Use and Selection of Bridges as Day Roosts by
Rafinesque’s Big-Eared Bats
FRANCES M. BENNETT
1
Department of Forestry and Natural Resources, Institute of Environmental Toxicology, Clemson University,
Pendleton, South Carolina 29670
SUSAN C. LOEB
2
USDA Forest Service, Southern Research Station, Clemson, South Carolina 29634
MARY S. BUNCH
South Carolina Department of Natural Resources, Pendleton, 29670
AND
WILLIAM W. BOWERMAN
Department of Forestry and Natural Resources, Institute of Environmental Toxicology, Clemson University,
Pendleton, South Carolina 29670
A
BSTRACT.—Rafinesque’s big-eared bats (Corynorhinus rafinesquii) use bridges as day roosts in
parts of their range, but information on bridge use across their range is lacking. From May to
Aug. 2002 we surveyed 1129 bridges (12.5%) within all 46 counties of South Carolina to
determine use and selection of bridges as day roosts by big-eared bats and todocumenttheir
distribution across the state. During summer 2003, we visited 235 bridges in previously occupied
areas of the state to evaluate short-term fidelity to bridge roosts. We found colonies and solitary
big-eared bats beneath 38 bridges in 2002 and 54 bridges in 2003. Construction type and size of
bridges strongly influenced use in both years; bats selected large, concrete girder bridges and
avoided flat-bottomed slab bridges. The majority of occupied bridges (94.7%)wereinthe
Upper and Lower Coastal Plains, but a few bridges (5.3%) were located in the Piedmont.
Rafinesque’s big-eared bats were absent beneath bridges in the Blue Ridge Mountains. We
established new records of occurrence for 10 counties. In the Coastal Plains, big-eared bats
exhibited a high degree of short-term fidelity to roosts in highway bridges. For bridges that were
occupied at least once, mean frequency of use was 65.9%. Probability of finding bats under a
bridge ranged from 0.46 to 0.73 depending on whether the bridge was occupied in the previous
year. Thus, bridges should be inspected three to five times in a given year to determine whether
they are being used. Regional bridge roost surveys may be a good method for determining the
distribution of C. rafinesquii, particularly in the Coastal Plains, and protection of suitable bridges
may be a viable conservation strategy where natural roost sites are limited.
INTRODUCTION
Rafinesque’s big-eared bat (Corynorhinus rafinesquii) is found in the southern and
midwestern United States (Fig. 1) and is one of the least studied bats in North America
(Harvey et al., 1999). Despite having a relatively widespread distribution, this species is
considered uncommon and is recognized as a species of special concern across most of its
range (Hurst and Lacki, 1999; Martin et al., 2002). However, because Rafinesque’s big-eared
bats are not easily captured or detected with standard methods (e.g., mist nets, acoustic
1
Present Address: Fish and Wildlife Branch, Saskatchewan Environment, 3211 Albert Street, Regina,
SK S4S 5W6, Canada
2
Corresponding author: e-mail: sloeb@fs.fed.us; Phone: 864-656-4865
Am. Midl. Nat. 160:386–399
386
DE- AI09-00SR22188 JOURNAL ARTICLE 2008 08-07-P
sampling), it has been difficult to estimate their relative abundance and determine their
geographic distribution.
Historical accounts, museum specimens and incidental capture records place this species
in the Blue Ridge Mountains, the Upper Coastal Plain and the Lower Coastal Plain
physiographic regions in North Carolina, South Carolina and Georgia (Handley, 1959;
Clark, 1990; Menzel et al., 2003). In the Carolinas, Rafinesque’s big-eared bat is commonly
associated with bottomland hardwood forests (Clark, 1990) which are most abundant within
the Upper and Lower Coastal Plains (Conner, 1993). Although bottomland hardwood
forests also occur in the Piedmont, this bat appears to be absent from this physiographic
region (Menzel et al., 2003). It is not clear whether the Piedmont truly does not support
populations of big-eared bats or whether there have been insufficient sampling efforts in
this region. A reliable method for locating Rafinesque’s big-eared bats is clearly needed to
determine their population status and distribution.
