Influence of Spatial and Temporal Use of Resources by a Top Predator on Use of Resources 
by Co-occurring Species of Mesopredators and Prey 
 
by 
 
Jesse Thomas Boulerice 
 
 
 
 
A thesis submitted to the Graduate Faculty of 
Auburn University 
in partial fulfillment of the  
requirements of the Degree of  
Master of Science 
 
Auburn, Alabama 
December 12th, 2011 
 
 
 
 
 
 
 
 
Approved by 
 
Todd D. Steury, Chair, Assistant Professor of Forestry and Wildlife Sciences 
James B. Grand, Professor of Forestry and Wildlife Sciences 
Troy L. Best, Professor of Biological Sciences 
 
 
 
 
ii 
 
Abstract 
 
 Use of resources by a species can be influenced by interactions with co-occurring species. 
Use of both space and time can be altered to putatively reduce interspecific competition or risk of 
predation.  I used occupancy modeling to measure the influence of competition with the coyote 
(Canis latrans) on co-occurring mesopredators.  Similarly, the influence of use of space by 
coyotes was analyzed in the context of a race for space between two species of prey with 
differing characteristics.  I did not detect support for competitive interactions between coyotes 
and mesopredators.  I suggest that differential use of prey and high availability of prey may 
reduce the need for differential use of resources among these predators.   I found evidence for 
contrasting responses of prey to use of space by coyotes.  I propose that traits related to mobility 
and reproductive output may influence the outcome of the race for space between predators and 
prey. 
 
iii 
 
Table of Contents 
 
 
Abstract ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ii 
 
List of Tables ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ...v 
 
Chapter 1: Influence of Intraguild Competition From Coyotes (Canis latrans) on Spatial and  
 Temporal Use of Resources in a Community of Mesopredators??????????.  1 
 
 Introduction ??????????????????????????????..1  
 
 Methods????????????????????????????????3  
 
  Study Sites????????????????????????????4  
   
  Field Surveys???????????????????????????4  
 
  Attributes of Microhabitat??????????????????????5  
 
  Occupancy Modeling ...??????????????????????? 5 
  
  Construction Spatial and Temporal Models ???????????????.7  
 
  Prediction Coyote Resource Use???????????????????..8  
 
  Modeling the Influence of Use of Resources by Coyotes??????????9  
  
 Results ????????????????????????????????10  
 
  Spatial Analysis ?????????????????????????..11  
 
  Temporal Analysis????????????????????????..12  
 
 Discussion ??????????????????????????????.. 14 
 
iv 
 
 Literature Cited ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ..18 
 
Chapter 2: Traits of Prey That Influence Use of Space Within Predator-prey Relationships ?? 25 
 
 Introduction ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? 25 
 
 Methods? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ..27 
 
  Study Sites? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ..27 
 
  Attributes of Habitats ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? .28 
 
  Field Surveys? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ..29 
 
  Occupancy Modeling ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? .29 
 
  Constructing Species-Specific Models? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ...30 
 
  Modeling the Spatial Association? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ...31 
 
 Results ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? 33 
 
  Constructing Species-Specific Models? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ...33 
 
  Modeling the Spatial Association? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ...34 
 
 Discussion ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ..34 
 
 Literature Cited ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ..39 
 
 
 
 
v 
 
List of Tables 
 
 
Table 1:  Univariate analysis of detection covariates from modeling spatial relationships  
 between coyote, white-tailed deer, and eastern cottontail? ? ? ? ? ? ? ? ? ? ? ? ..43 
 
Table 2: Estimated effects of model variables calculated from occupancy analysis of the spatial 
 relationship between coyote, white-tailed deer and eastern cottontail? ? ? ? ? ? ? ...44 
 
 
1 
 
Chapter 1:  Influence of Intraguild Competition From Coyotes (Canis latrans) on Spatial and 
Temporal Use of Resources in a Community of Mesopredators 
Introduction 
 
According to the mesopredator release hypothesis, top predators should negatively 
influence intermediate predators (mesopredators) in natural communities (Soule et al. 1988, 
Sieving 1992, Estes 1996, Terborgh et al. 1999, 2001).  Specifically, top predators may impact 
mesopredators either directly through intraguild predation (interference competition, Polis and 
Holt 1992; Holt and Polis 1997; Palomares and Caro 1999) or indirectly through use of a 
common, limited resource (exploitative competition, Holt et al. 1994; Linnell and Strand 2000).   
Theoretical studies provide strong support for purported competition among predators (Soule et 
al. 1988; Sieving 1992; Estes 1996; Terborgh et al. 1999; 2001) and empirical research suggests 
that competition among terrestrial mammalian predators may be common (Palomares and Caro 
1999; Donadio and Buskirk 2006).  However, several studies have reported a lack of support for 
negative interactions between species of predators that are expected to compete, which has 
drawn skepticism about appropriateness of the mesopredator release hypothesis across ecological 
systems (Wright et al. 1994; Litvaitis and Villafuerte 1996; Gehrt and Clark 2003). Thus, further 
study is required to elucidate the nature and strength of competitive interactions among 
carnivores. Unfortunately, comprehensive studies of competition among predators have been 
hampered by difficulties associated with research on such wide-ranging and secretive species 
(Fedriani et al. 2000).  
One putative means of overcoming the limitations associated with studying the 
mesopredator release hypothesis and competition among predators in general, may be to focus on 
selection of resources, rather than abundance of competitors.  Mesopredator release hypothesis 
 
2 
 
traditionally has focused on the effect of abundance of top predators on abundance of 
mesopredators (Crooks and Soule 1999; Rogers and Caro 1999; Henke and Bryant 1999).  
However, selection of resources by mesopredators also can be altered through competition with a 
top predator and such changes may have an equal or more negative effect on populations of 
mesopredators than simply changes in abundance (Schmitz et al. 1997; Werner and Peacor 
2003).  Therefore, selection of resources may be a solid alternative for measuring competition 
within a community of predators, while simultaneously overcoming limitations associated with 
estimating abundance of such hard-to-study species. 
Selection of resources can be defined as a disproportionate use of resources by a species 
relative to availability, typically in an attempt to maximize fitness (Boyce and McDonald 1999; 
Manly et al. 2002). In theory, resources are selected in a manner that maximizes benefits, such as 
energetic gain, while minimizing costs associated with predation and competition (Hirzel and Le 
Lay 2008; McLoughlin et al. 2010).  Among predators, competition for resources typically is 
related to dietary overlap where abundance of shared prey is the limiting factor (Mills and Biggs 
1993; Palomares and Caro 1999; Donadio and Buskirk 2006). For mesopredators, selection of 
resources spatially, temporally, or both may reflect this tradeoff between acquisition of food and 
exploitive or interference competition with a top predator.  Specifically, mesopredators can 
putatively decrease competitive interactions with top predators by using an area where top 
predators are rare (Palomares and Caro 1999; Durant 2000), or a time when activity of top 
predators is low (Johnson et al. 1996; Arjo and Pletcher 1999). Thus, differences in temporal or 
spatial selection between resources with and without top predators provide indirect support for 
the mesopredator release hypothesis and competition among predators.  
With removal of larger predators and natural expansion of geographic range, the coyote 
 
3 
 
(Canis latrans) is believed to have assumed a top-predator role and to exert competitive pressure 
on communities of mesopredators (Hill et al. 1987; Lovell et al. 1998; Gompper 2002; Laliberte 
and Ripple 2004).   Studies have elucidated negative correlations between populations of 
mesopredators and abundance of coyotes, suggesting that competitive interactions with coyotes 
may limit the number of mesopredators (Crooks and Soule 1999; Rogers and Caro 1999; Henke 
and Bryant 1999). Additionally, selection of spatial resources by some mesopredators such as 
foxes, tends to support selection against coyotes (Thompson and Gese 2007; Fedriani et al. 2000; 
Sovada et al. 1995; Gosselink et al. 2003; Farias et al. 2005) and competition from coyotes is a 
leading cause of mortality in some populations of foxes (Ralls and White 1995; Nelson et al. 
2007).  However, spatial overlap is common between coyotes and other mesopredators such as 
bobcats (Lynx rufus; Fedriani et al. 2000; Neale and Sacks 2001; Thornton et al. 2004), raccoons 
(Procyon lotor; Gehrt and Prange 2006; Gehrt and Clark 2003), and striped skunks (Mephitis 
mephitis; Gehrt 2005; Prange and Gehrt 2007), which suggests that competition with coyotes 
may be reduced among non-canid mesopredators. Given the growing geographic range and 
abundance of coyotes across North American communities (Hill et al. 1987; Laliberte and Ripple 
2004), greater insight into the influence that exploitative and interference competition with 
coyotes has on mesopredator communities is needed (Gompper 2002). 
In this study, I measured the influence of use of resources by coyotes on spatial and 
temporal selection of resources by a community of mesopredators, including bobcats, red foxes 
(Vulpes vulpes), gray foxes (Urocyon cinereoargenteus), raccoons (Procyon lotor), and Virginia 
opossums (Didelphis virginiana).  I predicted that each of these species would demonstrate either 
spatial or temporal selection (or both) against resources used by coyotes.  
Methods 
 
