Note Predator Exclusion As a Management Option for Increasing White-Tailed Deer Recruitment L. MIKE CONNER, 1 Joseph W. Jones Ecological Research Center, 3988 Jones Center Drive, Newton, GA 39870, USA MICHAEL J. CHERRY, Joseph W. Jones Ecological Research Center, 3988 Jones Center Drive, Newton, GA 39870, USA BRANDON T. RUTLEDGE, Joseph W. Jones Ecological Research Center, 3988 Jones Center Drive, Newton, GA 39870, USA CHARLES H. KILLMASTER, Wildlife Resources Division, Georgia Department Natural Resources, 2065 US Hwy 278, Social Circle, GA 30025, USA GAIL MORRIS, Joseph W. Jones Ecological Research Center, 3988 Jones Center Drive, Newton, GA 39870, USA LORA L. SMITH, Joseph W. Jones Ecological Research Center, 3988 Jones Center Drive, Newton, GA 39870, USA ABSTRACT Lethal control of coyotes ( Canis latrans ) is a mechanism for increasing white-tailed deer ( Odocoileus virginianus ) recruitment but can be difficult to implement and may be ineffective on small parcels of land because of coyote immigration. In 2003, we constructed 4 40-ha mesopredator exclosures with the objective of quantifying the influence of mesopredators, including coyotes, on select wildlife populations and communities. Camera trapping indicated neonate/adult female white-tailed deer ratios were approximately 2 times greater inside predator exclosures than in control plots. Hunter success data provided further evidence of increased recruitment associated with exclosures relative to the remainder of the study site. Because exclosures were located on the boundary of our study area and off-site white-tailed deer harvest records were not available, we used estimates of deer density, sex ratios, reproductive rates, relative use of exclosures, and neonate survival inside and outside of exclosures to parameterize a Monte Carlo simulation model to estimate the impact of our 4 exclosures on neonate recruitment into the start of the firearms hunting season on a 1,600- ha area. After 1,000 iterations, we estimated that predator exclusion provided an additional 18 0.13 ( x SE; median ¼ 15, interquartile range [IQR] ¼ 9–24) recruits/year above that expected without exclusion. Simulated neonate/adult female ratio was 0.74 0.03 (median ¼ 0.72, IQR ¼ 0.52–0.93) with exclosures and 0.41 0.008 (median ¼ 0.39, IQR ¼ 0.21–0.58) without. Simulated exclosure impact on recruitment was most sensitive to white-tailed deer relative use of exclosures and was least sensitive to neonate survival inside exclosures. We suggest that predator exclusion may be an effective mechanism for reducing neonate mortality and increasing white-tailed deer recruitment in areas where neonate survival is affected by predation. Exclosures may be particularly valuable on small parcels of land where predator removal would be offset by immigration. Finally, exclosures provide refuge from predators throughout the year, and we hypothesize exclosures may also contribute to adult survival and reduce impacts of predation risk that may also affect white-tailed deer population dynamics. Additional research incorporating variation in exclosure size and white-tailed deer density is needed to better evaluate predator exclosures for creating neonate refugia and mitigating sub-lethal impacts of predation. Ó 2015 The Wildlife Society. KEY WORDS Canis latrans , coyote, Odocoileus virginianus , predator, predator control, predator exclusion, white- tailed deer. The recent range expansion of the coyote ( Canis latrans ) into the eastern United States (Hill et al. 1987, Gompper 2002) has exposed many white-tailed deer ( Odocoileus virginianus ) populations to predation pressure that has largely been absent during the last century (Nowak 2002). Increasing coyote populations and declining white-tailed deer pop- ulations have led to concern that coyotes are negatively affecting neonate recruitment (Ballard et al. 1999; Kilgo et al. 2010, 2012). Coyote control studies (Howze et al. 2009, VanGilder et al. 2009) provide evidence that lethal coyote removal can have a positive effect on white-tailed deer recruitment, but recent work indicates white-tailed deer response to coyote control is site-specific (Gulsby et al. 2015) and may not be a viable option for most landowners (Kilgo et al. 2014). Kilgo et al. (2014) observed increased white-tailed deer recruitment during the first year of a 3-year intensive coyote removal effort. Recruitment dropped to pre-removal levels during the second year of removal and was intermediate Received: 16 January 2015; Accepted: 28 August 2015 1 E-mail: mike.conner@jonesctr.org The Journal of Wildlife Management 80(1):162–170; 2016; DOI: 10.1002/jwmg.999 162 The Journal of Wildlife Management 80(1) during the third year, leading to the conclusion that coyote removal was not an effective management tool for increasing white-tailed deer recruitment. Difficulty controlling