Divergent roosting habits of Rafinesque's big-eared bat and southeastern myotis during winter floods.
Tree roosts are a fundamental resource for forest dwelling bats, providing protection from the elements and predators, and serving as a location for many social and reproductive functions (Kunz and Lumsden, 2003). Although much effort has been devoted to describing summer tree roosts, few studies have described trees used as roosts during winter (Cryan and Veilleux, 2007). Winter roosts may be particularly important to survival because bats must cope with greater thermoregulatory challenges at a time of reduced food abundance (Speakman and Thomas, 2003) and because they are more susceptible to predators (Estok et al., 2010), disease (Bouma et al., 2010), and accidents (Cope, 1977) while hibernating. Given that the ecological constraints facing tree roosting bats are compounded by these thermoregulatory challenges during winter, winter roost selection strategies may differ from summer strategies.
The limited information available indicates that a number of tree roosting species use different roosts in winter than in summer. For example some lasiurine species use tree roosts in winter but move to leaf litter during periods of extreme cold (Mormann and Robbins, 2007; Hein et al., 2008). Low winter temperatures can also prompt silver-haired bats (Lasionycteris noctivagans) and evening bats (Nycticeius humeralis) to switch from tree roosts to rock crevices or burrows (Boyles et al., 2005; Perry et al., 2010). Understanding these seasonal changes in roost selection is vital to developing year round management strategies for tree roosting bats (Brigham, 2007).
Rafinesque's big-eared bat (Corynorhinus rafinesquii) and southeastern myotis (Myotis austroriparius) are restricted to the Southeast and lower Midwest regions of the United States (Jones, 1977;Jones and Manning, 1979). In the Coastal Plain, both species typically roost in hollow trees in baldcypress (Taxodium distichum)--tupelo (Nyssa spp.) swamps (Gooding and Langford, 2004). During summer both species commonly roost in large diameter water tupelo (N. aquatica) with basal openings, although Rafinesque's big-eared bat uses trees with a larger mean diameter (Carver and Ashley, 2008).
During winter cypress-gum swamps are subject to prolonged flooding, which can cover basal openings (Battle and Golladay, 2001). Therefore, trees used as roosts in summer may be unavailable for winter use and ground roosting strategies used by some other bats are untenable. Previous roost searches during winter were limited to times or locations with minimal flooding (Mirowsky and Homer, 1997; Stevenson, 2008; Rice, 2009; Loeb and Zarnoch, 2011). In those surveys, bats were commonly absent from summer roosts during winter. A telemetry study documented alternate Rafinesque's big-eared bat roosts in trees without basal openings in fall and early winter (Rice, 2009). However, roost searches have not been conducted during prolonged winter flooding. Therefore, our objectives were to locate winter roosts of Rafinesque's big-eared bat and southeastern myotis in an area subject to extensive flooding, evaluate the basis of roost selection in winter, and compare winter roosts to previously described summer roosts.
We conducted our research on River Bend Wildlife Management Area (WMA; 32[degrees]28'N, 82[degrees]50'W), a state-owned property in Laurens County, Georgia. During winter typical high temperatures are 14 to 21 C and typical low temperatures are 2 to 6 C. The site consists of pine (Pinus spp.) dominated uplands and a hardwood dominated river floodplain along the Oconee River. The floodplain is intermittently flooded by up to 0.5 m of water during winter, but surface water is limited to several oxbow lakes during summer. Accordingly, the floodplain supports diverse water tolerant tree species, including oaks (Quercus spp.), hickories (Carya spp.), red maple (Acer rubrum), and sweetgum (Liquidambar styraciflua), while the oxbow lakes are dominated by two hydrophilic tree species, water tupelo and baldcypress, that stand in 1-5 m of water throughout winter. We previously surveyed the study area during summer 2008 and determined that large diameter hollow trees were concentrated around two oxbow lakes, Troup Lake and Beacham Lake (Clement, 2011). Therefore, we limited our winter surveys to those lakes.
We conducted visual searches of potential roosts by boat from 23 Jan. to 16 Mar., 2010. We visually searched the interior of all trees with basal openings using a spotlight and mirror to determine if bats were present. Based on the high estimated probability of detection for roosting bats (Clement and Castleberry, 2013a), we assumed bats were not overlooked when present. We used standard arborist techniques to climb additional hollow trees that lacked basal openings but had higher openings (Jepson, 2000). We searched all trees that could be climbed safely and had an opening large enough (>25 cm) to allow visual inspection for roosting bats. Trees suitable for climbing were searched three times during the study to minimize false negatives due to bats switching roosts. Trees with basal openings were searched only once or twice because openings were covered by flood waters during most of the study.
We captured bats in single high mist nets (50 denier weight, 2 ply nylon, 38 mm mesh, Avinet Inc., Dryden, N.Y.) placed near Troup Lake from 3 Feb. to 7 Mar., 2010. Netting sites were selected on an ad hoc basis with the goal of maximizing captures. We attached 0.4 g radio transmitters (Lotek Wireless, Newmarket, Ontario) to bats with Torbot Liquid Bonding Cement (Torbot Group Inc., Cranston, RI). Radiotransmitters averaged 5.0% of body weight (range: 4.3-5.7%) for Rafinesque's big-eared bats and 6.2% of body weight (range: 6.1-6.3%) for southeastern myotis. Capture and handling protocols were approved by the University of Georgia Institutional Animal Care and Use Committee (approval no. A2007-10046-cl) and Georgia Scientific Collecting Permit #29-WCH-06-104. We radio tracked bats daily, beginning the day after capture on Feb. 3 and continued until Mar. 18. We located bats using a portable telemetry receiver (TRX 2000S; Wildlife Materials Inc., Murphysboro, IL) and a 3-element Yagi antenna (Advanced Telemetry Systems, Inc., Isanti, MI).
