2010 SOUTHEASTERN NATURALIST 9(4):773–780 Effects of Prescribed Fire and Predator Exclusion on Refuge Selection by Peromyscus gossypinus Le Conte (Cotton Mouse) Anna M. Derrick1, L. Mike Conner2, and Steven B. Castleberry1,* Abstract - Many small mammal species experience population declines following prescribed fire, presumably resulting from increased predation due to lack of cover. However, Peromyscus gossypinus (Cotton Mouse) typically shows a neutral or positive population response following fire. Because they typically spend diurnal hours in below-ground refuges, Cotton Mice may be less susceptible to predation following fire than other small mammals. We examined the effects of prescribed fire and exclusion of mammalian predators on selection of daytime refuges by Cotton Mice. We located daytime refuges of 12 radiotagged Cotton Mice in a fenced mesomammal-predator (hereafter, mesopredator) exclosure (23 refuge locations) and 9 Cotton Mice in an adjacent unfenced control plot (13 refuge locations) for one month prior to and one month after a prescribed fire in winter 2007. Refuge locations included Gopherus polyphemus (Gopher Tortoise) burrows (27.8%), other ground holes (44.4%), stump holes (25.0%), and holes at the base of trees (2.8%). Fire had little effect on refuge selection, likely because Cotton Mice primarily used below-ground refuges, which allowed them to avoid the direct effects of fire and predation following fire. Structure near the refuge, including burrows, stumps, and coarse woody debris, was important in selection of daytime refuges and was particularly important in the presence of mesopredators. Introduction Prescribed fire is critical to conservation and management of many North American ecosystems, including the Pinus palustris Miller (Longleaf Pine) ecosystem of the southeastern United States (Stoddard 1931, Wahlenberg 1946). Because prescribed fires typically result in a temporary reduction of cover, small mammals may be more susceptible to predation immediately following a prescribed fire event, and this mortality may result in temporary population declines (Bock and Bock 1978, Komarek 1963, Tester 1965, Whelan 1995). However, Peromyscus gossypinus Le Conte (Cotton Mouse) has shown neutral or positive immediate population responses to prescribed fire (Hatchell 1964, Jones 1992, Layne 1974, Shadowen 1963). Most studies examining the effects of prescribed fire have concluded that Cotton Mouse population increases are a result of individuals from surrounding unburned areas immediately immigrating into burned areas. Hatchell (1964) observed that Cotton Mice invaded recently burned areas in Louisiana pine stands, noting that the reduction in understory cover following fire had no apparent negative effect on population abundance. Similarly, 1Warnell School of Forestry and Natural Resources, University of Georgia, Athens, GA 30602. 2Joseph W. Jones Ecological Research Center, Route 2, Box 2324, Newton, GA 39870. *Corresponding author - scastle@warnell.uga.edu. 774 Southeastern Naturalist Vol. 9, No. 4 Layne (1974) attributed a Cotton Mouse population increase in burned flatwoods habitat in Florida relative to pre-burn levels to immigration and suggested that individuals from surrounding habitats invaded burned areas to take advantage of increased food availability in the form of seeds and insects. Cotton Mice may be adapted to a wider range of understory cover conditions than other small mammals, such as Sigmodon hispidus Say and Ord (Hispid Cotton Rat), which is negatively impacted by loss of cover following fire (Arata 1959, Layne 1974, Stoddard 1931). Cotton Mice commonly use below-ground refuges, such as stump holes, root boles, and Gopherus polyphemus Daudin (Gopher Tortoise) burrows (Frank and Layne 1992, Hinkelman and Loeb 2007, Layne and Ehrhart 1970, McCay 2000). Below-ground refuges provide lower temperatures during summer and escape cover from fire and predators (Frank and Layne 1992). Although refuges provide protection from predators, the type of refuge selected can influence predation susceptibility. For example, Frank and Layne (1992) found higher predation on