coyotes in forested ecosystems and the need to remove coyotes over very large areas contributed to reduced viability of lethal coyote control as a management option. Fences may be used as an alternative to lethal control of predators and are most often associated with prey species restoration (Treydte et al. 2001, Moseby and O’Donnell 2003, Richards and Short 2003), mitigating effects of introduced predators on native wildlife (Hayward and Kerley 2009), or for protecting existing populations of imperiled species (Larson et al. 2002, Murphy et al. 2003, Baskale and Kaska 2005). Fences have also been used as an experimental tool for studying predator impact on prey populations and communities (Krebs et al. 1995, Conner et al. 2010, Smith et al. 2013). Decreased predation risk associated with predator exclu- sion can result in a behavioral response of prey (Apfelbach et al. 2005, Schmidt et al. 2008, Zanette et al. 2011) as prey seek to balance risk with reward (Brown and Kotler 2004). This balancing act often includes increasing use of refuge sites during risky times (Lima and Dill 1990, Lima 1998, Brown and Kotler 2004, Orrock et al. 2013); all else being equal, prey should prefer the refuge provided by the predator exclosure. Use of predator exclosures for white-tailed deer management (i.e., fences that exclude predators but not white-tailed deer) has received little attention, but Teer et al. (1991) reported the positive effects of a 391-ha coyote exclosure to white-tailed deer and Carrera et al. (2015) discussed results of excluding predators on an enclosed mule deer ( Odocoileus hemionus ) population. In 2003, we initiated a large-scale predator exclosure experiment to better understand the role of mesomammal predators within a forested ecosystem (Conner et al. 2010; Morris et al. 2011 a , b , c ; Karmacharya et al. 2013). We immediately observed white-tailed deer use of exclosures to be greater than use of controls. Subsequent research (Nelson et al. 2015) suggested neonate survival within exclosures was greater than survival outside of exclosures. However, this research was not designed to assess effects of exclosures on neonate survival; thus, neonate captures were serendipitous resulting in small sample size ( n ¼ 4) within exclosures. We used existing camera trapping data (Cherry et al. 2015) to determine if there was evidence of greater recruitment within exclosures than in control sites, and we used hunter success data (Joseph W. Jones Ecological Research Center 2014) to determine if exclosures affected white-tailed deer recruitment within and surrounding exclosures. Finally, we used Monte Carlo simulations to estimate effects of exclosures on white-tailed deer recruitment. For simulations, we used site-wide white-tailed deer monitoring data (i.e., deer density, sex ratios, and corpora lutea (CL) counts; Joseph W. Jones Ecological Research Center 2014), indices of white-tailed deer use of predator exclosures relative to control areas, and known-fate neonate survival data (Nelson et al. 2015) to simulate the effect of our predator exclosures on recruitment. STUDY AREA The study took place on Ichauway, the 11,000-ha research site of the Joseph W. Jones Ecological Research Center in Baker County, Georgia, USA. Topography was flat to gently rolling with elevation ranging between 30 m and 70 m above sea level. Rainfall averaged 132 cm/year and the average daily temperature ranged from 11 8 C in winter to 27 8 C in summer. The site included approximately 7,250 ha of longleaf pine stands ( Pinus palustris ). Other forest types included slash ( P. elliottii ) and loblolly pine ( P. taeda ) forests, mixed pine and hardwood forests, lowland hardwood hammocks, oak barrens, and cypress–gum ( Taxodium ascendens – Nyssa biflora ) limesink ponds (Conner et al. 2010). Upland sites were subject to prescribed burning on an approximate 2-year rotation. This burn regime resulted in a relatively sparse mid-story and kept hardwoods, primarily oaks ( Quercus spp.), largely restricted to the shrub layer (i.e., height < 2 m). White-tailed deer abundance was managed using sport hunting to maintain the herd below carrying capacity. During the study, site-wide deer density averaged 6.4 deer/ km 2 with a 1.73:1.00 femal/male sex ratio (Joseph W. Jones Ecological Research Center 2014). Raccoons ( Procyon lotor ), Virginia opossums ( Didelphis virginiana ), striped skunks ( Mephitis mephitis ), gray foxes ( Urocyon cinereoargenteus ), red foxes ( Vulpes vulpes ), coyotes, and bobcats ( Lynx rufus ) were present on the area, but coyotes were the primary predator of white-tailed deer on our study site (Nelson et al. 2015). METHODS Predator Exclusion The predator exclusion experiment took place within longleaf pine-dominated sites with native ground cover (Conner et al. 2010; Morris et al. 2011 a , b , c ; Smith et al. 