For all searched trees and roosts located by telemetry, we recorded tree species and whether the tree was alive or dead. We recorded tree diameter at 0.5 m above the surface of the water using a diameter at breast height (dbh) tape (Spencer Products Co., Seattle, WA). The measurement typically was above the buttress swell common in hydrophilic trees, but the height of measurement varied with water level. We measured tree height above the water using a 400 LH laser hypsometer (Opti-Logic Corp., Tullahoma, TN), although height varied with water depth. We counted the number of visible cavity openings (>7.5 cm) on each tree, recorded height above water, height and width of each opening, and calculated the area of the opening. We recorded UTM coordinates of trees with an eTrex Venture HC global positioning system unit (Garmin Ltd., Olathe, KS).
We measured interior cavity diameter of hollow trees with a tape measure and interior cavity height using a tape measure or hypsometer. Cavity volume was calculated from cavity height and average diameter assuming a cylindrical shape ([h[pi][r.sup.2]). We estimated solid tree volume by subtracting estimated cavity volume from total bole volume, which we estimated from bole height and diameter, assuming a conical shape ([h[pi][r.sup.2]/3). We recorded whether or not the cavity had a chimney like opening at the top. We also characterized the interior surface of the cavity as rough (>50% of cavity surface covered with projections >2 cm) or smooth. We also obtained previously collected data on summer tree roosts of both bat species in River Bend WMA and seven other study sites in the Coastal Plain of Georgia (Clement, 2011). We returned to the study site and remeasured characteristics of winter roosts after flood waters had receded to allow comparisons between winter and summer roost trees.
For Rafinesque's big-eared bat, we used logistic regression to predict which trees were occupied in winter based on the measured characteristics of hollow trees (Hosmer and Lemshow, 2000). We considered trees that contained bats during any roost search or telemetry session to be occupied and trees that did not contain bats during any roost search to be unoccupied. Due to sample size constraints and the lack of prior knowledge about winter roosts, we only evaluated univariate models (Hosmer and Lemeshow, 2000). In addition to nine univariate models, we also evaluated a global model including all uncorrelated (Pearson [R.sup.2] < 0.25) predictor variables and a model with no predictors, indicating roost usage was random with respect to tree characteristics.
Before performing logistic regression on tree data for Rafinesque's big-eared bat, we transformed the predictor variables "tree cavity volume" and "solid tree volume" using the natural logarithm to ensure linearity in the logit function (Hosmer and Lemeshow, 2000). We also eliminated tree species as a predictor due to data separation (Hosmer and Lemeshow, 2000). For statistical analysis, we considered the tree to be the experimental unit. To compare trees that were occupied or unoccupied in winter, we used tree measurements with flood waters present to represent the trees as they would be encountered by bats during winter.
We evaluated goodness-of-fit of our global model using the Hosmero-Lemeshow statistic (Hosmer and Lemshow, 2000). We transformed log odds regression coefficients to odds ratios, which express how much more or less likely an outcome (i.e., bat occupancy) is as the predictor variable changes (Hosmer and Lemeshow, 2000). Because data were collected in a case control sampling scheme in which analyzed trees were not proportional to their presence in the population, analysis yielded unbiased estimates of coefficients and odds ratios, but biased intercept terms (Keating and Cherry, 2004). We used Akaike's Information Criterion corrected for small sample bias (AICc) to assess the fit of the candidate models, with the lowest AICc indicating the best supported model (Burnham and Anderson, 2002). We calculated a composite model from the best supported models, where appropriate (Burnham and Anderson, 2002). We calculated Nagelkerke's [R.sup.2] to quantify the variation explained by each model (Nagelkerke, 1991).
We used leave-one-out cross validation to estimate the top model's ability to predict bat presence (Efron, 1983). We selected a prediction cutoff equal to the proportion of trees occupied during winter (0.43), so that a predicted probability greater than 0.43 was considered a prediction of presence. The process was repeated for every data point and model predictions were compared to the actual states to calculate prediction and classification error rates.
For southeastern myotis roosts, our sample was smaller than the minimum recommendation of 10 observations per independent variable for logistic regression (Hosmer and Lemeshow, 2000). Therefore, we used t-tests and Fisher's exact test to identify continuous and binary tree characteristics that differed between occupied and unoccupied trees in winter. We conducted all analyses in Program R 2.11.1 (R Development Core Team, 2010).
We also examined differences between trees used in winter and trees used in summer using previously collected summer roost data from River Bend WMA and seven other sites. Because combining sites violated assumptions of many statistical tests, we qualitatively compared winter roosts to summer roosts from all eight sites by examining box-and-whisker plots. We then compared summer roosts from only River Bend WMA to winter roosts using the same statistical analyses described above (i.e., logistic regression, cross-validation, t-tests, and Fisher's exact tests). To compare winter and summer roosts, we used tree measurements with flood waters absent to obtain valid comparisons. For this analysis, we selected a prediction cutoff equal to the proportion of roosts that were winter roosts (0.58), so that a predicted probability greater than 0.58 was considered a prediction of a winter roost.
