Basic and Applied Ecology 43 (2020) 42–51 Vegetation characteristics influence fine-scale intensity of habitat use by wild turkey and white-tailed deer in a loblolly pine plantation Donald P. Chance a , Johannah R. McCollum a , Garrett M. Street a , Bronson K. Strickland a , Marcus A. Lashley a , b , ∗ a Department of Wildlife, Fisheries, and Aquaculture, Mississippi State University, Mississippi State, MS 39762, USA b Department of Wildlife Ecology and Conservation, University of Florida, Gainesville, FL 32611, USA Received 1 October 2018; accepted 24 October 2019 Available online 5 November 2019 Abstract Habitat quality is often evaluated based on food availability. However, ecological theory suggests cover should be a more important decision rule when food is not a proximate threat to fitness, as cover mediates predation risk as well as other important factors of fitness. In reality, vegetation characteristics related to food availability and cover are rarely coupled with animal use in the same space and time to determine their relative influences on habitat use. Using an array of 81 camera traps in a matrix of forest management strategies used to deliberately cause a wide disparity in vegetation characteristics, we monitored intensity of use by white-tailed deer ( Odocoileus virginianus ) and wild turkey ( Meleagris gallopavo ). We measured vegetation characteristics related to food and cover at each camera trap location then used a generalized additive model to determine how vegetation characteristics specific to the location affected intensity of habitat use by animals at the location. Consistent among both species, cover best explained intensity of habitat use. Contrastingly, food did not explain intensity of habitat use well for either species. Some vegetation simultaneously provides cover and food, and our data indicate that areas with vegetation characteristics providing both resources had the greatest influence on intensity of habitat use by both species. Our results suggest deer and turkey may perceive cover as a more important habitat component when food is not a proximate fitness threat. © 2019 Gesellschaft f ̈ ur ̈ Okologie. Published by Elsevier GmbH. All rights reserved. Keywords: Cover; Food; Habitat management; Habitat quality; Meleagris gallopavo ; Odocoileus virginianus Introduction Habitat quality is a foundational topic in wildlife ecology and management (Van Horne 1983). The relative importance of food availability and cover is often discussed in studies of habitat quality, leading many to focus on the impacts of ∗ Corresponding author. E-mail address: marcus.lashley@ufl.edu (M.A. Lashley). improving either food availability/quality or cover on use by animals (Morrison, Marcot, & Mannan 2006). Whether food or cover is more important for influencing habitat use is debat- able and if animal selection is a plastic trait, use is likely related to which factor is limiting individual fitness. Some studies report food availability as the primary habitat compo- nent affecting animal use (Clary & Larson 1971; Coulombe, Huot, MassØ, CôtØ, & De 2011; Martin, Palmer, Juhan, & Carroll 2012), while others report cover as more impor- https://doi.org/10.1016/j.baae.2019.10.007 1439-1791/© 2019 Gesellschaft f ̈ ur ̈ Okologie. Published by Elsevier GmbH. All rights reserved. D.P. Chance et al. / Basic and Applied Ecology 43 (2020) 42–51 43 tant (Hamerstrom, Mattson, & Hamerstrom 1957; Abramsky, Strauss, Subach, Riechman, & Kotler 1996; Creel, Winnie, Maxwell, Hamlin, & Creel 2005; Lashley Chitwood, Kays et al., 2015). In many cases, use by animals may depend on the landscape of fear created by the culmination of food and cover resource selection interacting in space and time with spa- tiotemporal variation in predation risks (LaundrØ, Hernández, & Altendorf 2001). Typical habitat quality studies are limited by time and labor leading to broad-scale data collection of habitat characteristics but relatively fine-scale data on habitat use from radiotagging individuals. This commonly leads to averaged vegetative characteristics for a treatment, assuming homogeneity throughout (Mysterud & Østbye 1999). This is especially problematic because heterogeneity within treat- ments may be one of the most important factors driving use of the treatment (Johnson 1980; Coulombe et al. 2011; Lashley Chitwood, Kays et al., 2015). A common way that habitat quality is measured is through the use of resource selection functions which often use pres- ence/availability data from an animal to infer preference (Boyce & McDonald 1999). While animal preference for a location or resource may not reflect its true quality in some cases (Johnson 2007), it remains a fundamental method of linking animal selection of resources to habitat quality and population level processes (Boyce et al. 2016). A com- mon use of resource selection function models is to sum the probabilities of use over space to get an estimate