 
4 
 
Study Sites 
To investigate selection of resources in communities of predators, 13 study sites were 
surveyed for carnivores as part of a larger state-wide biodiversity inventory (J.B. Grand, Auburn 
University, unpublished data).  Surveyed sites included state parks and wildlife management 
areas in Alabama.  Ecoregions contained within these sites included the Interior Plateau, Ridge 
and Valley, Southeastern Coastal Plains, Southwestern Appalachians, Southeastern Plains, and 
Piedmont (Griffith et al. 2001).   
For each study site, I superimposed a grid consisting of points spaced 250m apart.  I 
defined a potential sampling unit as a point and the area contained within 125m of that point.  
Using ArcGIS 9.2 (Environmental Systems Research Institute, Inc., Redlands, CA) and the 
National Land Cover Database (2001), I categorized the resulting potential sampling units into 
one of 33 types of land cover based on the majority type of land cover contained within each 
unit.  A stratified random-sampling design was used to select 10% of available units for survey 
where stratification was based on category of land cover.  Specifically, categories of land cover 
were surveyed proportional to availability within the 13 study sites.    
Field Surveys  
To estimate selection of resources by predators, presence-absence data for species were 
collected during August-December 2008-2010.  I obtained samples of predators using non-
invasive, infrared, digital trail cameras (RapidFireTM PC85, Reconyx, Inc., Holmen, WI) 
equipped to run for a 72-hour survey.  At each sampling unit, one camera was placed either at the 
center or 100m from center.  I administered commercially available scents (Caven?sTM, 
Minnesota Trapline Products, Pennock, MN) 2-3m in front of each camera.  Cameras were 
programmed to take photographs at 2-minute, time-lapse intervals, and when triggered by the 
 
5 
 
internal motion sensor, for the entire 72-hour survey.  I identified species captured in digital 
images.  Time of detection was recorded using the timestamp function; thus, providing a history 
of detection for each surveyed unit. 
Attributes of Microhabitat  
Because I expected that spatial use of resources by a species may be influenced by 
attributes of microhabitat, I collected site-specific measurements on vegetation at each camera 
including canopy cover, density of downed-woody debris, and understory visibility.  These data 
were collected only during surveys conducted in 2009 and 2010.  I estimated percentage canopy 
cover using point-quarter sampling (Brower and Zar 1977) by calculating the ratio between 
distance from camera to the nearest tree that provided the greatest canopy cover and distance 
over which the canopy of that tree extended.  This ratio was obtained in each of four quadrants 
(northwest, northeast, southwest, southeast) from the camera.  An index of downed-woody debris 
was calculated by measuring distance from camera to the first debris > 4.0cm in diameter 
encountered along each cardinal direction.  To index understory visibility, a modified density 
board was used (Nudds 1977).  At distances of 10, 15, and 20m in each cardinal direction, a 
visibility count of 56 8cm squares on a 128 by 64cm rectangular profile board was conducted.  
For all three vegetative characters, measurements along the four directions per quadrant were 
averaged to create an index for vegetation at each camera. 
Occupancy Modeling 
Because I anticipated that my methods (camera surveys) would provide imperfect rates of 
detection, I used occupancy analysis to examine patterns in selection of resources and 
associations between top predator and mesopredators (MacKenzie et al. 2002; 2004; 2005; 
MacKenzie and Royle 2005).  The value of occupancy modeling is that this approach recognizes 
 
6 
 
that detection of species rarely is perfect and adjusts estimates of use of resources to account for 
such imperfections.  Specifically, occupancy analysis allows the probability of use (?) of a 
resource to be estimated based on the history of detection (or the observed record of presence-
absence), while simultaneously accounting for probability of detection (p; or the probability that 
a species will be detected when present) of a species at resources. Both ? and p are estimated by 
conducting multiple surveys, or sampling occasions, over space or time, assuming unchanging 
use of the resource during sampling. When sampling units are smaller than the home range of the 
species investigated, as in this study, ? should be interpreted as the probability of use of a 
resource rather than occupancy of a resource (Kendall and White 2009).   
In occupancy analysis, probability of detection (p) is < 1, even when a species uses a 
resource.  Detection of a species during any sampling occasion will take one of three forms: 
resource is used and species is detected, resource is used but species is not detected, or resource 
is not used.  The probability that a species will use the resource and is detected is equal to the 
proportion of resources used (?) multiplied by the probability of detection (p), or ? x p.  In 
contrast, the probability that a species will use the resource but not be detected is ? x (1-p).  The 
probability that a species does not use a resource is equal to 1 ? ?.   
Additionally, use of, and detection at, a resource may vary as a function of environmental 
variables such as attributes of resource or co-occurrence of other species (MacKenzie et al. 2002; 
2004).  In occupancy analysis, estimates of use and detection of resources can be calculated 
while incorporating such covariates using the logistic model (MacKenzie 2006): 
 
 
7 
 
Where ? equals the parameter of interest (either ? or p) at the ith resource or sampling occasion, 
respectively, x represents the covariate values for the ith resource or sampling unit, and ? equals 
the estimates of coefficients for k covariates.   
Constructing Spatial and Temporal Models 
I constructed two series of occupancy models; one focused on spatial selection of 
resources and the other on temporal selection of resources within species.  For my evaluation of 
spatial selection of resources, the resource was each sampling unit.  I obtained the multiple-
sampling occasions of a sampling unit needed to estimate probabilities of detection within this 
analysis by dividing the 72-hour survey for each sampling unit into three 24-hour periods.  
Divisions were made mid-day, when least activity occurred.  Each period was considered to be 
an independent sampling occasion during which presence of a species was counted only once 
when detected.  I expected that spatial use of resources would vary with attributes of 
macrohabitat and microhabitat in relation to species-specific preferences for habitats.  To account 
for selection of macrohabitat, I reclassified the 33 types of land cover used to designate majority 
land cover for each sampling unit into 8 broader classes of habitat in an effort to generalize types 
of habitats and obtain reasonable samples at each class of habitat.  The classes of agriculture, 
coastal, developed, hardwoods, mixed, pine, riparian, and scrub were use as covariates of 
occupancy.  Similarly, the vegetative characteristics of understory visibility, percentage canopy 
cover, and density of downed-woody debris also were evaluated as covariates of occupancy to 
account for selection of microhabitat.  In addition, I also expected detection to vary with 
environmental variables associated with weather specific to each sampling occasion.  Therefore, 
I obtained data for temperature, wind speed, and precipitation for each occasion from the nearest 
weather station (average distance 24km, range 3-80km).  I also considered the effect of number 
 
8 
 
of days since the camera was deployed on probability of detection, anticipating that as time 
progressed, species might be more likely to encounter the bait or that deterioration of bait might 
influence rates of detection. 
In my evaluation of temporal selection of resources, the resource to be used was a 3-hour 
period during any date (i.e., 8 time blocks/day: 0000-0300, 0300-0600, 0600-0900 h CDT, etc.). 
Thus,  was the probability of use of a given block of time by the carnivore. Under this temporal 
framework, I obtained multiple sampling occasions by building a history of detection using 
spatial replicates of presence-absence of species across sampling units during each 3-hour block 
on each calendar day.  Species detected at multiple sampling units within a 3-hour block were 
counted only once.  Similar to the effect of preferences of habitat on spatial selection of 
resources, species-specific patterns of activity may influence temporal selection of resources.  To 
account for this effect, I considered each 3-hour block for inclusion as covariates of occupancy in 
models.  I also considered the effect of average temperature during each temporal block on 
occupancy.  I expected that ability to detect a species during a given 3-hour block may vary 
according to attributes of habitat of the sampling unit; therefore, I considered the eight classes of 
macrohabitat as covariates of detection in this temporal analysis. 
Predicting Coyote Resource Use 
 To estimate the influence of the coyote on use of resources in a community of 
mesopredators, I created a model that predicted probability of use for any resource (sampling 
unit or temporal block) where a coyote was not detected.  To build the spatial model of use of 
resources by coyotes, I used a combination of univariate tests (using a single variable in the 
occupancy model) and ecological reasoning to determine which covariates of occupancy and 
detection could be influential to spatial use of resource by coyotes.  All variables that were 
 
9 
 
significant at ? = 0.05 level in univariate analyses or believed to be important on ecological 
grounds were then combined to form a global model.  I then analyzed and evaluated the global 
model and all submodels using an AIC framework and calculated the probability of use of each 
resource using multimodel inference.  Specifically, model-averaged predictions were obtained 
for each resource via a summation of the product of the probability of use predicted by each 
model and the AICc weight of that model (Anderson et al. 2000; Burnham and Anderson 2002).  
The same procedure was used to construct a variable for probability of use by a coyote under the 
temporal framework.  Model-averaged predictions were used to predict probability of use for all 
resources (sampling units or time blocks) when a coyote was not detected; if a coyote was 
detected at the resource, then probability of use was recorded as 1.0. 
Modeling the Influence of Use of Resources by Coyotes 
 To evaluate the influence of use of resources by coyote on the other species of 
mesopredators, I constructed an occupancy model for each species for both spatial and temporal 
analyses.  Each model contained the predicted probability of use of resources by coyotes as a 
covariate of occupancy. However, observed associations between coyotes and mesopredators 
may be driven by similarities or differences in selection for specific attributes of resources, rather 
than avoidance of competitors. For example, even if mesopredators were avoiding coyotes, 
observed associations could be positive if both species select the same resource (Gilpin and 
Diamond 1984; Peres-Neto et al. 2001; MacKenzie et al. 2004).  Therefore, to investigate the 
influence of presence-absence of coyotes on use of resources by mesopredators after accounting 
for selection of resources, each species-specific spatial model included as covariates of 
occupancy all classes of macrohabitat (and variables of microhabitat if univariate analyses 
indicated they were important to the species). Similarly, each species-specific temporal model 
 
10 
 
contained all blocks of time to account for patterns of activity of species.  Species-specific 
models also contained those covariates of detection that were significant in preliminary 
univariate analyses or were believed to be important to the species.  A single-species approach to 
occupancy was used for these analyses, as opposed to the co-occurrence analyses of species 
(MacKenzie et al. 2004), because preliminary analyses indicated that the latter method typically 
failed to converge (unpublished data; MacKenzie et al. 2004).  Additionally, I only examined the 
effect of presence-absence of coyotes in a global model, rather than attempting to find a 
potentially more parsimonious submodel or averaging of models, because my goal was to 
estimate effects of competitors while accounting for selection of resources. I used the statistical 
package R (R Development Core Team 2005) and the unmarked package (Fiske and Chandler 
2010) to conduct modeling analyses. 
Results  
 