We identified 149 hollow trees with visible cavity openings during our winter surveys. We searched all 18 trees (12%) that had a basal opening exposed at some point during our surveys and climbed an additional 29 trees (19%) lacking basal openings, but with higher openings. Eleven (61%) of the trees with a basal opening also had an opening >4 m above the water, while the rest only had openings <2 m above the water. The remaining hollow trees (68%) were unsafe to climb or had small entrances and were not searched. Both searched and unsearched trees were dominated by large water tupelo but searched trees were larger in diameter (113 cm versus 84 cm), less likely to be dead (2% versus 6%), and less likely to be baldcypress (9% versus 13%). Three (17%) of the trees with basal openings were occupied by one or two Rafinesque's big-eared bats during [greater than or equal to] 1 survey. Two of these also had elevated openings, while one did not. Fifteen (52%) of the trees we climbed were occupied by one to nine Rafinesque's big-eared bats during [greater than or equal to] 1 survey. None of the trees searched held southeastern myotis.
We radio tagged eight Rafinesque's big-eared bats (five adult males with enlarged epididymides, three adult females) and two southeastern myotis (adult males with enlarged epididymides) captured in mist nets. We located the Rafinesque's big-eared bats on 112 of 122 tracking days (92%) for an average of 14.0 d per bat. For six bats, the distance from capture site to the initial roost was <150 m, while two bats moved 2.5 km to private land on the opposite (west) side of the Oconee River, yielding a mean of 718 m. On average, Rafinesque's big-eared bats switched roosts every 6.9 d (range: 1-22 d). Mean distance between successive roosts was 100 m (range: 3-210 m). Each bat used an average of 1.8 roosts, although radio tagged bats only used eight unique roosts because some were used by more than one bat. Six of these roosts only had elevated openings, one had both basal and elevated openings, and one only had a basal opening. Combining all search techniques and accounting for three roosts located by both telemetry and roost searches, we found 23 total roosts, all of which were water tupelo in the oxbow lakes or similar habitat on private property.
We located the two southeastern myotis on 17 of 20 tracking days (85%) for an average of 8.5 d each. The average distance from capture site to the initial roost was 208 m, while the mean distance between successive roosts was 716 m (range: 15-2,237 m). These bats switched roosts every 2.8 d (range: 1-11 d), and used six unique roosts, although we could not locate one roost on private property. In contrast to the Rafinesque's big-eared bat roosts, all southeastern myotis roosts were located in the floodplain surrounding Troup Lake. One roost had a large cavity with a basal opening that was occasionally submerged by flood waters. The other four roosts we located were small crevices at various heights above the ground.
The only differences between Rafinesque's big-eared bat winter roost trees and unoccupied trees were that occupied trees were less likely to have a low opening and more likely to be water tupelo (Table 1). The Hosmer-Lemeshow statistic was not significant ([chi square] = 11.95, d.f. = 8, P = 0.153) indicating adequate goodness-of-fit of the logistic regression models. The best supported model of winter roost occupancy by Rafinesque's big-eared bat, receiving 43% of the AICc weight, indicated that trees with low openings were less likely to be occupied (Table 2). The intercept-only model also received support and we discarded more complex models with less support from the data (Grueber et al., 2011), which left two models in the confidence set. In the composite model, every 1 m increase in height of the lowest opening made the odds of bat presence 1.05 times higher (90% confidence limits: 0.91-1.21). The top model was able to correctly identify 62% of winter roosts and 71% of unoccupied trees.
Characteristics of Rafinesque's big-eared bat winter roosts largely overlapped those of summer roosts (n = 170) across eight study sites (Fig. 1). One difference was that trees with no elevated entrance were commonly used in summer but rarely used in winter (Fig. 1). In addition winter roosts occurred in a narrower range of dbh and cavity volume sizes, with relatively small trees unused in winter. Winter roosts were also more likely to have a chimney opening and a rough interior. Summer roost data only from River Bend WMA (n = 15) confirmed many of these results, with significantly higher openings and more chimney openings on winter roosts (Table 3). The Hosmer-Lemeshow statistic was not significant ([chi square] = 7.28, d.f = 8, P = 0.506) indicating adequate goodness-of-fit of the global logistic regression model. The best supported model distinguishing between winter and summer roosts, receiving 62% of the AICc weight, indicated that winter roosts had higher openings (Table 4). The composite model also included chimney opening as a predictor variable. In the composite model, every I m increase in highest opening height made the odds of a roost being a winter roost 1.27 times higher (90% confidence limits: 0.97-1.66), while the presence of a chimney opening made the odds of a roost being a winter roost 1.54 times higher (90% confidence limits: 0.48-4.90). The top model identified 71% of winter roosts and 53% of summer roosts.
In contrast southeastern myotis winter roosts had smaller diameters, cavities, and openings than unoccupied trees (Table 1). Roost trees also lacked chimney openings and consisted of a sweetgum, a red maple, a water hickory (C. aquatica), an overcup oak (Q. lyrata), and an unidentifiable snag. Compared to summer roosts at eight field sites (n = 25), winter roosts generally had smaller diameters, heights, and cavity volumes, and had higher, but smaller openings (Fig. 2). Chimney trees were not used as roosts in either season. Southeastern myotis changed from using only water tupelo roosts in summer, to a variety of tree species, which did not include water tupelo, in winter. Considering summer roost tree data only from River Bend WMA (n = 2), dbh, cavity volume, and tree species differed between winter and summer roosts (Table 3).