of animal abundance (see Boyce et al. 2016; Street, Rodgers, Avgar, Vander Vennen, & Fryxell 2017). Estimating abundance in this way is based on the premise that intensity of use by ani- mals is correlated with habitat quality or at least the animal’s perception of quality. Food and cover resources can be manipulated through forest management activities. Implementing canopy reduc- tions allows light to reach the forest floor thereby improving understory food abundance and structure (Hurst, Campo, & Brooks 1980; Greenberg, Perry, Harper, Levey, & McCord 2011; Lashley, Harper, Bates, & Keyser 2011). However, without periodic disturbances, woody encroachment leads to degraded structure and forage availability for many wildlife species (Crawford 1971; Harper 2007; Jackson et al. 2007; Harper, Ford, Lashley, Moorman, & Stambaugh 2016). Coupling canopy reductions with periodic prescribed fire controls woody encroachment, creating a two-tiered forest structure with an open overstory and an understory con- sisting of grasses, forbs, and woody plants (Pack, Igo, & Taylor 1988; Masters, Lochmiller, Engle, Bulletin, & Winter 1993; Lashley et al. 2011; McCord, Harper, & Greenberg 2014). Another common management technique is combin- ing canopy reductions and thinning with selective herbicide application, which removes the hardwood midstory and increases herbaceous vegetation dominance in the understory (Godfrey & Norman 1999; Edwards, Demarais, Watkins, & Strickland 2004). A wide range of vegetation conditions can be created using forest management techniques alone and in conjunction with one another, thereby imposing heterogene- ity in habitat quality at fine spatial scales both across and within a forest management unit. This is important because fine-scale spatial heterogeneity influences wildlife space use and habitat quality (Boyce et al. 2003). By coupling fine- scale vegetation data from multiple treatment combinations (canopy thinning, prescribed fire, and herbicide application) with concurrent data on use by animals, we can deter- mine how vegetative characteristics affect intensity of use by wildlife. Camera trapping is a popular tool for monitoring animal use and selection and has been used since the early 1900s to collect animal data in the field (Chapman 1927). Camera traps are less invasive than many other techniques, allowing data to be collected with minimal influence on use by animals (Cutler & Swann 1999). Technology has improved drastically in the last 30 years, making camera trapping a more effec- tive and economical tool for ecological studies (Rowcliffe & Carbone 2008; McCallum 2013; Lashley et al. 2018). Further, due to their increasingly low cost, researchers are now able to increase sampling intensity and apply more cameras across a landscape (Rowcliffe & Carbone 2008). This provides a relatively low-cost opportunity to evaluate animal intensity of use at finer scales than was previously allowed with cam- era traps, enabling examination of relatively fine-scale spatial heterogeneity in habitat quality and environmental conditions on resulting patterns of wildlife space use. Radio-tagging individuals can provide similar information at similar spatial scales but it is important to consider how data from those two tools differ. That is, radiotags can provide detailed infor- mation about space use of an individual over time whereas camera traps characterize the use of a particular space by all individuals over time. Regardless of the tool used, spatial scale is important when evaluating habitat quality, but many studies fail to match the scale of animal choices with the resolution of habitat variables (Boyce 2006). Measuring veg- etation characteristics with intensity of use simultaneously being monitored affords the opportunity to determine what vegetation characteristics are contributing to the choice of how frequently to use it by animals. Given that fine-scale spatial habitat heterogeneity can influence realized patterns of animal space use (Boyce et al. 2003), and that forest management can evoke changes in vegetative structure (Harper 2007; Lashley et al. 2011), the natural conclusion is that common forest management strate- gies could induce changes in the distribution of associated wildlife communities. To explore this possibility, we imple- mented multiple forest management practices in a southern pine ( Pinus spp.) ecosystem to evoke increased spatial het- erogeneity in vegetative conditions within and across forest management units. Our intent was not to compare the forest management strategies in how they affect vegetation struc- ture, as that has been widely studied elsewhere (e.g., Edwards et al. 2004; Lashley et al. 2011; McCord et al. 2014; Lashley, Chitwood, Harper, DePerno, & Moorman 2015; Harper et al. 2016; Greene et al. 2016; Lashley, Chitwood, DePerno, & Moorman 2017), but instead to utilize previous knowledge 44 D.P. Chance et al. / Basic and