Hunter movement and habitat use affect observation rate of white-tailed deer (Odocoileus virginianus)

Introduction

Hunting by humans (hereafter referred to as hunting) is an important tool for wildlife population management (Kilpatrick and Walter 1999; Harden et al. 2005; Lebel et al. 2012). Humans can be a dominant and influential presence on the landscape (Laliberte and Ripple 2004; Montgomery et al. 2022) and are a major source of risk and population control for game species (Frid and Dill 2002; Marantz et al. 2016). Hunting affects the behavior of white-tailed deer (Odocoileus virginianus), leading to discussion among management agencies of how to best utilize hunters as a management tool (Hygnstrom et al. 2011; Lebel et al. 2012; Little et al. 2014, 2016; Marantz et al. 2016; Schuttler et al. 2017).

Current white-tailed deer research examines behavior in response to hunters as a source of predation (Hewitt 2011; Little et al. 2014, 2016; Marantz et al. 2016) with less examination of hunter behavior. Harvest success is generally measured by the number or relative percentage of hunters that harvest an animal, or by accounting for hunter effort (i.e. amount of time) in measures of success (e.g. number harvested per unit time; Grau and Grau 1980; Gratson and Whitman 2000; Kilpatrick et al. 2002; Iijima 2017; Rowland et al. 2023). Other studies have used observation rate as a proxy for harvest, given the assumption that the more animals a hunter observes, the more likely they are to harvest an animal (Jacques et al. 2011; Lebel et al. 2012; Little et al. 2014, 2016), but there also is a caveat that the likelihood of harvest success increases with hunter effort (Murphy 1965; Rowland et al. 2023). Hunter effort and observation rate are usually dependent on road access, landscape features, visibility, prey densities, and experience (Jacques et al. 2011; Lebel et al. 2012; Norum et al. 2015; Ranglack et al. 2017; O’Connor et al. 2018, Rowland et al. 2021).

Hunters often employ different hunting strategies to harvest prey, ranging from active methods (such as stalk hunting) to inactive methods (such as tree-stands or ground-blinds), which vary on the basis of region, prey species, and individual preference (Montgomery et al. 2022). However, there is limited research examining how hunting strategy affects harvest success. Whereas prey alter their behavior in response to hunting, it should follow that hunters adjust their behavior in response to prey, as has been observed in northern pike (Esox lucius) and three-spined stickleback (Gasterosteus aculeatus) predator–prey systems as well as wolf (Canis lupus) and white-tailed deer systems (McGhee et al. 2013; Johnson-Bice et al. 2023). Understanding what behaviors affect harvest success may increase hunter efficiency and effectiveness. Considering that hunting is both a major source of revenue as well as the primary tool for population control for many state agencies (Doerr et al. 2001; Harden et al. 2005), managers can utilize information of successful hunter behaviors to manage populations across the landscape by manipulating where and how hunting pressure influences prey (Cromsigt et al. 2013). Controlled hunts have been successful for managing overabundant urban and suburban deer populations (Green et al. 1997; Kilpatrick et al. 1997, 2002; Kilpatrick and Walter 1999). Urban and suburban environments limit hunters in where and when to hunt; thus, defining where hunter behavior overlaps with deer behavior creates a more targeted approach to hunting by identifying specific hunter behaviors and hunter habitat use that lead to greater harvest success. As white-tailed deer populations continue to grow across the United States (Urbanek et al. 2011), the ability to develop targeted management plans to control populations swiftly and effectively is increasingly vital (Green et al. 1997; Kilpatrick et al. 2007; Urbanek et al. 2011).

In this study, we examined human hunter movement and habitat use to determine which behaviors have the most significant effect on observation rate of white-tailed deer. Specifically, we examined hunter movement speed, path shape, and the proportion of a hunter’s time spent in different vegetation types, and the average distance from landscape features such as roads, water sources, etc. for each recorded hunt. Additionally, we examined hunter habitat use across the landscape at the population level. By identifying patterns of behavior that increase deer observation rates, we can determine opportunities to increase or maintain harvest success for management agencies to reach management goals and for hunters to improve their success.