Rafinesque's big-eared bats and southeastern myotis use roosts that differ subtly in diameter and cavity dimensions during summer (Carver and Ashley, 2008; Stevenson, 2008; Rice, 2009). However, during winter, roosting bats face different selection pressures due to colder temperatures and reduced food resources (Speakman and Thomas, 2003), which may induce bats to select different types of roosts during colder periods (Boyles and Robbins, 2006; Hein et al., 2008). In cypress-gum swamps, bats must also deal with flood waters that can limit roost availability or trap roosting bats (Rice, 2009). Presumably as a result of these environmental factors interacting with different roosting preferences of the species, we observed species specific seasonal patterns of roost tree selection.
For Rafinesque's big-eared bat during winter, the primary difference between used and unused trees was that used trees were less likely to have low openings. Bats may have avoided low openings due to the risk of flooding. However, the effect on occupancy was weak, with only a 5% increase in the odds of occupancy in the composite model for every 1 m in height of the lowest opening. The weak effect may be because most trees had an elevated opening, reducing risks posed by flooding. Other measured features did not differ between used and unused trees. Similarly, no differences between used and unused trees during winter were identified in Mississippi bottomlands, although entrance height was not examined (Stevenson, 2008). The similarity of used and unused trees suggests that many of the large trees in the oxbow lakes provided adequate roosting conditions and some trees that were unused during the study period may be used at other times during the winter. Although there were few differences between used and unused trees on the oxbow lakes, the large (mean dbh = 106 cm) trees used as roosts are rare on the landscape (Stevenson, 2008), indicating that roost trees were not typical of all trees at our site. The floodplains surrounding the lakes were within the range of roosting bats (0-2.5 km), but our telemetry data indicated they were unused, suggesting that large, hollow water tupelo were strongly preferred to other tree types.
Winter roosts were distinguished from summer roosts by the highest opening height, rather than the lowest. Trees lacking elevated entrances were commonly used in summer, but rarely in winter. The use of roosts with higher openings could be due to flood waters obscuring basal openings or because trees with chimney openings provide a more favorable microclimate during winter (Rice, 2009). If trees without basal openings provide a superior microclimate in winter, we would expect bats to avoid trees with basal openings in winter, regardless of water levels, and to avoid trees without basal hollows during summer. However, Rafinesque's big-eared bat regularly uses trees without basal hollows during summer (Trousdale and Beckett, 2005; Clement and Castleberry, 2013a). Furthermore, bats use trees with basal openings during winter, when not flooded (Stevenson, 2008; Rice, 2009). We suggest that any shift, during winter, from trees with basal openings to elevated openings is primarily due to flood waters submerging basal openings and the attendant risk of being trapped.
Another seasonal difference was that Rafinesque's big-eared bats avoided trees with rough cavity interiors in summer (Clement and Castleberry, 2013a), but were indifferent to interior surface in winter. During summer, smooth interior surfaces may be important as a barrier to snakes and other predators. Several bird species nest in trees with smooth bark that provide protection against snakes (Rudolph et al., 1990; Mullin and Cooper, 2002). During winter, however, snakes in the Coastal Plain are generally less active (e.g., Glaudas et al., 2007; Rudolph et al., 2007) and may pose less of a threat to roosting bats. If selection pressure exerted by snakes is reduced during winter, bats may not be constrained to select cavities with rough interiors for winter roosts.
In contrast to Rafinesque's big eared bats, southeastern myotis winter roosts differed from available trees on the oxbow lakes and from summer roosts. Summer roosts in Georgia were large diameter hydrophilic trees (Clement, 2011) consistent with those used at other sites (Hofmann et al., 1999; Gooding and Langford, 2004; Carver and Ashley, 2008; Stevenson, 2008). However, winter roosts at our site were small hardwood trees that shared few characters with summer roosts. Our results contrast with other studies that found winter roosts similar to summer roosts, with some trees used in both seasons. In Louisiana, almost all summer roosts were also used in winter and roost tree diameter was similar in both seasons (Rice, 2009). In Mississippi, individual trees also were used in both seasons, and diameter increased from summer to winter (Stevenson, 2008).
Although southeastern myotis roosts differ among seasons and locations, they never have chimney openings (Mirowsky and Homer, 1997; Hoffman, 1999; Gooding and Langford, 2004; Carver and Ashley, 2008; Stevenson, 2008; Rice, 2009; Clement, 2011). Southeastern myotis may prefer non chimney trees because reduced air flow provides a better microclimate (Stevenson, 2008; Rice, 2009). Although a chimney has a small effect on microclimate (Clement and Castleberry, 2013b), winter roosts differed in several other aspects known to affect microclimate, including tree size (Coombs et al., 2010), species (Kalcounis and Brigham, 1998), and health (Hosken, 1996). We expect that if microclimate were the primary factor in roost selection, at least some chimney trees would provide an acceptable microclimate, but the universal rejection of chimney trees suggests that another factor drives southeastern myotis roost selection. We hypothesize that the preference for non chimney trees is related to the roosting substrate. In contrast to Rafinesque's big-eared bats that usually roost on cavity walls, southeastern myotis always roosts on the cavity ceiling (Mirowsky and Homer, 1997; Carver and Ashley, 2008; Stevenson, 2008; Rice, 2009; Clement, 2011). Therefore, only non chimney trees provide a roosting substrate for southeastern myotis. Roosting on cavity ceilings may help southeastern myotis avoid predators, take flight, maintain a grip on the roosting substrate, or provide some other unknown benefit.