Applied Ecology 43 (2020) 42–51 Fig. 1. Study area and data collection design. We established three 30-m point intercept transects to measure understory plant composition (A), a visual profile board in four cardinal directions at 10 m from plot center to measure visual obstruction (B), a 1/25th-ha midstory tree sapling survey (C), a 25-m fruit abundance transect (A) and a camera trap to monitor animal use (D). GPS coordinates of the property: 33.412849, − 88.944231. that forest management strategies can be used to generate a wide disparity in vegetation conditions both within and across management units. We then measured vegetation char- acteristics and animal intensity of use at camera trap sites to determine how vegetation characteristics at a camera site influence intensity of use of that site by white-tailed deer ( Odocoileus virginianus; hereafter, deer) and wild turkey ( Meleagris gallopavo ; hereafter, turkey). We considered this scale to fit in well with the third order of selection described by Johnson (1980) as use of habitat components by animals within the home range. However, rather than measure habitat use of specific individuals as would be done with radiotag- ging, we monitored intensity of space use of the population. We used these species due to their prevalence on the site, their economic and recreational significance, and most importantly because of contradictory information regarding which habi- tat components are generally most important to these species. For example, traditional management for deer focused on maximizing forage (Chabreck & Mills 1980; Wood 1988); however, Lashley Chitwood, Kays et al. (2015) demonstrated that females with neonates apparently avoided areas of high- quality forage to instead select areas of cover. Similarly, turkey management traditionally focused on food produc- tion (Chabreck & Mills 1980); however, more recent research has demonstrated the greater apparent importance of cover (Miller, Leopold, & Hurst 1998; Hubbard, Garner, & Klaas, 2001). Based on recent observations of habitat use by these species, we predict that the intensity of use (an indicator of perceived habitat quality) of camera sites by both deer and turkey will be best explained by vegetation characteristics that can simultaneously provide cover and food. Materials and methods Study site Our study was conducted at Andrews Forestry and Wildlife Laboratory in Oktibbeha County, Mississippi, a 200-ha, 27- year-old loblolly pine ( Pinus taeda ) forest (Fig. 1). The study area was within the flatwoods topographic region with gently sloping terrain at an elevation ranging from 90 to 101 m. The dominant soil type present on the study area was Falkner silt loam with 0–5% slope and the average annual rainfall was ∼ 140 cm (U.S. Climate Data 2016). Predators of deer and turkeys on the study area include coyotes ( Canis latrans ) and human hunters but this study occurred in its entirety outside legal hunting seasons. White-tailed deer were the only large mammalian herbivore in the study area and wild turkeys were the only large generalist avian species. Combinations of canopy reduction, herbicide, and fire were implemented within forest management units to impose a wide range of vegetation compositional and structural dis- parity both within and across forest management units. In D.P. Chance et al. / Basic and Applied Ecology 43 (2020) 42–51 45 2014, 18 management units ranging from 5 to 22 ha in size were thinned to target basal areas (BA) of 9, 14, or 18 m 2 /ha producing a gradient in basal area ranging from 5–30 m 2 /ha at camera trap sites within those forest management units. In three management units, imazapyr and metsulfuron-methyl were applied to remove understory and midstory vegetation in October 2014. In May 2015, prescribed fire was implemented in six management units receiving thinning and all three man- agement units also receiving herbicide applications. Thus, combining the wide variation in basal area within and across management units with variation in fire and herbicide appli- cation generated a wide variation in vegetation characteristics at each camera trap location. Field methods The study area was systematically divided into a 200- m sampling grid with permanent sampling plots occurring at intersections of the grid lines ( n = 81). In June and then again in August of 2016, we measured understory composi- tion, visual obstruction, midstory density and composition, basal area, and soft mast production (Fig. 1). We measured each variable twice to minimize successional changes in vegetation between vegetation and animal intensity of use measurements throughout the study period. We grouped plant species into categories based on growth form to include grass, forb, brambles, woody which included trees and woody vines with each individual classification potentially contributing simultaneously to food and cover. Also, for deer we grouped plants that are generally