Cotton Mice using Gopher Tortoise burrows than other refuge types. Because refuges provide Cotton Mice a place to seek shelter from fire, in addition to protection from predators, the influence of fire and predators likely interact in selection of refuge type. Therefore, our objective was to examine the effects of prescribed fire and mesomammal-predator (hereafter mesopredator) exclusion on Cotton Mouse selection of daytime refuges. We hypothesized that daytime refuges would be surrounded by more vegetative cover and coarse woody debris in areas where mammal predators were present and after fire. Study Area Our study site was located at the Joseph W. Jones Ecological Research Center at Ichauway, an 11,700-ha research center located in Baker County, GA. The Jones Center is located on the Upper Coastal Plain and is dominated by the Longleaf Pine-Aristida stricta Michaux (Wiregrass) forest type. The Jones Center consists of a traditional management zone that is managed for sustainable, multiple-use practices such as Colinus virginianus L. (Northern Bobwhite) hunting, and a benchmark management zone focusing on conservation of natural biota and restoration of pre-settlement land-use patterns. Prescribed fire, primarily dormant season, is utilized throughout the property on two-year intervals. Our research plots were located on a 2897-ha area of continuous benchmark management zone. Methods As part of a larger study examining mesopredator exclusion, 4 mesopredator exclosures and 4 control research plots were established in 2002. We chose one exclosure and one control for our study based on vegetation similarity, Cotton Mouse abundance, and their close proximity to each other. The 36.2-ha exclosure plot was enclosed with woven-wire fencing with wire attached to an E2000 electrical fence charger (Twin Mountain Fence Co., San Angelo, TX) along the top, middle, and bottom to deter mammal predators 2010 A.M. Derrick, L.M. Conner, and S.B. Castleberry 775 from climbing or digging under the fence. The adjacent 36.0-ha control plot was unfenced. Fences were monitored twice weekly for dig-unders and electrical failure. Mesopredator monitoring was conducted seasonally in all exclosures and controls using track counts and thermal camera surveys between 13 July 2004 and 21 March 2007. Mammal exclusion was not total, as Lynx rufus Schreber (Bobcat), Mephitis mephitis Schreber (Striped Skunk), and Didelphis virginiana Kerr (Virginia Opossum) were known to breach the fence. However, we only recorded 4 mesopredator detections in all exclosures versus 44 in controls during our surveys. We captured Cotton Mice on trapping grids established in the exclosure and control, and at trap locations in study plots not restricted to grids from 16 January–15 March 2007. Sherman live traps (7.6 cm x 8.9 cm x 30.5 cm) baited with a mixture of oats and birdseed were placed at each trap location. Cotton Mice >22 g were selected for radio-transmitter attachment. Mice were anesthetized with isoflourane (McColl and Boonstra 1999), and 1.1-g transmitters (Advanced Telemetry Systems, Isanti, MN) were attached to the neck using monofilament. Four males and 5 females were tagged in the control, and 8 males and 4 females were tagged in the exclosure. Animal capture and handling followed recommendations of the American Society of Mammalogists (Gannon et al. 2007) and were approved by the University of Georgia Animal Care and Use Committee (approval no. A2006-10207-c1). We tracked mice for one month before and one month after prescribed fires conducted in the exclosure and control on 15 February 2007. The goals of the prescribed fire were consistent with the overall goals of the fire program at Ichauway, which are fuel reduction, woody vegetation control, perpetuation of fire-dependent species, and wildlife habitat management. Although we did not quantify vegetation cover prior to the fire, cover was considerably lower during the one-month tracking period after the fire than before the fire. We attempted to obtain 1 daytime refuge location/week for each radiotagged mouse. Because some mice died or slipped collars, the same mice were not tracked