The preference of southeastern myotis for non chimney trees likely explains the differences between winter and summer roost trees at our study site. The basal openings of nearly all large water tupelo that we located in summer were submerged and unavailable during winter. Therefore, southeastern myotis likely used the only available non chimney trees, even though they differed in size, species, and other characteristics. In other surveys that focused on times or locations with less flooding, summer roosts remained available in winter and roosts showed fewer seasonal differences (Stevenson, 2008; Rice, 2009). We suggest that southeastern myotis share a similar preference for large water tupelo with basal openings throughout the Coastal Plain, but they will use other non chimney trees if preferred trees are flooded.
We identified seasonal differences in roost selection for Rafinesque's big-eared bat and southeastern myotis that are likely due to differences in roost entrances and preferred roosting substrate. Due to their reliance on non chimney trees, southeastern myotis changed roosts substantially in winter, moving from the lake to the floodplain, switching tree species, and using smaller trees. In contrast the ability of Rafinesque's big-eared bat to use roosts with chimney openings allowed them to use a subset of summer roosts and maintain a presence in a flooded oxbow lake in winter. Due to our small sample of southeastern myotis roosts, we likely did not capture the full variation of winter roosts. Nonetheless, we demonstrated that under some circumstances, southeastern myotis use smaller trees and cavities than reported elsewhere and that they may move from cypress-gum swamps to floodplains supporting a different suite of tree species. Accordingly, management that focuses solely on conserving summer roosting habitat may be adequate for Rafinesque's big-eared bat, but inadequate for southeastern myotis.
Acknowledgments.--Funding was provided by the Georgia Department of Natural Resources Wildlife Resources Division and the Daniel B. Warnell School of Forestry and Natural Resources at the University of Georgia. The Georgia Department of Natural Resources provided access and housing at River Bend WMA. We thank C. Carpenter, V. Kinney, and C. Bland for field assistance.
SUBMITTED 7 AUGUST 2012
ACCEPTED 7 FEBRUARY 2013
BATTLE, J. AND S. GOLLADAY. 2001. Hydroperiod influence on breakdown of leaf litter in cypress-gum wetlands. Am. Midl, Nat., 146:128-145.
BOUMA, H. R., H. V. CAREY, AND F. G. M. KROESE. 2010. Hibernation: the immune system at rest. J Leukocyte Biol, 88:619-624.
BOYLES, J., B. MORMANN, AND L. ROBBINS. 2005. Use of an underground winter roost by a male evening bat (Nycticeius humeralis). Southeast. Nat., 4:375-377.
-- AND L. ROBBINS. 2006. Characteristics of summer and winter roost trees used by evening bats (Nycticeius huraeralis) in southwestern Missouri. Am. Midl. Nat., 155:210-220.
BRIGHAM, R. M. 2007. Bats in forests: what we know and what we need to learn, p. 1-16. In: M.J. Lacki, J. P. Hayes, and A. Kurta (eds.). Bats in forests: conservation and management. Johns Hopkins University Press, Baltimore, Maryland. 329 p.
BURNHAM, K. P. AND D. R. ANDERSON. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd ed. Springer-Verlag, New York. 488 p.
CARVER, B. D. AND N. ASHLEY. 2008. Roost tree use by sympatric Rafinesque's big-eared bats (Corynorhinus rafinesquii) and southeastern myotis ( Myotis austroriparius). Am. Midl. Nat., 160:364-373.
CLEMENT, M. J. 2011. Roosting ecology of Rafinesque's big-eared bat and southeastern myotis in the coastal plain of Georgia. Ph.D., Dissertation, University of Georgia, Athens. 185 p.
-- AND S. B. CASTLEBERRY. 2013a. Summer tree roost selection by Rafinesque's big-eared bat. J. Wildl. Manag., 77:414-422.
-- AND --. 2013b. Tree structure and cavity microclimate: implications for bats and birds. Int. J. Biometeorol, 57:437-450.
COOMBS, A. B., J. BOWMAN, AND C.J. GARROWAY. 2010. Thermal properties of tree cavities during winter in a northern hardwood forest. J. Wildl. Manage., 74:1875-1881.
COPE, J. B. AND S. R. HUMPHREY. 1977. Spring and autumn swarming behavior in Indiana bat, Myotis sodalis. J. Mammal., 58:93-95.
CRYAN, P. M. AND J. P. VEILLEUX. 2007. Migration and use of autumn, winter, and spring roosts by tree bats, p. 153-176. In: N.J. Lacki, J. P. Hayes, and A. Kurta (eds.). Bats in forests: conservation and management. Johns Hopkins University Press, Baltimore, Maryland. 329 p.
EFRON, B. 1983. Estimating the error rate of a prediction rule--improvement on cross-validation. J. Am. Stat. Assoc., 78:316-331.
ESTOK, P., S. ZSEBOK, AND B. M. SIEMERS. 2010. Great tits search for, capture, kill and eat hibernating bats. Biol. Letters, 6:59-62.