During the 3 years of surveying, I collected data for 446 sampling units across 13 study 
sites.  This was 958,300 digital images, of which 7,938 captured species of interest (<1%).  I 
detected coyotes at least once at 77 of 446 sampling units (17.3%).  I did not detect bobcats at 
sampling units designated as agricultural or coastal.  Because rates of detection can only cannot 
be calculated where a species is detected at least once, I removed these 25 units from analysis. 
We detected bobcats at 17 of the remaining 421 units (4.0%).   For raccoons, opossums, and 
foxes, I used data from 360 sampling units (second and third seasons) because of the importance 
of attributes of microhabitat to selection of resources by the species. Raccoons were detected at 
117 (32.5%) and opossums at 120 (33.3%) of these sampling units.  Neither species of fox was 
detected at sampling units designated as riparian; therefore, I removed all riparian sites (20 sites) 
from analysis of foxes.  Because of the small number of red foxes detected (3 of 340 sampling 
 
11 
 
units, 0.009%), I combined detections of red foxes with those of gray foxes and constructed 
models to incorporate both species.  I believe this was appropriate because both species of foxes 
were expected to respond similarly to competition from coyotes (Gosselink et al. 2003).  
Combined, I detected foxes at 31 (9.1%) of the remaining sampling units.   
Spatial Analysis 
 In preliminary univariate analyses of probability of spatial use by coyotes, covariates of 
detection for minimum temperature (-0.051, SE = 0.011, p = <0.001) and day of survey (2nd Day:  
0.582, SE = 0.293, p = 0.047) had a significant influence on detection of coyotes.  No variable 
for microhabitat was a significant predictor of spatial selection of resources by coyotes in 
univariate analysis (p > 0.064).  My global model for predicting occupancy by coyotes, therefore, 
included these covariates of detection, and eight covariates of occupancy (each habitat class).  
Analyses suggested that the best model describing spatial use of resources by coyotes, among 
those considered, included covariates of occupancy for coastal + developed + hardwoods + 
mixed types of habitats and covariates of detection for minimum temperature and 2nd day.   
However, the Akaike weight of this model was only 0.017, supporting use of model-averaging. 
Compared to the weight of 4.82 x 10-7 for the null model, this best model was 35,269 times 
better.  Our model-averaged predicted probability of spatial use of resources by coyotes (for 
those sampling units where coyotes were not detected) ranged from 0.381 to 0.581.   
In preliminary analyses of use of resources by mesopredators, I detected the following 
microscale covariates of occupancy to be significant predictors of spatial use of resources:  for 
foxes, understory visibility (0.078, SE = 0.037, p = 0.035); for raccoons, understory visibility 
(0.065, SE = 0.027, p = 0.016) and canopy cover (1.85, SE = 0.812, p = 0.023); and for Virginia 
opossums, canopy cover (2.23, SE = 0.625, p < 0.001).  Therefore, I only used data from 2009 
 
12 
 
and 2010 in subsequent analyses for these species.  For detection of mesopredators, the following 
variables were significant:  for foxes, average temperature (-0.087, SE = 0.022, p < 0.001); for 
raccoons, day of survey (2nd Day: 0.444, SE = 0.213, p = 0.037); and for Virginia opossums, day 
of survey (2nd Day:  0.450, SE = 0.219, p = 0.040; 3rd Day: 1.045, SE = 0.233, p < 0.001).  No 
microscale covariate of occupancy or detection had a significant influence in the occupancy 
model for bobcats (all p > 0.174).   
 The effect of use of resources by coyotes on spatial use of resources by mesocarnivores, 
after accounting for selection of resources by species, was not significant for any species (all p > 
0.151). Additionally, estimates of the effect were small and positive for most species, counter to 
predictions; bobcats, foxes, and raccoons were 6.34 (0.51 ? 78.85; 95% CL; p = 0.151), 1.74 
(0.29 ? 10.31; p = 0.540), and 2.85 (0.43 ? 19.09; p = 0.280) times as likely to use a spatial 
resource used by coyotes.  The estimated effect on spatial use of resources by Virginia opossums 
was also non-significant, but slightly negative.  Virginia opossums were 1.43 (0.23 ? 2.15; p = 
0.534) times as likely to use to resources that were not used by coyotes. 
Temporal Analysis     
 In preliminary analysis of probability of temporal use of resources by coyotes, 3-hour 
blocks from 0900 to 1800 h CDT (day) were not significantly different from each other (partial 
likelihood ratio test; p = 0.289). Similarly, blocks from 1800 to 0900 (night) were not 
significantly different from each other for predicting occupancy by coyotes (p = 0.717).  
However, there was a significant difference between day and night (1.96, SE = 0.372, p < 0.001).  
Therefore, I reduced the eight 3-hour blocks into two 12-hour blocks (day and night).  Average 
temperature during each block of time also had a significant effect on probability of temporal use 
of resources by coyotes (-0.085, SE = 0.017, p < 0.001).  Analyses also indicated that pine, scrub, 
 
13 
 
and riparian habitats were not significantly different from each other (partial likelihood ratio test; 
p > 0.3); that agriculture, hardwoods, and mixed-forest habitats were significantly different from 
the other classes (p < 0.04), but not from each other (p > 0.3); and that coastal and developed 
habitats were significantly different from all other classes of habitats (p < 0.001 and 0.004, 
respectively).  Therefore, my global, temporal-occupancy model included variables of occupancy 
for the two time blocks (night vs. day) and average temperature, and a detection variable for the 
four grouped habitats (agriculture + hardwoods + mixed, coastal, developed, other). Analyses 
suggested that the best model describing temporal use of resources by coyotes, among those 
considered, included covariates of occupancy for blocks of time and average temperature of 
blocks of time and covariates of detection for the four grouped habitats. The Akaike weight of 
this model was 0.99, and I based predictions on this best model rather than using multi-model 
inference. Weight of the null model was 3.21 x 10-19, making the best model 3.08 x 1018 times 
better. The predicted probability of temporal use of resources by coyotes (for those temporal 
resources where coyotes were not detected) ranged from 0.018 to 0.808.   
Preliminary analyses indicated that macrohabitat had a significant effect only on 
detection for the Virginia opossum and the most-parsimonious grouping of those classes was 
group1: agriculture, pine, mixed, scrub, and coastal (p > 0.3 for test of significance within 
group); group 2: developed and riparian (p = 0.630 for test of significance within group), and 
hardwoods (p  < 0.001 for test of significant difference from other groups).  
 My analyses indicated that the effect of use of resources by coyotes on temporal use of 
resources by mesocarnivores, after accounting for patterns of activity for each species, was not 
significant for any species (all p > 0.147). The bobcat was the only species to exhibit a large 
association with use of resources by coyotes; bobcats were less than 0.001 (<0.0001 ? 112.57; 
 
14 
 
95% CL; p = 0.147) times as likely to use a temporal period also used by a coyote.  Estimates of 
the effect were small and negative for the other species; foxes, raccoons, and Virginia opossums 
were 0.871 (0.10 ? 7.25; p = 0.898), 0.612 (0.10 ? 3.77; p = 0.597), and 0.692 (0.21 ? 2.25; p = 
0.541) times as likely to use a temporal resource used by coyotes.  
Discussion 
Theory predicts that exploitive or interference competition with top predators, or both, 
will negatively impact mesopredators and may force these species to adjust spatial or temporal 
use of resources to putatively reduce the costs associated with such competitive interactions.  In 
this study, I measured the influence of use of spatial and temporal resources by a purported top 
predator, the coyote, on use of the same resources by bobcats, foxes, raccoons, and Virginia 
opossums.  I predicted that spatial or temporal use of resources by mesopredators would be less 
likely if those resources also were used by coyotes.  I observed little evidence for an effect of 
coyotes on spatial use of resources by mesopredators, as estimates of selection for such resources 
were non-significant and positive (with the exception of the Virginia opossum, for which the 
estimated effect was negative, but still non-significant), which contradicts my prediction. 
Similarly, there was little evidence for an effect of coyotes on temporal use of resources by 
mesopredators; estimates of selection for such resources were non-significant for all species. 
However, the estimated effect of temporal use of resources by coyotes on use of resources by 
bobcats was large, suggesting that the effect might have been statistically significant with larger 
samples and could be biologically important.  
Lack of support for a negative association between spatial and temporal use of resources 
by a top predator and mesopredators could be due to several factors.  For example, the lack of 
association may be related to differential use of prey by predators or high availability of shared 
 
15 
 
prey.  Effective partitioning of prey can serve to diminish both the interference and exploitative 
competition that smaller-bodied predators experience from top predators (Koehler and 
Hornocker 1991; Palomares and Caro 1999; Heithaus 2001).  If mesopredators can minimize 
competition for a shared prey by switching to alternative prey, costly interactions from top 
predators should be less common, and the need for mesopredators to use different spatial or 
temporal resources from top predators would be abated.  Although literature is lacking on the 
dietary habits of predators within Alabama, differential use of prey frequently occurs among 
these species in other regions (Litvaitis 1981; Cypher 1993; Neale and Sacks 2001; Azedo et al. 
2006).  Likewise, whereas interspecific competition among predators is typically most frequent 
and intense when shared resources are limited, top predators are often more tolerable of 
mesopredators as availability of prey increases (Koehler and Hornocker 1991; Palomares and 
Caro 1999; Heithaus 2001). Thus, avoidance of spatial and temporal resources used by coyotes 
may be limited because of a combination of differential use of prey and high availability of prey, 
reducing the need for negative associations in use of resources. 
Prey partitioning may also explain why we determined that bobcats were the only 
mesopredator to display a large negative association in temporal use of resources with the coyote 
(albeit non-significant). Body size of predators typically is correlated with use of prey; predators 
of similar size are expected to display greater dietary overlap (Rosenzweig 1966; Gittleman 
1985).  Of all the mesopredators considered here, the bobcat is the most similarly sized predator 
relative to the coyote.  Whereas body size may enable the other smaller mesopredators to readily 
partition prey with coyotes (i.e, by targeting smaller species), coyotes and bobcats both feed on 
deer and lagomorphs (as well as rodents), although typically in different proportions (Gese and 
Grothe 1995; Neale and Sacks 2001; Thorton et al. 2004).  This overlap may require bobcats to 
 