considered high-forage quality into a forage category to represent the availability of high-quality forages (Table 1; Miller & Miller 2005; Lashley et al. 2011; Lashley, Chitwood, Street, Moorman, & Deperno 2016). Also we used empirical counts of fruits at each site to rep- resent fruit availability which is a primary food of wild turkey (Baughman & Guynn 1993). The forage and fruit classifications included plants from each growth form and were intended to be a measure primarily of food whereas plant height and visual obstruction were considered primarily factors of cover. We measured understory ( ≤ 1.5 m) compo- sition and height (cm) using 3 30-m point-intercept transects. Visual obstruction was measured in each cardinal direction using a vegetation profile board and averaging the proportion obstructed by vegetation of each of five sections up to 2.5 m in height as described by Nudds (1977). We measured midstory density and composition using a 400 m 2 fixed-radius plot, wherein we identified all midstory trees to species and mea- sured their diameter (DBH). Basal area was measured using a 10-basal area factor (BAF) prism. Soft mast was sampled monthly using the fruit count method described in Lashley, Thompson, Chitwood, and Deperno (2014). A camera trap was deployed at each sampling location to monitor animal intensity of use (Fig. 1) continuously from May 15, 2016, until September 15, 2016. Camera traps were set to record activity at any time of day or night and with a one-minute Table 1. Plants selected as forage for white-tailed deer base on Miller & Miller 2005 and Lashley et al. 2011, 2016. Common name Species Red maple a Acer rubrum Common ragweed Ambrosia artemisiifolia Toothcup Ammania spp. Alabama supplejack Berchemia scandens Spanish needles Bidens spp. Trumpet creeper Campsis radicans Partridge pea Chamaecrista fasciculata Tick-trefoil Desmodium spp. Swamp sunflower Helianthus angustifolius Wild lettuce Lactuca spp. Lespedeza Lespedeza spp. Chinese privet Ligustrum sinense Japanese honeysuckle Lonicera japonica Blackgum Nyssa sylvatica Winged sumac Rhus copallinum Blackberry Rubus spp. Greenbriar Smilax spp. Poison ivy Toxicodendron radicans Winged elm a Ulmus alata Summer grape Vitis aestivalis a Coppice growth. sampling delay between detections. We used 100 detections as a threshold for the total needed for the study to ensure robust sample size to model intensity of use. While no thresh- olds in sample size have been established specifically for robust intensity of use estimates, they recently have been for estimating other types of animal activity in two camera trap studies one of which was for the same studies species herein (Rowcliffe, Kays, Kranstauber, Carbone, & Jansen 2014; Lashley et al. 2018). Additionally, we make the assumption that intensity of use reflects the perception of habitat quality by the animals. While this may not be a direct measure of quality and could have some biases associated with factors that cause animals to choose poor quality habitat (Johnson 2007), using animal use data to define habitat quality remains a foundational practice in ecology (Boyce et al. 2016; Street et al. 2017). Data analysis To account for varying detection probabilities occurring at different sampling locations, we implemented an oppos- ing camera design for a subset of the study on a subset of the camera trap locations. We used a stratified random sample to select opposing camera locations ( n = 20). Five of the selected locations represented each of the 4 subdivi- sions in visual obstruction. The subdivisions were determined by natural breaks in our observed visual obstruction data (McCollum 2018). Opposing cameras were deployed approx- imately 15 m from the original camera trap so that they were observing approximately the same space for 30 trap nights. 46 D.P. Chance et al. / Basic and Applied Ecology 43 (2020) 42–51 Fig. 2. Mean beta coefficients for variables influencing deer habitat use. Values above baseline 0 indicate positive influence on use and values below indicate negative influence. Error bars represent 2.5% and 97.5% empirical quantiles. Midstory indicates the number of sapling stems detected in the 1/25th-ha survey. Percent grass, woody, brambles, and forbs indicates the relative proportion of the respective plant group in the understory based on point intercept transects. Percent forage indicates the relative proportion of the understory plant community considered high quality forages for white-tailed deer based on point intercept transects. Avg height refers to the average plant height in the understory based on point intercept transects. Basal area refers to the basal area of overstory trees at the sample site. Percent invasive indicates the relative proportion of the understory plant community that is a nonnative plant based on point intercept transects. Vis Obstruct refers to the average proportion of density profile board obstructed from