GLAUDAS, X., K M. ANDREWS, J. D. WILLSON, AND J. W. GIBBONS. 2007. Migration patterns in a population of cottonmouths (Agkistrodon piscivorus) inhabiting an isolated wetland. J. Zool., 271:119-124.
GOODING, G. AND J. R. LANGFORD. 2004. Characteristics of tree roosts of Rafinesque's big-eared bat and southeastern bat in northeastern Louisiana. Southwest. Nat., 49:61-67.
GRUEBER, C. E., S. NAKAGAWA, R. J. LAWS, AND I. G. JAMIESON. 2011. Multimodel inference in ecology and evolution: challenges and solutions. J. Evol. Biol., 24:699-711.
HEIN, C. D., S. B. CASTLEBERRY, AND K. V. MILLER. 2008. Male Seminole bat winter roost-site selection in a managed forest. J. Wildl. Manage., 72:1756-1764.
HOFFMAN, V. E., III. 1999. Roosting and relative abundance of the southeastern myotis, Myotis austroriparius, in a bottomland hardwood forest. M.Sc. Thesis, Arkansas State University, Jonesboro. 35 p.
HOFMAN, J. E., J. E. GARDNER, J. K. KREJCA, AND J. D. GARNER. 1999. Summer records and a maternity roost of the southeastern myotis (Myotis austroriparius) in Illinois. Trans. Illinois State Acad. Sci., 92:95-107.
HOSKEN, D.J. 1996. Roost selection by the lesser long-eared bat, Nyctophilus geoffroyi, and the greater longeared bat, N. major (Chiroptera: Vespertilionidae) in Banksia woodlands. J. R. Soc. West. Aust., 79:211-216.
HOSMER, D. W. AND S. LEMESHOW. 2000. Applied logistic regression. 2nd ed. Wiley, New York. 392 p.
JEPSON, J. 2000. The tree climber's companion. 2nd ed. Access Publications, 104 p.
JONES, C. 1977. Plecotus rafinesquii. Mammal. Spec., 69:1-4.
-- AND R. W. MANNING. 1989. Myotis austroriparius. Mammal. Spec., 332:1-3.
KALCOUNIS, M. C. AND R. M. BRIGHAM. 1998. Secondary use of aspen cavities by tree-roosting big brown bats. J. Wildl. Manage., 62:603-611.
KEATING, K. A. AND S. CHERRY. 2004. Use and interpretation of logistic regression in habitat selection studies. J. Wildl. Manage., 68:774-789.
KUNZ, T. H. AND L. F. LUMSDEN. 2003. Ecology of cavity and foliage roosting bats, p. 3-89. In: T. H. Kunz and M. B. Fenton (eds.). Bat ecology. University of Chicago Press, Chicago, Illinois. 798 p.
LOEB, S. C. AND S.J. ZARNOCH. 2011. Seasonal and multiannual roost use by Rafinesque's big-eared bats in the coastal plain of South Carolina, p. 111-122. In: S. C. Loeb, N.J. Lacki, and D. A. Miller (eds.). Conservation and management of eastern big-eared bats. General Technical Report SRS-145 edition. U.S. Department of Agriculture, Forest Service, Southern Research Station, Asheville, North Carolina. 157 p.
MIROWSKY, K. M. AND P. HORNER. 1997. Roosting ecology of two rare vespertilionid bats, the southeastern myotis and Rafinesque's big-eared bat, in east Texas. 1996 Annual Report (20 Jun 1997), Texas Parks and Wildlife Department, Resource Protection Division, Austin. 45 p.
MORMANN, B. M. AND L. W. ROBBINS. 2007. Winter roosting ecology of eastern red bats in southwest Missouri. J. Wildl. Manage., 71:213-217.
MULLIN, S.J. AND R.J. COOPER. 2002. Barking up the wrong tree: climbing performance of rat snakes and its implications for depredation of avian nests. Can. J. Zool., 80:591-595.
NAGELKERKE, N.J.D. 1991. A note on a general definition of the coefficient of determination. Biometrika, 78:691-692.
PERRY, R. W., D. A. SAUGEY, AND B. G. CRUMP. 2010. Winter roosting ecology of silver-haired bats in an Arkansas forest. Southeast. Nat., 9:563-572.
R DEVELOPMENT COPE TEAM. 2010. R: a language and environment for statistical computing, ver. 2.11.1. R Foundation for Statistical Computing, Vienna, Austria.
RICE, C. L. 2009. Roosting ecology of Corynorhinus rafinesquii (Rafinesque's big-eared bat) and Myotis austroriparius (southeastern myotis) in tree cavities found in a northeastern Louisiana bottomland hardwood forest streambed. M.Sc. Thesis, University of Louisiana at Monroe, Monroe. 124 p.
RUDOLPH, D. C., H. KYLE, AND R. N. CONNER. 1990. Red-cockaded woodpeckers vs rat snakes: the effectiveness of the resin barrier. Wilson Bull., 102:14-22.
--, R. R. SCHAEFER, S.J. BURGDORF, M. DURAN, AND R. N. CONNER. 2007. Pine snake (Pituophis ruthveni and Pituophis melanoleucus lodingi) hibernacula. J. Herpetol., 41:560-565.