16 
 
use the same space as coyotes to capture similar prey, but direct contact with the larger predator 
likely is undesirable (Anderson 1986; Wilson et al. 2010). By using similar spaces but at 
different times, bobcats may be able to reduce competitive interactions as a result of direct 
contact with coyotes, while still obtaining access to shared prey.  However, because the power of 
my measured effect of use of resources by coyotes remains low, I cannot be certain of the 
strength of this temporal relationship. 
I acknowledge a second, albeit opposing, explanation to my observed (or lack thereof) 
patterns in use of resources by predators.  Where poor environmental conditions lead to low 
availability of prey, patches within a landscape may not contain the necessary dietary resources 
to fulfill energetic needs of mesopredators, even when prey are partitioned.  Instead, 
mesopredators are forced to compete for those few prey that remain available by congregating at 
patches with the highest availability of prey, even under threat of interspecific competition with 
larger predators such as the coyote (Sergio et al. 2003; Wilson et al. 2010,).  This type of 
congregation would explain the observed (albeit non-significant) positive spatial association 
between coyotes and mesopredators.  However, given the large geographic area and time-scale 
surveyed here, as well as the detection of a large proportion of prey by our cameras (unpublished 
data), the likelihood that low availability of prey is to blame seems doubtful.  
My results hint at the possibility that coyotes in the communities we examined may not 
serve the moderating role of a top predator as suggested by the mesopredator-release hypothesis.   
The lack of significant negative association between predators suggests that none of the 
mesopredators avoided coyotes in space or time, as would be expected if coyotes were dominate 
competitors.  Although coyotes have been considered as top predators elsewhere (Crooks and 
Soule 1999; Rogers and Caro 1999; Henke and Bryant 1999), previous studies may have 
 
17 
 
occurred where differences in regions or communities resulted in lower availability of prey or a 
reduced ability for mesopredators to effectively partition prey from coyotes.  If this supposition 
is the cause for the variations in results among sites, it would support the growing body of 
evidence suggesting that community stability, including those likely to experience release of 
mesopredators, depends on both top-down and bottom-up forces (Wilmers et al. 2006; Elmhagen 
and Rushton 2007).   
As conservation and management strategies have began to consider the important link 
between top predators and release of mesopredators (Ritchie and Johnson 2009), especially in 
relation to coyotes (Gompper 2002), the value of understanding competition among predators is 
great. Here, I failed to detect support of competitive interactions among predators that were 
expected to compete and suggest that partitioning and availability of prey may be responsible.  
While the mesopredator-release hypothesis may accurately describe the dynamics of some 
communities, my results hint that other variables may influence the strength and frequency of 
competitive interactions among predators.    
 
18 
 
Literature Cited 
 
Anderson, D. R., Burnham, K.P., and Thompson, W.L. 2000. Null hypothesis testing: problems, 
prevalence, and an alternative. J. Wildl. Manag. 64: 912-923. 
Anderson, E.M. 1986. Bobcat behavioral ecology in relation to resource use in southeastern 
Colorado. Ph.D. dissertation, Colorado State University, Fort Collins. 
Arjo, W.M., and Pletscher, D.H. 1999.  Behavioral responses of coyotes to wolf recolonization in 
northwestern Montana.  Can. J. Zool. 77: 1919-1927. 
Azevedo, F.C. C., Lester, V., Gorshch, W., Lariviere, S., Wirsing, A.J., and Murray, D. L. 2006.  
Dietary breadth and overlap among five sympatric carnivores. J. Zool. (London) 269: 
127-135. 
Boyce, K.A., and Barry, R.E. 2007.  Seasonal home range and diurnal movements of Sylvilagus 
obscurus (Appalachian cottontail) at Dolly Sods, West Virginia. Northeastern Naturalist 
14: 99-110. 
Brower, J.E., and Zar, H.J. 1977.  Field and laboratory methods for general ecology. W. C. 
Brown Co., Dubuque, Iowa. 
Burnham, K.P., and Anderson, D.R. 2002.  Model selection and multimodel inference: a practical 
information-theoretic approach.  2nd edition. Springer-Verlag.  New York, USA. 
Crooks, K.R., and Soule, M.E. 1999.  Mesopredator release and avifaunal extinctions in a 
fragmented system.  Nature 400: 563-566. 
Cypher, B.L. 1993.  Food item use by three sympatric canids in southern Illinois. Trans. Ill. State 
Acad. Sci. 86: 139-144. 
Donadio, E., and Buskirk, S.W.  2006.  Diet, morphology, and interspecific killing in Carnivora. 
Am. Nat. 167: 524-536. 
 
19 
 
Durant, S.M. 2000. Predator avoidance, breeding experience, and reproductive success in 
endangered cheetahs, Acinonyx jubatus.  Anim. Behav. 60: 121-130. 
Elmhagen, B., and Rushton, S.P. 2007.  Trophic control of mesopredators in terrestrial 
ecosystems: top-down or bottom-up?  Ecological Letters 10: 197-206. 
Estes, J.A. 1996. Predators and ecosystem management. Wildlife Society Bulletin 24: 390-396. 
Farias, V., Fuller, T.K., Fedriani, J.M., Wayne, R.K., and Sauvajot, R.M. 2005.  Survival and 
cause-specific mortality of gray foxes (Urocyon cinereoargenteus) in southern California.  
J. Zool. (London) 266: 249-254. 
Fedriani, J.M., Fuller, T.K., Sauvajot, R.M., and York, E.C. 2000.  Diets of three sympatric 
carnivores in the Santa Monica Mountains of California:  the importance of 
habitat,human presence and interspecific competition.  Oecologia (Berlin) 125: 258-270. 
Fiske, I., and Chandler, R. 2011. Unmarked: an R package for fitting hierarchical models of 
wildlife occurrence and abundance.  Journal of Statistical Software 43: 1-23. 
Gehrt, S.D., and Clark, W.R. 2003.  Raccoons, coyotes, and reflections on the mesopredator 
release hypothesis.  Wildlife Society Bulletin 31: 836-842. 
Gehrt, S.D., and Prange, S. 2006.  Interference competition between coyotes and raccoons: a test 
of the mesopredator release hypothesis.  Behavioral Ecology 18: 204-214. 
Gese, E.M., and Grothe, S. 1995.  Analysis of coyote predation on deer and elk during winter in 
Yellowstone National Park, Wyoming.  Am. Midl. Nat. 133: 36-43 
Gilpin, M.E., and Diamond, J.M. 1984.  Are species co-occurrences on islands non-random, and 
are null hypotheses useful in community ecology?  In Ecological Communities:  
Conceptual Issues and the Evidence.  Edited by D.R. Strong, D. Simberloff, L.G. Abele, 
and A.B. Thistle. Princeton University Press. pp. 297-315 
 
20 
 
Gittleman, J.L. 1985.  Carnivore body size: ecological and taxonomic correlates.  Oecologia 
(Berlin) 67: 540-554. 
Gompper, M.E. 2002.  Top carnivores in the suburbs? Ecological and conservation issues raised 
by colonization of northeastern North American by coyotes. BioScience 52: 185-190. 
Gosselink, T.E., Van Deelen, T.R., Warner, R.E., and Joselyn, M.G. 2003.  Temporal habitat 
partitioning and spatial use of coyotes and red foxes in east-central Illinois. J. Wildl. 
Manag.67: 90-103. 
Griffith, G.E., J.M. Omernik, J.A. Comstock, S. Lawrence, G. Martin, A. Goddard, V.J. Hulcher, 
and Foster, T. 2001. Ecoregions of Alabama and Georgia. US GeologicalSurvey, Reston, 
Virginia. 
Henke, S.E., and Bryant, F.C. 1999.  Effects of coyote removal on the faunal community in 
western Texas.  J. Wildl. Manag. 63: 1066-1081. 
Hill, E.P., Sumner, P.W., and Wooding, J.B. 1987.  Human influences on range expansion of 
coyotes in the Southeast.  Wildlife Society Bulletin 15: 521-524. 
Hirzel, A.H., and Le Lay, G. 2008.  Habitat suitability modeling and niche theory. J. Appl. Ecol. 
45: 1372-1381. 
Holt, R.D., and Polis, G.A. 1997.  A theoretical framework for intraguild predation.  Am. Nat. 
149: 745-764. 
Holt, R.D., Grover, J., and Tilman, D. 1994.  Simple rules for interspecific dominance in systems 
with exploitative and apparent competition.  Am. Nat. 144: 741. 
Johnson, W.E., Fuller, T.K., and Franklin, W.L. 1996.  Sympatry in canids: a review and 
assessment.  In Carnivore behavior, ecology, and evolution. Edited by J.L. Gittleman.  
Cornell University Press, Ithaca, NY. pp. 189-218. 
 
21 
 
Kendall, W.L., and White, G.C. 2009.  A cautionary note on substituting spatial subunits for 
repeated temporal sampling in studies of site occupancy.  J. Appl. Ecol. 46: 1182-1188. 
Koehler, G.M., and Hornocker, M.G. 1989.  Influences of seasons on bobcats in Idaho.  J. Wildl. 
Manag.53: 197-202. 
Laliberte, A.S., and Ripple, W.J. 2004.  Range contractions of North American carnivores and 
ungulates. BioScience 54: 123-138. 
Linnel, J.D., and Strand, O. 2000.  Interference interactions, co-existence and conservation of 
mammalian carnivores.  Diversity and Distributions 6: 169-176. 
Litvaitis, J.A. 1981.  A comparison of coyote and bobcat food habits in the Wichita Mountains, 
Oklahoma.  Proc. Okla. Acad. Sci. 61: 81-82.  
Litvaitis, J.A., and Villafuerte, R. 1996.  Intraguild predation, mesopredator release, and prey 
stability. Conservation Biology 10: 676-677. 
Lovell, C.D., Leopold, B.D., and Shropshire, C.C. 1998.  Trends in Mississippi predator 
populations, 1980-1995.  Wildlife Society Bulletin 26: 552-556. 
MacKenzie, D.L. 2006.  Occupancy estimation and modeling: inferring patterns and dynamics of 
species occurrence. Academic Press.  Burlington, MA. USA. 
MacKenzie, D.L., and Royle, A. 2005.  Designing occupancy studies: general advice and 
allocating survey effort.  J. Appl. Ecol. 42: 1105-1114. 
MacKenzie, D.L., Bailey, L.L., and Nichols, J.D. 2004.  Investigating species co-occurrence 
patterns when species are detected imperfectly.  J. Anim. Ecol. 73: 546-555. 
MacKenzie, D.L., Nichols, J.D., Sutton, N., Kawanishi, K., and Bailey, L.L. 2005. Improving 
inferences in population studies of rare species that are detected imperfectly. Ecology 86: 
1101-1113. 
 