2013). We selected 8 40-ha sites that were completely surrounded by roads, of similar habitat composition, and located in the northern third of the study area (to avoid conflicts with other long-term research or conservation projects). We randomly selected 4 of these plots to receive mesopredator exclusion; the remaining 4 unfenced plots served as controls. At sites chosen for mesopredator exclusion, we constructed a woven-wire (1.22 m tall, 10 20-cm mesh) fence with electric wire attached to E2000 electrical fence chargers (Twin Mountain Fence Company, San Angelo, TX, USA) along the top, middle, and bottom to deter mesopredators from climbing over or digging under fences. We removed all trees outside of exclosures that had branches overlapping the fence to ensure predators did not use overhanging branches to trespass. At the onset of the study, we trapped mammalian predators from within exclosures using a combination of soft-catch (Woodstream Corp., Lititz, PA, USA) and cage (Tomahawk Live Trap Company, Tomahawk, WI, USA) traps. We relocated captured animals just outside of the exclosure in which the animal was captured. Mesopredators targeted for capture included raccoons, Virginia opossums, striped skunks, gray foxes, red foxes, coyotes, and bobcats. We Conner et al. Predator Exclosures 163 19372817, 2016, 1, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/jwmg.999 by University Of Florida, Wiley Online Library on [11/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License trapped each exclosure periodically to remove any meso- predators that may have been able to breach the fence; the majority of the removals were Virginia opossums that were small enough to fit through the woven wire mesh (Conner et al. 2010). We monitored fences twice weekly for dig-ins and to ensure the electric wires were functioning. Animal capture and handling followed recommendations of the American Society of Mammalogists (Sikes et al. 2011) and took place under Georgia Scientific Collecting Permit number 29-WJH-13-203. We used thermal camera surveys, sampled by driving around each plot at 10 km/hr and recording all mammals detected (Boonstra et al. 1994) using a Palm IR450 thermal camera (Ratheon Company, Waltham, MA, USA), and track count surveys (Frey et al. 2003) to monitor mesopredator relative abundance within predator exclosures and control sites. Track count stations consisted of 5, 1 2- m areas along the inside edge of exclosures and along the roads serving as boundaries of control sites. We sampled each plot 3 times/season using track counts and thermal camera surveys during 2004–2007; sampling since 2007 has consisted of track counts only. Complete mesopredator sampling methods are described in Conner et al. (2010). We detected only 1 coyote trespass; this occurred when a flood event resulted in a week-long electrical failure of one of the exclosures. Moreover, global positioning system (GPS) monitoring of coyotes during 2011–2013 as part of a companion study suggested coyote use in the vicinity of exclosures, but no coyote locations were obtained inside predator exclosures (Fig. 1). Finally, we opportunistically observed white-tailed deer jumping exclosure fences, and our observations indicated that ability to jump the fence occurred at approximately 12 weeks of age. White-Tailed Deer Recruitment During Augusts of 2011 and 2012, we placed 2 motion sensitive cameras (Cuddeback Capture Trail Camera, Non Typical, Green Bay, WI) within each predator exclosure and control (1 camera/20 ha) to measure white-tailed deer antipredator behaviors (Cherry et al. 2015). We baited cameras with shelled corn and monitored them for 2 weeks during each sample interval. Although neonates may have been too young to travel with their dams in August (McCoy et al. 2011) resulting in underrepresentation of neonates at bait stations, we assumed this effect would be the same regardless of whether neonates were in exclosures or controls. Therefore, for each trial, we counted the number of adult female white-tailed deer and neonates/juveniles detected by our cameras using individual exclosures and controls as our experimental units. From each set of counts, we calculated a neonate/adult female ratio and used a general linear model to determine if the ratio differed as a function of treatment while blocking by year. The camera trapping data allowed assessment of recruit- ment only relative to our exclosure and control plots. Because we also sought to determine if exclosures had a demonstrable impact on white-tailed deer abundance, we used white-tailed deer harvest data to determine if predator exclosures resulted in a measurable change in hunter success. On our study area, deer hunters were assigned a hunting block and were required to sign in before hunting and to check all harvested animals. Additionally, hunting hours were restricted for safety reasons (i.e., hunting was allowed for approx. 