SPEAKMAN, J. R. AND D. W. THOMAS. 2003. Physiological ecology and energetics of bats, p. 430-492. In: T. H. Kunz and M. B. Fenton (eds.). Bat ecology. University of Chicago Press, Chicago, Illinois. 798 p.
STEVENSON, C. L. 2008. Availability and seasonal use of diurnal roosts by Rafinesque's big-eared bat and southeastern myotis in bottomland hardwoods of Mississippi. M.Sc. Thesis, Mississippi State University, Starkville, Mississippi. 109 p.
TROUSDALE, A. W. AND D. C. BECKETT. 2005. Characteristics of tree roosts of Rafmesque's big-eared bat (Corynorhinus rafinesquii) in southeastern Mississippi. Am. Midl. Nat., 154:442-449.
MATTHEW J. CLEMENT (1,2) AND STEVEN B. CASTLEBERRY
D.B. Warnell School of Forestry and Natural Resources, University of Georgia, Athens 30602
(1) Present address: USGS Patuxent Wildlife Research Center, Laurel, MD 20708
(2) Corresponding author: e-mail: firstname.lastname@example.org
TABLE 1.--Mean [+ or -] standard deviation of characteristics of unoccupied hollow trees, trees occupied by Rafinesque's big- eared bat (Cmynorhinus rafinesquii), and trees occupied by southeastern myotis (Myotis austroriparius) at River Bend Wildlife Management Area, Georgia during winter 2010. P-values indicate results of t-tests or Fisher's exact tests compared to unoccupied trees Unoccupied Variable Mean (n = 30) Tree height (m) 16.9 [+ or -] 5.6 dbh (cm) 117.8 [+ or -] 48.7 Cavity volume (L) 1,157 [+ or -] 1,509 Solid tree volume (L) 7,841 [+ or -] 11,554 Area of openings ([m.sup.2]) 0.22 [+ or -] 0.23 Widest opening (cm) 32.5 [+ or -] 11.9 Highest opening (m) 6.08 [+ or -] 5.16 Lowest opening (m) 1.73 [+ or -] 2.61 Total openings (no.) 2.23 [+ or -] 1.28 Live (Y/N) 0.97 [+ or -] 0.18 Tupelo tree (Y/N) 0.80 [+ or -] 0.41 Chimney (Y/N) 0.63 [+ or -] 0.49 Rough interior (Y/N) 0.50 [+ or -] 0.51 Rafinesque's big-eared bat Variable Mean (n = 23) P Tree height (m) 16.3 [+ or -] 4.1 0.673 dbh (cm) 105.9 [+ or -] 34.8 0.327 Cavity volume (L) 1,038 [+ or -] 604 0.252 Solid tree volume (L) 4,341 [+ or -] 5,945 0.613 Area of openings ([m.sup.2]) 0.19 [+ or -] 0.12 0.623 Widest opening (cm) 36.4 [+ or -] 13.1 0.256 Highest opening (m) 5.88 [+ or -] 2.70 0.866 Lowest opening (m) 3.56 [+ or -] 2.50 0.013 Total openings (no.) 2.26 [+ or -] 1.66 0.946 Live (Y/N) 0.96 [+ or -] 0.21 0.999 Tupelo tree (Y/N) 1.00 [+ or -] 0.00 0.030 Chimney (Y/N) 0.83 [+ or -] 0.39 0.140 Rough interior (Y/N) 0.41 [+ or -] 0.50 0.577 Southeastern myotis Variable Mean (n = 5) P Tree height (m) 13.4 [+ or -] 8.0 0.187 dbh (cm) 33.2 [+ or -] 18.5 <0.001 Cavity volume (L) 26.7 [+ or -] 55.3 <0.001 Solid tree volume (L) 1138 [+ or -] 1069 0.007 Area of openings ([m.sup.2]) 0.03 [+ or -] 0.03 0.050 Widest opening (cm) 13.9 [+ or -] 10.4 0.001 Highest opening (m) 3.95 [+ or -] 4.73 0.281 Lowest opening (m) 3.82 [+ or -] 4.85 0.362 Total openings (no.) 1.20 [+ or -] 0.45 0.107 Live (Y/N) 0.80 [+ or -] 0.45 0.245 Tupelo tree (Y/N) 0.00 [+ or -] 0.00 <0.001 Chimney (Y/N) 0.00 [+ or -] 0.00 0.003 Rough interior (Y/N) 0.33 [+ or -] 0.58 0.999 TABLE 2.--Predictor variables, number of model parameters (K), Akaike's Information Criterion adjusted for small sample size (AIC,), difference between model and top model (AAICc), model weight ([w.sub.I]), and Nagelkerke's [R.sup.2] for logistic regression models of winter 2010 roost use by Rafinesque's big-eared bat (Corynorhinus raftnesquia) at River Bend Wildlife Management Area, Georgia [DELTA] Variables K AICc AICc [w.sub.I] [R.sup.2] Lowest opening 2 66.15 0.00 0.434 0.131 Intercept-only 1 69.01 2.86 0.104 0.000 Cavity volume 2 69.78 3.63 0.071 0.038 Widest opening 2 69.94 3.79 0.065 0.034 Diameter 2 70.01 3.85 0.063 0.032 Chimney 2 70.10 3.95 0.060 0.029 Rough interior 2 70.49 4.34 0.050 0.019 Solid tree volume 2 70.92 4.76 0.040 0.007 Opening area 2 71.00 4.85 0.039 0.005 Highest opening 2 71.17 5.02 0.035 0.000 Height 2 71.19 5.03 0.035 0.000 Global 7 75.05 8.90 0.005 0.216 TABLE 3.