats for reintroduced breeding
whooping cranes in Central Wisconsin, as determined in the
previous section. We analyzed habitat characteristics of East
Wisconsin with a moving window analysis in FragStats. The
moving window was a 3.66-km
2
circle, which was the mean
home range size rounded to the 30-m resolution of the raster.
The moving window produced a raster map depicting the
percent for each land cover class within the window for each
pixel in the raster. We then summarized these data into a
habitat suitability map by multiplying the percent of each land
cover class at each pixel by its Jacob’s index selection coefficient
and summing the results for all land cover classes. We excluded
from the calculation land cover classes that were not
significantly preferred or avoided.
Because this map was a sum of whooping cranes’ preference
and avoidance metrics, it can be interpreted as a map of the
Van Schmidt et al. Whooping Crane Habitat Use and Suitability 5

relative likelihood of a crane selecting each pixel for the
center of a circular home range of average size. For ease of
interpretation, we used the transformation (suitability 
minimum suitability)/maximum suitability to reclassify our
suitability ranks on a relative scale from 0 (very unsuitable) to
1 (very suitable). Qualitative assessments of how these
relative rankings related to actual suitability were based on
the range of suitability values observed in existing home
ranges (see Results for details).
Finally, because open water was not preferred or avoided in
central Wisconsin and thus not included in the original
moving window analysis, we separately set any large deep-
water ponds, lakes, and rivers (waters that were too deep for
whooping cranes to wade in) to 0 suitability using the United
States Geological Survey open water dataset (U.S. Geologi-
cal Survey 1989).
RESULTS
Habitat Use and Selection
We analyzed data from 16 unique home ranges (averaged
from 44 home range-years). Mean (SD) size was
3.68 0.95 km
2
for home ranges, 2.69 1.03 km
2
for nesting
territories, and 1.27 0.96 km
2
for off-territory elements.
Emergent herbaceous wetland was the largest component of
habitat in both home ranges and nesting territories, and was
the only land cover class used more than expected based on its
availability (P¼0.009; Table 2). Whooping cranes used
woody wetlands, open water, and upland open habitats in
proportions as expected based on their availability. All other
habitat classes were used by whooping cranes significantly less
than expected based on their availability (Table 2).
Habitat selection differed significantly between nesting
territories and off-territory elements of the home ranges
(Table 3). As cranes moved off nesting territories, they
selected less of all 3 wetland-type habitats and more upland
open habitats. Their avoidance of other land cover classes did
not change between territories and off-territory elements.
Habitat Suitability
Habitat suitability rankings in the Central Wisconsin study
area (Fig. 3) were heterogeneous, a result of the tight linking
of wetlands, forests, and scrublands. Observed home ranges
generally contained raster cells ranked as high relative
suitability for the establishment of home ranges: the mean
(SD) maximum value within home ranges was 0.75 0.05
and the lowest maximum value within a home range was
0.63. We report maximum values here because the value of a
raster cell indicates the suitability of a 3.66-km
2
circular
home range centered on that cell, and thus mean values are
negatively biased by the inclusion of cells on the margins of a
home range that include within their 3.66-km
2
window
many unsuitable habitats that were not within the home
range. Based on these values, we refer to areas with suitability
Table 2. Mean (SD) total area, percent of total area, and Jacob’s index for habitat types in reintroduced breeding whooping crane home ranges (n¼16) in
central Wisconsin, April–July 2005–2010. Positive values for Jacob’s index indicate preference and negative values indicate avoidance.
Land cover class Area (ha) % Jacob’s index
Emergent herbaceous wetlands 80.51 51.86 21.69 13.65 0.37 0.43

Woody wetlands 50.64 39.23 13.98 12.72 0.01 0.45
Open water 35.58 31.10 9.658.84 0.04 0.63
Upland open 103.98 62.95 28.65 19.28 0.08 0.46
Shrub/scrub 7.98 7.35 2.11 1.96 0.26 0.38

Forest 74.89 61.80 19.71 15.78 0.48 0.36

Barren land 1.75 6.84 0.70 2.73 0.78 0.61

Developed low 13.13 5.03 3.52 0.90 0.27 0.14

Developed high 0.00 0.00 0.00 0.00 1.00 0.00


Significant at P<0.05.