22 
 
MacKenzie, D.L., Nichols, J.D., Lachman, G.B., Droege, S., Royle, A., and Langtimm, C.A. 
2002. Estimating site occupancy rates when detection probabilities are less than one.  
Ecology 83: 2248-2255. 
Manly, B.F.J., McDonald, L.L., Thomas, D.L., McDonald, T.L., and Erickson, E.P. 2002.  
Resource selection by animals: statistical design and analysis for field studies. Second 
edition. Kluwer Academic Publishers, Boston, Massachusetts, USA. 
McLoughin, P.D., Morris, D.W., Fortin, D., VanderWal, E., and Contasti, A.L. 2010. Considering 
ecological dynamics in resource selection functions.  J. Anim. Ecol. 79: 412. 
Mills, M.G.L., and Briggs, H.C. 1993.  Prey apportionment and related ecological relationships 
between large carnivores in Kruger National Park.  Symposium of the Zoological Society 
of London 65: 253-268. 
Neale, J.C.C., and Sacks, B.N. 2001.  Resource utilization and interspecific relations of 
sympatric bobcats and coyotes.  Oikos 94: 236-249. 
Nelson, J.L., Cypher, B.L., Bjurlin, C.D., and Creel, S. 2007.  Effects of habitat on competition 
between kit foxes and coyotes.  J. Wildl. Manag.71: 1467-1475. 
Nudds, T.D. 1977.  Quantifying the vegetative structure of wildlife cover.  Wildlife Society 
Bulletin 5: 113-117. 
Palomares, F., and Caro, T.M. 1999.  Interspecific killing among mammalian carnivores. Am. 
Nat. 153: 492-508. 
Peres-Neto, P.R., Olden, J.D., and Jackson, D.A. 2001.  Environmentally constrained null 
models: site suitability as occupancy criterion. Oikos 93: 110-120.  
Polis, G.A., and Holt, R.D. 1992.  Intraguild predation: the dynamics of complex trophic 
interactions.  Trends in Ecology and Evolution 7: 151-155. 
 
23 
 
Ralls, K., and White, P. J. 1995.  Predation of San Joaquin kit foxes by larger canids.  J. 
Mammal. 76: 723-729. 
Ritchie, E.G., and Johnson, C.N. 2009.  Predator interactions, mesopredator release and 
biodiversity conservation.  Ecology Letters 12: 982-998. 
Rogers, C.M., and Caro, M.J. 1998.  Song sparrows, top carnivores and nest predation: a test of 
the mesopredator release hypothesis.  Oecologia (Berlin)116: 227-233. 
Rosenzweig, M.L. 1966.  Community structure in sympatric Carnivora. J. Mammal. 47: 602-612.   
Schmitt, R. J. 1987.  Indirect interactions between prey: apparent competition, predator 
aggregation, and habitat segregation.  Ecology 68: 1887-1897. 
Sergio, F., Marchesi, L., Pedrini, P., and Penteriani, V.  2007.  Coexistence of a generalist owl 
with its intraguild predator: distance-sensitive or habitat-mediated avoidance.  Anim. 
Behav. 74: 1607-1616. 
Sieving, K.E. 1992.  Nest predation and differential insular extinction among selected forest 
birds of central Panama.  Ecology 73: 2310-2328. 
Soule, M.E., Bolger, D.T., Alberts, A.C., Wright, J., Sorice, M., and Hill, S. 1988. Reconstructed 
dynamics of rapid extinctions of chaparral-requiring birds in urban habitat islands. 
Conservation Biology 2: 75-92. 
Sovada, M.A., Roy, C.C., Bright, J.B., and Gillis, J.R. 1998.  Causes and rates of mortality of 
swift foxes in western Kansas.  J. Wildl. Manag. 62: 1300-1306. 
Terborgh, J., Estes, J., Paquet, P., Ralls, K., Boyd, D., Miller, B., and Noss, R. 1999.  Role of top 
carnivores in regulating terrestrial ecosystems.  In Continental conservations: Scientific 
foundations of regional reserve networks.  Edited by M. Soule and J. Terborgh. Island 
Press, Covela, California. pp. 39-64.  
 
24 
 
Terborgh J., Lopez, L., Nunez, P.V., Rao, M., Shahabuddin, G., Orihuela, G., Riveros, M., 
Ascanio, R., Alder, C.H., Lambert, T.D., and Baldas, L. 2001.  Ecological meltdown in 
predator-free forest fragments. Science 294: 1923-1926. 
Thompson, C.M., and Gese, E.M. 2007.  Food webs and intraguild predation: community 
interactions of a native mesocarnivore.  Ecology 88: 334-346. 
Thornton, D.H., Sunquist, M.E., and Main, M.B. 2004.  Ecological separation within newly 
sympatric populations of coyotes and bobcats in south-central Florida.  J. Mammal. 85: 
973-982. 
Werner, E.E., and Peacor, S.D. 2003.  A review of trait-mediated indirect interactions in 
ecological communites.  Ecology 84: 1083-1100. 
Wilmers, C.C., Post, E., Peterson, R.O., and Vucetich, J.A. 2006.  Predator disease out-break 
modulates top-down, bottom-up and climatic effects on herbivore population dynamics. 
Ecological Letters 9: 383-389. 
Wilson, R.R., Blankenship, T.L., Hooten, M.B., and Shivik, J.A. 2010. Prey-mediated avoidance 
of an intraguild predator by its intraguild prey.  Oecologia (Berlin) 164: 921-929. 
Wright, S.J., Gompper, M.E., and DeLeon, B. 1994.  Are large predators keystone species in 
Neotropical forests? Oikos 71: 279-294. 
 
 
 
25 
 
Chapter 2:  Traits of Prey That Influence Use of Space Within Predator-prey Relationships 
 
Introduction  
Use of space within a landscape by an individual is believed to be dictated largely by 
efforts to maximize fitness. For most species, fitness primarily is governed by the need to acquire 
resources and to avoid predation (Lima and Dill 1990; Lima 1998).  Given that areas with high 
availability of resources often impart the greatest risk of predation, use of space by prey 
constitutes a balance between maximizing acquisition of resources and minimizing the likelihood 
of encountering a predator (Lima and Dill 1990; Sih 1992; Hugie and Dill 1994; Molvar and 
Bowyer 1994; Bleich et al. 1997; Nicholson et al. 1997; Werner and Peacor 2003; Lima and 
Steury 2005; Bell et al. 2009).   However, for predators, the sought-after resource is prey.  
Therefore, predators attempt to use space that provides the greatest likelihood of capturing prey, 
typically areas where abundance of prey is greatest, or potentially where prey are more 
vulnerable to predation (Burger et al. 2004; Honer et al. 2005; Lima 2002; Veit et al. 1993).  As 
prey seek to avoid predators while predators seek to find prey, the ensuing two-player game 
becomes a predator-prey space race in which predators and prey must spatially respond to 
actions of each other to eat or avoid being eaten (Sih et al. 1998; Lima 2002, Sih 2005). 
Within this predator-prey space race, responses of both species shape spatial associations 
between predator and prey (Sih 1984; Lima 2002; Bell et al. 2009). For example, when predators 
outmaneuver efforts of prey at spatial avoidance, predator and prey will be positively associated 
and occur together more often than expected by chance.  Where this occurs, the predator is 
believed to win the race.  When response of prey dominates and prey win the race by 
successfully avoiding the predator, the spatial association between the two will be negative, as 
predator and prey will occur together less often than expected by chance (Sih 2005).  
 
26 
 
Although multiple efforts have been made to model dynamics of this race (Sih 1984; 
Hugie and Dill 1994; Mitchell and Lima 2002; Rosenheim 2004), few models have addressed 
how traits of the species involved might influence which species successfully responds to the 
other and wins the race (Sih 2005; Bell et al. 2009).  However, Sih (2005) suggested that traits 
related to costs and benefits of responding to the opposing species may strongly influence how a 
species reacts and, thus, the outcome of the race.  For example, if one species is less mobile than 
the other, relative energetic cost of movements in reaction to the other species may outweigh 
benefits of such movements. Consequently, the more mobile species may be able to outmaneuver 
the less mobile species and win the space race (Sih 2005).  Other behavioral traits that may 
spatially anchor a species, such as denning, nesting, mating, territories, or parental care (Sih 
2005; Thomson et al. 2006; Thaker et al. 2011) similarly may increase the costs of responding 
spatially, giving the unanchored species the advantage in the predator-prey space race (Sih 2005).  
Smaller territory or home range, for example, may serve as a spatial anchor by conveying a cost 
in terms of energetics, unfamiliarity of new terrain, or intraspecific competition, if a response to 
actions of the other species would require movement outside of the established home range.  
However, while the influence of such traits on spatial associations makes theoretical sense, few 
empirical tests have been conducted in regards to such predictions.   
Here, I investigate how differences in traits of prey influence the outcome of the predator-
prey race.  Specifically, I measured patterns of spatial association between a predator, the coyote 
(Canis latrans), and two species of prey that display distinct characteristics, the white-tailed deer 
(Odocoileus virginianus) and the eastern cottontail (Sylvilagus floridanus).  White-tailed deer 
and eastern cottontails exhibit contrasting traits in terms of mobility and sizes of home range 
(Smith 1991; Chapman et al. 1980).  Being of large body size, white-tailed deer are accustomed 
 