for the entire study period. Refuges for some individual mice were located only during the pre- (1 exclosure, 5 control) or post-fire (5 exclosure, 3 control) periods, whereas locations for others were obtained during both periods (6 exclosure, 1 control). Although fences were permeable to Cotton Mice, we did not document mice crossing the fence during our study. Refuge locations were recorded with a GPS Pathfinder Pro XRS Receiver (Trimble Navigation Limited, Sunnyvale, CA). At daytime refuge locations, we measured habitat characteristics known to be biologically important to Cotton Mice (Frank and Layne 1992, Loeb 1999, McCay 2000, Mengak and Guynn 2003, Wolfe and Linzey 1977). We recorded refuge type (Gopher Tortoise burrow, stump hole, hole in tree, or other ground hole), measured distance to nearest coarse woody debris (CWD), and estimated percent ground cover within a 4-m radius of the refuge. We assessed availability of structure near the refuge by recording presence/absence of ≥1 Gopher Tortoise burrow, stump, or log within 4 m of the refuge. Each daytime refuge was paired with a 4-m radius circular plot 776 Southeastern Naturalist Vol. 9, No. 4 randomly located 10–100 m from the refuge. Data collection at random sites followed protocols for refuge sites. Sexes were combined for analysis. We developed 12 predictive models using logistic regression with refuge or random site as the binary response. Habitat variables and their interactions with fire and mesopredator exclusion treatments were included as explanatory variables (Table 1). Observations were categorized as either before or after the prescribed fire and within either the predator exclosure or control. Because we were interested in how predator exclusion and fire affected refuge selection, fire and predator exclusion effects were only included in interaction terms in models and not as main effects. We evaluated models with Akaike’s Information Criterion adjusted for small sample sizes (AICc; Burnham and Anderson 2002). We calculated AICc, ΔAICc, and Akaike weights (ωi) for each model to identify models best supported by our data. We considered the best model(s) as those with ΔAICc < 2. All models with 2 < ΔAICc < 4 were considered to have marginal support. We calculated sum of model weights (Σωi) for each variable to determine their importance in daytime refuge selection (Burnham and Anderson 2002). Results We located and characterized 36 daytime refuge locations (23 exclosure, 13 control) from 12 and 9 Cotton Mice in the exclosure and control, respectively. Seventeen refuge locations (9 exclosure, 8 control) were obtained pre-fire and 19 (14 exclosure, 5 control) were obtained post-fire. In the exclosure, 8 (35%) refuges were in Gopher Tortoise burrows, 13 (57%) were in other ground holes, 1 (4%) was in a stump hole, and 1 (4%) was in a hole at the base of a tree. In the control, 8 (62%) refuges were in stump holes, 3 (23%) were in other ground holes, and 2 (15%) were in Gopher Tortoise burrows. The models Structure and Structure*Exclusion received the strongest support from our data (ΔAICc < 2; Table 2) with greater use of structure at Table 1. Definitions of candidate models used to compare characteristics of daytime Cotton Mouse refuges with random locations at the Jones Ecological Research Center, Baker County, GA, 2007. Model Definition Cover Percent vegetation cover within 4-m radius of refuge CWD Distance from refuge to nearest coarse woody debris Structure Presence/absence of a Gopher Tortoise burrow, stump, or fallen log within 4 m of refuge Cover*Exclusion Interaction between Cover and predator-exclusion treatment Cover*Fire Interaction between Cover and fire treatment CWD*Exclusion Interaction between CWD and predator-exclusion treatment CWD*Fire Interaction between CWD and fire treatment Structure*Exclusion Interaction between Structure and predator-exclusion treatment Structure*Fire Interaction between Structure and fire treatment Main effects model Model including Cover, CWD, and Structure (no interactions) Global model Model including Cover, CWD, and Structure and their interactions with fire and exclusion