Rafinesque’s big-eared bats are non-migratory and use tree cavities, caves, mines,
buildings and other man-made structures for roosting (Barbour and Davis, 1969). Like most
cavity-roosting species, Rafinesque’s big-eared bats that use tree cavities and bridges
frequently switch roost sites (Lance et al., 2001; Trousdale and Beckett, 2005), whereas cave
roosting Rafinesque’s big-eared bats rarely switch roosts (Hurst and Lacki, 1999). In the
Coastal Plains, naturally occurring structures include cavities in large diameter gum (Nyssa
sp.) and cypress (Taxodium sp.) trees (Clark, 1990; Gooding and Langford, 2004; Trousdale
and Beckett, 2005). Artificial sites are structurally similar to natural cavities, and include
dimly lit areas in abandoned buildings, cisterns, wells and highway bridges (Barbour and
Davis, 1969; Clark, 1990; Lance et al., 2001; Mirowsky et al., 2004; Trousdale and Beckett,
2002, 2004; Ferrara and Leberg, 2005a). Both artificial and natural structures are used as day
and night roosts year-round, but frequency of use in anthropogenic structures peaks during
May–Aug. when maternity colonies appear (Felts and Webster, 2003; Trousdale and Beckett,
2004). Thus, summer is the most appropriate time to conduct surveys in artificial structures,
particularly bridges (Ferrara and Leberg, 2005b).
FIG. 1.—Upper right: range of Rafinesque’s big-eared bats. Enlargement shows the four physiographic
provinces of South Carolina and the locations of bridges used by Rafinesque’s big-eared bats, late May
through mid-Aug. 2002–2003
2008 BENNETT ET AL.: BAT ROOSTS 387
The relative use of artificial versus natural structures may depend upon the availability of
structures in each physiographic region. Rafinesque’s big-eared bats more commonly roost
in artificial structures in the southern portion of their range, and in natural roost sites in the
northern portion of their range ( Jones, 1977). The Coastal Plain lacks an abundance of
natural roost sites (i.e., large trees) because many were harvested over a century ago.
However, artificial structures now are widespread and are frequently used as roost sites. By
contrast, in the northern portions of the range natural roost sites such as tree cavities, rock
houses, abandoned mines and caves are more frequently used (Bunch et al., 1998; Hurst and
Lacki, 1999), possibly because of their higher occurrence on the landscape.
Type of bridge construction is the strongest predictor of bridge occupancyby
Rafinesque’s big-eared bats (Lance et al., 2001; McDonnell, 2001; Trousdale and Beckett,
2002). Bats roost in the space between girders on the underside of bridges and have not
been observed in enclosed and concealed expansion joints (Ferrara and Leberg, 2005a). In
South Carolina, three main bridge types are present: flat-bottomed slab bridges, multi-beam
(MB) girder bridges and T-beam (TB) cast-in-place girder bridges (L. R. Floyd, South
Carolina Department of Transportation, unpubl.). MB bridges are variable in structure, but
generally consist of parallel beams that span the entire length of the bridge and sometimes
are referred to as I-beam or channel beam bridges. TB bridges also have parallel beams that
span the entire length of the bridge, but the support beams are intersected at right angles by
cross beams. Although Rafinesque’s big-eared bats most frequently use girder bridges in
Louisiana and North Carolina (Lance et al., 2001; McDonnell, 2001), it is not known
whether they select either of the two girder type bridges found in South Carolina.
Most studies of bridge use by Rafinesque’s big-eared bats have been conducted within
relatively small geographic areas restricted to the Coastal Plain regions (Lance et al., 2001;
McDonnell, 2001; Trousdale and Beckett, 2002; Felts and Webster, 2003). Limited bridge
surveys have been conducted in South Carolina, but no day-roosting bats were found under
the 44 bridges examined (Keeley and Tuttle, 1999). The objectives of our study were to: (1)
document the use of bridges by Rafinesque’s big-eared bats in South Carolina, (2) conduct a
statewide bridge survey to determine their distribution across the state, (3) evaluate bridge
attributes such as size and type that influence occupancy and (4) determine short term
bridge fidelity and the number of visits needed to document presence. By identifying bridge
types used by Rafinesque’s big-eared bat and its fidelity to these structures, it may be possible
to improve survey methods across its range.