27 
 
to traveling greater distances than eastern cottontails in search of resources.  Similarly, size of 
home ranges of white-tailed deer are 60 to 520 times larger than that of eastern cottontails.  Thus, 
energetic costs related to mobility are likely far less for white-tailed deer than for eastern 
cottontails, and deer are less likely to be spatially anchored by smaller home ranges. 
I hypothesized that character traits related to mobility and size of home range of prey, 
relative to the common predator, would dictate the outcome of the predator-prey space race.  I 
predicted that 1) the spatial association between white-tailed deer and coyotes would be negative 
(white-tailed deer would win the space race) because the high mobility and large home range of 
deer relative to coyotes should facilitate avoidance of coyotes by deer; 2) the spatial association 
between eastern cottontails and coyotes would be positive (eastern cottontails would lose the 
space race) because the low mobility and small home range of cottontails relative to coyotes 
should hinder avoidance of coyotes by cottontails; and 3) owing to differences in mobility of 
prey, the spatial association between white-tailed deer and coyotes would be significantly 
different than the spatial association between eastern cottontails and coyotes. 
Methods 
Study Sites 
To investigate patterns of spatial association between the coyote and the two species of 
prey of interest, white-tailed deer and eastern cottontails, 13 study sites were surveyed across 
Alabama as part of a larger state-wide biodiversity inventory (J.B. Grand, Auburn University, 
unpublished data).  Surveyed sites encompassed state parks and wildlife management areas.  
Ecoregions contained within these properties included the Interior Plateau, Ridge and Valley, 
Southeastern Coastal Plains, Southwestern Appalachians, Southeastern Plains, and Piedmont 
(Griffith et al. 2001).   
 
28 
 
For each study site, I superimposed a grid consisting of points spaced 250m apart.  I 
defined a sampling unit as a point and the area contained within 125m of that point.  Using 
ArcGIS 9.2 (Environmental Systems Research Institute, Inc., Redlands, CA) and the National 
Land Cover Database 2001, I categorized all of the resulting sampling units into one of 33 types 
of land cover based on the majority type of land cover contained within each unit.  A stratified 
random-sampling design was employed to select 10% of the available units for survey where 
stratification was based on category of land cover.  Specifically, categories of land cover were 
surveyed proportionally to availability within the 13 study sites.    
Attributes of Habitats  
 Observed spatial associations between species should be shaped by avoidance of 
predators by prey and pursuit of prey by predators.  However, observed associations also may be 
driven by similarities or differences in selection for specific attributes of habitats. For example, if 
predator and prey are positively associated with each other (occur together more often than 
predicted by chance), they could simply be selecting for similar habitats rather than the 
association being a result of predators doing a better job at finding prey than prey are at avoiding 
predators (although such patterns may be part of predator-prey co-evolution).  In fact, even if 
prey are succeeding in avoiding predators, observed associations can be positive if both predator 
and prey select for the same habitat (Gilpin and Diamond 1984; Peres-Neto et al. 2001; 
MacKenzie et al. 2004).  Therefore, to investigate the influence of space use of predator on space 
use of prey after accounting for selection of habitat, I included attributes of habitats as covariates 
in analyses.  I reclassified the 33 types of land cover used to designate majority land cover into 
six broader classes of habitats and assigned the appropriate class to each sampling unit.  These 
classes included agriculture, hardwoods, mixed, pine, riparian, and scrub.   
 
29 
 
Field Surveys  
To investigate spatial associations between predator and prey, presence-absence data were 
collected during August-December 2008-2010.  We sampled for coyotes, white-tailed deer, and 
eastern cottontails using non-invasive, infrared, digital trail cameras (RapidFireTM PC85, 
Reconyx, Inc., Holmen, WI) equipped to run for a 72-hour survey.  At each sampling unit, one 
camera was placed either at the center or 100m from center.  We administered commercially 
available predator lures (Caven?sTM, Minnesota Trapline Products, Pennock, MN) 2-3m in front 
of the camera.  Although these lures were designed to target predators, we were able to also 
generate relatively high capture rates for non-predator species such as white-tailed deer and 
eastern cottontails (unpublished data).   Cameras were programmed to take photographs at 2-
minute intervals, and when triggered by the internal motion sensor, for the duration of the 72 
hour survey.  Time of detection was recorded using the timestamp function; thus, providing a 
history of detection for each surveyed unit. 
Occupancy Modeling 
Because I anticipated that my methods (camera surveys) would give rise to imperfect 
rates of detection, I used analysis of occupancy to examine patterns of selection of resources 
patterns and associations between predator and prey (MacKenzie et al. 2002; 2004; 2005; 
MacKenzie and Royle 2005).  The value of occupancy modeling is that this approach recognizes 
detection of species rarely is perfect, and adjusts estimates of use of resources and occupancy to 
account for such imperfect detections.  Specifically, analysis of occupancy allows the probability 
of use (?) of a resource to be estimated based on the history of detection (or the observed record 
of presence-absence) while simultaneously accounting for probability of detection (p; or the 
probability that a species will be detected when present) of a species at sampled resources. Both 
 
30 
 
? and p are estimated by conducting multiple surveys, or sampling occasions, over space or time, 
assuming unchanging use of a resource during sampling.   When sampling units are smaller than 
size of the home range of the species investigated, as is the case in this study for white-tailed 
deer and coyotes, ? should be interpreted as the probability of use of resources, rather than 
occupancy of resources (Kendall and White 2009).   
In occupancy analysis, probability of detection (p) is less than one even when species use 
a resource.  Detection during any sampling occasion will take one of three forms: resource is 
used and species is detected, resource is used but species is not detected, or resource is not used.  
The probability that a species will use the resource and is detected is equal to the proportion of 
resources used (?) multiplied by the probability of detection (p), or ? x p.  In contrast, the 
probability that a species will use the resource but not be detected is ? x (1-p).  The probability 
that a species does not use a resource is equal to 1 ? ?.   
Additionally, use and detection at a site may vary as a function of environmental 
variables such as attributes of resources or co-occurrence of other species (MacKenzie et al. 
2002; 2004).  In analysis of occupancy, estimates of use of resources and detection can be 
calculated while incorporating such covariates using the logistic model (MacKenzie et al. 2006): 
    
Where ? equals the parameter of interests (either ? or p) at the ith resource or sampling occasion, 
respectively, x represents the covariate values for the ith resource or sampling unit, and ? equals 
the coefficient estimates for k covariates.   
Constructing Species-Specific Models 
To account for imperfect detection in my analysis of spatial associations between 
predator and prey, I built species-specific models of occupancy to estimate probability of 
 
31 
 
occupancy by a species at each sampling unit where that a species was not detected.  To obtain 
the multiple sampling occasions necessary to estimate probabilities of detection, I divided each 
72-hour survey from each sampling unit into three 24-hour blocks.  Divisions were made mid-
day, when the least amount of activity occurred.  Each temporal block was considered as an 
independent sampling occasion during which presence of a species was counted only once when 
detected.  I expected use of resources to vary with attributes of habitats and I also expected 
detection to vary with environmental variables associated with weather.  Therefore, I obtained 
data for temperature, wind speed, and precipitation for each survey from the nearest weather 
station (average distance 24km, range 3-80km).  I also considered the effect of number of days 
since the camera was activated, anticipating that as time progressed, either species might be more 
likely to encounter the bait or that deterioration of bait could influence rates of detection. 
To build each species-specific model, I examined the influence of each covariate of 
detection in univariate analyses.  All variables of detection that were significant at ? < 0.05 level 
were included in a global model.  This model also included all variables of occupancy (classes of 
habitats) because I wanted to account for similarities and differences in species-specific 
preferences in habitat. I then fit all submodels of that global model using an AIC framework and 
calculated the probability of use of each sampling unit by a given species using multimodel 
inference.  Specifically, the model averaged predictions for probability of use of each sampling 
unit equaled the sum of individual predictions of models weighted by the AICc weight for that 
model (Anderson et al. 2000; Burnham and Anderson 2001).   
Modeling the Spatial Association 
Once I obtained the species-specific probability of use of each sampling unit, I was able 
to estimate spatial associations of predator and prey.  I constructed a new occupancy model for 
 
32 
 
the predator, whereby occupancy by coyotes of a given sampling unit was a function of 
probability of use by white-tailed deer and eastern cottontails at each sampling unit.  Specifically, 
if a prey was detected at a sampling unit, then the probability of use of that sampling unit was 
1.0; if the species was not detected, then the probability of use equaled the model-averaged 
prediction for the probability of use.  As above, this model also included all variables for 
occupancy of habitats to estimate the association between coyote and white-tailed deer or eastern 
cottontails after accounting for any effects of selection of habitats. The model also included those 
variables of detection we determined were significant predictors of detection of coyotes in 
univariate analyses.  We used a single-species approach to occupancy as opposed to the co-
occurrence analysis of species (MacKenzie et al. 2004) because preliminary analyses indicated 
that the latter method typically failed to converge (unpublished data; MacKenzie et al. 2004).  
Additionally, I chose to model the predator-prey relationship from the point of view of the coyote 
because doing so facilitated the testing of my prediction that spatial associations with coyotes 
would be different between white-tailed deer and eastern cottontails (prediction 3).  For testing 
prediction 3, I compared the model for coyotes that contained separate probabilities of use for 
white-tailed deer and eastern cottontails with a second model for coyotes that contained a single 
variable describing the probability of use by either species of prey. This probability was 
calculated using the formula (equation 1): 
? ? ?1 1 1pre y de e r c ot t ont ai l 
As above, this probability was incorporated at each sampling unit where neither species of prey 
was detected, and a value of 1.0 was used where either species was detected.  I used a partial-
likelihood ratio test to determine if the more complicated two-prey model was a significant 
improvement over the simpler, single-prey model, and thereby test if the effect of use by white-
 