Materials and methods

Study area

The Noble Research Institute’s Oswalt Road Ranch (ORR) is in Love County, Oklahoma, USA (Fig. 1). The ORR property is an 1861-ha ranch within the Cross Timbers and Prairie eco-region, consisting of mixed wooded areas, bottomlands, uplands and rangeland (Gee et al. 2011). To mitigate the carryover effect of previous hunting on the ORR, lease hunting (x̅ = 5 hunters) ceased after the 2006 hunting season. No hunting occurred on the ORR until our study began in 2008. Additionally, the ORR management team did not conduct any cattle grazing or prescribed fire management on the ranch during our study. Refer to Little et al. (2014) for additional study area details.

Fig. 1.

The Noble Research Institute’s Oswalt Road Ranch located in Love County, Oklahoma, USA, with treatment compartments delineated. Left: 2008 treatment compartments where control (C) = no hunters on 679 ha; low-risk (L) = 1 hunter/101 ha on 585 ha total; and high-risk (H) = 1 hunter/30 ha on 583 ha total. Right: compartments were shifted clockwise for the 2009 hunting season where control (C) = no hunters on 586 ha; low-risk (L) = 1 hunter/101 ha on 583 ha total; and high-risk (H) = 1 hunter/30 ha on 679 ha total. Map produced in ArcGIS 10.6.1 (ArcGIS® software by Esri).

Analytic approach

Results

During the 2008 and 2009 hunting seasons, 83 individual hunters recorded deer observation and location data for 516 unique hunts, collecting >140,000 GPS locations. Between the two hunting seasons, 29 animals were harvested by hunters (2008: 1 male, 10 females; 2009: 4 males, 14 females). Of the 516 records collected for this study, 29 were removed from analysis with associated errors (e.g. GPS recording error, missing observation data, hunter behavior did not follow instruction parameters, e.g. spending <1 h afield, leaving the assigned compartment, or using horses or ATVs) for a total of 487 recorded hunts used for all analyses, resulting in 111,706 GPS locations used for analyses.

Across all hunting methods, hunters spent an average of 3.7 h (s.e. = 0.1 h) afield during each recorded hunt, traveled 2085 m (s.e. = 79.0 m), and observed 2.7 deer/hunt (s.e. = 0.15). Hunters primarily used still and stand hunting methods, with many hunters using a combination of methods. During each recorded hunt, hunters spent an average 3.7 h afield (s.e. = 0.1; 879 total h) for still hunting, 3.8 h (s.e. = 0.1; 1256 total h) for stand hunting, and 4.7 h (s.e. = 0.7; 52 total h) when using multiple methods. Overall, hunters observed 1336 deer (male: 303; female: 696; fawns: 194; and unknown: 143) with an average rate of observation for males (0.3/h, s.e. = 0.02), females (0.6/h, s.e. = 0.04), and fawns (0.2/h, s.e. = 0.02). Still hunters observed 0.9 deer/h (s.e. = 0.1), stand hunters observed 0.7 deer/h (s.e. = 0.05), and hunters using multiple methods observed 0.4 deer/h (s.e. = 0.2). Between each GPS fix (1 fix per min), the mean step length per recorded hunt averaged at 9.4 m (s.e. = 0.3) and the mean absolute turning angle of hunters per recorded hunt averaged 0.3 radians (s.e. = 0.01). Mean residence time averaged 46%, indicating that hunters spent 46% of their time within 10 m (±3.6 s.e.) of the previous GPS location. The percentage of use for different vegetation cover classes per hunt time averaged 16% forested cover (s.e. = 0.01), 22% mixed cover (s.e. = 0.01), and 38% grassland cover (s.e. = 0.02), and 24% riparian cover; topographic roughness averaged 5.8 (s.e. = 0.1; Table 1).