Significant at P<0.01.
Table 3. Habitat use and selection differences between nesting territory and off-territory home range elements (n¼15) of reintroduced breeding whooping
cranes in central Wisconsin, April–July 2005–2010. Values are mean (SD) total area, percent of nesting territory or off-territory element area, and Jacob’s
index (for which positive values indicate preference and negative values indicate a version).
Land cover class
Nesting territories Off-territory elements
Area (ha) % Jacob’s index Area (ha) % Jacob’s index
Emergent herbaceous wetlands 72.84 56.17 27.3118.72 0.45 0.42

11.37 17.11 7.06 7.78 0.31 0.57

Woody wetlands 36.32 31.45 12.828.82 0.01 0.42

13.27 20.86 10.7020.61 0.41 0.64

Open water 35.94 31.14 12.5610.73 0.18 0.59

4.15 6.93 3.075.11 0.56 0.61

Upland open 51.72 35.83 18.9112.52 0.31 0.39

67.91 61.43 52.8233.04 0.30 0.68

Shrub/scrub 5.79 5.94 2.945.32 0.27 0.45 3.20 3.67 1.841.36 0.31 0.42
Forest 57.23 51.29 22.0819.97 0.44 0.46 27.81 28.71 19.3113.32 0.49 0.32
Barren land 0.28 1.08 0.100.40 0.87 0.50 2.45 9.50 1.545.95 0.87 0.52
Developed low 9.93 5.28 3.271.45 0.33 0.28 4.91 4.08 3.652.59 0.32 0.36
Developed high 0.00 0.00 0.000.00 1.00 0.00 0.00 0.00 0.000.00 1.00 0.00

Significantly different between nesting territories and off-territory elements at P<0.05.

Significantly different between nesting territories and off-territory elements at P<0.01.
6 The Journal of Wildlife Management 9999

ranks of 0.667 as highly suitable. Although most observed
home ranges were therefore found in areas deemed highly
suitable, the pair that nested successfully at the far north end
of the Central Wisconsin study area was a notable exception
(Fig. 3). This pair’s nesting territory was in an area classified
as fairly unsuitable: approximately 0.4 suitability at the nest
locations, which was below the minimum suitability value in
most home ranges (mean SD 0.50 0.07; these values do
not include deep-water habitats that were manually set to 0
suitability). A large amount of habitat ranked as high
suitability also existed to the northwest of the main breeding
population but was not used by whooping cranes.
The habitat suitability map of East Wisconsin (Fig. 4)
contained 2 broad clusters of habitat ranked as highly
suitable. As with Central Wisconsin, several large publicly
owned conservation areas were judged highly suitable for
whooping crane use in East Wisconsin (Table 4). The largest
single area of highly suitable habitat was Horicon Marsh
along the Rock River, with an additional large suitable
habitat patch north of it in El Dorado State Wildlife Area
(SWA). Highly suitable habitat also occurred along the Fox
River in the western half of the study area, including a large
patch along White River Marsh SWA. A third, smaller
complex of highly suitable habitat patches ran along the Wolf
River in the north, separated from the Fox River swath by
Lake Poygan. Overall, larger expanses of the East Wisconsin
study area were ranked >0.5 suitability compared to the
Central Wisconsin study area because East Wisconsin had
lower proportions of forests and scrublands; only the cities
and the more forested northwest and western sections of East
Wisconsin had large areas of values <0.5.
DISCUSSION
Home Range and Nesting Territory Composition
Our results are consistent with previous descriptions of
whooping crane nesting territories in Wood Buffalo National
Park (Timoney 1999). Cranes nesting there preferred open
habitat with large amounts of open water, featuring shallow
lakes and ponds, emergent herbaceous marshes, mudflats,
and sedge meadows, and avoided shrubby bogs and upland
forest. Croplands were not present in the Wood Buffalo
landscape. Breeding whooping cranes in the EMP extensively
preferred emergent herbaceous wetland habitat, as expected
Figure 3. Map of habitat suitability for nesting whooping cranes for the extent of the landscape delineated as available to breeding whooping cranes in the
Central Wisconsin study area based on crane location data from 2005 to 2010. Major lakes and rivers are labeled. Open water is black. Whooping crane home
ranges are from the year with the most location data for each pair.
Van Schmidt et al. Whooping Crane Habitat Use and Suitability 7