1.5 hr each morning and afternoon during deer season), allowing us to quantify hunter effort based on number of hunts rather than hunter-hours. We classified hunting blocks into those that either included or bordered predator exclosures (hereafter northern zone; 2,800 ha) and those that did not (hereafter southern zone; 8,200 ha). For each hunting season between 1994 and 2014, we calculated hunter success as the number of deer harvested/100 hunts for both the northern and southern zones. We then compared hunter success before and after exclosure construction in northern and southern zones using a before-after-control-intervention (BACI) model (Stewart-Oaten et al. 1986). A significant interaction between zones (northern or southern) and time (before or after exclosure construction) indicates an exclosure effect. We conducted statistical analyses in R (R 2.15, Development Core Team 2013) using an alpha value of 0.05. Simulations Exclosures were located in a relatively isolated portion of our study area and near its boundary, and it was not possible to quantify white-tailed deer harvest on adjacent lands. Lack of off-site estimates of harvest and lack of data specifically collected to assess recruitment (e.g., known-fate survival analyses associated with exclosures) precluded our ability to explicitly quantify number of white-tailed deer recruited into the hunted population. Therefore, we relied on long-term monitoring data and data collected in association with other Figure 1. Coyote global positioning system locations recorded in the area of mesopredator exclosures at the Joseph W. Jones Ecological Research Center in Baker, County, Georgia, USA during 2012–2013. No coyote locations occurred inside mesopredator exclosures, providing evidence that coyotes were effectively excluded. 164 The Journal of Wildlife Management 80(1) 19372817, 2016, 1, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/jwmg.999 by University Of Florida, Wiley Online Library on [11/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License on-site research projects (Table 1) to parameterize a model of annual recruitment (Fig. 2). We then used Monte Carlo simulations to quantify how white-tailed deer recruitment was affected by predator exclosures. To perform simulations, we assumed a 1,600-ha study area such that our exclosures constituted 10% of the area of interest (i.e., 160 ha, which is the same total area as our exclosures). We used annual (2002–2013) white-tailed deer density (number/km 2 ) estimates obtained from spotlight counts and distance sampling along 2, 19.2-km transects (Joseph W. Jones Ecological Research Center 2014). Ovulation rates were estimated using CL counts (Mansell 1971) of 39 deer harvested during the hunting seasons of 2005–2006, 2006–2007, and 2012–2013 (Cherry 2014). Neonate survival estimates outside of exclosures were derived from known fate survival estimates calculated for 47 radio-tagged neonates (Nelson et al. 2015). We considered known-fate survival estimates of neonates inside exclosures unreliable because of small sample size ( n ¼ 4; all survived); therefore, we based survival estimates inside exclosures on the na € ıve survival rate of monitored neonates (Nelson et al. 2015) after excluding neonates that were depredated. When hunters discharged a firearm, they were required to report the shot, provide any harvested animal for biological data collection, and report number, age (juvenile or ad), and sex of white-tailed deer seen during the hunt. We used these hunter observations to estimate sex ratios during 2002–2014 (Joseph W. Jones Ecological Research Center 2014). We initially sampled white-tailed deer use of meso- predator exclosure and control plots using both thermal imagery and track count methods as described above. Both methods suggested white-tailed deer used exclosures more than controls. However, we had no thermal data during peak parturition seasons and thermal camera sampling was abandoned after 4 years of sampling. Therefore, we used track count data to compare white-tailed deer relative use of exclosures and controls. We calculated use of exclosures relative to controls during each parturition season Table 1. Variables, parameter estimates ( x SD), source of data used to derive estimates, and statistical distributions used in Monte Carlo simulations to estimate annual impact of 4 40-ha mesopredator exclosures on white-tailed deer neonate recruitment into the opening of the firearm hunting season at the Joseph W. Jones Ecological Research Center in Baker County, Georgia, USA. Variable Parameter estimate Data source Distribution a Density (deer/km 2 ) 6.4 1.7 Distance sampling b N(6.4, 1.7, 4, 8) Sex ratio (F:M) 1.73 0.17 Hunter observation c N(1.73, 0.17) Relative use of exclosure 4.85 3.33 Track