33 
 
tailed deer on use of habitats by coyotes was significantly different from the effect of use by 
eastern cottontails.  I used the statistical package R (R Development Core Team 2005) and the 
unmarked package (Fiske and Chandler 2010) to conduct all modeling analyses. 
Results 
 During my study, I collected usable data from 420 sampling units across 13 study sites. 
Coyotes were detected at least once at 67 (16%), white-tailed deer at 140 (33.3%), and eastern 
cottontails at 22 (5.2%) of those units. 
Constructing Species-Specific Models 
 In construction of each species-specific model of occupancy, the following variables 
influenced (p < 0.05) probabilities of detecting species (summarized in Table 1): for coyotes, 
minimum temperature (? = -0.052 ? 0.013; ? SE, p < 0.001) and number of days from start of 
survey (2nd Day: ? = 0.623? 0.315, p < 0.048); for white-tailed deer, maximum temperature (? = 
-0.044? 0.009, p < 0.001); and for eastern cottontails, no variable significantly influenced 
probability of detection (p > 0.211).   I constructed species-specific global models that included 
each variable that influenced probability of detecting a species and the six classes of habitats as 
covariates to occupancy.  Analyses suggested that the best model for describing spatial use of 
resources by coyotes included covariates of occupancy for coastal + developed + hardwoods + 
mixed habitats and covariates of detection for minimum temperature and 2nd day .  For white-
tailed deer, the best model included no covariate for occupancy and maximum temperature for 
detection, while the best model for eastern cottontails contained the mixed-forest habitat and no 
covariate of detection.  The Akaike weights for these best models were 0.017, 0.132, and 0.132 
respectively, supporting use of model-averaging. Relative to weights in null models, the best 
models for the coyote, white-tailed deer, and eastern cottontail were 35,269 (weight of null 
 
34 
 
model = 4.82 x 10-7), 2.3 x 106 (weight = 5.73 x 10-8), and 1.57 (weight = 0.084) times better.  
Model-averaged predictions estimated that probability of use for those sampling units where 
species went undetected ranged from 0.341 to 0.435 for coyotes, 0.598 to 0.705 for white-tailed 
deer, and 0.070 to 0.167 for eastern cottontails.  Using equation 1, probability of use of sampling 
units by either species of prey ranged from 0.648 to 0.748. 
Modeling the Spatial Association 
 Our analysis indicated that use of resources by white-tailed deer was not significantly 
associated with occupancy by coyotes (p = 0.147). However, the large estimated effect (? = -1.43 
? 0.984), coupled with my relatively limited number of detections of coyotes, suggest that my 
test may suffer from low power and the effects could be biologically important. Estimates 
indicate that sampling units without white-tailed deer were 4.17 (0.61 - 28.75; 95% CL) times as 
likely to have coyotes as sampling units with deer (Table 2). Similarly, use of resources by 
eastern cottontails was not significantly associated with occupancy by coyotes (p = 0.142). 
Again, the large estimated effect (? = 1.28 ? 0.869) and limited number of detections of coyotes 
and eastern cottontails suggest that my test may suffer from low power and effects could be 
biologically important. Estimates indicate that sampling units with eastern cottontails were 3.60 
(0.65 ? 19.75) times as likely to have coyotes as sampling units without eastern cottontails. In 
testing prediction 3, the partial-likelihood ratio test comparing the model with variables for both 
use of resources by white-tailed deer and eastern cottontails with a model that contained a single, 
combined variable for use of resources by prey was significant ( 2 = 3.816, p = 0.05), indicating 
that the more complex model was superior and that use of sampling units by deer and cottontails 
had significantly different associations with use of habitats by coyotes.   
Discussion 
 
 
35 
 
The outcome of the predator-prey space race likely is influenced by traits of the species 
involved.  Relative mobility between predator and prey is believed to influence how species 
respond to each other within a space race (Sih 2005).  When the mobility of prey is equal to or 
greater than that of the predator, prey should be able to spatially respond and avoid areas of high 
likelihood for encounters with predators.  For less mobile prey, however, spatial response to a 
mobile predator requires greater effort on the part of the prey, leaving the predator with an 
advantage.   In comparing the relationship between a predator and two species of prey of 
differing levels of mobility, my evidence supports this tendency.  Specifically, the spatial 
associations between each of the two prey (white-tailed deer and eastern cottontails) and the 
common predator (coyote) were significantly different from each other (prediction 3).  Such 
differences in spatial association suggest that traits of the prey may influence how prey respond 
to predators and, thus, the outcome of the predator-prey space race. 
As a large-bodied species capable of long-distance travel and relatively large home 
ranges (Smith 1991), the mobility exhibited by white-tailed deer should enable the use of spatial 
avoidance of predators such as coyotes.   The benefits to fitness (i.e., reduced rates of predation, 
reduced levels of stress) of shifting use of space to avoid encounters with coyotes likely are 
greater than the energetic costs of movement accrued by white-tailed deer.  Likewise, large home 
ranges may allow white-tailed deer to reduce risks of predation through shifts in use of space 
while remaining within the familiarity of their home range.  For white-tailed deer, and other 
mobile prey, low costs of movement should make spatial avoidance a viable solution to risks of 
predation, and can give the prey the edge in predator-prey races.   Although results here are non-
significant (albeit with low power and large size of effect), I observed a negative trend in spatial 
association between white-tailed deer and coyotes after accounting for species-specific 
 
36 
 
preferences in habitat, a relationship supported by other studies (Lingle 2002).  The observed 
association tends to suggest (with statistical caution) that white-tailed deer are likely to use space 
not being used by coyotes; thereby, reducing predation by coyotes and winning the race with 
coyotes.   
Conversely, the small body and minuscule home range (relative to coyotes) that describes 
the eastern cottontail suggests that limited mobility would make spatial avoidance of a predator 
less likely.  Being less than 7% the size of coyotes (Bekoff 1977; Chapman et al. 1980), the 
energetic costs for eastern cottontails to move far enough to spatially avoid coyotes likely 
outweigh the benefits of avoiding coyotes.  Likewise, the small home ranges of eastern 
cottontails make an individual cottontail vulnerable to predation by coyotes throughout its range, 
and spatial avoidance would require the individual to leave its established home range to avoid 
encounter with coyotes.  My examination of the spatial association between eastern cottontails 
and coyotes again revealed a non-significant (with low power and large effect of size) trend, 
whereas in this case, the association was positive.  Unlike the more mobile white-tailed deer, this 
association may suggest that eastern cottontails use the same space as coyotes, and are losing the 
predator-prey space race. 
  Of course, I cannot be certain that the observed difference in spatial associations between 
the two predator-prey pairs were due to difference in size and rates of movement of the prey.  
Besides factors related to mobility and size of home range, response of prey to use of space by 
predators may be influenced by reproductive behaviors that affect the cost-benefit tradeoff of 
avoiding predators.  For species that have relatively few offspring and require large investments 
in parental care, such as the white-tailed deer, spatial avoidance of a predator increases the 
probability that young will survive.  For the parent, reproductive fitness requires surviving long 
 
37 
 
enough to maximize production of viable offspring, an effort that is enhanced by avoiding 
predation at an early age.  Consequently, such trends may be reversed for species bearing 
multiple precocial offspring such as the eastern cottontail.  Because many young are produced at 
once, loss of a few to predation is not detrimental to reproductive success.  Likewise, the need 
for long-term survival in adults is reduced because several litters can be reared after only a few 
reproductive seasons.  Such species may benefit more by concentrating efforts toward 
reproduction rather than accruing the costs associated with spatial avoidance of predators.  
Although I could not discern whether differences in reproductive traits played a role in shaping 
the predator-prey associations in this study, the patterns detected here would support such 
predictions. 
 I acknowledge the possibility that the spatial associations of this study are the results of 
population-level processes rather than behavioral responses between predator and prey.  For 
example, if coyotes were highly effective at capturing white-tailed deer but less effective at 
capturing eastern cottontails, I might expect to see the same patterns in use of space (small 
populations of white-tailed deer and large populations of eastern cottontails in areas used by 
coyotes).  However, I believe this is unlikely because behavioral responses typically occur at 
more rapid rates than population-level processes.  Likewise, expansion of the geographic range 
of coyotes into Alabama is a relatively recent event (Hill et al. 1987; Lovell et al. 1998), making 
it doubtful that enough time has passed for coyotes to reduce populations of white-tailed deer. 
 Use of space by predators and prey can be influenced by several factors besides the 
availability of resources.  I demonstrate here that traits, such as mobility, size of home range, and 
possibly reproductive behavior, may influence how prey spatially respond to predators.  
Although I did not investigate how traits of predators might influence the predator-prey race, 
 
38 
 
trends are fully expected to be similar. Any trait of either predator or prey for which a cost for 
fitness is accrued due to spatial movement may affect ability of a species to spatially respond to 
the other and influence the outcome of a predator-prey race.   Further research on additional traits 
in other prey, as well as in predators, could greatly benefit our understanding of use of space by 
species within the context of scenarios associated with the predator-prey space race.  Such traits 
could be incorporated in efforts to model use of space by species, a valuable tool for 
conservation and management endeavors.       
 