Figure 4. Map of habitat suitability predicted for nesting whooping cranes in the East Wisconsin study area. Open water is black and major lakes and rivers are
labeled. Cities are labeled in bold. Public lands >1km
2
in size are outlined in white and those >25 km
2
are labeled.
Table 4. Habitat suitability for nesting whooping cranes of publicly owned lands (National Wildlife Refuges [NWRs], State Wildlife Areas [SWAs], and a
state forest) >25 km
2
within Central and East Wisconsin study areas. For scale, the average crane home range size was 3.68 km
2
.
Public land unit Area (km
2
) Mean suitability rank
a
Amount of high suitability habitat
b
in km
2
(% of total area)
Horicon NWR 136.9 0.797 108.3 (79.1)
White River Marsh SWA 48.1 0.798 37.4 (77.7)
Necedah NWR 247.4 0.520 24.6 (9.9)
El Dorado SWA 26.0 0.817 21.2 (81.4)
Grand River Marsh SWA 28.3 0.615 15.6 (55.2)
Kettle Moraine State Forest 120.5 0.514 3.2 (2.7)
a
Mean value for a raster cell within each land unit; range 0–1, with 1 being most suitable.
b
Relative suitability rank of 0.667, weighted by each raster cell’s suitability rank.
8 The Journal of Wildlife Management 9999