counts d b -PERT(2, 3, 13, 4) Ovulation rate 1.49 0.43 Harvest data e N(1.49, 0.43, 0, 2) Neonate survival inside 1.0 Telemetry f U(0.60, 1.0) Neonate survival outside 0.29 0.29 Telemetry N(0.29, 0.29, 0, 0.6) a Distribution used to generate estimates for simulation. N ¼ normal distribution ( x , SD, min., max.); if no minimum and maximum values are listed, the distribution was not truncated. b -PERT ¼ b -PERT distribution (min., modal value, max., l ; Roseboom et al. 1959). U ¼ uniform distribution (min., max.). b Deer density (deer/km 2 ) calculated from observations obtained from spotlighting or thermal imaging along 2, 19.2-km sections of road. Sampling occurred annually after deer season, 2002–2013. c Data obtained from annual hunter logs, 2002–2013. d Track count stations consisted of 5, 1 2-m raked areas within each mesopredator exclosure and control. Stations were sampled 3 times during summer, 2002–2013. e Corpora lutea counts obtained from hunter harvested deer. f Telemetry used to determine known-fate fawn survival estimates to the opening day of firearms deer season (Nelson et al. 2015). We excluded neonates killed by predators to obtain the estimate inside the predator exclosure (na € ıve survival was 0.85). A limited sample ( n ¼ 4) of neonates were born inside exclosures and all survived, setting the upper bound of the simulated distribution. Figure 2. Graphical model used in Monte Carlo simulations to estimate impact of mesopredator exclosures on white-tailed deer neonate recruitment until the first day of firearm hunting season at the Joseph W. Jones Ecological Research Center in Baker County, Georgia, USA. Conner et al. Predator Exclosures 165 19372817, 2016, 1, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/jwmg.999 by University Of Florida, Wiley Online Library on [11/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License (2003–2014) as the ratio of average number of track plots containing white-tailed deer tracks within exclosures to the average number of plots containing tracks in the controls. We used RiskAMP 4.12 (Structured Data, LLC, San Francisco, CA) and parameter estimates (Table 1) to perform Monte Carlo simulations of a model (Fig. 2) estimating the number of white-tailed deer recruited into the population on the first day of the firearms hunting season (i.e., third Saturday in Oct). We also used the model to estimate the neonate/adult female ratio, an index of recruitment (Jacobson et al. 1997) used by managers. Our model started with deer density at the beginning of parturition season. We used estimated sex ratio to calculate the number of adult females. We then estimated number of offspring using simulated ovulation rates, and calculated number of offspring born within predator exclosures using a simulated preference value for exclosures during the parturition season. We multiplied number of offspring by the appropriate simulated survival rate to determine number of recruits inside and outside exclosures; we summed numbers to estimate recruitment in presence of exclosures and calculated the neonate/adult female ratio. We assumed that adult female survival during the period of interest was the same inside and outside our exclosures. Because adult female survival during this period was generally high (i.e., > 98%) on our study area (100% of 32 monitored adult females survived; L. M. Conner, Joseph W. Jones Ecological Research Center, unpublished data) and elsewhere (Campbell et al. 2005, Webb et al. 2010), we ignored female survival to simplify simulations. To simulate recruitment outside exclosures, we used the product of simulated number of adult females, ovulation rate, and neonate survival in the absence of exclosures. We considered the difference between recruit- ment with and without exclosures to be the impact of exclosures on recruitment. We performed 1,000 iterations of the model to calculate summary statistics ( x , SE, median, and interquartile range [IQR]) associated with exclosure impact on number of white-tailed deer neonates recruited into the fall hunting season and impacts on neonate/adult female ratios at the start of the fall hunting season. We evaluated effects of input parameter estimates on number of additional neonates recruited by determining the minimum, mean, and maximum values that actually occurred (i.e., values actually drawn from the specified distribution) in simulations. For each minimum, mean, and maximum value of a simulated variable, we performed 1,000 additional simulations allowing the remaining variables to vary as based on the distributions used in the overall simulation (Table 1). We then plotted change in neonate recruitment relative to the minimum, mean, and maximum value for each variable. For example, the theoretical maximum value used to simulate relative use by white- tailed deer of exclosures was 13.0, the