the observer at plot center in four cardinal direction. We defined a detection (1) as an event when both cameras cap- tured the same individual; a non-detection (0) was assigned in instances that the original camera detected an animal and thus it was known to be present, and the paired opposing cam- era failed to detect that same animal. We used the detection data to calculate detection probabilities p for each camera location using logistic regression with animal size (turkey vs. deer), midstory density, % cover grass, % cover forbs, % cover woody plants, % cover brambles as predictor variables (Version 3.3.1, R Development Core Team, 2016). We then re-sampled 10,000 possible models using the vector of esti- mated coefficients and the variance-covariance matrix of the model to create distributions of model coefficients and cal- culated p from each of the 10,000 models for each sampling location. This resulted in 10,000 unique detection probabili- ties for each of the 81 sampling locations, which empirically accommodates error in the detection process. Counts of animals (i.e., intensity of use) at a sampling location will be biased low if the probability of detection is less than 100%. We thus calculated a corrected use met- ric c cor by dividing the number of detections c raw of a given species at the i -th sampling location by the j -th detection prob- ability estimated for that location (i.e., c cor,i,j = c raw,i /p i,j , i = { 1, 2, , 81 } , j = { 1, 2, , 10,000 } ; McCollum 2018). This resulted in 10,000 corrected use values for each loca- tion (i.e. 10,000 corrected use datasets) and each species to accommodate species-specific error in the detection model (i.e., detection probabilities for each species did not influ- ence the empirical estimates for the other; McCollum 2018). We fit a generalized additive model for location, scale, and shape (GAMLSS) with a zero-adjusted gamma distribution (ZAGA; Stasinopoulos & Rigby 2007) to each of the 10,000 sets of corrected use values to model the effect of vegetation characteristics on animal use given a particular realization of the detection model. By invoking the Central Limit Theo- rem, the peak of the distribution of count model coefficients becomes the best estimate of the effect of a vegetation covari- ate on animal intensity of use given the error introduced into the count model due to error in the detection model (i.e., a hierarchical model). We thus fit 10,000 corrected use GAMLSS models and averaged their beta coefficients and then calculated the 2.5% and 97.5% empirical quantiles D.P. Chance et al. / Basic and Applied Ecology 43 (2020) 42–51 47 Fig. 3. Mean beta coefficients for variables influencing turkey habitat use. Values above baseline 0 indicate positive influence on use and values below indicate negative influence. Error bars represent 2.5% and 97.5% empirical quantiles. Midstory indicates the number of sapling stems detected in the 1/25th –ha survey. Percent grass, woody, brambles, and forbs indicates the relative proportion of the respective plant group in the understory based on point intercept transects. Percent forage indicates the relative proportion of the understory plant community considered high quality forages for white-tailed deer based on point intercept transects. Avg height refers to the average plant height in the understory based on point intercept transects. Basal area refers to the basal area of overstory trees at the sample site. Percent invasive indicates the relative proportion of the understory plant community that is a nonnative plant based on point intercept transects. Vis Obstruct refers to the average proportion of density profile board obstructed from the observer at plot center in four cardinal direction. to assess the central tendency and variance of each model coefficient (midstory density; percent cover by grass, woody vegetation, brambles, forbs, and invasive species; average vegetation height; basal area; visual obstruction as indicated by a Nudds board; for deer, percent cover by forage; for turkeys, fruit/soft mast abundance; Table S1). Results Proportional cover of plant functional groups used in our model was highly variable (Table S2) indicating the intent of using forest management to generate a wide disparity in con- ditions for the purposes of our analyses had been successful. During the 123-day sampling period, we recorded 189 pho- tographs of turkeys (0.019 observations/trap night) and 2915 photographs of deer (0.293 observations/trap night). Presence of coyote predators was relatively low with 112 detections total (McCollum 2018). Although 9% of the possible combi- nations of variables had a correlation of 50% or higher (Table S1), there was relatively low multicollinearity overall. Deer and turkey use corrected for detection error was positively associated with percent cover of grasses, woody species, and brambles (Table S3, Figs. 2, 3 ), all of which constitute cover. Both