(Timoney 1999). Woody wetland habitats were used by cranes
but not in proportions significantly greater than their availability.
Likewise, open water was neither significantly preferred nor
avoided, though this land cover class contains no information on
water depth. Open water habitats in Necedah NWR range from
water depths that are shallow enough for wading cranes to use
(i.e., <0.5 m) to being so deep that wading is impossible. In
addition, water depths in the Necedah NWR flowages can vary
from year to year and thus even areas classified as water too deep
forwadinginsomeyearsmayhavebeenaccessibletowhooping
cranes in other years. Shallow open water may be preferred by
cranes, whereas deep water is avoided, but our land classification
was unable to resolve this difference.
Open water may also be preferred at specific times of the
summer and not at other times. Whooping cranes undergo
a remigial molt every 2 or 3 years that renders them
flightless(Folketal.2008).During2011,theremigialmolt
began in early June and was completed in July for 6 adults in
central Wisconsin (Lacy and McElwee 2014). Open water
was important for cranes during the remigial molt and
mean home range size of these 6 birds during molt was
0.47 km
2
(estimated using minimum convex polygons and
roost-to-roost tracking; Garland 2012), almost an order of
magnitude less than that measured in our study from
April–July. Use of open water and smaller territories also
occurred with molting nonmigratory whooping cranes in
Florida(Folketal.2008),thoughthedatesoftheremigial
molt were April–June.
Although most nesting territories were tightly associated
with emergent herbaceous wetland areas, off-territory home
range elements included more upland open habitat. This is
consistent with observations of whooping cranes on their Texas
winteringgrounds,wheretheyhavebeenknownto
occasionally move into upland areas to forage (Chavez-
Ramirez et al. 1996). As in Texas, cranes in Wisconsin often
congregated in groups with few antagonistic interactions while
using off-territory upland areas. Off-territory upland habitats
were most often cropland; 12 of 18 crane pairs left their nesting
territories at some point during the study to spend time in
upland areas to the south of Necedah NWR, especially in corn
fields(e.g.,Fig.2).Usingcropfieldswithinbroaderhome
ranges may reflect important differences in foraging versus
nesting habitat selection. Territories used while nesting may be
more constrained than areas used in home ranges. Typically,
nesting birds stay within a given distance from their nest
because they must provide for nest defense as well as forage
(Smith 1995). Although uplands may be suboptimal for
nesting territories, upland habitats could be an important
supplemental habitat, especially if food resources are limited
during the early nesting season on nesting territories.
Cranes often used croplands after nest abandonment,
which could reflect a lack of sufficient food resources located
within nesting territories and could have contributed to nest
abandonment. This hypothesis can be tested as more nests
are established near upland cropland and grassland in future
years, especially associated with the reintroductions in
eastern Wisconsin. The crane pair that nested farthest
from Necedah NWR did so in an area dominated by
cropland, and successfully fledged a chick (Garland 2012). If
cropland is conducive to nesting success, its abundance in
eastern Wisconsin, often adjacent to wetlands, may make this
area more productive. However, this might also lead to
increased human-wildlife conflicts as well.
Whooping cranes avoided all land cover classes that were not
wetland, open water, or upland open for both home ranges and
nesting territories. Low levels of development were present in
many territories, but these bits of development were over-
whelmingly roads, often gravel, cutting through otherwise
undeveloped habitats. No cranes outside of the refuge formed
home ranges near highly developed areas, despite many
developed areas having adjacent wetlands. Whooping cranes
also avoided both forest and shrub habitats, but did use wetlands
located between forested tracks. We observed individual cranes
over the course of this study moving through forested areas on
foot to access scattered wetlands within a home range.
Model Performance
Independent data for validation of this model were not
available because of the restricted range and small size of the
EMP. Likewise, hindcasting was not an option due the limited
historical data on Wisconsin whooping crane distribution.
Comparing model predictions from Central Wisconsin to new
territories found within the East Wisconsin study area as
reintroduction efforts advance will provide independent
validation. First nesting attempts by reintroduced whooping
cranes in eastern Wisconsin occurred in 2014. Extensive areas
northwest of Necedah NWR were also ranked as highly
suitable (Fig. 4), and observing whether or not cranes begin to
use these areas as the breeding population expands will provide
an additional opportunity for validation.
Because most of the nesting territories were located in the 2
main flowages at Necedah NWR, measured habitat composi-
tion was fairly uniform, but this consistency may be biased by
the philopatric nature of whooping cranes. We attempted to
correct for philopatric bias by weighting habitats closer to the
whooping crane release site as more available, thus decreasing
its influence on the selection index. Furthermore, some crane
pairs established territories up to 40km from their release sites
and even those cranes that nested within Necedah NWR often
had off-territory elements of home ranges far (>20 km) from
Necedah NWR. Thus, cranes selected habitats beyond
geographic areas defined simply by their release site. Off-
refuge territories may be more indicative of the birds’ innate
preferences because they represent the selection of habitats
made by birds dispersing away from more immediate natal
areas. As only 3 such nesting territories had been established,
an analysis to test this hypothesis was not possible; future
analyzes should examine off-refuge territories separately. The
crane pair that formed a nesting territory farthest from
Necedah NWR did so in an area rated as low relative suitability
(approx. 0.4 around the nests; Fig. 3)—cropland with a riparian
corridor running through it and only a small patch of woody
wetland—indicating that the cranes have a degree of flexibility
in habitat selection as noted in Allen (1952).
We did not include in our model of habitat suitability those
habitats for which Jacob’s index was not statistically
Van Schmidt et al. Whooping Crane Habitat Use and Suitability 9