M
ETHODS
STUDY AREA
South Carolina consists of four physiographic regions: the Blue Ridge Mountains, the
Piedmont, the Upper Coastal Plain and the Lower Coastal Plain (Fig. 1). The climate of
South Carolina is warm temperate to subtropical and is characterized by short, mild winters
and long, hot and humid summers. Rainfall occurs throughout the year, but peak levels
occur during the winter months in the mountains, and in Mar. and Jul. throughout the rest
of the state. Average monthly rainfall amounts range from 11.4 cm to 17.3 cm in the
mountains, 7.4 cm to 11.7 cm in the Piedmont and 6.0 cm to 16.6 cm in the Coastal Plains.
The Blue Ridge Mountain region, a part of the southern Appalachian Mountains, is
situated in the upper northwestern portion of South Carolina. This region covers
approximately 1.9% of the state, has a mountainous topography and ranges in elevation
from 366 to 1067 m. Oak-hickory (Quercus sp. - Carya sp.), oak-pine (Quercus sp. Pinus sp.)
388 T
HE AMERICAN MIDLAND NATURALIST 160(2)
and loblolly-shortleaf pine (P. taeda – P. echinata) are the dominant forest types (Conner,
1993).
The Piedmont region is adjacent to the Blue Ridge Mountains and covers 31.9% of
South Carolina. It has a rolling topography and ranges in elevation from 91 to 366 m.
Urbanization and agriculture are common in this region; the dominant forests are loblolly-
shortleaf pine forests. Localized stands of mixed pine-hardwoods and bottomland hardwood
forests consisting of oak-bald cypress-tupelo gum (Quercus sp. - Taxodium distichum - Nyssa
sp.) trees also are found in the Piedmont, but are concentrated in areas adjacent to the
Upper Coastal Plain (Conner, 1993).
The Upper and Lower Coastal Plain provinces cover the largest area of South Carolina
(66.2%), extending 193 to 241 km inland from the Atlantic Ocean. The topography of this
region is flat; the highest elevation is 91 m. Forests in both Coastal Plain regions are
dominated by loblolly-shortleaf and longleaf-slash pine (P. palustris - P. elliotti) forests;
however, bottomland hardwood forests are more extensive in these physiographic provinces
than any other in the state (Conner, 1993). The Upper Coastal Plain has comparatively
more urban, agriculture and other non-forest cover types than the Lower Coastal Plain
region.
2002 STATEWIDE BRIDGE SURVEY
We conducted a county-by-county survey from 22 May to 8 Aug. 2002. Bridge data
including structure type, construction material, latitude/longitude, feature crossed (i.e.,
waterway), unique identification number and bridge length and width were obtained from
the South Carolina Department of Transportation (SC DOT; L. R. Floyd, South Carolina
Department of Transportation, unpubl.). For each of the 46 counties in South Carolina, we
grouped and surveyed bridges according to type (slab, MB and TB). Because bats rarely
roost in bridges over roadways and train tracks (Erickson, 2002) we only surveyed bridges
over water bodies. We surveyed bridges on public roads, including those on National Forests
and National Wildlife Refuges. For safety reasons, we did not survey bridges on interstate
highways. Each bridge was surveyed once.
For the first 9 d of the survey, we used a stratified random sampling design based on
bridge type and inspected bridges in proportion to their occurrence. Slab bridges were the
most common bridge type over water in South Carolina (n 5 4025), followed by MBs (n 5
1616) and TBs (n 5 676). Based on the results of this initial sampling period (Bennett,
2004) and data from previous studies (Lance et al., 2001; McDonnell, 2001), we modified
the study design to increase the likelihood of locating bats under bridges. For the remainder
of the survey, we inspected bridges in the following order of precedence: TB, MB, and slab.
We attempted to inspect every TB bridge over water and simultaneously surveyed a
randomly generated subset of MB and slab bridges.