We confirmed the relationship between observation rate and harvest success. The probability of harvest increased as the total number of deer observed increased (P < 0.01, s.e. = 0.04, z = 4.398) and as observation rate increased (P < 0.01, s.e. = 0.14, z = 3.916) (Fig. 2).

Fig. 2.

Plots of binomial logistic regression models of white-tailed deer harvest by hunters. (a) Total deer observed during a hunt (number of deer); and (b) observation rate (number of deer/h).

Deer observation rates

From the GLMs, using a Gaussian distribution, the best covariates for each movement suite and functional form included linear mean step length, mean turning angle as a quadratic function, and mean residence time as a quadratic function (observed deer/h ~ mean step length + mean turning angle + mean turning angle² + mean residence time + mean residence time²; Table 2). There were two models within 2 ΔAIC_c units of the top model for hunter movement; therefore, we selected the most parsimonious model. This model involved linear mean step length and mean turning angle as a quadratic function (observed deer/h ~ mean step length + mean turning angle + mean turning angle²).

Table 2.Standardized coefficient estimates, standard errors, P-values, and t-values for the GLM using a Gaussian distribution of the top movement model from 2008 to 2009 white-tailed deer hunts at Oswalt Road Ranch (ORR), Love County, Oklahoma.

Covariate	Estimate	s.e.	t-value	P-value
Intercept	0.58	0.07	8.24	<0.01
Mean step length	0.35	0.14	2.55	0.01
Mean turning angle	0.06	0.10	0.62	0.54
(Mean turning angle)2	0.31	0.12	2.66	<0.01
Mean residence time	0.01	0.14	0.06	0.95
(Mean residence time)2	0.52	0.20	2.59	<0.01

The best covariates for each habitat use suite and functional form included linear proportion for use of grasslands, forests, mixed, and mean roughness (observed deer/h ~ proportion of hunt spent in grasslands + proportion of hunt spent in forests + proportion of hunt spent in mixed habitat + mean roughness; Table 3). Distance covariates did not show a significant effect on observation rate. There were five models within 2 ΔAIC_c units of the top model for hunter habitat use (Table 4); therefore, the most parsimonious model was selected, containing only the proportion of forested cover used (observed deer/h ~ proportion of hunt spent in forest).

Table 3.Standardized coefficient estimates, standard errors, P-values and t-values for the GLM using a Gaussian distribution of the habitat use model from 2008 to 2009 white-tailed deer hunts at Oswalt Road Ranch (ORR), Love County, Oklahoma.

Covariate	Estimate	s.e.	t-value	P-value
Intercept	0.79	0.04	17.91	<0.01
Proportion of grassland use	−0.04	0.1	−0.42	0.67
Proportion of forest use	0.4	0.1	3.79	<0.01
Proportion of mixed use	0.06	0.1	0.59	0.56
Mean roughness	0.11	0.09	1.17	0.24

Table 4.Top five performing Gaussian GLMs for hunter habitat use and observation of deer per hour from 2008 to 2009 white-tailed deer hunts at Oswalt Road Ranch (ORR), Love County, Oklahoma.

Model	d.f.	logLik	ΔAICc	Weight
Deer/h ~ forest	3	−674.82	0.00	0.26
Deer/h ~ forest + roughness	4	−674.08	0.55	0.19
Deer/h ~ forest + mixed	4	−674.43	1.25	0.14
Deer/h ~ forest + grassland	4	−674.53	1.45	0.12
Deer/h ~ forest + mixed + roughness	5	−673.72	1.89	0.10

The top combined model for hunter movement and habitat use included all of the best competing covariates: linear mean step length, mean turning angle as a quadratic function, and linear proportion of hunt spent in forested cover (observed deer/h ~ mean step length + mean turning angle + mean turning angle2 + proportion of hunt spent in forested cover; Table 5). As average step length increased, the model predicted the number of deer observed per hour increased (standardized coefficient estimate ± s.e.; 0.3 ± 0.09; Fig. 3). The predicted number of deer observed was the lowest when the mean absolute turning angle was approximately 0.3 radians, with predicted observations increasing as the mean absolute turning angle increased, or approached 0 (0.06 ± 0.1; Fig. 3). As the mean proportion of hunt time spent within forested cover increased, the number of deer observed per hour increased (0.42 ± 0.09; Fig. 3).