theoretical minimum value was 2.0, and the modal value was 3.0; of the 1,000 values simulated from this distribution, the actual mini- mum, mean, and maximum values were 2.01, 4.5, and 10.8, respectively. We performed 1,000 simulations for each of these fixed values of preference while allowing simulated values for other parameters to vary and calculated mean impact on neonate recruitment for each value. RESULTS We used camera trapping within exclosures and controls for 2 weeks during the Augusts of 2011 and 2012 and observed images of 7,033 adult female (4,560 in exclosures and 2,473 in controls) and 1,345 noeonate white-tailed deer (1,112 in exclosures and 233 in controls). The average annual neonate/ adult female ratio was greater ( F 1,13 ¼ 5.02, P ¼ 0.04) in exclosures (0.19 0.04) than in controls (0.09 0.04). Hunter success prior to construction of predator exclosures (1994–2002) was 3.0 0.3 ( x SE) white-tailed deer harvested/100 hunts in the northern zone and 16.0 1.0 in the southern zone. After predator exclosures were constructed (2002–2014), hunter success was to 7.0 1.0 deer/100 hunts in the northern zone and 8.0 1.0 in the southern hunting zone. There was a significant interaction ( F 1,37 ¼ 24.29, P < 0.001) between zones and presence of predator exclosures (before and after construction), indicat- ing that hunter success increased in association with construction of predator exclosures. Simulation results indicated that increased white-tailed deer recruitment associated with mesopredator exclosures was highly skewed (Fig. 3) with 18 0.13 (median ¼ 15, IQR ¼ 9–24) neonates/year above that expected without exclosures (average of 41 neonates annually recruited into the fall population with exclosures and only 23 without on our simulated 1,600 ha area). Simulated neonate/adult female ratio was 0.74 0.03 (median ¼ 0.72, IQR ¼ 0.52–0.93) with exclosures and 0.41 0.008 (median ¼ 0.39, IQR ¼ 0.21–0.58) without. Our analyses of impacts of variation of input parameters (Table 2) suggested that white-tailed deer relative use of exclosures resulted in the greatest variation in change in recruitment, whereas simulated neonate survival inside the exclosures had the least impact on variation (Fig. 4). Figure 3. Number of additional white-tailed deer recruited to the first day of firearm hunting season at the Joseph W. Jones Ecological Research Center in Baker County, Georgia, USA as a result of mesopredator exclosures. Data are based on 1,000 Monte Carlo simulations. 166 The Journal of Wildlife Management 80(1) 19372817, 2016, 1, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/jwmg.999 by University Of Florida, Wiley Online Library on [11/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License DISCUSSION Empirical data indicated that mesopredator exclosures positively affected the white-tailed deer population on our study site. Camera trapping showed that the neonate/adult female ratio was approximately 2 greater inside predator exclosures than in controls. Hunter success increased in the vicinity of the predator exclosures but declined elsewhere. Although we lacked data that would allow for direct quantification of effects on white-tailed deer recruitment on our study site, simulations indicated that exclosures contributed an average of 18 extra recruits into the fall hunting season. Because there is much interest in the effects of coyotes on white-tailed deer populations and in managing white-tailed deer recruitment in the presence of coyotes (Saalfeld and Ditchkoff 2007, McCoy et al. 2013, Kilgo et al. 2014, Robinson et al. 2014, Chitwood et al. 2015), there also may be interest in extrapolating our modeling results to other areas. Therefore, we emphasize that our simulations were used to estimate effects of mesopredator exclosures on white- tailed deer recruitment into the hunted population of our study site. As such, our parameter inputs into the model were derived from study area-specific estimates and may not apply to other sites. Still, all distributions used in our simulations were within the ranges reported for other white-tailed deer populations in the southeastern United States (Ditchkoff 2011, Kilgo et al. 2012). Many of the distributions used in our simulations were conservative. For example, our data indicated that relative use of exclosures during the parturition season was approximately 5 times greater than for control plots with a range from 2.0 to 13, yet the distribution used to simulate preference resulted in lower use relative to that of a truncated normal distribution; indeed, 75% of all simulated values were < 5.5. Additionally, neonate survival estimates outside of exclosures were derived from left-truncated data (i.e., neonates were not captured at the moment of birth and significant mortality can occur between birth and when neonates enter the marked sample), and