treatments Null model Model containing no variables 2010 A.M. Derrick, L.M. Conner, and S.B. Castleberry 777 refuges in the control plot than exclosure. Structure*Fire, all main effects, and CWD*Exclusion were marginally supported by the data (2 > ΔAICc > 4). Use of structure was marginally greater following fire, and use of CWD was marginally greater at refuges in the control plot than exclosure. Sum of model weights that contained the variable Structure was 0.909. There was less support for distance to nearest CWD as an important characteristic of selected refuges (Σωi = 0.163). Vegetation cover received virtually no support from our data. Sum of model weights for models that included vegetation cover was 0.088, and ΔAICc for all 3 models that examined cover without structure or CWD was >4. Models with predator-exclusion treatment interactions (Σωi = 0.313) had greater support from the data than fire treatment interactions (Σωi = 0.154). Evidence for a predator-exclusion effect was supported when interacting with Structure (Structure*Exclusion) and distance to the nearest CWD (CWD*Exclusion), whereas evidence for the fire treatment was only supported when interacting with structure (Structure*Fire). Discussion We found that presence of structure within 4 m of the refuge, in the form of Gopher Tortoise burrows, stumps, or CWD, was the primary criterion in refuge selection by Cotton Mice. Other studies have noted a similar positive relationship between Cotton Mice and the presence of structure (Hinkelman and Loeb 2007, Loeb 1999, McCay 2000). Additionally, strong support for our Structure*Exclusion model suggests that structure may be especially important in refuge selection in the presence of mesopredators. Mesopredators had been excluded from exclosure plots for approximately 5 years prior to our study, which was likely sufficient time for Cotton Mice to exhibit a behavioral response to mesopredator absence. Although we did not examine avian and reptilian predators in our study, absence of mesopredators may cause compensatory shifts among predator taxa (Derrick et al., in press). Because of differences in feeding strategies, Table 2. Models, number of parameters in the model (K), Akaike’s information criterion adjusted for small sample size (AICc), AICc difference between a model and the model with the lowest AICc (ΔAICc), and weights (ωi) of models used to examine characteristics of Cotton Mouse refuges at the Jones Ecological Research Center, Baker County, GA, 2007. Model K AICc ΔAICc ωi Structure 2 92.65 0.00 0.43 Structure*Exclusion 3 93.78 1.14 0.25 Structure*Fire 3 94.78 2.14 0.15 Main Effects 4 95.97 3.32 0.08 CWD*Exclusion 3 96.39 3.72 0.07 CWD 2 100.19 7.55 0.01 Null 1 101.87 9.23 0.00 Cover 2 102.23 9.59 0.00 CWD*Fire 3 102.25 9.61 0.00 Cover*Exclusion 3 104.35 11.70 0.00 Cover*Fire 3 104.38 11.74 0.00 Global 10 109.11 16.47 0.00 778 Southeastern Naturalist Vol. 9, No. 4 selection pressures exerted by avian and reptilian predators in the absence of mesopredators may result in differences in refuge selection. For example, we found less use of Gopher Tortoise burrows as refuges in the control (15%) relative to the exclosure (35%). Cotton Mice using Gopher Tortoise burrows as refuges may be more susceptible to mammalian predation than when using other structure types (Frank and Layne 1992). Cotton Mice appeared to shift refuge selection in the absence of mesopredators, suggesting an adaptive advantage of using Gopher Tortoise burrows when birds and reptiles are the primary predators. Similar to other studies of Cotton Mouse use of daytime refuges, we documented almost all refuges in below-ground structures. Frank and Layne (1992) found that 94.7% of refuges in Florida were found in below-ground structures, with 70.5% in Gopher Tortoise burrows. Although only 27.8% of refuges observed in our study were associated with Gopher Tortoise burrow openings, many of the other holes used as refuges were small tunnels that likely connected to burrows (chimneys). In contrast, on a South Carolina study site where the Gopher Tortoise was absent, stump holes created by rotting stumps and root boles were the