Table 5.Standardized coefficient estimates, standard errors, P-values, and t-values for the GLM using a Gaussian distribution of the top full movement and habitat-use model from 2008 to 2009 white-tailed deer hunts at Oswalt Road Ranch (ORR), Love County, Oklahoma.

Covariate	Estimate	s.e.	P-value	t-value
Intercept	0.70	0.05	<0.01	13.54
Mean step length	0.30	0.09	<0.01	3.16
Mean turning angle	0.06	0.1	0.51	0.65
(Mean turning angle)2	0.34	0.1	<0.01	3.01
Proportion of forest use	0.42	0.09	<0.01	4.83

Fig. 3.

Plots of three covariates included in the top model of hunter movement, habitat use and observation rate of white-tailed deer. (a) Mean turning angle (radians); (b) mean step length (m); and (c) proportion of use of forested habitat.

Hunter use of landscape

We found that the best competing functional forms for each covariate included in our hunter resource selection models included the following: distance to ponds and roads as pseudothresholds, linear distance to anthropogenic features, forest and riparian cover as quadratic functions, mixed cover as a pseudothreshold, and roughness as a pseudothreshold (Used ~ Log(distance to ponds ± 0.001) + Log(distance to roads + 0.001) + Distance to anthropogenic features + Forest cover + Forest cover² + Riparian cover + Riparian cover² + Log(mixed cover +0.001) + Log (roughness + 0.001); Table 6). The most supported model included each of the covariates (model weight = 1), with the next highest model within 41.21 ΔAIC_c units. Predicted hunter use of forested cover was greatest at 40% forested habitat within the shooting window (Fig. 4). Predicted hunter use of mixed cover quickly increased before reaching an asymptote at 5% mixed cover within the shooting window (Fig. 4). Predicted hunter use of riparian cover was lowest at 50% riparian cover within the shooting window, with use increasing as availability either increased or approached 0% (Fig. 4). Predicted hunter use decreased as distance to ponds increased, quickly reaching an asymptote in predicted use after approximately 200 m of distance (Fig. 4). Predicted hunter use increased as distance to roads increased, quickly reaching its threshold at approximately 50 m of distance (Fig. 4). Predicted hunter use increased as roughness increase, and distance to anthropogenic features increased (Fig. 4). Hunters showed greater use of forested cover in proportion to availability than other vegetation cover classes (P < 0.01, s.e. = 0.10, t = 3.227; Fig. 5).

Table 6.Standardized coefficient estimates, standard errors, P-values, and z-values for the logistic regression of the top resource selection model from 2008 to 2009 white-tailed deer hunts at Oswalt Road Ranch (ORR), Love County, Oklahoma.

Covariate	Estimate	s.e.	P-value	z-value
Intercept	−0.78	0.01	<0.01	−119.96
Distance to anthropogenic features	−0.12	0.01	<0.01	−8.72
Log(distance to ponds)	−0.22	0.01	<0.01	−32.99
Log(distance to roads)	0.26	0.01	<0.01	26.13
Forested habitat	0.31	0.01	<0.01	26.99
(Forested habitat)2	−0.31	0.01	<0.01	−21.96
Log(mixed habitat)	0.43	0.01	<0.01	43.58
Riparian habitat	−0.16	0.01	<0.01	−16.61
(Riparian habitat)2	0.11	0.02	<0.01	5.97
Log(roughness)	0.34	0.01	<0.01	34.50

Fig. 4.

Plots of seven covariates included in the top model of population level resource selection of white-tailed deer hunters. (a) Percentage forested cover of shooting window; (b) percentage mixed cover of shooting window; (c) percentage riparian cover of shooting window; (d) available roughness (s.d. of elevation); (e) available distance to anthropogenic features (m); (f) available distance to ponds (m); and (g) available distance to roads (m).