We inspected the underside of each bridge during the day for presence of bats with
1,000,000 candle-power flashlights. Data collected included: date, county, latitude and
longitude, physiographic region, bridge type and material, number of Rafinesque’s big-
eared bats present, number and species of other bats present, presence of bat feces and
disturbance level. If bats were found under a bridge, we recorded details of the roost
location and the group type (maternity colony or solitary). Where possible, independent
counts of pups and adults were made by at least two field personnel and compared to ensure
the most accurate count of bats. In some instances, total counts were not made to reduce
disturbance to the bats. In these instances, we recorded an approximate range of the
numbers of bats present. For data analysis, we used the lowest estimate.
2008 B
ENNETT ET AL.: BAT ROOSTS 389
We rated disturbance beneath each bridge on a discrete scale of 0–3. Bridges with no
obvious disturbances were given a rating of 0, low levels of disturbance were recorded as 1,
medium levels of disturbance were recorded as 2, and bridges with high levels of
disturbance were given a rating of 3. Disturbance factors included presence or evidence of
humans such as trash, vandalism, footprints, all terrain vehicle tracks and heavy vehicular
traffic on the surface of the structure.
2003 BRIDGE SURVEYS AND ROOST MONITORING
We conducted bridge surveys from 23 May to 1 Aug. 2003 using the same methods as in
2002. Although the 2003 field survey was similar in execution to the 2002 statewide survey,
there were two important differences. First, we did not survey the entire state. Instead, we
focused surveys in areas where big-eared bats were found in 2002. Second, we inspected
bridges occupied in 2002 several times in 2003; most bridges with big-eared bats were
surveyed every 2–3 wk so that bridges were examined up to five times. We also inspected
additional bridges over water that were not visited in 2002, but were within occupied areas of
the state. If a bridge was occupied in 2003 but not in 2002, we also monitored it regularly.
However, due to restricted access, some bridges (,10) were only inspected once in 2003.
DATA ANALYSIS
We used likelihood ratio chi-square tests (PROC FREQ; SAS, 2002) to determine
the association between the presence of big-eared bats and qualitative attributes of
bridges (type, physiographic region and disturbance) in 2002 and 2003. Dueto
small sample size, we used a Fisher’s exact test to evaluate the association between the
presence of big-eared bats and the occurrence of other bat species in 2002, and big-eared
bat presence and disturbance in 2003 (Freeman and Halton, 1951). Associations between
the presence of bats and quantitative attributes of bridges (length, width and area) were
assessed using the Kruskall-Wallis one-way analysis of variance. We were unable to attain
bridge size for 11 bridges in 2002 and eight bridges in 2003. Because of differences in
sampling procedures between years, we analyzed data from 2002 and 2003 separately.
Furthermore, because our sampling procedure was biased against slab bridges and no bats
were found under these structures, we also ran the above analyses after excluding slab
bridges from the dataset. We were unable to determine the association between bridge
material (concrete, timber, steel alloy) and presence of bats because material and bridge
type were not independent.
We used logistic regression analysis with a stepwise selection process (a 5 0.05) to
determine bridge attributes selected or avoided by Rafinesque’s big-eared bats (PROC
LOGISTIC). We determined the goodness of fit of the logistic regression equations for
binary response models (Hosmer and Lemeshow, 2000). Models were run with and without
slab bridges. Because bridge area was highly correlated with bridge length(r. 0.95) it was
not included in the models. We used an a # 0.05 to determine statistical significance for all
tests. Data are presented as the mean 6
SD throughout the results.
We used program PRESENCE (MacKenzie et al., 2002) to estimate the probability of
detecting bats (p) under a bridge and bridge occupancy (Y) for the 2003 sampling period.
We used estimates of p to determine the minimum number of times a bridge needs to be
inspected to determine whether it is occupied. However, because the bridges were not
chosen randomly, p and Y are biased. Models were run on all bridges inspected once in
2002 and $2 times in 2003. We included bridge use in 2002 as a covariate to test whether
previous occupation of a bridge was an important variable in detection probabilities
and occupancy in 2003. We compared models using Akaike’s Information Criterion
390 T
HE AMERICAN MIDLAND NATURALIST 160(2)