this results in a 31% overestimate of survival (Gilbert et al. 2014). Further, neonate survival estimates were truncated between 0.0 and 0.6. This upper limit would be among the greatest reported for white-tailed deer neonates in the Southeast in the presence of coyotes (Kilgo et al. 2012, McCoy et al. 2013). Finally, the mean simulated ovulation rate was 1.37 CLs/ female, less than the observed mean (1.49 CLs/female) for our population. Simulated neonate survival inside exclosures had the least impact on change in recruitment, but it is obvious this value must be greater than survival outside of exclosures for there to be an exclosure benefit. We suggest the chosen input distribution for this parameter was also conservative because the calculated na € ıve survival estimate of 0.85 was greater than the midpoint of the simulated distribution (0.80). Addition- ally, coyote predation was the greatest source of neonate mortality on our study area (Nelson et al. 2015), and an earlier coyote removal experiment (Howze et al. 2009) provided evidence that predator-induced mortality was largely additive. Our track count data and coyote GPS locations indicated fences were effective at excluding coyotes; thus, coyote predation of neonates inside exclosures should have been virtually eliminated. A limited sample of 4 radio- monitored neonates born inside exclosures corroborates this idea; all survived the period of interest. Finally, Teer et al. (1991) reported a near doubling of the neonate/adult female ratio following construction of a single large coyote exclosure. This increase is very similar to our simulated results, providing further evidence that our simulated survival rates inside exclosures were reasonable. We did not account for correlations among parameter estimates used in our simulations and this could have also affected our results. This is particularly relevant regarding relative use of exclosures and neonate survival outside of exclosures because they are likely inversely correlated; increased predation risk should cause increased use of exclosures as refuge sites (Lima and Dill 1990, Brown and Kotler 2004). Therefore, our model would underestimate benefit of exclosures under great predation risk and overestimate benefit if predation risk is low. However, overestimation of benefit should not be problematic because if predation risk is low, managing predation should not be a concern. Reduced antlerless deer harvest can not only maintain some white-tailed deer populations exposed to coyote predation but may also reverse population declines (Robinson et al. Table 2. Minimum, mean, and maximum values used to evaluate impact of simulated values for variables used to estimate annual impact of 4 40-ha mesopredator exclosures on white-tailed deer neonate recruitment into the opening of the firearm hunting season at the Joseph W. Jones Ecological Research Center in Baker County, Georgia, USA. For each minimum, mean, and maximum value, 1,000 simulations were performed while allowing values of remaining variables to vary from their simulated distribution. Variable Distribution a Min. x Max. Density (deer/km 2 ) b N(6.4, 1.7, 4, 8) 4.0 6.0 7.75 Sex ratio (F:M) c N(1.73, 0.17) 1.15 1.72 2.22 Relative use d b -PERT(2, 3, 13, 4) 2.03 4.52 10.8 Ovulation rate e N(1.49, 0.43, 0, 2) 0.21 1.37 2.0 Neonate survival inside f U(0.60, 1.0) 0.6 0.80 1.0 Neonate survival outside N(0.29, 0.29, 0, 0.6) 0.0 0.30 0.6 a Distribution used to generate estimates for simulation. N ¼ normal distribution ( x , SD, min., max.); if no minimum and maximum values are listed, the distribution was not truncated. b -PERT ¼ b -PERT distribution (min., modal value, max., l ;Roseboom et al. 1959). U ¼ Uniform distribution (min., max.). b Deer density (deer/km 2 ) calculated from observations obtained from spotlighting or thermal imaging along 2, 19.2-km sections of road. Sampling occurred annually after deer season, 2002–2013. c Data obtained from annual hunter logs, 2002–2013. d Relative use of exclosures determined using track counts. Track count stations consisted of 5, 1 2-m raked areas within each mesopredator exclosure and control. Stations were sampled 3 times during summer, 2002–2013. e Corpora lutea counts obtained from hunter harvested deer. f Telemetry used to determine known fate fawn survival estimates to the opening day of firearms deer season (Nelson et al. 2015). We excluded neonates killed by predators to obtain the estimate inside the predator exclosure (na € ıve survival was 0.85). A limited sample ( n ¼ 4) of neonates were born inside exclosures and all survived, setting the upper bound of the simulated distribution. Conner et al. Predator Exclosures 167 19372817, 2016, 1, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/jwmg.999 by University Of Florida, Wiley Online Library on [11/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 