Fig. 5.

Functional response of hunter use versus compartment availability. (a) Proportion of hunter use of forested habitat over proportion of forested habitat available within a compartment; (b) proportion of hunter use of mixed habitat over proportion of mixed habitat available within a compartment; (c) proportion of hunter use of grassland habitat over proportion of grassland habitat available within a compartment; (d) proportion of hunter use of riparian habitat over proportion of riparian habitat available within a compartment; (e) mean hunter distance from anthropogenic features (m) over mean compartment distance from anthropogenic features (m); (f) mean hunter distance from ponds (m) over mean compartment distance from ponds (m); (g) mean hunter distance from roads (m) over mean compartment distance from roads (m); and (h) mean hunter roughness (s.d. of elevation) over mean compartment roughness (s.d. of elevation).

Discussion

Most research has focused on the effect of hunting pressure on prey behavior, whereas little research has analyzed hunter behavior and its potential effects on harvest success. Hunter behavior, in terms of how movement and habitat use influence observation of deer, has not been previously measured on such a fine temporal and spatial scale. Our results suggest that hunter movement and habitat use influenced the number of deer observed during a hunt. Hunters that used forested cover in greater proportions were more likely to observe deer than were hunters that used forested cover in lesser proportions. Additionally, hunters selected mixed cover more consistently than forested cover, relative to availability. Hunters with greater movement rates were more likely to observe deer than were hunters with lower rates of movement. Walking at approximately 25 m/min (1 mph), the model predicts that hunters will observe one deer per hour. During movement, hunters that either did not turn or turned more sharply were more likely to observe deer than were hunters that turned more moderately, suggesting that decisive, directional movements may increase likelihood of observation. However, it is important to note that not every observation results in a harvest opportunity (e.g. observing a fleeing animal). Although the probability of harvest generally increases with higher observation rates, this is highly dependent on the specific scenario (Fig. 2). For instance, a hunter may observe more deer, but they could simply be fleeing, which does not guarantee an increased harvest rate.

Previous research found that hunter effort and observation rate were dependent on road access, landscape features, visibility, prey densities, and experience (Jacques et al. 2011; Lebel et al. 2012; Norum et al. 2015; Ranglack et al. 2017; O’Connor et al. 2018), and prey altered their vigilance behaviors, movement patterns, and habitat use (Kufeld et al. 1988; Kilgo et al. 1998; Conner et al. 2001; Stankowich 2008; Bonnot et al. 2013; Little et al. 2016; Marantz et al. 2016). In response to hunting pressure, some ungulates increased selection for areas of greater cover (Kufeld et al. 1988; Kilgo et al. 1998; Bonnot et al. 2013; Lone et al. 2015; Ranglack et al. 2017), whereas some selected for more open habitats with greater visibility within flight distance to cover (Swenson 1982; Stankowich 2008). Prey distribution changed over the hunting season as prey increased awareness of predation risk and altered movement and behavior (Little et al. 2014); for example, prey species become more vigilant and are more likely to flee when there is a perceived risk (Frid and Dill 2002; Stankowich 2008; Schuttler et al. 2017). During the study period, deer avoided hunting pressure by increasing selection for forested cover (Little 2011), decreasing diurnal movement (Little et al. 2016), and increasing site fidelity (Marantz et al. 2016). Deer increased use of forested cover by 1.7–2.5 times and mixed cover by 1.4–2.3 times in response to predation pressure (Little 2011). Collared deer that were unobserved by hunters moved 38.3% less than did observed collared deer, suggesting that greater movement increased observability (Little 2011).