 
39 
 
Literature Cited 
Anderson, D.R., Burnham, K.P., Thompson, and W. L. 2000. Null hypothesis testing: problems, 
prevalence, and an alternative. J. Wildl. Manag. 64: 912-923. 
Bell, A.V., Rader, R.B., Peck, S.L., and Sih, A. 2009. The positive effects of negative 
interactions: can avoidance of competitors or predators increase resource sampling 
byprey?  Theoretical Population Biology 76: 52-58. 
Bleich, V.C., Bowyer, R.T., and Wehausen, J.D. 1997.  Sexual segregation in mountain sheep: 
resources or predation?  Wildl. Monogr. 134: 3-50. 
Burger, A.E., Hitchcock, C.L., and Davoren, G.K. 2004.   Spatial aggregations of seabirds and 
their prey on the continental shelf off SW Vancouver Island.  Marine Ecology-Progress 
Series 283: 279-292. 
Burnham, K.P., and Anderson, D.R. 2002.  Model selection and multimodel inference: a practical 
information-theoretic approach.  2nd edition. Springer-Verlag.  New York, USA. 
Chapman, J.A., Hockman, J.G., and Ojeda, M.M. 1980.  Sylvilagus floridanus.  Mammalian 
Species 136: 1-8. 
Fiske, I., and Chandler, R. 2011. Unmarked: an R package for fitting hierarchical models of 
wildlife occurrence and abundance.  Journal of Statistical Software 43: 1-23. 
Gilpin, M.E., and Diamond, J.M. 1984.  Are species co-occurrences on islands non-random, and 
are null hypotheses useful in community ecology?  In Ecological Communities: 
Conceptual Issues and the Evidence.  Edited by D.R. Strong, D. Simberloff, L.G. 
Abele,and A.B. Thistle. Princeton University Press. pp. 297-315 
Griffith, G.E., J.M. Omernik, J.A. Comstock, S. Lawrence, G. Martin, A. Goddard, V.J. Hulcher, 
and Foster, T. 2001. Ecoregions of Alabama and Georgia. US Geological Survey, Reston, 
 
40 
 
Virginia. 
Hill, E.P., Sumner, P.W., and Wooding, J.B. 1987.  Human influences on range expansion of 
coyotes in the Southeast.  Wildlife Society Bulletin 15: 521-524. 
Honer, O.P., Wachter, B., East, M.L., Runyoro, V.A., and Hofer, H. 2005.  The effects of prey 
abundance and foraging tactics on the population dynamics of a social, territorial 
carnivore, the spotted hyena.  Oikos 108: 544-554. 
Hugie, D.M., and Dill, L.M. 1994.  Fish and game: a game theoretic approach to habitat selection 
by predators and prey.  J. Fish. Biol. 45: 151-169. 
Kendall, W.L., and White, G.C. 2009.  A cautionary note on substituting spatial subunits for 
repeated temporal sampling in studies of site occupancy.  J. Appl. Ecol. 46: 1182-1188. 
Lima, S.L. 1998.  Stress and decision making under the risk of predation: recent developments 
from behavioral, reproductive, and ecological perspectives. Adv. Study Behav. 27: 215-
290. 
Lima, S.L. 2002.  Putting predators back into behavioral predator-prey interactions. Trends in 
Ecology and Evolution 17: 70-75. 
Lima, S.L., and Dill, L.M. 1990.  Behavioral decisions made under the risk of predation: a 
review and prospectus.  Can. J. Zool. 68: 61-640. 
Lima, S.L., and Steury, T.D. 2005.  Perception of predation risks: the foundation of nonlethal 
predator-prey interactions. In Ecology of predator-prey interactions.  Edited by P. 
Barbosa, and I. Castellanos. Oxford University Press, New York, NY. pp. 166-188 
Lingle, S. 2002.  Coyote predation and habitat segregation of white-tailed deer and mule deer. 
Ecology 83: 2037-2048. 
Lovell, C.D., Leopold, B.D., and Shropshire, C.C. 1998.  Trends in Mississippi predator 
 
41 
 
populations, 1980-1995.  Wildlife Society Bulletin 26: 552-556. 
MacKenzie, D.L. 2006.  Occupancy estimation and modeling: inferring patterns and dynamics of 
species occurrence. Academic Press.  Burlington, MA. USA. 
MacKenzie, D.L., and Royle, A. 2005.  Designing occupancy studies: general advice and 
allocating survey effort.  J. Appl. Ecol. 42: 1105-1114. 
MacKenzie, D.L., Bailey, L.L., and Nichols, J.D. 2004.  Investigating species co-occurrence 
patterns when species are detected imperfectly.  J. Anim. Ecol. 73: 546-555. 
MacKenzie, D.L., Nichols, J.D., Sutton, N., Kawanishi, K., and Bailey, L.L. 2005. Improving 
inferences in population studies of rare species that are detected imperfectly. Ecology 86: 
1101-1113. 
MacKenzie, D.L., Nichols, J.D., Lachman, G.B., Droege, S., Royle, A., and Langtimm, C.A. 
2002. Estimating site occupancy rates when detection probabilities are less than one. 
Ecology 83: 2248-2255. 
Mitchell, W.A., and Lima, S.L. 2002.  Predator-prey shell games: large-scale movement and its 
implications for decision-making by prey.  Oikos 99: 249-259. 
Molvar, E.M., and Bowyer, R.T. 1994.  Cost and benefits of groups living in a recently social 
ungulate: the Alaskan moose.  J. Mammal. 75: 621-630. 
Nicholson, M.C., Bowyer, R.T., and Kie, J.G. 1997.  Habitat selection and survival of mule deer: 
tradeoffs associated with migration.  J. Mammal. 78: 483-504. 
Peres-Neto, P.R., Olden, J.D., and Jackson, D.A. 2001.  Environmentally constrained null 
models: site suitability as occupancy criterion. Oikos 93: 110-120.  
Rosenheim, J. A. 2004.  Top predators constrain the habitat selection games played by 
intermediate predators and their prey.  Isr. J. Zool. 50: 129-138. 
 
42 
 
Sih, A. 1984.  The behavioral response race between predator and prey.  Am. Nat. 123: 143-150. 
Sih, A. 1987.  Predators and prey lifestyles: an evolutionary and ecological overview. In 
Predation: direct and indirect impacts on aquatic communities.  Edited by W.C. Kerfoot 
and A. Sih. University Press of New England, Lebanon, NH. pp. 203-223. 
Sih, A. 1992.  Prey uncertainty and the balancing of antipredator and feeding needs.  Am. Nat. 
139: 1052-1069. 
Sih, A. 2005.  Predator-prey space use as an emergent outcome of a behavioral response race. In 
Ecology of predator-prey interactions.  Edited by P. Barbosa, and I. Castellanos. Oxford 
University Press, New York, NY. pp. 240-255. 
Sih, A., Ziemba, R., and Harding, K.C. 2000.  New insights on how temporal variation in 
predation risk shapes prey behavior.  Trends in Ecology and Evolution 15: 3-4. 
Smith, W. P. 1991. Odocoileus virginianus.  Mammalian Species 388: 1-13. 
Thaker, M., Vanak, A.T., Owen, C.R., Ogden, M.B., Niemann, S.M., and Slotow, R. 2011. 
Minimizing predation risk in a landscape of multiple predators: effects on the spatial 
distribution of African ungulates.  Ecology 92: 398-407. 
Thomson, R.L., Forsman, J.T., Sarda-Palomera, F., and Monkkonen, M. 2006.  Fear factor: prey 
habitat selection and its consequences in a predation risk landscape.  Ecography 29: 507-
514. 
Viet, R.R., Silverman, E.D., and Everson, I. 1993.  Aggregation patterns of pelagic predators and 
their principal prey, Antarctic Krill, near South-Georgia.  J. Anim. Ecol. 62: 551-564. 
Werner, E.E., and Peacor, S.D. 2003.  A review of trait-mediated indirect interactions in 
ecological communites.  Ecology 84: 1083-1100. 
 
43 
 
Table 1:  Estimated size of effect, standard error, and p-values as calculated from univariate 
modeling of each variable of detection (p).  Variables that approached a significant level of effect 
(? < 0.05) were selected for use in subsequent models and are denoted in bold. 
 Coyote Deer Rabbit 
Variable Estimate  (SE) p -value Estimate  (SE) p -value Estimate  (SE) p -value 
Maximum 
Temperature* 
-0.036 
(0.012) 0.002 
-0.044 
(0.009)  <0.001 
0.020 
(0.021) 0.360 
Average 
Temperature* 
-0.052  
(0.013) <0.001 
-0.044 
(0.009)  <0.001 
0.002 
(0.023) 0.910 
Minimum 
Temperature* 
-0.052 
(0.013) <0.001 
-0.031 
(0.009)  <0.001 
< -0.001 
(0.020) 0.976 
Precipitation -0.024 (0.023) 0.301 -0.031 (0.020)  0.118 -0.071 (0.074) 0.340 
Maximum 
Wind Speed 
0.033 
(0.025) 0.172 
0.013 
(0.020)  0.505 
-0.070 
(0.056) 0.211 
Average 
Wind Speed 
0.058 
(0.044) 0.183 
-0.030 
(0.040)  0.458 
-0.105 
(0.100) 0.293 
2nd Day of 
Survey 
0.623 
(0.315) 0.048 
-0.137 
(0.235) 0.558 
< -0.001 
(0.558) 0.999 
3rd Day of 
Survey 
0.453 
(0.319) 0.156 
-0.286 
(0.240) 0.234 
0.280 
(0.529) 0.597 
*To avoid confounding effects, I selected the most statistically significant measurement of 
temperature when multiple measures were significant (? < 0.05). 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
44 
 
Table 2:  Estimated size of effect, standard errors, odds ratios, and p-values of all variables of 
occupancy (?) and detection (p) included in the model used to estimate spatial associations 
between coyotes, white-tailed deer, and eastern cottontails. 
Type Variable Estimate SE Odds Ratio p-value 
? Probability of use for White-tailed Deer -1.427 0.984 0.240 0.147 
? Probability of use for Eastern Cottontails 1.277 0.869 3.586 0.142 
? Agriculture 1.806 1.197 6.086 0.131 
? Hardwoods 1.761 0.854 5.818 0.039 
? Mixed 1.803 0.893 6.068 0.044 
? Pine 0.948 0.928 2.581 0.307 
? Riparian 1.182 0.999 3.261 0.237 
? Scrub -1.402 1.140 0.246 0.219 
p Minimum Temp -0.053 0.013 0.948 <0.001 
p 2nd Day 0.657 0.322 1.929 0.039