significant (woody wetlands, open water, and upland open).
Because the magnitude of their Jacob’s indices were small,
their inclusion or exclusion would not have greatly changed
model results. However, had we included these, modeled
habitat suitability would have been more similar between
Central Wisconsin and East Wisconsin because upland open
had a small negative coefficient and was abundant in East
Wisconsin. Our model did not distinguish between areas
ranked as moderately suitable because they were homo-
geneously neither preferred nor avoided (e.g., a large patch of
upland open habitat) and areas ranked as moderately suitable
because they were a heterogeneous patchwork of suitable and
unsuitable habitats (e.g., a mix of emergent herbaceous
wetlands and forest). Given that we observed home ranges
that represented both of these scenarios, we believe this
assumption to be reasonable. Nevertheless, if such differ-
ences are important, our model would have missed them.
Future Assessments of Whooping Crane Habitat
Given their current rate of population recovery, a full
assessment of habitat suitability for whooping cranes nesting
in the historical upper tall grass prairie region of North
America (Allen 1952) will take decades; therefore, earlier
habitat assessments are warranted even if potentially biased
by philopatry. As the population naturally expands from
release sites, future analysis and modeling should be able to
distinguish philopatry and habitat covariance from habitat
selection. Habitat suitability assessments can be tested by
directing dispersion of cranes through releasing young birds
in new areas thought to be appropriate based on the analysis
presented here. Such feedback mechanisms should increase
the rate at which the appropriateness of a given habitat can be
assessed. This will also allow for verification and refinement
of the model presented here.
The map produced here is a coarse first approximation with
respect to ecological variables of interest, designed primarily
to select future reintroduction sites that facilitate dispersion
of whooping cranes into optimal habitat as quickly as
possible. Analyzing more precisely characterized habitat
types (e.g., types of cropland, water depth, or productivity of
wetlands) for habitat suitability as measured by survival rates
and reproductive success can provide a more rigorous
estimate of what habitat is truly suitable versus simply used.
Examining the role of food availability in particular may be
important, as cranes in our study often moved to cropland to
congregate in large foraging groups following nest abandon-
ment. Future studies in Wisconsin, and more widely in the
upper Midwest, should focus on specific promising wetland
complexes and precisely identify ecologically fine-scale
suitability of these potential habitats. Systematic collection
of habitat use data incorporating daily rhythms, instead of
the daily locations used here, could also more precisely
estimate habitat use and selection by whooping cranes in this
region with better temporal resolution. This would allow for
examination of behavioral importance, monthly changes in
home range and habitat use as the breeding season
progresses, and whether juxtaposition of habitats matters
(e.g., Rothstein et al. 1984).
MANAGEMENT IMPLICATIONS
Based on this study, the Whooping Crane Eastern Partnership
began releasing whooping cranes into Horicon Marsh (NWR
and SWA) and White River Marsh SWA in 2011. Site visits
were conducted to those areas of eastern Wisconsin estimated
as highly suitable in this study to assess their reintroduction
utility based on management potential, future development
trends, and sociopolitical issues (e.g., anticipated crop damage
problems). The Whooping Crane Eastern Partnership
identified Horicon Marsh and White River Marsh SWA as
the best potential reintroduction sites as they encompassed
large areas of highly suitable habitat and had extensive public
lands where infrastructure could support reintroduction efforts
(Fig. 4). Both areas had a greater total and proportional
amount of land ranked highly suitable by our model than
Necedah NWR (Table 4). Furthermore, neither of these
wetlands was affected by other factors thought to limit
reproduction in the Central Wisconsin population, most
notably black flies (Adler 2010). Although the cause of the low
productivity is an open question, there may be problems with
the specific conditions in the Necedah area that can only be
assessed once whooping cranes begin nesting in new
landscapes. Cranes released from White River Marsh SWA
and Horicon Marsh will still interact with cranes in central
Wisconsin; assisting the cranes’ dispersal into hypothetically
more suitable habitats will more quickly test the suitability of a
range of landscapes that are representative of the Upper
Midwest while maintaining continuity with past efforts.
ACKNOWLEDGMENTS
The Whooping Crane Eastern Partnership, whose founding
members include International Crane Foundation, Interna-
tional Whooping Crane Recovery Team, National Fish and
Wildlife Foundation, Natural Resources Foundation of
Wisconsin, Operation Migration, Patuxent Wildlife Re-
search Center, United States Geological Survey (USGS)
National Wildlife Health Center, United States Fish and
Wildlife Service, USGS, and Wisconsin Department of
Natural Resources supported this research. M. Duong, F.
East, A. Fasoli, M. Fitzpatrick, L. Fondow, M. Frakes, N.
Frey, A. Gossens, P. Leibl, S. Kerley, R. King, T. Kitagawa,
T. Love, M. Lupek, K. Maguire, C. Malachowski, M. Moss,
K. Norris, B. Paulan, M.Putnum, D. Staller, M. Strausser, E.
Szyszoksi, J. Thompson, T. Trestor, R. Urbanek, M. Waage,
L. Wargowski, C. Wisinski, and S. Zimorski assisted with
data collection. We thank S. Hull, J. Langenberg, L. Hall,
and D. King for review comments on this manuscript.
LITERATURE CITED
Aarts, G., M. MacKenzie, B. McConnell, M. Fedak, and J. Matthiopoulos.
2008. Estimating space-use and habitat preference from wildlife telemetry
data. Ecography 31:140–160.
Adler, P. H. 2010. Black fly surveys and pilot Bti study, with references to the
species that attack whooping cranes in Wisconsin. <http://www.bring-
backthecranes.org/technicaldatabase/#Research>Accessed 3 Dec 2013.
Allen, R. P. 1952. The whooping crane. Research report 3. National
Audubon Society, New York, New York, USA.
Armstrong, D. P., and P. J. Seddon. 2008. Directions in reintroduction
biology. Trends in Ecology & Evolution 23:20–25.
10 The Journal of Wildlife Management 9999