We observed a positive relationship between the number of deer observed and harvest success, which supports our use of observation rate as a proxy to harvest success. Although, it is important to consider that increased observations do not necessarily equate to increased harvest in all cases, because hunters may simply be observing a moving deer that cannot be ethically and legally harvested, or they choose not to harvest the specific animal. For the 2008 and 2009 hunting seasons, ORR had a harvest quota of 20 antlerless deer each year, with three mature, un-collared antlered deer in 2008, and four mature, un-collared antlered deer in 2009. This showed limited harvest on the property, while observations were unlimited. Only 29 harvests occurred during the study period, whereas 1333 deer were observed, leading to the relationship being highly variable. Further research, with higher rates of harvest success, is necessary to more confidently determine the strength of the relationship between observation of prey and likelihood of harvest.

Given that deer increase their use of concealment cover at the onset of the hunting season (Swenson 1982; Kufeld et al. 1988; Kilgo et al. 1998; Paton et al. 2017), visual obstruction can be a determining factor in harvest success (Lone et al. 2015). However, our results showed that as deer increase the use of forested cover and decrease movement, hunters that increase the use of forested cover and increase movement are more likely to observe a deer by overlapping habitat use and potentially triggering a flight response (Grau and Grau 1980; Little et al. 2014, 2016). Prey perceive higher rates of movement and more direct trajectories as more threatening, triggering flight behavior (Grau and Grau 1980). As flight behavior is attention-attracting, deer that had greater movement during the hunting period were considered more observable (Little et al. 2014). Thus, the increased probability of observing deer as mean step length and mean absolute turning angle increased may be a result of triggering flight behavior in deer.

Interestingly, our results contradicted previous studies that have showed increased visibility to be a main indicator of harvest success (Swenson 1982; Jacques et al. 2011; Lebel et al. 2012; Plante et al. 2017). On ORR, visual obstruction increased from grasslands to forests (Little 2011), which should indicate a decrease in observation rate for hunters that used more forested cover, although this is dependant on deer density. The RSF determined that the predicted use of forested habitat was highest at 40% forested habitat within the shooting window of 150 m, with decreased predicted use as availability increased, whereas predicted use of mixed habitat reached a plateau after 5% mixed habitat within the shooting window (Fig. 4). This suggests that hunters selected mixed habitat when available more reliably than they did forested habitat. Mixed habitat had less visual obstruction than forested habitat, which may account for the selection of larger areas of mixed habitat, but this selection did not significantly affect observation rate. Deer altered their habitat use and movement behaviors following the initial 3-day exposure to predation pressure and selected more strongly for forested habitat than mixed habitat (Little 2011), suggesting that once deer are aware of predation pressure, overlapping habitat use becomes a more important factor in determining observation success than is visibility, because deer alter their movement and resource selection behaviors to avoid risk but learn to adapt their behaviors to avoid harvest (Little et al. 2014, 2016; Marantz et al. 2016).

Access roads have been shown to be a large influence on the number of hunting days per harvest, with hunters selecting for shorter distances from roads (Kilgo et al. 1998; Ranglack et al. 2017), which our data did not reflect. The RSF determined that predicted use increased as distance to roads increased (Fig. 4). However, this relationship did not show a significant effect on observation rate. Previous research has illustrated that most hunters focus their activity within as little as 100 m of a road or trail (Diefenbach et al. 2005; Lebel et al. 2012; Rowland et al. 2021). Thus, the probability of harvest generally increases as the amount of access roads increases within hunting areas, dependent on the degree of visibility, traffic and density of roads (Lebel et al. 2012; Paton et al. 2017; Plante et al. 2017; O’Connor et al. 2018). ORR did not have substantive internal road systems for hunters to utilize during their hunts. Compartment boundaries were often marked by pathways and simple, two-track roads, which limited hunter use as well. Additionally, the truncation of movement to be within assigned compartments removed nearly all use of roads by eliminating transit to and from the compartment. Although roads may be a significant source of hunter decision on other properties and within other studies (Kilgo et al. 1998; Gratson and Whitman 2000; Ranglack et al. 2017), the layout of ORR and the study design largely eliminated the use of roads by hunters. Our findings reflect this, because we did not find a significant effect on observation rate. Hunters also showed a strong preference in their selection of areas with higher roughness (Fig. 4), which aligns with literature suggesting varying topography to be an indicator of success (Swenson 1982; Iijima 2017). However, the low variation of the topography of the study area led to no significant effect of roughness on observation rate of deer. Given that the elevation of the study area ranged from 233 to 300 m and the slope averaged 4.12 degrees (s.e. = 0.02) (S. Webb, unpubl. data), habitat composition had a greater effect on visibility than topography in terms of observation rate.