most important refuge locations (Hinkelman and Loeb 2007, McCay 2000). Frank and Layne (1992) suggested that selection of refuge type by Cotton Mice likely reflects availability of structures. Gopher Tortoise burrows and associated holes were commonly available structures for small mammal refuges in the Longleaf Pine forests on our study site. Jones and Franz (1990) suggested that Gopher Tortoise burrows provide thermal refuge and escape from fire for Podomys floridana Chapman (Florida Mouse). Because temperature decreases at a constant rate along the length of the burrow, they suggested that mice avoid surface temperature extremes and select a preferred temperature within the burrow. Although Cotton Mice appear to have a close association with Gopher Tortoise burrows in sympatric areas, they exhibit considerable plasticity in refuge selection across their range (Hinkelman and Loeb 2007). Many small mammals use structures, such as logs, as travel routes presumably because they provide a quieter substrate than leaf litter (Barnum et al. 1992, Graves et al. 1988, Planz and Kirkland 1992). Our data marginally supported the CWD*Exclusion model, providing limited evidence that Cotton Mice select daytime refuges in close proximity to CWD when exposed to mesopredators. Mice may select daytime refuges in habitat with more CWD which allows them to travel more quietly and provides more cover. McCay (2000) found that Cotton Mice in Coastal Plain pine stands were associated with CWD, but use of CWD for travel was less than reported for Peromyscus species in other habitats. He speculated that Cotton Mice in pine stands use CWD less for travel because pine litter is quieter than deciduous leaf litter and less likely to attract predators. Lower use of CWD in pine stands relative to other habitat types may explain why the CWD*Exclusion model was only marginally supported by our data. Studies on other Peromyscus species have shown drastic population declines following fire events (Tester 1965, Tevis 1956). Population declines in these studies appeared to have occurred from a combination of direct 2010 A.M. Derrick, L.M. Conner, and S.B. Castleberry 779 fire-induced mortality, increased predation following fire, and emigration. In our study, we observed no direct mortality from fire, and no radiotagged mice were depredated while under study. Additionally, no radiotagged mice emigrated to unburned areas during or after the fire. Cotton Mice in our study used below-ground refuges, which allowed them to avoid the direct effects of fire and predation following fire despite the reduction in vegetation cover. Cotton Mice in the Longleaf Pine ecosystem evolved with frequent fires and likely have evolved behavioral mechanisms (e.g., selecting below-ground refuges) to avoid the temporary population reduction following fire reported for other Peromyscus species (Tester 1965, Tevis 1956). None of the 3 models that examined vegetation cover in the absence of structure were supported by our data, suggesting that vegetation cover has minimal importance in daytime refuge selection by Cotton Mice. Similarly, Hinkelman and Loeb (2007) found that refuges with less vegetation cover were used more frequently than expected. They suggested that Cotton Mice may avoid areas of high vegetative cover surrounding the refuge because densely vegetated areas provide ambush sites for predators. The same selection criterion likely is exhibited by mice on our study area. We additionally suggest that pine-litter substrate, which makes less noise and is less likely to attract predators than deciduous leaf litter (McCay 2000), reduces dependence on vegetation cover for concealment from predators. Acknowledgments We thank J. Rutledge, G. Morris, and E. Wilichowski for field assistance. We thank G. Barrett for an earlier review of this manuscript. Funding was provided by the Joseph W. Jones Ecological Research Center and Warnell School of Forestry and Natural Resources. Literature Cited Arata, A.A. 1959. Effects of burning on vegetation and rodent populations in a Longleaf Pine-Turkey Oak association in north-central Florida. Quarterly Journal of the Florida Academy of Sciences 22:94–104. Barnum, S.A., C.J. Manville, J.R. Tester, and W.J. Carmen. 