Bo
¨rger, L., N. Franconi, G. De Michele, A. Gantz, F. Meschi, A. Manica, S.
Lovari, and T. Coulson. 2006. Effects of sampling regime on the mean and
variance of home range size estimates. Journal of Animal Ecology
75:1393–1405.
Canadian Wildlife Service. U.S. Fish and Wildlife Service. 2005.
International recovery plan for the whooping crane. Recovery of
Nationally Endangered Wildlife (RENEW), Ottawa, Canada. U.S.
Fish and Wildlife Service, Albuquerque, New Mexico, USA.
Cannon, J. R. 1999. Wisconsin whooping crane breeding site assessment:
final report. Canadian-United States Whooping Crane Recovery Team.
<http://www.bringbackthecranes.org/technicaldatabase/#Research>
Accessed 3 Dec 2013.
Chavez-Ramirez, F., H. E. Hunt, R. D. Slack, and T. V. Stehn. 1996.
Ecological correlates of whooping crane use of fire-treated upland habitats.
Conservation Biology 10:217–223.
Converse, S. J., J. A. Royle, and R. P. Urbanek. 2012. Bayesian analysis of
multi-state data with individual covariates for estimating genetic effects on
demography. Journal of Ornithology 152:561–572.
Curtis, J. T. 1959. The vegetation of Wisconsin: an ordination of plant
communities. University of Wisconsin Press, Madison, USA.
Folk, M. J., S. A. Nesbitt, J. M. Parker, M. G. Spalding, S. P. Baynes, and
K. L. Candelora. 2008. Feather molt of nonmigratory whooping cranes in
Florida. Pages 128–132 in Proceedings of the Tenth North American
Crane Workshop. B. K. Hartup, editor. North American Crane Working
Group, 7–10 Feb 2006, Zacatecas City, Zacatecas, Mexico.
Folk, M. J., J. A. Rodgers, Jr., T. A. Dellinger, S. A. Nesbitt, J. M. Parker,
M. G. Spalding, S. B. Baynes, M. K. Chappell, and S. T. Schwikert. 2010.
Status of non-migratory whooping cranes in Florida. Pages 118–123 in
Proceedings of the Eleventh North American Crane Workshop. B. K.
Hartup, editor. North American Crane Working Group, 23–27 Sep 2008,
Wisconsin Dells, Wisconsin, USA.
Fry, J., G. Xian, S. Jin, J. Dewitz, C. Homer, L. Yang, C. Barnes, N. Herold,
and J. Wickham. 2011. Completion of the 2006 National Land Cover
Database for the Conterminous United States. Photogrammetric
Engineering & Remote Sensing 77:858–864.
Garland, J. 2012. Whooping Crane Eastern Partnership 2011 Annual
Report. Whooping Crane Eastern Partnership. <http://www.bring-
backthecranes.org/whatwedo/wcep11.html>Accessed 3 Dec 2013.
Harrell, Wade. 2014. Winter 2013–2014 Whooping Crane Survey Results.
U.S. Fish and Wildlife Service, <http://www.fws.gov/nwrs/threecolumn.
aspx?id¼2147544385>Accessed 17 Apr 2014.
Horne, J. S., and E. O. Garton. 2006. Likelihood cross-validation versus
least squares cross-validation for choosing the smoothing parameter in
kernel home-range analysis. Journal of Wildlife Management 70:641–648.