--Mean [+ or -] standard deviation of roost characteristics for Rafinesque's big-eared bat (Corynorhinus raftnesquii) and southeastern myotis (Myotis austroriparius) at River Bend Wildlife Management Area, Georgia. Winter roosts were measured in summer 2011 and summer roosts were measured in summer 2008 (Clement, 2011). P-values indicate results of t-tests or Fisher's exact tests Rafinesque's big-eared bat Variable Winter (n = 23) Summer (n = 15) P Tree height (m) 18.7 [+ or -] 3.9 17.7 [+ or -] 6.2 0.571 dbh (cm) 147.5 [+ or -] 29.8 142.9 [+ or -] 34.3 0.670 Cavity volume (L) 3,229 [+ or -] 1,301 2,773 [+ or -] 1,486 0.267 Solid tree 8,037 [+ or -] 5,474 8,767 [+ or -] 8,554 0.946 volume (L) Area of 0.37 [+ or -] 0.36 0.20 [+ or -] 0.17 0.101 openings ([m.sup.2]) Widest 46.0 [+ or -] 26.8 32.4 [+ or -] 18.3 0.098 opening (cm) Highest 7.51 [+ or -] 2.74 3.97 [+ or -] 3.95 0.003 opening (m) Lowest 1.96 [+ or -] 3.35 0.99 [+ or -] 2.00 0.328 opening (m) Total 3.52 [+ or -] 2.04 2.93 [+ or -] 2.34 0.427 openings (no.) Live (Y/N) 0.95 [+ or -] 0.22 0.93 [+ or -] 0.26 0.999 Tupelo tree (Y/N) 1.00 [+ or -] 0.00 0.93 [+ or -] 0.26 0.417 Chimney (Y/N) 0.81 [+ or -] 0.40 0.40 [+ or -] 0.51 0.017 Rough interior 0.38 [+ or -] 0.50 0.13 [+ or -] 0.35 0.142 (Y/N) Southeastern myotis Variable Winter (n = 5) Summer P Tree height (m) 13.4 [+ or -] 8.0 18.5 [+ or -] 2.1 0.436 dbh (cm) 33.2 [+ or -] 18.5 111.4 [+ or -] 31.8 0.008 Cavity volume (L) 26.7 [+ or -] 55.3 2,242 [+ or -] 1,922 0.022 Solid tree 1138 [+ or -] 1069 4,000 [+ or -] 1,048 0.182 volume (L) Area of 0.03 [+ or -] 0.03 0.09 [+ or -] 0.02 0.082 openings ([m.sup.2]) Widest 13.9 [+ or -] 10.4 19.0 [+ or -] 8.5 0.574 opening (cm) Highest 3.95 [+ or -] 4.73 0.05 [+ or -] 0.07 0.321 opening (m) Lowest 3.95 [+ or -] 4.73 0.05 [+ or -] 0.07 0.321 opening (m) Total 1.20 [+ or -] 0.45 2.00 [+ or -] 0.00 0.062 openings (no.) Live (Y/N) 0.80 [+ or -] 0.45 1.00 [+ or -] 0.00 0.999 Tupelo tree (Y/N) 0.00 [+ or -] 0.00 1.00 [+ or -] 0.00 0.048 Chimney (Y/N) 0.00 [+ or -] 0.00 0.00 [+ or -] 0.00 0.999 Rough interior 0.33 [+ or -] 0.59 0.00 [+ or -] 0.00 0.999 (Y/N) TABLE 4.--Predictor variables, number of model parameters (K), Akaike's Information Criterion adjusted for small sample size ([AIC.sub.c]), difference between model and top model ([DELTA][AIC.sub.c]), model weight ([w.sub.i]), and Nagelkerke's [R.sup.2] for logistic regression models distinguishing winter and summer roost use by Rafinesque's big-eared bat (Corynorhinus rafinesquit) at River Bend Wildlife Management Area, Georgia, 2007-2010 [DELTA] Variables K AICc AICc [w.sub.i] [R.sup.2] Highest opening 2 44.42 0.00 0.620 0.293 Chimney 2 46.81 2.39 0.187 0.221 Opening area 2 49.69 5.27 0.044 0.127 Widest opening 2 49.89 5.47 0.040 0.121 Rough interior 2 50.42 5.99 0.031 0.102 Intercept-only 1 51.02 6.60 0.023 0.000 Cavity volume 2 51.97 7.55 0.014 0.048 Low 2 52.18 7.76 0.012 0.040 Height 2 52.92 8.50 0.009 0.013 Diameter 2 53.07 8.65 0.008 0.007 Solid tree volume 2 53.26 8.84 0.007 0.000 Global 7 55.32 10.90 0.003 0.370
|Printer friendly Cite/link Email Feedback|
|Author:||Clement, Matthew J.; Castleberry, Steven B.|
|Publication:||The American Midland Naturalist|
|Date:||Jul 1, 2013|
|Previous Article:||Do native warm-season grasslands near airports increase bird strike hazards?|
|Next Article:||Demographic effects of white-tailed deer (Odocoileus virginianus) exclosures on white-footed mice (Peromyscus leucopus).|