species experienced a slight positive influence of basal area and midstory density likely related to ther- moregulation benefits. Deer use was negatively associated with percent cover of preferred food plants and high levels of visual obstruction but had no relationship with percent cover of forbs and average understory vegetation height (Fig. 2). Turkey use was negatively associated with fruit abundance but was not affected by visual obstruction and percent cover of forbs (Fig. 3). Use by both species was negatively asso- ciated with percent cover of non-native species (Fig. 2). On average, across all vegetation variables considered and rela- tive influences on intensity of use by both species, the canopy reduction with fire treatment tended to maximize vegetation characteristics positively influencing use while minimizing characteristics negatively influencing use (Table S3). 48 D.P. Chance et al. / Basic and Applied Ecology 43 (2020) 42–51 Discussion Our results indicate both deer and turkey intensity of habi- tat use was most positively influenced by cover rather than food during the study period on this site, which may be a reflection of the importance of predator avoidance to prey decision rules (Goertz 1964; Cimino & Lovari 2003; Lashley, Chitwood, Harper et al., 2015; Lashley Chitwood, Kays et al., 2015). In fact, McCollum (2018) reported a high correlation between habitat use of coyotes with deer and turkeys. As a result, predation risk may represent a greater proximate fit- ness cost than suboptimal acquisition of resources (Brown & Kotler 2004). Coulombe et al. (2011) demonstrated this point well with white-tailed deer when they eliminated predation risk and observed that deer use transitioned to be associ- ated with forage availability. Similar decision rules related to predation risk and foraging have been demonstrated in red deer ( Cervus elaphus ) which selected areas with better cover (Jayakody, Sibbald, Gordon, & Lambin 2008) when preda- tion risk is perceived to be higher. In this case, inncreased feeding rates may counteract the more limited food resources in safe patches (Illius & Fitzgibbon 1994; Fortin et al. 2004). Also, in this study many of the vegetation characteristics we measured may simultaneously provide food and cover, and those variables tended to be most favored, which may reflect the animals choosing to simultaneously acquire food and cover or that food is not limited in areas with adequate cover. Decoupling food and cover can be difficult, particularly with generalist herbivores that can utilize plants from differ- ent growth forms for food and cover simultaneously. In this case, grasses, brambles, shrubs and young trees could provide adequate cover and food resources simultaneously (Miller & Miller 2005). The negative association we observed with deer use and forage availability could be an artifact of deer using cover for forage. However, the plants we chose to represent forage availability are commonly regarded as high-quality forages often selected by deer (Table 1; Miller & Miller 2005; Lashley et al. 2011, 2016), indicating even if deer can con- sume their cover, they were choosing to do so at the expense of exploiting the highest availability of high-quality forages. Again, this may be because the availability of high-quality forages was not limited in areas with suitable cover. Simi- larly, turkeys selected areas high in grasses, brambles, and woody species, which provide high-quality brood-rearing cover (Godfrey & Norman 1999; McCord et al. 2014) and potentially foraging values provided by seeds, soft mast, and invertebrates (Shelton & Edwards 1983; Baughman & Guynn 1993). However, they still avoided areas with the highest soft mast abundance, which may reflect the preference for cover over food. On our study area during the time frame of observa- tion, deer and turkey may have been able to effectively forage within cover due to cover resources simultaneously provid- ing some food and thermoregulation benefits (Ockenfels & Brooks 1994; Millspaugh, Raedeke, Brundige, & Willmott 1998). However, there was a tradeoff given cover and food were not perfectly correlated and both species appeared to trade the opportunity to exploit maximum food resources for areas with high-quality cover. Considering life history when evaluating importance of cover to use by animals is important. Our sampling period encompassed the nesting and brood-rearing seasons for turkey and fawning season for deer (Miller et al. 1998; Campbell et al. 2016). Selection of cover is often strongest during reproduction because of the relative vulnerability of young animals to predation risk (Kunkel & Mech 1994; Miller et al. 1998; Paisley, Wright, Kubisiak, & Rolley 1998; Chitwood, Lashley, Kilgo, Moorman, & Deperno 2015; Lashley, Chitwood, Harper et al., 2015; Lashley Chitwood, Kays et al., 2015). Similarly, thermoregulation costs are high during the summer, so use of