Jacobs, J. 1974. Quantitative measurement of food selection: a modification of
the forage ratio and Ivlev’s electivity index. Oecologica (Berlin) 14:413–417.
Keys, J. Jr., C. A. Carpenter, S. L. Hooks, F. G. Koenig, W. H. McNab,
W. E. Russell, and M. L. Smith. 1995. Ecological units of the eastern
United States–First Approximation (ARCINFO format, selected imag-
ery, and map unit table. U.S. Department of Agriculture, Forest, Service
Atlanta, Georgia, USA.
King, R. S., J. J. Trutwin, T. S. Hunter, and D. M. Varner. 2013. Effects of
environmental stressors on nest success of introduced birds. Journal of
Wildlife Management 77:842–854.
Lacy, A., and D. McElwee. 2014. Observations of molt in reintroduced
whooping cranes in Proceedings of the Twelfth North American Crane
Workshop. D. Aborn, editor. North American Crane Working Group,
13–16 March 2011, Grand Island, Nebraska, USA.
Maguire, K. J. 2008. Habitat selection of reintroduced whooping cranes,
Grus americana, on their breeding range. Thesis, University of Wisconsin,
Madison, USA.
Melvin, S. M., R. C. Drewien, S. A. Temple, and E. G. Bizeau. 1983. Leg-
band attachment of radio transmitters for large birds. Wildlife Society
Bulletin 11:282–285.
Moore, C. T., S. J. Converse, M. J. Folk, M. C. Runge, and S. A. Nesbitt.
2012. Evaluating release alternatives for a long-lived bird species under
uncertainty about long-term demographic rates. Journal of Ornithology
152:S339–S353.
Rothstein, S. I., J. Verner, and E. Steven. 1984. Radio-tracking confirms a
unique diurnal pattern of spatial occurrence in the parasitic brown-headed
cowbird. Ecology 65:77–88.
Smith, J. P. 1995. Foraging flights and habitat use of nesting wading birds
(Ciconiiformes) at Lake Okeechobee, Florida. Colonial Waterbirds
18:139–158.
Stehn, T. 2012. Whooping crane (Grus americana) 5-year review: summary
and evaluation. U.S. Fish and Wildlife Service, Aransas, National Wildlife
Refuge, Autswell, Texas, USA.
Timoney, K. 1999. The habitat of nesting whooping cranes. Biological
Conservation 89:189–197.
U.S. Geological Survey. 1989. Wisconsin Open Water from 1:100,000-scale
sources (polygon features). Wisconsin Department of Natural Resources,
Madison, Wisconsin, USA.
Urbanek, R. P., S. E. Zimorski, A. M. Fasoli, and E. K. Szyszkoski. 2010.
Nest desertion in a reintroduced population of migratory whooping cranes.
Pages 133–141 in Proceedings of the Eleventh North American Crane
Workshop. B. K. Hartup, editor. North American Crane Working Group,
23–27 Sep 2008, Wisconsin Dells, Wisconsin, USA.
Wickham,J.D.,S.V.Stehman,J.A.Fry,J.H.Smith,andC.G.Homer.2010.
Thematic accuracy of the NLCD 2001 land cover for the conterminous
United States. Remote Sensing of Environment 114:1286–1296.
Associate Editor: David King.
Van Schmidt et al. Whooping Crane Habitat Use and Suitability 11