It is important to note that hunters participating in this study were limited in their decision-making options regarding behavioral choices. Because this study was originally designed to control for hunting pressure across a landscape by assigning hunters to compartments (Little 2011), hunters were limited to what was available to them within the compartment. Compartments could vary greatly in habitat composition and landscape features. Coupled with the truncation of GPS location points, this could account for the sharp threshold reached in predicted use as distance to ponds and roads increased, as well as the negative linear relationship of distance to anthropogenic features in the RSF. The available riparian cover ranged from 0 to 73%, with an average of 22% of the compartment composition. Despite the high variability of available riparian habitat, predicted use of riparian cover varied only from 0.33 to 0.38 (Fig. 4), suggesting that hunters did not select strongly for riparian areas and this selection did not have a significant effect on observation rate. The available forested cover ranged from 0 to 45%, with an average of 18% of the compartment composition; however, hunters used forested cover in higher proportions to compartment availability (Fig. 5). Given that use of forested habitat was one of the most important covariates in predicting the number of deer observed, hunters that had limited availability of forested cover at their disposal during their hunts did not necessarily observe less deer because of active behavioral decisions, but rather because of predetermined compartment.

However, our results are context dependent, with an 18% average forest cover in our hunting compartments, compared with states in eastern United States that are predominantly forested. Therefore, future research should continue to explore this relationship between deer-hunters and land-use patterns.

Management implications

One issue currently facing management agencies is the low recruitment and retention of hunters (Ryan and Shaw 2011). Hunting is an activity based long in tradition, with methods and locations passed down through the generations (Gratson and Whitman 2000) and, often, knowledge can be a barrier to new hunters in an area (Ryan and Shaw 2011). Additionally, hunter satisfaction is linked with harvest success, which influences hunter opinion of management efforts by agencies (Miller and Graefe 2001). Therefore, providing hunters with behavioral tools to increase the likelihood of harvest success increases hunter satisfaction in agency management of wildlife while also maintaining hunting as a viable management tool.

White-tailed deer populations are increasing throughout their range across the United States (Urbanek et al. 2011), and urbanization and habitat fragmentation have pushed deer into more conflict with humans (Green et al. 1997; Kilpatrick et al. 2007; Urbanek et al. 2011). There is a growing need to improve hunting as a management tool to more effectively manage growing populations (Cromsigt et al. 2013). By identifying and optimizing where hunter and prey behaviors interact, managers have the potential to create more targeted approaches to hunting and population management. The ability to determine what behaviors and areas maximize the probability of harvest, for example, movement through forested cover, may increase the efficiency and effectiveness of hunting as a tool. In the case of urban and suburban deer management where agencies must consider how to remove a specific number of animals from an area with limited time and space availability (Hansen and Beringer 1997; Kilpatrick and Walter 1999; Kilpatrick et al. 2002, 2007), the ability to maximize probability of behavioral overlaps between deer and hunters could increase the efficiency of a targeted hunt. Our observations made during a highly regulated hunt on private property could also be of use to private landowners seeking to more effectively manage populations on their properties. Local wildlife agencies and private landowners can more strategically plan hunts on the basis of habitat composition of the property to meet management goals. Conversely, if the goal of management is to increase prey populations by decreasing harvest success, identifying what hunter behaviors lead to greater harvest success would delineate methods that could decrease the level of hunter harvest. Limiting the overlaps of habitat use by increasing road closures or adjusting hunting areas would likely decrease hunter success by limiting where hunter behaviors and prey behaviors overlap (Kilgo et al. 1998; Ranglack et al. 2017).