cover can limit stress related to thermoregulation (Millspaugh et al. 1998). The strong use of areas with taller understory vegetation (up to 1.5 m height) and areas with a more developed understory woody component, may reflect the utility of those vegeta- tion characteristics as high quality cover to turkeys during nesting and brood-rearing (Hubbard et al. 2001; Streich, Little, Chamberlain, Conner, & Warren 2015). Because poult and neonate predation are important to population dynam- ics (Vangilder & Kurzejeski 1995; Roberts & Porter 1996; Rolley and Kubisiak 1998; Chitwood et al. 2015), the evolved response of increased use of cover to maximize recruitment is to be expected. In fact, these behaviors can be so evo- lutionarily ingrained that animals may use cover perceived as high-quality, even if it predisposes animals to higher rates of predation (i.e., evolutionary trap, Chitwood, Lashley, Moorman, & DePerno 2017). Although we conclude in this study area over the sam- pling period that cover was most likely driving intensity of use trends of both deer and turkey, we acknowledge that food availability is an important habitat component. Our study area may have had relatively high quality cover and food resources compared to the surrounding landscape of loblolly pine ( Pinus taeda ) plantation because the lack of canopy reduction and fire, the treatments that tended to pro- mote the greatest values in vegetation characteristics related to both habitat components in our study. If both resources are in relatively high abundance, animals may choose cover in the landscape context of this study. However, in different landscape configurations or under differing conditions, it is entirely plausible that food could become limiting. For exam- ple, during times of extreme drought the proportion of plant species that provide high-quality forage may be restricted (Lashley & Harper, 2012). Similarly, vegetation communi- ties dominated by plant species that provide only high-quality cover may be limited in food resources. In the case that food becomes limiting, we expect the relative selection of food and cover to reflect the change in limitation. This high- lights the need for spatial and temporal replication, which we acknowledge is a limitation of this study. Ultimately, spatial and temporal replication across differing landscape D.P. Chance et al. / Basic and Applied Ecology 43 (2020) 42–51 49 configurations and environmental conditions are needed to understand how those factors influence the potential tradeoff between habitat components. Understanding the underlying mechanisms for how and why resource limitations change can be important for predicting what efforts are needed to conserve targeted populations. Conclusions Deer and turkey simultaneously exploited cover and food resources. Combining canopy reductions and prescribed fire created the most favorable conditions for both species by maximizing abundance of grasses, woody plants, and bram- bles while minimizing vegetation characteristics that they avoided. We suggest that managers should attempt to create a well-developed understory consisting of grasses, woody plants, and brambles to maximize food availability within cover. Declaration of interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. The authors declare the following financial inter- ests/personal relationships which may be considered as potential competing interests: Acknowledgements We would like to thank the Mississippi Agricultural and Forestry Experiment Station (MAFES) and the Mis- sissippi State University Forestry and Wildlife Research Center for funding this research. We would also like to thank volunteers and technicians that assisted in data col- lection. Specifically, we thank John DelPapa, Ashley Jones, Marlee Fuller-Morris, John McCollum, Rainer Nichols, Kelsey Paolini, Bonner, Powell, Trevon Strange, and Seth Wong. Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/ j.baae.2019.10.007. References Abramsky, Z., Strauss, E., Subach, A., Riechman, A., & Kotler, B. P. (1996). The effect of barn owls ( Tyto alba ) on the activity and microhabitat selection of Gerbillus allenbyi and G. pyramidum Oecologia , 105 , 313–319. Baughman, W. M., & Guynn, D. C. (1993). Wild turkey food habits in pine plantations in South Carolina. Proceedings of the Southeastern Association of Fish and Wildlife Agencies , 47 , 163–169. Boyce, M. S., & McDonald, L. L. (1999). Relating populations to habitats using resource selection functions. Trends in Ecology & Evolution , 14 , 268–272. Boyce, M. S. (2006). Scale for resource selection functions. Diver- sity and Distributions , 12 (3), 269–276. Boyce, M. S., Mao, J. S., Merrill, E. H., Fortin, D., Turner, M. G., Fryxell, J., et al. (2003). Scale and heterogeneity in habitat selection by elk in Yellowstone National Park. Ecoscience , 10 , 421–431. Boyce, M. S., Johnson, C. J., Merrill, E. H., Nielsen, S. E., Solberg, E. J., & Van