2014). However, in some cases, female harvest may already be negligible, rendering further harvest reductions ineffective (Kilgo et al. 2014, Chitwood et al. 2015). In other cases, regulatory attempts to reduce antlerless harvest may be ineffective, particularly when illegal harvest is a substantial mortality source (Patterson et al. 2002). Therefore, when white-tailed deer recruitment is insufficient to meet management goals and further reducing antlerless harvest is not possible, reducing predation through predator control or exclusion may be the only viable management option. If white-tailed deer are limited by predation on neonates, reducing predation by lethally controlling predators can increase recruitment (Howze et al. 2009, VanGilder et al. 2009, Gulsby et al. 2015). However, there are instances when lethal control of coyotes would be expected to be ineffective. For example, if lethal predator removals occur in a small area from a predator population that is continuously distributed over a larger area, then control efforts constitute a spatially explicit harvest from a continuous population (McCullough 1996). Thus, immigration from surrounding areas can render removal efforts ineffective (Conner and Morris 2015). This is particularly apparent when attempting to manage predation on small parcels of land because immigration from surrounding areas can keep pace with removals (Bodey et al. 2011). In these cases, predator exclusion may be the only effective approach for managing predation. Coyotes are capable of preying on large cervids including moose ( Alces alces ; Benson and Patterson 2013) and mule deer (Lingle and Pellis 2002) and are predators of adult white- tailed deer in northern temperate climates (Whitlaw et al. 1998, Patterson et al. 2002). There is growing evidence that coyotes also prey on adult white-tailed deer in the Southeast (Chitwood et al. 2014). The ability of coyotes to prey on adult deer suggests that deer must respond to coyotes throughout the year and not only during the fawning period. Impacts of predation risk are well recognized in ungulates (Hunter and Skinner 1998), and predation from coyotes can negatively affect white-tailed deer foraging behavior (Cherry et al. 2015) and reproduction (Cherry 2014). Therefore, predator exclosures may also reduce the need for energetically costly antipredator behaviors (e.g., increased vigilance while foraging) by providing year-round reduction in predation risk. We propose white-tailed deer are able to detect reduced predation risk associated with mesopredator exclosures. This leads to increased use of exclosures by adult female white- tailed deer, which subsequently give birth to young. Young born inside predator exclosures experience increased survival due to lack of predation. These young then disperse back into the larger landscape resulting in increased recruitment in the vicinity of the exclosures. Each step in our proposed mechanism can be empirically investigated, providing rich opportunities for future research. Specifically, research using methods explicitly designed for the task, as opposed to retrospective analyses and simulation modeling, is needed. Additionally, our exclosures were all approximately the same Figure 4. Average simulated minimum (bottom of line), mean (triangle), and maximum (top of line) number of additional white-tailed deer neonates recruited into the fall firearm hunting season relative to minimum, mean, and maximum values of variables used to simulate change in white-tailed deer neonates recruited into the firearm hunting season due to presence of 4 40-ha mesopredator exclosures at the Joseph W. Jones Ecological Research Center in Baker County, Georgia, USA. Each minimum, mean, and maximum value is based on 1,000 simulations while allowing values of remaining variables to vary from their simulated distribution. Density (deer/km 2 ) was simulated from N(6.4, 1.7, 4, 8), sex ratio (F:M) from N(1.73, 0.17), relative use of exclosures from beta- PERT(2, 3, 13, 4) (see Roseboom et al. 1959), ovulation rate from N(1.49, 0.43, 0, 2), neonate survival inside exclosures from U(0.60, 1.0), and neonate survival outside exclosures from N(0.29, 0.29, 0, 0.6). Minimum, mean, and maximum values used in simulations to determine average minimum, mean, and maximum values were (4.0, 6.0, 7.75) for density, (1.15, 1.72, 2.22) for sex ratio, (2.03, 4.52, 10.8) for preference for exclosures, (0.21, 1.37, 2.0) for ovulation rate, (0.6, 0.8, 1.0) for neonate survival inside exclosures, and (0.0, 0.30, 0.60) for neonate survival outside exclosures. 168 The Journal of Wildlife Management 80(1) 19372817, 2016, 1, Downloaded from https://wildlife.onlinelibrary.wiley.com/doi/10.1002/jwmg.999 by University Of Florida, Wiley Online Library on [11/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/