1992. Path selection by Peromyscus leucopus in the presence and absence of vegetative cover. Journal of Mammalogy 73:797–801. Bock, C.E., and J.H. Bock. 1978. Response of birds, small mammals, and vegetation to burning sacaton grasslands in southeastern Arizona. Journal of Range Management 31:296–300. Burnham, K.P., and D.R. Anderson. 2002. Model Selection and Multimodel Inference: A Practical Information-theoretic Approach. 2nd Edition. Springer-Verlag, New York, NY. 454 pp. Derrick, A.M., L.M. Conner, and S.B. Castleberry. In press. Effects of prescribed fire and mammalian predators on Hispid Cotton Rat (Sigmodon hispidus) survival and cause-specific mortality. Journal of Wildlife Management. Frank, P.A., and J.N. Layne. 1992. Nests and daytime refugia of Cotton Mice (Peromyscus gossypinus) and Golden Mice (Ochrotomys nuttalli) in south-central Florida. American Midland Naturalist 127:21–30. 780 Southeastern Naturalist Vol. 9, No. 4 Gannon, W.L., R.S. Sikes, and the Animal Care and Use Committee of the American Society of Mammalogists. 2007. Guidelines of the American Society of Mammalogists for the use of wild mammals in research. Journal of Mammalogy 88:809–823. Graves, S., J. Maldonado, and J.O. Wolff. 1988. Use of ground and arboreal microhabitats by Peromyscus leucopus and Peromyscus maniculatus. Canadian Journal of Zoology 66:277–278. Hatchell, G.E. 1964. Small-mammal species and populations in the Loblolly-Shortleaf Pine forest type of Louisiana. US Department of Agriculture Forest Service, Southern Forest Experiment Station, New Orleans, LA. Research Paper SO-10. 12 pp. Hinkelman, T.M., and S.C. Loeb. 2007. Effect of woody debris abundance on daytime refuge use by Cotton Mice. Southeastern Naturalist 6:393–406. Jones, C.A. 1992. Review of the effects of fire on Peromyscus and Podomys. Florida Scientist 55:75–84. Jones, C.A., and R. Franz. 1990. Use of Gopher Tortoise burrows by Florida mice (Podomys floridanus) in Putnum County, Florida. Florida Field Naturalist 18:45–68. Komarek, E.V., Sr. (Ed.). 1963. Fire, research, and education. Proceedings of the 2nd Tall Timbers Fire Ecology Conference, Tall Timbers Research Station, Tallahassee, fl. 192 pp. Layne, J.N. 1974. Ecology of small mammals in a flatwoods habitat in north-central Florida, with emphasis on the Cotton Rat (Sigmodon hispidus). American Museum Novitates 2544:1–48. Layne, J.M., and L.M. Ehrhart. 1970. Digging behavior of four species of deer mice (Peromyscus). American Museum Novitates 2429:1–16. Loeb, S.C. 1999. Responses of small mammals to coarse woody debris in a southeastern pine forest. Journal of Mammalogy 80:460–471. McCay, T.S. 2000. Use of woody debris by Cotton Mice (Peromyscus gossypinus) in a southeastern pine forest. Journal of Mammalogy 81:527–535. McColl, C.J., and R. Boonstra. 1999. Physiological effects of three inhalant anesthetics on Arctic Ground Squirrels. Wildlife Society Bulletin 27:946–951. Mengak, M.T., and D.C. Guynn, Jr. 2003. Small mammal microhabitat use on young Loblolly Pine regeneration areas. Forest Ecology and Management 173:309–317. Planz, J.V., and G. Kirkland, Jr. 1992. Use of woody ground litter as a substrate for travel by the White-footed Mouse, Peromyscus leucopus. Canadian Field- Naturalist 106:118–121. Shadowen, H.E. 1963. A live-trap study of small mammals in Louisiana. Journal of Mammalogy 44:103–108. Stoddard, H.L. 1931. The Bobwhite Quail: Its Habits, Preservation, and Increase. Charles Scribner’s, New York, NY. 559 pp. Tester, J.P. 1965. Effects of a controlled burn on small mammals in a Minnesota oak savanna. American Midland Naturalist 74:240–243. Tevis, L. 1956. Effect of slash burn on forest mice. Journal of Wildlife Management 20:405–409. Wahlenberg, W.G. 1946. Longleaf Pine: Its Use, Ecology, Regeneration, Protection, Growth, and Management. Charles Lathrop Pack Foundation, Washington, DC. 429 pp. Whelan, R.J. 1995. The Ecology of Fire. Cambridge University Press, Cambridge, UK. 360 pp. Wolfe, J.L., and A.V. Linzey. 1977. Peromyscus gossypinus. Mammalian Species No. 70:1–5.