Moorter, B. (2016). Can habitat selection predict abundance? The Journal of Animal Ecology , 85 , 11–20. Brown, J. S., & Kotler, B. P. (2004). Hazardous duty pay and the foraging cost of predation. Ecology Letters , 7 , 999–1014. Campbell, K. L., Strickland, B. K., Demarais, S., Jones, P. D., Wang, G., Dacus, C. M., et al. (2016). Adjusting for seasonal harvest bias in the lactation index for white-tailed deer management. Wildlife Society Bulletin , 40 , 754–757. Chabreck, R. H., & Mills, R. H. (Eds.). (1980). Integrating timber and wildlife management in Southern forests. In 29th Annual forestry symposium Baton Rouge, LA: Louisiana State Univer- sity. Chapman, F. M. (1927). Who treads our trails? National Geographic Magazine , 52 (3), 330–345. Chitwood, M. C., Lashley, M. A., Kilgo, J. C., Moorman, C. E., & Deperno, C. S. (2015). White-tailed deer population dynamics and adult female survival in the presence of a novel predator. The Journal of Wildlife Management , 79 , 211–219. Chitwood, M. C., Lashley, M. A., Moorman, C. E., & DePerno, C. S. (2017). Setting an evolutionary trap: could the hider strategy be maladaptive for white-tailed deer? Journal of Ethology , 35 , 251–257. Cimino, L., & Lovari, S. (2003). The effects of food or cover removal on spacing patterns and habitat use in roe deer ( Capreolus capre- olus ). Journal of Zoology , 261 , 299–305. Clary, W. P., & Larson, F. R. (1971). Elk and deer use are related to food sources in Arizona ponderosa pine, USDA Forest Service Research Note. Rocky Mountain Forest and range experiment station Coulombe, M., Huot, J., MassØ, A., CôtØ, S. D., & De, C. (2011). Influence of forage biomass and cover on deer space use at a fine scale: A controlled-density experiment. Ecoscience , 18 , 262–272. Crawford, H. S. (1971). Wildlife habitat changes after intermediate cutting for even-aged oak management. The Journal of Wildlife Management , 35 , 275–286. Creel, S., Winnie, J. J., Maxwell, B., Hamlin, K., & Creel, M. (2005). Elk alter habitat selection as an antipredator response to wolves. Ecology , 86 , 3387–3397. Cutler, T. L., & Swann, D. E. (1999). Using remote photogra- phy in wildlife ecology: A review. Wildlife Society Bulletin , 27 , 571–581. Edwards, S. L., Demarais, S., Watkins, B., & Strickland, B. K. (2004). White-tailed deer forage production in managed and unmanaged pine stands and summer food plots in Mississippi. Wildlife Society Bulletin , 32 , 739–745. 50 D.P. Chance et al. / Basic and Applied Ecology 43 (2020) 42–51 Fortin, D., Fryxell, J. M., Fryxell Fortin, J. M., Fortin, D., et al. (2004). Foraging costs of vigilance in large mammalian herbi- vores. Oikos , 107 , 172–180. Godfrey, C. L., & Norman, G. W. (1999). Effect of habitat and move- ment on wild turkey poult survival. Proceedings of the Annual Conference Southeastern Association of Fish and Wildlife Agen- cies , 53 , 330–339. Goertz, J. W. (1964). The influence of habitat quality upon density of cotton rat populations. Ecological Monographs , 34 , 359–381. Greenberg, C. H., Perry, R. W., Harper, C. A., Levey, D. J., & McCord, J. M. (2011). The role of young, recently disturbed upland hardwood forest as high quality food patches. In C. H. Greenberg, B. S. Collins, & F. Thompson III (Eds.), Sus- taining young forest communities: Ecology and management of early successional habitats in the central hardwood region, USA Springer Science & Business Media. Greene, R. E., Iglay, R. B., Evans, K. O., Miller, D. A., Wigley, T. B., & Riffell, S. K. (2016). A meta-analysis of biodiversity responses to management of southeastern pine forests—Opportunities for open pine conservation. Forest Ecology and Management , 360 , 30–39. Hamerstrom, F., Mattson, O., & Hamerstrom, F. (1957). A guide to prairie chicken management, Technical Wildlife Bulletin Wis- consin: Madison. Harper, C. A. (2007). Strategies for managing early succession habi- tat for wildlife. Weed Technology: A Journal of the Weed Science Society of America , 21 , 932–937. Harper, C. A., Ford, W. M., Lashley, M. A., Moorman, C. E., & Stambaugh, M. C. (2016). Fire effects on wildlife in the Central Hardwoods and Appalachian regions, USA. Fire Ecology , 12 , 127–159. Hubbard, M. W., Garner, D. L., & Klaas, E. E. (2001). Factors influencing wild turkey poult survival in southcentral Iowa. Pro- ceedings of the National Wild Turkey Symposium , 8 , 167–171. Hurst, G. A., Campo, J. J., & Brooks, M. B. (1980). Deer forage in a burned and burned-thinned pine plantation. Proceedings of the Southeastern Association of Fish and Wildlife Agencies , 34 , 474–481. Illius, A., & Fitzgibbon, C. (1994). Costs of vigilance in foraging ungulates. Animal Behaviour , 47 , 481–484. Jackson, S. W., Basinger, R. G., Gordon, D. S., Harper, C. A., Buckley, D. S., & Buehler, D. A. (2007). Influence of silvicul- tural treatments on eastern wild turkey habitat characteristics in eastern Tennessee. Proceedings of the National Wild Turkey Symposium